JUNE 2022 ISSN 1030-2662 06 9 771030 266001 11 NZ 12$ 50* $ 90 INC GST INC GST Spectral Sound MIDI Synthesiser with timbre morphing and 18-note polyphony Buck-Boost LED Driver drive 12V LED panels, charge batteries and convert 12V ↔ 24V Arduino Programmable Load a clever shield to test power supplies Metal Oxide Air Quality Sensors for detecting CO2, NOx and VOCs IFnatebgrraitcedaCtiricounit Copyright © 2022 SILICON CHIP Publications. Downloaded by Mike Blake (#19283)
Build your own Dew Heater This project brings together a few Arduino-compatible modules and some other parts to create a versatile tool. Inspired by the Dew Heaters used on telescopes, it senses ambient temperature and humidity to control a small heater. Use this on anything that needs to avoid condensation and on your telecscope too (if you have one). SKILL LEVEL: ADVANCED For step-by-step CLUB OFFER instructions & materials BUNDLE DEAL scan the QR code. $5995 For step-by-step instructions & materials www.jaycar.com.au/diy-telescope-dew-heater SAVE 30% KIT VALUED AT $91.83 1MM TRACK TRACE WIDTH FROM ONLY ONLY ONLY $595 $1295 $2995 $3995 Blank Fibreglass PCB Polymorph Pellets PCB Etching Kit Silver Pure copper bonded to quality fibreglass Commercial grade thermoplastic that Complete with assortment of double-sided Conductive Pen base. Single or double sided. 150x75mm to softens to be formed into any shape at copper boards, etchant, working bath and Apply instant traces 300x300mm available. around 62 - 65°C. 100g bag of 3mm tweezers. HG9990 on most surfaces HP9510-HP9515 pellets. NP4260 e.g glass, plastic, metal, epoxy etc. NS3033 $100 Got a great Looking for project or kit idea? your next build? gift card If we produce or publish your electronics, Arduino or Silicon Chip projects: Pi project, we’ll give you a complimentary $100 gift card. jaycar.com.au/c/silicon-chip-kits Upload your idea at projects.jaycar.com Kit back catalogue: jaycar.com.au/kitbackcatalogue Awesome 1800 022 888 projects by www.jaycar.com.au On Sale 24 May 2022 Shop online and enjoy 1 hour click & collect to 23 June 2022 or free delivery on orders over $99* Copyright © 2022 SILICON CHIP Publications. *Exclusions apply - see website for full T&Cs. Downloaded by Mike Blake (#19283)
Contents Vol.35, No.6 June 2022 12 IC Fabrication, Part 1 63(&75$/6281' We take an in-depth look at how silicon chips, also known as integrated 0,', circuits, are made. ICs form the lifeblood of most modern technology, from computers to medical devices. 6<17+(6,6(5 SDJH By Dr David Maddison Semiconductors page 40 38 Radar Coach Speed Detector The Radar Coach is ideal for measuring the speed of cricket, baseball and footballs. It can also be used to measure your own sprint, or even a car! By Allan Linton-Smith Speed detector review 72 MOS Air Quality Sensors MOS (metal oxide semiconductor) modules are air quality sensors that rely on the behaviour of metal oxide in the presence of air to measure gas levels. By Jim Rowe Low-cost electronic modules 84 Altium Designer 22 LBEuDckD-rBiovoesrt We use Altium Designer for all our project PCBs and so with the release of AD22, we wanted to see what new features are available. By Tim Blythman Software review 24 Spectral Sound MIDI Synthesiser The Spectral Sound MIDI Synth is easy to build and can be connected to any Altium MIDI compatible device. It can play up to 18 different notes simultaneously, providing you with a device that can create rich and detailed sounds. Designer 22 By Jeremy Leach Musical instrument project review on page 84 40 Buck-Boost LED Driver This high-power project drives ridiculously bright 12V LED panels. It delivers 2 Editorial Viewpoint up to 8A with adjustable current and voltage. You can even use it to charge 4 batteries from a DC source, or as a 12 ↔ 24V DC converter. 88 Mailbag By Tim Blythman LED/regulator project 92 Circuit Notebook 98 48 Arduino Programmable Load 106 1. RF burst power meter 108 2. Artificial candle using a “real” flame A variable load is indispensible when testing power supplies, driver circuits 111 3. Digital volume control with discrete logic and the like. Our Arduino shield can handle up to 70W continuous at 15V 112 4. An easy way to measure SMDs and 4.7A, with a load resistance between 3.1W and 47W in 15 steps. Serviceman’s Log By Tim Blythman Arduino project Vintage Television 61 500W Power Amplifier, Part 3 Admiral 19A11S TV by Dr Hugo Holden Follow these assembly, testing and calibration instructions to finish building the 500W Power Amplifier. Online Shop By John Clarke Audio project Ask Silicon Chip 81 Revised Battery Charge Controller Market Centre Due to the unavailability of the Si8751 Mosfet driver, we have redesigned Advertising Index our 2019 Universal Battery Charge Controller to use alternative parts. By John Clarke Project update Copyright © 2022 SILICON CHIP Publications. Downloaded by Mike Blake (#19283)
SILICON Editorial Viewpoint CHIPwww.siliconchip.com.au Shutting down our old website Publisher/Editor Nicholas Vinen From around 2000 to 2012, our website was run by a third party and not under our direct control. When Technical Editor I started working at Silicon Chip, it was apparent John Clarke – B.E.(Elec.) that we needed to build our own website for various reasons. For example, our subscription system was Technical Staff completely separate from the website, so there was Jim Rowe – B.A., B.Sc. no good way for people to renew their subscriptions online (or change their address etc). Bao Smith – B.Sc. There were a lot of other reasons to take control, such as being able to Tim Blythman – B.E., B.Sc. sell items like PCBs from the website, which is now a critical service that Nicolas Hannekum – Dip.Elec.Tech. we provide, along with other parts. It would also give us better control over how our articles were presented online. It just made so much more sense to Advertising Enquiries handle it ourselves. Glyn Smith When we set up the new website, we had to decide what to do about peo- ple who had paid for access to articles or magazine issues through the old Mobile 0431 792 293 one. We realised that we had to provide continuity, so everyone who had [email protected] access to a magazine through the old method was given perpetual access to the same issue on the new website. Regular Contributors We also kept the old website going as-is to provide the best transition possi- Allan Linton-Smith ble for our readers, allowing them to decide when they wanted to switch over. Dave Thompson But as time goes on, there seems to be less point in keeping that old website David Maddison – B.App.Sc. (Hons 1), (http://archive.siliconchip.com.au) going. By early 2020, we finished adding PhD, Grad.Dip.Entr.Innov. all the back issues on our current website, back to the very first issue (Novem- ber 1987). That’s all the content that was on our old website, and much more. Geoff Graham Our main website – www.siliconchip.com.au (or www.siliconchip.au if you Associate Professor Graham Parslow prefer a slightly shorter URL), does everything the old site did and so much more. So I think the time is approaching to shut the archive server down. Dr Hugo Holden – B.H.B, MB.ChB., With PDFs now being available for the latest issues to subscribers, and even FRANZCO older issues for those who’ve purchased the PDFs on USB collection (or paid for separate back issues), there is even less reason to keep the archive site up. Ian Batty – M.Ed. The presentation of articles in our PDFs is so much better than on the Phil Prosser – B.Sc., B.E.(Elec.) archive website, where the articles were converted to HTML format and dia- grams were rasterised, often making them blurry or pixelated. Cartoonist So I am writing this to give anyone who objects to that an opportunity to Louis Decrevel contact us and explain why they think we should keep the archive server up. As the saying goes, “speak now or forever hold your peace”. loueee.com If you’re wondering why I want to shut it down, part of the reason is that we didn’t develop any of the code, and it is now on a very old platform that Former Cartoonist sees few updates. I’m concerned about the security implications of keeping Brendan Akhurst such old software running. It is isolated from the rest of our infrastructure, but a breach could still reveal some customer information such as names Founding Editor (retired) and e-mail addresses. Leo Simpson – B.Bus., FAICD There are also costs associated with keeping it going, including some of our time and hosting expenses that I would rather spend on our current web- Silicon Chip is published 12 times site and producing new content. a year by Silicon Chip Publications I hope that, by now, all our readers have switched over to using the new Pty Ltd. ACN 626 922 870. ABN 20 website. If not, please give it a go as we believe it is a significant improve- 880 526 923. All material is copy- ment over the old one. right ©. No part of this publication Unless we are given good reasons to keep it going, we plan to shut the may be reproduced without the written archive site down by the end of July 2022. consent of the publisher. Australia's electronics magazine by Nicholas Vinen Subscription rates (Australia only) 6 issues (6 months): $65 siliconchip.com.au 12 issues (1 year): $120 24 issues (2 years): $230 Online subscription (Worldwide) 6 issues (6 months): $50 12 issues (1 year): $95 For overseas rates, see our website or email [email protected] Recommended & maximum price only. Editorial office: Unit 1 (up ramp), 234 Harbord Rd, Brookvale, NSW 2100. Postal address: PO Box 139, Collaroy Beach, NSW 2097. Phone: (02) 9939 3295. ISSN: 1030-2662 Printing and Distribution: 24-26 Lilian Fowler Pl, Marrickville 2204 2 Silicon Chip Copyright © 2022 SILICON CHIP Publications. Downloaded by Mike Blake (#19283)
Copyright © 2022 SILICON CHIP Publications. Downloaded by Mike Blake (#19283)
MAILBAG your feedback Letters and emails should contain complete name, address and daytime phone number. Letters to the Editor are submitted on the condition that Silicon Chip Publications Pty Ltd has the right to edit, reproduce in electronic form, and communicate these letters. This also applies to submis- sions to “Ask Silicon Chip”, “Circuit Notebook” and “Serviceman’s Log”. The early years of radio and TV Mouse Club on the TV he made; it had a war-surplus I am writing regarding the Vintage Radio article on the round green-phosphor screen. He made a fridge in the 1930s, when most still had ice cabinets. He went to a lot Phenix Ultradyne L-2 by Dennis Jackson (March 2022; of trouble getting the cabinetry all vitreous enamelled. siliconchip.au/Article/15248). Dad made a “Neutrodyne” Receiver some years back, shown in the photos. He also In Melbourne, before Farnell and Radio Spares moved made the “Queen Anne” legs on the cabinet. in, we had a local supplier as well as Radio Parts. It was called “Stewart’s Electronics”. Stewart Day would call in He was born in 1904 and died in 1994. I saw in “Per- at the Moorabbin Radio Club for the old timers’ morning cy’s” Log Book that he had still been active on the radio tea; not bad for business! Stewart was talking to Dad and a week or so before he passed. said we must get some of this down for the record, and took a few notes. The receiver remains for (public) inspection at the His- toric Homestead of “Mont DeLancey” at Wandin in the During the WW2, dad had to parcel all his gear up and Dandenong Ranges. You can lift the lid with the speaker surrender it to the post office until the war ended. trumpet out of the way! Mum used the battery compart- ment for glassware and various other table items. Robert Sebire, Emerald, Vic. Comment: we have reproduced some of the photos pro- He built his own amateur radio rig and wound all the vided to us by Robert Sebire of his father on this page various transformers. He made a reel-to-reel tape recorder, and overleaf. LCR Bridge with magic-eye tuning and the beam rotator More on Noughts & Crosses design transmission propagation indicator, driven from a pul- ley with thin stainless rope on a drum. The display was Thanks for publishing a brief description of my entry backlit, with Melbourne at its centre point. in the Dick Smith Noughts & Crosses competition (April 2022, page 84). Regarding the “impractical to build it” His longest-lasting communications receiver was a comment you added at the end, I’m not sure why that Hammarlund Super Pro. was necessary, but I understand the editor is entitled to their opinion. When he got around to making a new transmission tower in his retirement, he had to make an arc welder Yes, it would be a beast to build the entire system using first. When the tower was finished, he had to dig a hole TTL logic-gate ICs. But remember that Dick Smith made about eight feet deep to meet the council regulations. An such a machine out of disused telephone exchange switch- inspector from Moorabbin council came and measured ing gear. As Dick explains in his autobiography, building the depth for compliance. such a beast has great benefits from a pedagogical and self-confidence perspective. When he was transmitting, the dimmed shack flickered with the blue light of the mercury arc rectifiers, conduct- The ICs are organised into modules so that each module ing with his voice modulation. He earned a certificate for can be built and tested separately, then integrated into a making contact with one amateur operator in each of the complete system in a step-wise manner. A teacher could 50 United States plus Washington DC and Puerto Rico. As a boy, I watched Casey Jones, Seahunt and Mickey 4 Silicon Chip Australia's electronics magazine siliconchip.com.au Downloaded by Mike Blake (#19283) Copyright © 2022 SILICON CHIP Publications.
Copyright © 2022 SILICON CHIP Publications. Downloaded by Mike Blake (#19283)
even organise a class of students to work in teams, each Australia, May 1983. To reduce the cost, I used parts I team working on one of the modules and learning how had on hand. This called for some changes in the output the systems integration process works. specification and circuit design. I replaced the nominated transformer with a 31V, 4A toroidal type, effectively giv- The full design is intended to adjudicate a game between ing me a 44V/4A power supply. competing automated players. That is why there are 12 PCBs of TTL ICs. The design deliberately avoids the use I replaced the main MJ15004 switching transistor with of a synchronising clock so that the only limit to how fast a BDX62A Darlington and the volt and amp panel meters the system can adjudicate a game between two players is with a single, switchable digital meter to display either the propagation delay of the TTLs. volts or amps. I built a small 9V DC supply to run the digital meter from a small, light winding wound on the It is possible to build a cut-down version of the design toroidal core. with only two PCBs, a bit like your runner-up #1. Since the design uses a common bus interface between the mod- The switchmode supply runs quite cool with no fan ules, a minimal system would be the human-interface required. The picture (at lower left) shows the power sup- module plugged into the Arduino module. I would sug- ply with the top cover removed. gest that as a good starting point for a school or class- room-based project. Mauri Lampi, Glenroy, Vic. Simulating boat sounds For anyone interested in such a project, I have uploaded the article describing the full system (https://moonbounce. In the April issue of your magazine, on page 118 (Ask com.au/tictactoe.html). That page also includes a Java Silicon Chip), G. C. asked about simulating steamboat Script version of the suggested logic that can be played sounds. in most web browsers. If anyone has questions regarding the design or PCBs, I can be reached by e-mailing Silicon The website www.component-shop.co.uk has a catalog Chip with a request to forward the query. link at the bottom of the opening page. There, you will find quite a range of simulated boat sound devices (start- Dr George Galanis, ing on page 68 at the time of writing this). They also sell Emerald, Vic. many other small items that may interest electronic minds. Power supply one-upmanship I read with interest the comment by Greig Sheridan in I get a lot of enjoyment reading your magazine most the October 2021 issue (Mailbag, page 12). It referred to months, even though the electronics are past my skill an Electronics Australia May 1987 lab power supply he level. I gave up when valves left the scene. had built. I built a switchmode 50V/5A laboratory power supply Graeme Baker, Grovedale, Vic. based on the design by Jeff Skeen published in Electronics Migrating from Microchip mpasm to pic-as I have some information that might help G. C. of Ran- giora, NZ, who wrote the letter “Advice on coding PICs and using MPLAB” in the April 2022 Ask Silicon Chip section, on page 117. Yesterday, I worked on an assembler program for the first time since installing MPLAB X v5.50. It seems that since about v5.40, Microchip has replaced the old mpasm assembler with pic-as(v2.32) and is trying to move users to relocatable code instead of absolute location code. We now have to use PSECT instructions in our code to achieve this. If I just added a PSECT code at the top, all the “phase error” errors went away and it built successfully. But looking at the .hex file, no line began “:02000” indi- cating programming at location 0x0000 – the reset vector. In my case, what had been created was code that was relocated to 0x0FBE. The ORG statements were being treated as offsets from the relocated origin, not absolute addresses. To get it to produce a .hex file that looks like it 6 Silicon Chip Australia's electronics magazine siliconchip.com.au Copyright © 2022 SILICON CHIP Publications. Downloaded by Mike Blake (#19283)
KEEP PACE WITH AN EVER CHANGING MARKET, WORK WITH: • A global distributor of technology products, services and solutions that provides you with local support • A reliable partner that offers you a broad range of products • An industry leader providing access to specialised products Single Board Computing Semiconductors Leading distributor of Over 120,000 semiconductors in stock cutting edge technology at competitive prices Passives Electromechanical Over 140,000 Over 60,000 passive products electromechanical components Interconnect Test and Tools Over 550,000 connector, Full range of products to support electronic cable and wire products design and preventative maintenance Authorised distributor of Downloaded by Mike Blake (#19283) Contact us now au.element14.com | 1300 361 005 Copyright © 2022 SILICON CHIP Publications.
will work, I had to restructure my code as follows. You also want to include baffles to prevent the air from circulating inside the product – in your design, you need processor 12F683 to block off the area between the fans and the chassis rear, #include <xc.inc> and probably at the top too. A sheet of plastic is all that is required. The rack case that it goes into also needs simi- PSECT resetVec, class=CODE, reloc=2 larly sized vents on the side. ORG 0 ; reset vector resetVec: I’d also like to comment about SMDs. I’ve been work- goto MAIN ing with them since we started designing products with them in the 90s. I agree they’re intimidating at first, but ORG 8 ; bytes, not words you get used to them, opening the packets, using tweezers goto INTERRUPT and making sure you don’t breathe too hard. PSECT code The smallest I generally design with these days is impe- INTERRUPT: rial 0603 (M1608). Still, I use the occasional 0402 (M1206) retfie – you can save a fair bit of space using 0402-sized 100nF decoupling capacitors, for example, because you use so MAIN: many of them. You can also fit them in easily, nice and ...... close to the noise source. END resetVec You also need to go to the Project tab in the box a bit Parts on both sides of the board is becoming the norm below “File” in the top left (it may show as “Pro…”). these days, which can make it hard to get decent-sized Right-click your project name, go to the bottom, select power tracks. I have a soldering station with a hot air “Properties” and click “pic-as Linker”. Find “Additional blower but don’t use it much – just a normal iron from options:” roughly in the centre of the screen and insert Jaycar. “-Wl,-presetVec=0h” and click OK (That’s a lower-case L after the W). D. T., Sylvania, NSW. Now build, and you should get a :020000 line in the Comments: while we’ve mounted the 500W amplifier in a .hex file, and it should work. rack case, that’s mainly because it’s the most reasonably- David Heckingbottom, St Ives, NSW. priced, sturdy case that’s large enough to fit the complete amplifier and not too heavy as it’s made from aluminium. Migrating from PIC16F88 to newer 8-bit PICs It also looks pretty good. I sent an e-mail to Silicon Chip about two months ago It could be rack-mounted, but you are correct that the with a suggestion for an improved Solar Charger. I took cooling system design is not optimised for that. We mainly your feedback onboard and contemplated why I was still intended it to be used in spaces like entertainment centres dealing with an ancient PIC16F88. I only had one left, where there will be space above it for air to flow into the and when I checked to see how available they were from top on one side, through the fans, over the heatsink and major suppliers, I was in for a bit of a surprise! then out the lid on the other side. That’s why we used quiet fans at low speed; so it can be used in a listening room. I researched some possible substitutes from Microchip and found a suitable pin-for-pin enhanced mid-range If it were placed in a rack with very little (or no) air replacement. Microchip had them available from Singa- gap above, you are right that the holes in the side would pore, and they came in under two weeks. But I couldn’t need to be enlarged. As you say, ducting would also be find any decent practical guide online on how to go about required to ensure sufficient airflow over the heatsink and such a migration process, as the new enhanced mid-range minimal recycling of hot air back into the intake. In that architecture means old code is incompatible. case, it would also be a good idea to use fans that spin faster and move more air. I made notes as I went and produced a migration doc- Are “repair programs” useful? ument. I’m sending you the PDF in case it’s helpful to your readers. According to one source, it seems that your campaign for repairability has as least two adherents whose names Note: the supplied document is available for download rather surprise me. In this month’s IEEE Spectrum jour- at siliconchip.au/Shop/6/6489 nal is an article “A Laptop That’s Fit to be Fixed”, which features a repairable Dell laptop, and also contains these Phil Nicholson, Mentone, Vic. words: 500W Amplifier cooling efficacy “Apple, too, is preparing a Self-Service Repair program I used to design products for a major equipment sup- that will sell parts for iPhones, iPads and Macs directly to consumers. Owners will be able to fix their devices with plier, so I’d like to make some comments regarding your new, official repair manuals.” Won’t that turn the Apple new 500W Amplifier (siliconchip.com.au/Series/380). repairers’ network upside down – or was it a wind-up? Those vent holes on the side are grossly inadequate for anything other than if the amplifier is used for home, Alan Ford, Salamander Bay. where its high power rating is more of a way to achieve Comments: as Louis Rossmann runs an Apple repair busi- a high dynamic range. ness, we think he is in a good position to comment on this repair program. You can see his opinion in the YouTube For fans to be effective, the inlet area should be about video at https://youtu.be/agG108sxkyo the same size as the fan blade area, and the exhaust area 20% bigger to allow for expansion of the air. To summarise, the available parts are pretty limited (for example, spare charge ports are not on offer, even though they wear out). He thinks that the purpose of this 8 Silicon Chip Australia's electronics magazine siliconchip.com.au Copyright © 2022 SILICON CHIP Publications. Downloaded by Mike Blake (#19283)
END OF“Setting the standard for Quality & Value” FINANCIAL YEAR SALEOTRHDEER BIENFORDEUJUSNTER30YTHS’ CHOICE! Established 1930 CNC Metal Sheet Metal Wood Workshop Lifting Cutting Machine Tool Measuring Machinery Working Fabrication Working & Automotive Handling Tools Accessories Equipment 31-180 33-234 Pick Up Tool with 4 Piece Telescopic Digital Caliper Gear Tooth Vernier Caliper Magnetic Head Inspection Mirror Set Splash Proof • 1 - 26 Module • 1 x LED light • Mirror extends 240 - 920mm • 150mm / 6\" • Two adjustable verniers • 620mm length • Magnetic pick up tool • Hardened S/S mechanism • Flexible spring jaw head • Metric/Imperial Settings Order Code: Q234 • Claw can grip up to extends 165 - 695mm • Four way measurement • 3.6kg fixed & swivel $149 30mm round objects Order Code: Q180 • Ø14.8mm magnetic head head magnetic pick up tools SAVE $27 • 2.3kg fix head magnetic $98 Order Code: M0010 Order Code: M0009 SAVE $17.50 $12 $18 SAVE $4.50 SAVE $6.20 70-670 - Refractometer Deburring Tool LED Rechargeable HL-72L - 14W LED Work Light Handheld Work Light with 2.25X Magnifier • Scale 0-18 Brix • Pen Style • Automatic temperature compensation • HSS Blades • Max output: 400 lumens • 14W - 5700K LED • Differentiates • Ø11.5 x 104mm • Rechargeable via USB • Dimmer control • 3 hours operating time • Swivel & pivoting arm between old and long hexagon body • 4 Modes • Includes magnified lens new coolants • Spring loaded tip • Magnetic base & torch arm end • 240V / 10amp • pocket clip Order Code: Q670 Order Code: T9501 HINEACDLUDDUESSTLCIOGVHETR Order Code: D040 $112 $33 $16.50 SAVE $20 SAVE $6.60 SAVE $3.30 TPS-20 Order Code: L2821 Touch Point Sensor Measuring Kit - 4 Piece $187 • Accuracy within 0.005mm • 0 - 25mm micrometer • Ø10mm hardened & • 150mm / 6\" rule SAVE $33 • 150mm / 6\" vernier ground ball end • 100 x 70mm square • LED Light & Order Code: M012 Beeper Sensor Order Code: M690 $56 $84 SAVE $10 SAVE $15 VS-1000 TU-1503V Hex Keys Set with 50-520 - Portable, Portable Video Inspection Opti-Turn Bench Lathe - Mini T-Bar Handle Digital Hardness Tester Camera • Ø150 x 300mm turning capacity • Chrome vanadium steel • Large LCD display screen • 12mm IP67 camera with 1M cable • 11mm spindle bore • Adjustable 3 detent positions on • Hardness scale: HL, HB, HRB, HRC, • LED lighting • 80mm 3 jaw chuck • Includes cable, magnetic pick up & • Electronic variable speed T-bar handle HRA, HV, HS • 0.45kW, 240V motor Metric • Displays scales for Leeb, Brinell, Rockwell mirror tool • 2, 2.5, 3, 4, 5, 6, • A,B,C, Vicker & Shore • Large memory COHMAFNODRGTARBIPLE 8, 10mm storage of Order Code: H820 100 groups • Communication Imperial interface: USB • 5/64, 3/32, 1/8, 5/32, 3/16, 1/4, 5/16, 3/18\" Order Code: H821 Order Code: M697 Order Code: L685 Torx® Order Code: Q520 • T10, T15, T20, T25, $149 $979 $1,402 T27, T30, T40, SAVE $27 SAVE $66 T45, T50 $79OrdeOr rCdeordCe:odHe8:2M2 697 SAVE $20 SAVE $248 BUY & INSTALL BEFORE JUNE 30TH TO MAKE THE MOST OF THE INSTANT ASSET WRITE-OFF NOW!* CONTACT YOUR ACCOUNTANT OR BUSINESS ADVISER TO DISCUSS YOUR PARTICULAR CIRCUMSTANCE. www.machineryhouse.com.au/signup FOR SPECIAL DEALS VISIT UNIQUE PROMO CODE MACHINERYHOUSE.COM.AU$70 FREE 04_SC_260522 SC0622 DISCOUNT VOUCHERS 04_SC_260522 (0S2Y)w9D8w9N0wE91.Y1m1 acThHVBiEI(0nESR7eW)EI3Sr7Iy1BTA5hEAN2oM2ND0uE0SPsOeUNR.cCLoIHMNmAE(E0S.3LaEA) B9uT2O1/2SU4CR42N02 6E22 1N/2SWWPinEd(s0Ro2rT)Rd9H,8N9o0rth9m11e1ad Q625ELBDsotau(bn0dl7ias)hry3eR7dd11,5C93o2o02p0e0rs Plains ONLINE OR INSTORE! V(I0C8) (90337)3992919294422 WA (08) 9373 9999 4 Abbotts Rd, Dandenong 11 Valentine Street Kewdale Copyright © 2022 SILICON CHIP Publications. Specifications & Prices aSrepescuifibcjaeticotnsto&cPhriacnesgearewsituhbojeuctt tnoocthifaicnagteiownit.hAoullt pnoritcifeicsatiionnc.luAdll eprGiceSsTinacnluddevGalSidT aunndtivl a2li7d-u0n6t-il2227-06-22 Downloaded by Mike Blake (#19283)
program is more to get good PR than actually help repair devices. See the video for the complete analysis. On inventions, risk and old batteries In the January Editorial Viewpoint, you imply that large American companies are not risk averse. I disagree. Like most companies everywhere, they risk other people’s money. Why would the management want to risk their own wealth? Let investors and shareholders carry the risk. There is one thing in law that is more valuable than gold that reduces the risk of failure, and that is a patent. An invention must be enshrined in a patent, and not just an Australian patent and/or an American patent, but a patent in every country that supports them. Huge American companies own large numbers of pat- ents and actively pursue smaller companies to obtain more patents. I believe that without patents, America would not be the powerhouse that it has been and still is now. Concerning inventions, my ex-boss recommended that after I had created some invention and developed it fully (a must), I should sell it for as much as I could to who- ever will pay the price and let the buyer worry about the headaches that follow. This is advice borne from many years of working very long hours creating and running a medical technology company in Australia. Dr Maddison’s articles on batteries (January-March 2022; siliconchip.au/Series/375) were good. I know I am showing my age, but I owned a few of those very early cells; it is a pity I didn’t keep them. I remember that either large lead-acid cells or large Edison cells were made using a moulded glass case. As well, I can remember timber cases being used for some batteries. While at primary school, I discovered that the PMG was throwing old batteries into their rubbish bin. These were the large zinc-carbon cells as shown in Fig.13 (Jan- uary 2022, page 17) but with the standard EverReady red labelling. After recovering over a hundred of these, I rediscovered the arc lamp and arc cutting of steel sheets. They made my childhood that much better. George Ramsay, Holland Park, Qld. A plethora of information I love your magazine. I have read every issue since 2018 and wish I knew about it earlier as you have a lot of very useful information. On the subject of information, I have just stumbled across the website www.bitsavers.org and was blown away by the wealth of data sheets and design specifications available. They have a huge cache of documents, including CMOS and TTL data sheets, data books, magazines and applica- tion notes. I am having a blast going through all of this, and I know your readers will really appreciate the detail and diversity offered. My favourite is this one on op amps: siliconchip.au/link/abem (Fairchild, 1979). Ben Dempsey, Waimate, NZ. Comment: there does seem to be a lot of interesting infor- mation on that website. The Fairchild data book has details on quite a few classic devices. Good suggestions for RF Prescaler I recently built your High Performance 6GHz RF Pres- caler (May 2017; siliconchip.com.au/Article/10643), and 10 Silicon Chip Australia's electronics magazine siliconchip.com.au Copyright © 2022 SILICON CHIP Publications. Downloaded by Mike Blake (#19283)
it works very well. I am using it with my old EA frequency Helping to put you in Control meter which only goes up to 500MHz. I sprayed the case satin black as I reckon it looks a lot better than unpainted. LabJack T7 Data Acquisition Module I have a few comments/observations. A USB/Ethernet based multifunction data The SMA connector specified (Molex 073391032) is a acquisition and control device. It features reverse-polarity SMA connector similar to those used on high data acquisition rates with a high WiFi modems etc. I believe the correct part is 0733910083. resolution ADC of 4 ksamples/s at 18 bits to 50 ksamples/s at 16 bits. The MMBT3640 PNP transistors (12V 200mA 500MHz) are obsolete, so I used 50A02CH-TL-E (50V 500mA SKU: LAJ-045 690MHz) and these seem to work OK. Price: $902.00 ea + GST I noticed that the divide-by-5 chip and the counter tend Temperature probe 5m Teflon Cable to oscillate with no signal at 450MHz and 150MHz, respec- tively. An onsemi application note AND8020 mentions RTD probe with magnet fixing for surface this on page 17 and recommends offsetting one input up temperature measurement. -50 to 200 ºC to 50mV with a high-value resistor. range and 5meter teflon cable. I piggy-backed a 47kW resistor to ground on the 10nF SKU: CMS-007T capacitor at pin 2 of IC3. I also added a 2kW resistor from Price: $153.95 ea + GST pin 23 of IC4 to Vcc, which fitted neatly between the 100nF capacitor and L4. These stopped the self-oscillations. J Thermocouple Temperature probe with magnet fixing I chose 47kW as it was the only high-value resistor I had This J type Thermocouple sensor has magnet in M2012/0805 size, and for IC4, resistors much above fixing for surface temperature measurement. 2kW did not stop the self-oscillation. The 2 wire sensor has a silicone cable which is 3m long. Temperature range is -50 to 200 I would be interested to know if you think this is the best ºC. Class B. solution. I would also be interested to know the reasoning behind using 1.4V as the bias voltage for the inputs of IC4 SKU: CMS-017J rather than something closer to Vbb (measured at 1.84V). Price: $142.95 ea + GST Most of my test equipment was/is home built from EA/ 400W ACM Brushless AC Servo Motor ETI/Silicon Chip and I have had many years of good use from them. Congratulations on a great magazine; I hope Leadshine ACM604V60-T-2500 400W it continues for many years to come. brushless AC servo motor with 2500 line encoder suitable to work with the ACS806 Kind regards, brush-less drive. Mike Hammer. Comments: thanks for your findings which seem thor- SKU: MOT-450 ough and well-researched. We think you are right that Price: $347.60 ea + GST those changes are the simplest way to stop self-oscillation. It’s true that a standard-polarity SMA socket would ACS806 Brushless Servo Motor Drive probably make more sense. We used a reverse-polarity socket handy because it mated with most of the SMA Brushless servo motor driver for 50 to 400 cables we had on hand, but perhaps that was not a sound basis for the choice. SWK,UA:CSMbrCu-s4h10less motors with encoders. We generally don’t concern ourselves too much if pres- Price: $319.00 ea + GST calers and counters self-oscillate because it’s arguably a trade-off between that and sensitivity. In other words, any LCD Temperature and Humidity Sensor changes to prevent self-oscillation are likely to reduce sensitivity because they push the device’s operating con- The Pronem Midi from Emko Elektronik are ditions further from the switching point. microprocessor based instruments that incorporate Also, you don’t generally use a device like this without high accurate and stable sensors that convert a signal applied. Your changes will probably not desen- ambient temperature and humidity to linear 4 to 20 sitise it all that much. The reason for using 1.4V for the bias voltage for IC4’s mA. Dimensions are only 60x 126 x 35mm. clock inputs is that it is right in the middle of IC4’s logic low-level voltage range (1355-1675mV for Vcc = 3.3V; see SKU: EES-020A the data sheet, p6). We used a single bias voltage for sim- Price: $241.95 ea + GST plicity and thought it was best to have these inputs rest at a low level in the absence of an input signal (although TxIsoloop-1 Single Loop Isolator that’s somewhat irrelevant if IC3 is going to self-oscillate). Biasing them to Vbb as you suggest (halfway between the Loop isolators provide signal protection by low and high levels) would likely give better sensitivity. electrically isolating the 4-20mA input signal We found the output swing from IC3 sufficient to trigger IC4 with reasonable signal levels over the intended oper- fSrKoUm: SthIGe-240-21 0mA output. ating frequency range, but increased sensitivity would be Price: $168.19 ea + GST welcome above 1GHz (and would likely overcome any reduction in sensitivity due to your other changes). SC For Wholesale prices Contact Ocean Controls Ph: (03) 9708 2390 oceancontrols.com.au Prices are subjected to change without notice. siliconchip.com.au Australia's electronics magazine June 2022 11 Copyright © 2022 SILICON CHIP Publications. Downloaded by Mike Blake (#19283)
IC Fabrication Image Source: https://pr.tsmc.com/english/gallery-fabs-inside – Taiwan Semiconductor Manufacturing Co., Ltd. from inception to cutting-edge technology We take an in-depth look at the technology this magazine is named after: silicon chips, also known as integrated circuits or ICs. They are critical to most modern technology, and it has taken decades to get these devices to the pinnacle of performance they have achieved. But the technology has not stopped advancing yet! Part 1 – History & Manufacturing – By Dr David Maddison T his three-part series describes IC focus on how IC technology has manufacturing and some other inter- (integrated circuit) technology. improved over time, including pro- esting aspects of the field. Given their incredible complex- duction nodes, transistor counts, and ity, we can only really scratch the wafer sizes. It will then describe the The third and final part will cover surface of the fascinating and highly extreme UV (EUV) lithography tech- the latest IC technology such as Fin- advanced manufacturing methods nology that is the current top-tier tech- FETs, GAAFETs, stacked dies and required. Arguably, this technology nology behind advanced ICs like com- multi-chip modules. It will also dis- is the most advanced ever developed. puter CPUs. cuss the challenges of improving this technology into the future. This first article covers the early his- That article will also look at tory of ICs, IC design, silicon wafer pro- what components can be fabricated The transistor’s development duction, fabrication and lithography. within an IC and how they are made, how ICs are packaged, Australian The development of the planar tran- Next month, the second part will sistor was a prerequisite for successful 12 Silicon Chip Australia's electronics magazine siliconchip.com.au Copyright © 2022 SILICON CHIP Publications. Downloaded by Mike Blake (#19283)
integrated circuit construction. We Early IC history not get the acknowledgement it deserves. covered the history of transistors in Three significant factors associated with detail, including planar transistor Combining several electronic compo- manufacturing, in the March, April nents into a single physical device was the commercialisation of ICs toward the end and May 2022 issues (siliconchip.au/ tried with valves in the 1920s to evade a of 1958 were: Series/378). “tube tax” in Germany. Reducing the num- ber of valves meant less tax on the radio. a) The development of a hybrid IC by Jack The field of integrated circuits is Kilby of Texas Instruments; patent awarded vast and developing rapidly. We can- For example, the Loewe 3NF contained in 1964 (siliconchip.au/link/abdp). not possibly mention every possible three triode valves, two capacitors and technology, as a comprehensive sur- four resistors in one glass envelope (July Unlike Johnson’s device and modern ICs, vey would require thousands of pages 2020; siliconchip.au/Article/14513). this one relied on manually placed wires of text! But we will attempt to cover between the devices. Nevertheless, Kilby’s all the critical aspects of IC design and In 1949, just one year after the transis- device is usually regarded as the first IC, and fabrication. tor was patented, German Werner Jacobi he was awarded the Nobel Prize for Physics filed a patent (published 1952) for an IC in 2000 for his efforts. Making a single IC is a long, multi- style transistor amplifier. step process. Advanced chips like b) Kurt Lehovec of Sprague Electric Com- computer CPUs (central processing On May 7th, 1952, British engineer pany developed a way to electrically isolate units) and GPUs (graphics process- Geoffrey Dummer proposed a device with individual electronic components on an IC ing units) are reported to take up to several discrete components on a single using “P-N junction isolation”. The device 15 weeks. Over the last few years, the semiconductor wafer. He wrote: “With the is surrounded by a material with the oppo- industry average for advanced 7nm, advent of the transistor and the work on site doping to the substrate. A reverse-bias 10nm and 14nm devices has been semi-conductors generally, it now seems voltage is applied to the junction, creating a 11-13 weeks. possible to envisage electronic equipment region with few charge carriers. A patent for in a solid block with no connecting wires. this was awarded to Lehovec in 1962 (see Those times are for the actual man- The block may consist of layers of insulat- siliconchip.au/link/abdq). ufacturing process, from the growth of ing, conducting, rectifying and amplifying the silicon crystal that forms the wafer materials, the electronic functions being c) Fairchild co-founder Robert Noyce to the finished product being ready for connected directly by cutting out areas developed the concept of the monolithic sale. But they do not include the tens of the various layers.” IC with diodes, transistors, capacitors and of thousands (or much more) hours of resistors in silicon, with aluminium intercon- research and development for the chip This is regarded as the first descrip- nects and a protective silicon dioxide coat- design itself. tion of the modern IC. However, he did not ing (see the diagram at lower left reproduced claim to be the inventor of the IC. from siliconchip.au/link/abdr). Noyce died There is an adage that the first chip in 1990; otherwise, he might have shared the costs millions of dollars to produce Sidney Darlington of Bell Labs was Nobel Prize with Kilby. (due to the cost of research and fab- awarded a patent in 1953 for a monolithic rication equipment), but subsequent device with more than one device on a sin- In addition to the above patent, Jean copies cost only cents or perhaps dol- gle semiconductor crystal (siliconchip. Hoerni developed the planar process for lars per piece (depending upon com- au/link/abdn). That patent would be one fabricating transistors and other semicon- plexity). of the first for an IC had the patent lawyer ductor devices (the patent was awarded in not insisted on limiting it to two devices. 1962; siliconchip.au/link/abds). This pro- The modern production of ICs is cess was critical for Noyce’s work and he almost entirely automated, using ultra- In 1957, Yasuo Tarui of Japan produced improved the process. pure materials in extremely clean facil- a similar device, a “quadrapole” transis- ities with extensive atmospheric and tor, but unlike modern ICs, the transistors The “traitorous eight” other controls to eliminate dust con- were not electrically isolated. tamination. One advantage of auto- Jean Hoerni initially worked for William mation is that it means fewer work- In 1957, Harwick Johnson was awarded Shockley, but Shockley’s behaviour led ers shedding skin cells, hair or other a patent for a “Semiconductor phase shift Hoerni, along with seven others, to leave detritus that would affect production! oscillator and device” (siliconchip.au/ Shockley in 1957 to found Fairchild Semi- link/abdo), on one ‘chip’ of semiconduc- conductor. They became known as the “trai- Why use integrated circuits? tor material, in accordance with the mod- torous eight” (see https://w.wiki/522K). ern concept of an IC. This invention does Compared to devices built with dis- crete components, ICs allow for much Diagrams from Robert Noyce’s The “traitorous eight”, from left to smaller, simpler, more reliable and less (Fairchild Semiconductor) US right: Gordon Moore, C. Sheldon expensive devices. This is because Patent 2,981,877 (filed 1959, Roberts, Eugene Kleiner, Robert most or all of the parts can be made awarded 1961) for “Semiconductor Noyce, Victor Grinich, Julius Blank, in a single process. Device-and-Lead Structure”. This is Jean Hoerni and Jay Last. Source: regarded as the first practical IC. Wayne Miller, Magnum Photos Also, modern devices with (https://w.wiki/53GC) extremely high numbers of compo- Australia's electronics magazine nents (in the billions), such as com- June 2022 13 puters and mobile phones, would be practically impossible to make with- out ICs. They would be incredibly expensive and huge, even if it were possible to build them. siliconchip.com.au Copyright © 2022 SILICON CHIP Publications. Downloaded by Mike Blake (#19283)
Fig.1: a die photo of the Micrologic The first operational IC possible, as we will describe next uL903 from 1960, one of Fairchild’s month. But first, we’ll explain how an first commercially produced ICs. It is The first operational IC was pro- IC is made, as the limitations of that a 3-input NOR gate used in the Apollo duced on the 27th of September, 1960, process determine how these compo- guidance computer. It contains four by a group at Fairchild. They were led nents must be fabricated. resistors and three transistors. by Jay Last and used ideas from Noyce (monolithic IC) and Hoerni (Fig.1). Silicon doping This led to a patent dispute with Texas Instruments, which held Kilby’s Like transistors and diodes, inte- hybrid IC patent. This was eventually grated circuits are mainly made of P resolved by industry cross-licensing (positive) and N (negative) doped sil- in 1966. icon, conductive metals like alumin- ium and copper, and insulators like sil- Historians do not share a strong icon dioxide. We covered doping in the consensus on whether a specific indi- aforementioned series on transistors, vidual invented the IC or whether the so we will only briefly cover it here. honour should go to multiple inven- tors. This author thinks multiple con- Doping alters the electrical conduc- tributions should be acknowledged. tivity and other properties of the semi- conductor material. The semiconduc- The first commercial IC was released tor is typically silicon but may also be: to the general public in March 1961, a type F flip-flop under the Micrologic • silicon-germanium brand, followed by more types in 1962 • gallium arsenide, in microwave – see Fig.2. integrated circuits, infrared LEDs, Texas Instruments released their laser diodes and solar cells first commercial devices in Octo- • gallium nitride, in blue LEDs ber 1961, the Series 51 DCTL “fully- and other opto-electronic, high- integrated circuit” family (siliconchip. frequency and high-power devices au/link/abdt). • cadmium telluride in photovolta- ics and infrared optical windows Components in ICs • gallium phosphide, as used in LEDs As you would expect, transistors can Doping involves introducing dif- be fabricated in ICs, including bipolar ferent metals into the silicon crystal transistors, Mosfets and JFETs. Most structure, from around one atom in modern processes can produce either 100 million for “light” doping to one polarity of each device (ie, NPN, PNP, in 10,000 for “heavy” doping. Either N-channel or P-channel). way, only trace amounts of the dop- ants are used. Naturally, diodes can also be made, Metals (conductors) conduct elec- as they are usually just a single P-N tricity because of the free electrons junction. That can include zener provided by each atom in a metal crys- diodes, depending on the fabrication tal structure. Semiconductors lack free process being used. electrons, but doping the semiconduc- tor with metal atoms introduces extra But to make a truly useful IC, it is charge carriers. Therefore, doping also necessary to include other com- ponents like resistors, capacitors and inductors, and that is certainly Fig.2: IC die patterns from Fairchild One of the pickup tools used to move groups of wafers around the factory. Semiconductor, released in October Picture: Bosch 1962 following the uL903, including (B) a buffer, (C) counter adaptor, (F) Australia's electronics magazine siliconchip.com.au flip-flop, (G) gate, (H) half-adder, S) half shift register. Some time after that, they added the 4-input gate (G1) and dual 2-input gate (D). Source: Fairchild Semiconductors siliconchip.au/link/abej 14 Silicon Chip Copyright © 2022 SILICON CHIP Publications. Downloaded by Mike Blake (#19283)
Fig.3: an overview of the VLSI design process. VHDL and Verilog are hardware description languages (HDL). Original source: www.eng.auburn. edu/~strouce/class/elec4200/CADtools.pdf RTL is Register Transfer Level and AUSIM and PSPICE are both circuit simulators. increases the electrical conductivity The conductivity of semiconductors The designer specifies what is of the semiconductor. in integrated circuits can also be con- required using a language like Ver- trolled by nearby electric fields (as in ilog or VHDL, and the computer then It is possible to make a heavily- Mosfets) or by charge carrier injection figures out what combination of tiles doped semiconductor conduct almost (as in bipolar transistors). This means provides an equivalent function. It lays as well as some metals. This means that the current flow through a junc- the tiles out on a grid, calculates the that it is possible to replace metal tion can be electronically controlled, routing between the tiles and generates tracks with heavily-doped semicon- either continuously in an analog cir- the physical structure. The result is a ductor material in integrated circuits. cuit, or in an on/off fashion in a dig- set of masks that can be run through ital circuit. simulations to verify that the chip will Unlike metals, where the charge behave as expected. carrier is almost always an electron, IC design in semiconductors, the charge carrier These masks or photomasks are then can be an electron or the absence of Before a chip can be made, it must be used to transfer the design to silicon. an electron, called a “hole”. designed. As the most complex VLSI An IC mask layout view of a simple designs now contain billions of tran- operational amplifier is shown in N-type (negative) doping means the sistors, the process is heavily reliant Fig.4, while an actual mask is shown majority charge carrier is a negatively on computers and software tools. The in Fig.5. charged electron. P-type doping is exact design procedures are many and where the majority charge carrier is a varied and beyond the scope of this The highest performance chips hole with a positive charge. article, but an overview is provided require significant ‘bottleneck’ areas in Fig.3. (such as multiplier-accumulators) to Typical P-type dopants used for sili- be designed by hand as they can be con are boron, aluminium, gallium and Briefly, the design process is usu- made smaller, faster and more effi- indium, while N-type dopants are anti- ally a combination of computer-aided cient. These hand-made pieces can mony, arsenic, bismuth, lithium and and manual design. Simpler, less- be integrated into the synthesised phosphorus. They have advantages in demanding digital chips can be made designs. It is also possible to manu- different applications. almost entirely using a ‘tile-based’ ally modify a synthesised design or scheme. Each tile might be a different give the software ‘hints’ to produce a Other semiconductors use dopants type of logic gate, memory cell, multi- more optimal result. such as carbon, chromium, germa- plexer, adder, multiplier etc. nium, lithium, magnesium, nitrogen, The industry-standard digital file phosphorus, selenium, sodium, sulfur, tellurium, tin and zinc. Fig.4: a mask layout of a simple IC, an operational amplifier. Red is polysilicon; Fig.5: an IC photomask. Source: blue is metal layer 1; green is N-doped Si; brown is P-doped Si and the Xs Wikimedia user Peellden (CC BY-SA are cross-layer “vias”. The large square on the right is a capacitor. Source: 3.0) Wikimedia user Atropos235 (CC BY-SA 2.5) June 2022 15 siliconchip.com.au Australia's electronics magazine Copyright © 2022 SILICON CHIP Publications. Downloaded by Mike Blake (#19283)
Fig.7: the process starts with purified silicon rods (left). Silicon from ► trichlorosilane gas is deposited onto them (centre), then they are broken up and formed into large silicon crystals by the Czochralski process (right). Source: Silicon Products Group GmbH ► Fig.6: a 3D view of a small “cell” (a standard design element of an IC) generated with the ShapeshifteR software from GDSII mask files. There are three metal layers plus vertical interconnects, with silicon gates in a reddish colour on top of the multi-coloured bulk silicon. The insulating material has been removed from this image. Source: David Carron (public domain) format for masks which can be trans- implemented in silicon. However, gaseous hydrochloric acid to form ferred from designer to foundry is companies like Intel also specialise in trichlorosilane, HCl3Si. This is a gas called “Graphic Design System” (GDS, both design and fabrication. at the temperatures used so that it introduced 1971) and “GDSII” (intro- can be further purified by fractional duced in 1978). Since 2004, OASIS Silicon wafer manufacturing distillation. (Open Artwork System Interchange Standard) has been used, which can Apart from design, the first stage of The purified trichlorosilane gas is handle much larger mask sizes than IC manufacture for silicon devices is then mixed with hydrogen in a cham- GDSII. to grow a near-perfect silicon crystal. ber with purified silicon rods electri- Quartz ore called quartzite (basically cally heated to 1150°C. It decomposes The GDSII files for the mask descrip- silicon dioxide, SiO2) is the major and is deposited as pure silicon on tion of a ‘system-on-a-chip’ device like component of most beach sands. It the rod surfaces to make polysilicon a mobile phone processor (as an exam- is extracted from quartz mines and (many crystals as opposed to a single ple) can exceed 200GB. refined to make silicon. crystal) with a purity of 99.99999% (“seven nines”) or even ten or eleven Fig.6 shows a 3D view of a ‘cell’ Quartzite is crushed and then nines. The polysilicon is then broken within a silicon wafer produced by mixed with coke (coal that had previ- up to make a feedstock for the crystal software called ShapeshifteR that ously been heated without oxygen). growing process. takes a mask design from a GDSII file The mixture of quartzite and coke and renders it into a 3D representa- is added to an electric arc furnace Dopant metals such as antimony, tion and cross-section of the actual where high temperatures of around arsenic, boron or phosphorus are chip. See http://shapeshifter.free.fr/ 2000°C are produced. The carbon in added to the polysilicon to give the index.htm the coke reacts with the oxygen in the silicon the required electrical proper- quartzite, removing it. The result is an ties. This is called the Siemens process A ‘fabless’ design house does not impure form of silicon that needs fur- (Fig.7). It is the most commonly used manufacture chips but sends its ther refining. process, but it uses a lot of energy; mask files to a ‘pure play’ (fabrica- other processes have been developed, tion only) foundry to have its design The silicon is then mixed with Fig.8: the Czochralski process for growing single large pure silicon crystals. 16 Silicon Chip Australia's electronics magazine siliconchip.com.au Downloaded by Mike Blake (#19283) Copyright © 2022 SILICON CHIP Publications.
such as a fluidised bed reactor. Picture: Taiwan Semiconductor Picture: Bosch Once purified polysilicon is broken Manufacturing Co., Ltd. Wafers come in multiple different sizes from 25mm up to 450mm- up, it is melted in an inert atmosphere wafers. A typical wafer cut from a diameter. The thickness of the wafer and a ‘seed’ crystal attached to a puller 300mm diameter crystal is 0.775mm is important as it needs to be strong rod is introduced into the melt and thick, weighs 125g and 640 10mm x enough not to break during handling; slowly withdrawn. The melted silicon 10mm dies (chips) can be made on it. a typical thickness for 300mm wafers solidifies and crystallises onto the seed is 775μm. Wafers can be stored in crystal (set up with a preferred crystal The planar process “desiccants” or transfer machines as orientation). shown above. The key concept of IC fabrication is The growing crystal is withdrawn the “planar process”. This was initially Conductors on ICs can be made by from the melt as the rod is raised (see developed by Fairchild Semiconduc- the deposition of metals or the selec- Figs.8 & 9). This is called the Czochral- tor in 1959, and it involves consider- tive doping of semiconductor areas ski process. It is economically bene- ing the construction of an IC as one (eg, silicon). Insulators can be made ficial, up to a certain point, to make (or, nowadays, a series of) 2D plane(s). by oxidising silicon to produce silicon crystals with as large a diameter as dioxide or using the technique of P-N possible to maximise the number of Individual areas within each plane junction isolation. A silicon dioxide devices that can be made at once on are either joined together or insulated insulator can also be etched to expose a single slice of crystal, known as a from each other. Various types of junc- underlying material for alteration in wafer – see Fig.10. tions can be created this way, such as various ways. P-N, N-P-N or P-N-P. Silicon wafer preparation Semiconductor junctions can be This planar approach means that made by doping specific regions and Once the crystal has been grown, it lithography can be used, where images depositing additional material on top is sliced into thin wafers, and the sur- are projected onto the wafer to form the of those regions. face and edges are ground, polished circuit with the aid of light-sensitive and cleaned to make uniformly-sized chemicals and photoresist coatings. Wafer processing steps This involves selective etching, depo- sition, implantation and other alter- Processing a wafer to produce ICs ations of desired areas of the wafer. involves four categories of operations as follows. These may be done multi- ple times, up to around 300 steps for the most complex devices, and in var- ious orders. #1 Deposition This involves depositing coatings ► Fig.9: a silicon crystal grown by the Czochralski process at Raytheon in 1956. The melt is heated by the coils of the induction heater; here, the temperature is being measured. In this case, a 25mm diameter crystal was grown, but today 300mm diameter is typical, and 450mm is under development. Source: Radio and Television News, May 1956 (public domain) Fig.10: a silicon ingot on display at the ► Intel Museum, 300mm in diameter. That is one of the current industry standards, starting around 2002, and it is a compromise between size and productivity. Source: Wikimedia user Oleg Alexandrov (CC BY-SA 3.0) siliconchip.com.au Australia's electronics magazine June 2022 17 Copyright © 2022 SILICON CHIP Publications. Downloaded by Mike Blake (#19283)
Fig.11: one method of exposing individual areas of a silicon wafer with a mask. The lens shrinks the image from the mask to the die size. A more advanced process is ‘step and scan’, where an individual die is exposed through a narrow slit which is scanned to obtain tighter focus and smaller feature size. onto the wafer, such as oxidising the that either totally block light or let it all around the 1990s, the aligner was silicon to create an insulating silicon through, unlike a monochrome photo- replaced with a “stepper”. Only a sin- dioxide layer (passivation), deposition graph/slide, where there are grey areas gle die image was produced at a time, of metal conductors, silicon, or other of partial light transmission. with the light focused onto a single semiconductor materials. area on the die. The mask is moved Before the 1980s, an “aligner” was (stepped) across the wafer to repeat The processes to do this are varied used that had large masks containing the pattern. and include: many duplicate die images so that an entire wafer could be exposed at In the pursuit of even higher resolu- • Physical and chemical vapour one time. tion, since the 2000s, the stepper has deposition been replaced with “step and scan” The pursuit of higher resolution systems where only a small portion of • Electrochemical deposition (smaller feature size) meant that • Molecular beam epitaxy • Atomic layer deposition A photo of a clean room at Bosch’s semiconductor factory in Dresden, Germany. • Thermal oxidation of the entire Picture: Bosch – www.bosch-presse.de/pressportal/de/en/bosch-semiconductor- manufacturing-in-dresden-225609.html wafer • LOCOS (local oxidation of sil- Australia's electronics magazine siliconchip.com.au icon), where individual areas of the chip are selectively con- verted to a silicon dioxide insu- lating layer #2 Patterning This involves laying down the desired circuit pattern on the wafer or deposited or etched materials. This is done using a photographic-l ike pro- cess called lithography (see Figs.11- 13). The mask, which is like an old photographic slide or negative, is placed between an appropriate light source and the die, and an image is projected onto the die, which has been coated with photoresist. Note that the mask is much larger than the die size; a reducing lens is used to shrink the mask size to the die size. Also, the mask usually has areas 18 Silicon Chip Copyright © 2022 SILICON CHIP Publications. Downloaded by Mike Blake (#19283)
Fig.12: the basic process of photolithography using photoresist. Original systems. Electron beams can be used Source: Wikimedia user May lam (CC BY-SA 4.0) as an alternative to light sources. a mask is exposed at one time, enabling After certain photoresist regions are Electron beam lithography provides better focusing. washed away, the wafer itself can be a high resolution, but it has a low etched in those areas or processed in throughput, so it is mainly used for Before patterning, the die will have some way, such as being doped. After low-volume production of semicon- been prepared with a light-sensitive that, the remaining photoresist can be ductors and the production of pho- coating called photoresist. In the case removed. tomasks. of a positive photoresist, the photore- sist regions that are exposed to the light Using shorter wavelengths of light There may be multiple masks used become soluble and can be washed allows for higher pattern resolutions. and multiple exposures between addi- away, leaving the unexposed photo- These days, the density is so high that tional etching, deposition and other resist behind. A negative photoresist the light is typically in the UV spec- procedures. will do the opposite. trum, or extreme UV (EUV) in the latest Another possible process is contact lithography, but it is not used for mass production. Figs.14 & 15 will give you an idea of the complexity of the built-up lay- ers of an IC. Figs.16 & 17 are mask and die images of the world’s first micro- processor, the Intel 4004, designed by hand and released in 1971. Consider that modern chips are many orders of magnitude more complicated than that! Other lithographic processes of note, but not currently used for mass production, are: • displacement Talbot lithography (DTL) for periodic patterns • thermal scanning probe lithog- raphy (t-SPL), where nanoscale structures are generated with a heated probe moved over the sur- face of a resist coating which is then etched • UV flood exposure, to expose individual wafers on a small R&D scale Fig.13: a simplified version of the etching process using a positive photoresist. Cr is chromium on the mask, while PR stands for photoresist. Source: Wikimedia user Cmglee (GNU FDL V1.2) Fig.14: a simplified version of the processes to produce a portion of a CMOS IC. Note that the gate, source and drain contacts are not usually in the same plane in real devices. Source: Anonymous Wikimedia user (CC BY-SA 3.0) siliconchip.com.au Australia's electronics magazine June 2022 19 Copyright © 2022 SILICON CHIP Publications. Downloaded by Mike Blake (#19283)
Fig.15: a cross-section of a multi-layer CMOS chip with five metal layers, • direct laser lithography, a form of denoted Layer 1 to Layer 5. There’s a legend at the top; STI is shallow maskless lithography, for small trench isolation, FEOL is front-end of line and BEOL is back-end of line. scale R&D use Original Source: Wikimedia user Cepheiden (CC BY-SA 3.0) • nanoimprint lithography, in Fig.18: light is diffracted which nanoscale patterns are as an incident wavefront imprinted into a resist by a mould of a beam of light (eg, from a laser) passes by an edge, causing with the desired pattern and then potentially unwanted secondary wavefronts and thus light spreading. In etched photolithography, the edge would be part of the mask pattern. #3 Removal 20 Silicon Chip Australia's electronics magazine Material is removed from the silicon die by wet or dry etching processes Copyright © 2022 SILICON CHIP Publications. or a combination of chemical and mechanical polishing (called CMP for chemical-m echanical planarisation). The polishing is also used to ensure that the surface of the wafer is atomi- cally flat before the next layer is added. #4 Modification of electrical properties This involves processes such as dop- ing selected areas by methods such as diffusion or ion implantation to cre- ate the sources or drains of transis- tors, with P- or N-type dopants, or the creation or modification of insulating areas, such as through oxidation. Ion implanation is a method of dop- ing in which a beam of dopant ions from a particle accelerator is scanned over the wafer, implanting ions in the areas not covered by the photoresist to a controllable depth. The wafer is then annealed in an oven, reforming the crystal structure and ensuring that the ions are evenly distributed. Alternatively, dopants can be intro- duced to the surface of the wafer via gas-phase or solid diffusion, followed by ‘drive-in’, where the dopants are diffused deeper into the semiconduc- tor material. The wafer is placed in a furnace with an inert atmosphere and heated, diffusing the dopants through- out the areas on which they have been deposited. Similar furnaces can be used to also convert the top layer of semiconduc- tive silicon to the insulator silicon dioxide by heating the wafer in an oxygen-rich atmosphere. Front-end-of-line and back-end-of-line The term “front-end-of-line” (FEOL) refers to the initial part of the fabri- cation process, where the individ- ual components such as capacitors, diodes, resistors and transistors are formed. But it is before metal inter- connect layers are deposited to join them electrically. siliconchip.com.au Downloaded by Mike Blake (#19283)
The FEOL process for CMOS (com- plementary metal oxide semiconduc- tor) includes the following steps: 1. preparation of the wafer 2. electrical isolation of trenches or other selected areas by oxida- tion of silicon to silicon dioxide or deposition of other dielectric materials 3. well formation (the well is the first layer fabricated of a CMOS IC and may comprise an N-doped well in a P-type substrate; see Fig.15) 4. gate module formation 5. source and drain module for- mation The gate, source and drain referred to above are the main parts of a field- effect transistor or FET. “Back-end-of-line” (BEOL) refers to the second main stage of IC fabrica- tion, where the interconnection of the devices formed in the FEOL process takes place by adding metal layers. It also includes the addition of insu- lating layers, vias (vertical conducting elements to connect between layers; see Fig.15) or bonding sites for chip- to-package connections. Many metal layers can be added in multiple pro- cessing steps. You can think of these a bit like the copper patterns on a PCB. Wavelength of light for lithography Over time, as the number of transis- tors on a chip has increased, lithogra- phy has required shorter and shorter wavelengths of light to produce the smaller IC feature sizes. We’ll have some details on the light sources used when we discuss the shrinking process nodes in part two, next month. Features smaller than the Figs.16 & 17: images of the Intel 4004 microprocessor from 1971 showing wavelength of light a composite image of the masks (light colour) and the die (dark colour). It was a 12mm2 4-bit microprocessor with 2250 transistors and it started an As you can see from the above, IC electronic revolution. Source: Tim McNerney (http://alumni.media.mit. feature sizes are now much smaller edu/~mcnerney/2009-4004/) than the wavelength of light passing Fig.19: an illustration of the Rayleigh Criterion, the theoretical limit of through the mask and illuminating the resolution. The two blue peaks merge to form a single large (red) peak when wafer. You might expect that diffrac- they are close together but become separately resolved as they move apart. tion effects (spreading out the light Original source: Wikimedia user Mpfiz (public domain) and causing images to be indistinct) would prevent accurate patterns from being made on the wafer, and this is indeed the case. So how is this prob- lem overcome? There is a limit to how short the light wavelength can be (to make smaller feature sizes), so there is obviously the desire to minimise this effect. Note also that EUV equipment is expensive. siliconchip.com.au Australia's electronics magazine June 2022 21 Copyright © 2022 SILICON CHIP Publications. Downloaded by Mike Blake (#19283)
Fig.20: (a) a conventional binary mask, (b) an alternating phase-shift mask and (c) an attenuated phase-shift mask. The latter two types can provide finer details for the same wavelength of light. Original source: Wikimedia user Oleg Alexandrov (public domain) Diffraction (see Fig.18) is the pro- #1 Phase-shift masks with non-attenuated light from other duction of secondary wavefronts that Phase-shift masks make diffraction areas to enhance contrast and resolu- occurs at the edges of an opening work for you, not against you. Interfer- tion – see Fig.21(c) & (d). when the primary wavefront of a light ence is generated by phase differences beam passes through. This happens brought about by different thicknesses The half-tone mask has transparent with projection lithography, which is or translucency in parts of the mask, and semi-transparent material regions the dominant form, but does not hap- to improve contrast on the photoresist that cause light interference, enhanc- pen much with contact lithography, and thus resolution. Fig.21 shows the ing contrast and resolution. These although that is not suitable for mass behaviour of light energy with various masks are easier to make than alter- production. mask types. nating phase-shift masks. A conventional binary mask either There is a fundamental physical transmits light or doesn’t, depending #2 Photoresists limit to resolution defined by the on the region, as shown in Fig.21(a). To achieve higher resolution, new Rayleigh Criterion. The web page at In alternating phase-shift masks, photoresists have had to be developed. siliconchip.au/link/abdv states, “The some regions are made thicker and The following factors have to be con- Rayleigh criterion for the diffraction others thinner. When the thickness is sidered in developing a photoresist: limit to resolution states that two appropriately chosen, the light going • contrast between exposed and images are just resolvable when the through modified areas of the mask centre of the diffraction pattern of one interferes with the light going through unexposed portions is directly over the first minimum of unmodified regions, improving con- • sensitivity to the wavelength of the diffraction pattern of the other.” - trast and resolution – see Fig.21(b). see Fig.19. In attenuated phase-shift masks, the light used (the shorter the light is allowed to pass through partic- wavelength of light, the less It is not simply a matter of making ular mask sections but attenuated due absorption of light energy) a design and specifying it be made to partial transmittance of the mask • viscosity smaller; significant new problems material. The small amount of light • adherence to the substrate have had to be overcome each time the allowed through will not cause a pat- • the ability to resist etching process size has been shrunk. The var- tern on the wafer, but it will interfere • surface tension ious techniques that have been used to achieve feature sizes smaller than #3 High numerical aperture lenses the wavelength of light are: Note the projection (also called objective) lens in Fig.11. The lens Fig.21: different mask types with the resulting patterns that appear on the wafer. Note the middle detail missing on the wafer for the binary mask and the added detail in the phase-shift masks. Source: Wikimedia user Shigeru23 (GNU FDL V1.2) 22 Silicon Chip Australia's electronics magazine siliconchip.com.au Copyright © 2022 SILICON CHIP Publications. Downloaded by Mike Blake (#19283)
should gather the diffracted light from Note the overhead system as TSMC’s facility and the row of DD-1223V the mask. The higher the numerical “12-inch” wafer furnaces. Picture: Taiwan Semiconductor Manufacturing aperture (NA) of the lens (similar to Co., Ltd. the f-number in photographic lenses), the more diffracted light that it will resolution than could be achieved with used to create a second set of features. gather and the higher the resolution a single step. Therefore, the distance between the of the image produced. As an example, a double patterning features can then be less than the min- However, the higher the NA, the process results in a 30% smaller fea- imum pitch of the lithographic system smaller the depth of focus, requiring ture size, but the number of process – see Fig.23. extremely precise mask alignment to steps and therefore cost is increased. avoid parts being out of focus. Next month Double patterning is used to make #4 Immersion lithography NAND flash memory (as used in SSDs Next month, we will discuss how Another technique is to use immer- and SD cards), random access memory feature sizes have changed over time sion lithography, in which the light (RAM) and the fins in FinFETs, used and what advances that progress has passes through water rather than air. in many cutting-edge computer chips. allowed, including Moore’s Law. We’ll The higher index of refraction of water also go into more detail about the sili- means an effective decrease in wave- There are many different methods con wafer sizes and extreme UV (EUV) length of about 33%, enabling smaller of multiple patterning. Double pattern- lithography, plus describe IC packag- feature sizes. ing in its original form was also called ing and the various components that pitch splitting. can be created using the IC fabrication #5 Optical proximity correction process described above. OPC (optical proximity correction) is Two adjacent features cannot be a method to compensate for errors due made closer together than the mini- There is more to come after that, to diffraction or other reasons – Fig.22 mum pitch allowed by the lithographic including the latest 3D stacking and shows one example. Calculating the system; therefore, one set of features multi-chip module technologies. SC correct patterns for OPC is extremely is made first, and the second mask is computationally intensive and can occupy compute clusters for days. #6 Multiple patterning Double or multiple patterning, also known as self-aligned multiple patterning (SAxP), is a complicated and expensive process. It is used to produce photomasks for the highest possible feature density. In multiple patterning, multiple lithography and etch steps are used to achieve higher Fig.22: in optical proximity Australia's electronics magazine Fig.23: a form of correction, the image is ‘pre- multiple patterning corrected’, so the projected pattern called pitch splitting. distorted by projection is the desired one. The thicker areas are the desired Three trenches pattern; the thinner wavy lines do not are first etched, print. They are called sub-resolution then covered in assist features (SRAFs) and improve photoresist (top). depth of focus. Source: Wikimedia Then a second user LithoGuy (CC BY-SA 3.0) exposure is made, siliconchip.com.au and a second set of trenches is etched (middle). The photoresist is washed away, resulting in pairs of trenches that are closer together than a single exposure would allow (bottom). Original Source: Wikimedia user Wdwd (GNU FDL V1.2) June 2022 23 Copyright © 2022 SILICON CHIP Publications. Downloaded by Mike Blake (#19283)
QRWH -HUHP\\/HDFKÍV 0,', 86% 3RO\\SKRQ\\ ,QSXWV 0,', +DUPRQLF /RZ/DWHQF\\ 6\\QWKHVLV 6<17+(6,6(5 $XGLR '7LPEUH 2QERDUG 0RUSKLQJ 3DWFKLQJ This advanced MIDI synthesiser is easy to build and can be hooked up to any MIDI compatible device. It lets you explore the broad range of acoustic elements that capture the characteristics of real and imaginary instruments. It is more than an experiment – it is a full-blown instrument capable of forming rich, detailed sounds using a plethora of settings, envelopes and waveforms – a blank canvas. The Spectral Sound and imaginary instruments. As such, Manual for those who want to explore Synthesiser uses seven it is a wonderful way of appreciating more deeply. dsPIC chips running at 70 or 40MIPS types of sound, how musical instru- each in combination to produce ments work and why some instruments The whole topic of sound synthe- sounds digitally. This gives it 18 note are notoriously difficult to emulate. sis, interwoven with music history, is polyphony (ie, the ability to play 18 a rich and intensely interesting evolu- different notes simultaneously) and While it is a working device as pre- tionary journey, driven by our instinc- complex sound creation, with ‘tim- sented, it’s also a great way to exper- tive desire to understand and create bre morphing’ being the module’s iment with audio synthesis. This is sound. The Spectral Sound Synthe- key feature. a fun and stimulating pursuit, with siser taps into that desire. an endearing interplay between dig- It also has low latency, which is ital waveform generation and human An overview of the system important since you don’t want an perception. apparent delay between pressing a key Fig.1 shows how the overall system on a keyboard and the sound being The module’s design focuses on true works. It has a MIDI input for receiving produced. parallel processing by splitting the MIDI note and control messages (eg, computational load across six ‘Tone from a keyboard), a USB input for con- Being a standalone sound module, it Processor’ chips. figuration by the Windows software, has the tantalising possibility of being and a stereo line-level audio output built into custom DIY musical instru- All the source code is available for jack, for hooking it up to an amplifier. ments without the need for a computer. the firmware and the accompanying Windows desktop software. However, You can create patches with the The module is an adventure into the software is also available pre-built, Windows software, and a certain num- real-time sound synthesis, exploring including any version updates. ber of these patches are loaded onto the broad range of acoustic elements the module and stored internally. You that capture the characteristics of real This is an advanced project, and can send any tweaks to patch settings 24 Silicon Chip there is even a Technical Reference siliconchip.com.au Australia's electronics magazine Copyright © 2022 SILICON CHIP Publications. Downloaded by Mike Blake (#19283)
immediately to the module and hear For samples of what the the result. Synthesiser can do, visit siliconchip.au/link/abeo The ‘Master Controller’ is a PIC18LF25K50 8-bit micro with useful Fig.1: this block diagram shows how the Spectral Sound Synthesiser works. The USB connectivity. This chip is com- Master Controller receives MIDI messages from the MIDI In port and patch data mon in embedded systems requiring from the computer via USB. It commands the six Tone Processors to generate USB. It functions as a hub, processing sounds based on the stored patches and possibly stored performance data. incoming USB and MIDI messages. It These sounds are fed to the Mixer and then to the analog audio output. also allocates processors to tasks in the rest of the system. sound’s frequency spectrum and how transfer functions, but this idea soon it changes over time. became widespread, from predicting The six Tone Processors are tides to planetary motion, and much dsPIC33EP512MC502-I/SP 16-bit Additive synthesis is a method later, audio synthesis. chips running identical code to gen- of creating and modulating timbres erate digital sound samples. Each cal- based on the fact that any periodic The simplicity of the idea is appeal- culates up to three live note instances function can be expressed as the sum ing because it means that complex tim- at once, so the system has a maximum of a series of sinewaves – the ‘Fourier bres can be constructed just by adding of 6 x 3 = 18 note polyphony. Each series’, described by Joseph Fourier sinewaves with appropriate weights Tone Processor holds a single patch, in 1822. He was using it to solve heat and phases. but different ICs can have different patches, making this MIDI instrument ‘multi-timbral’. A single ‘Mixer’ chip, a 16-bit dsPIC33FJ128GP802-I/SP, mixes the samples from all the Tone Processors, limits the generated audio level using automatic gain control (AGC), then passes the audio out through its inbuilt stereo DAC to an MCP6022 op amp. The output is ‘pseudo stereo’, using a well-known audio trick called the Haas effect, where delaying a copy of a signal from one ear to the other gives a very convincing impression of a stereo field! The module can hold several patches and ‘performances’ in a 24LC512 EEPROM IC. What is additive synthesis? Ongoing research into hearing and human perception reveals that we are still a long way from completely understanding how our brains pro- cess and identify sound. A key ele- ment is timbre, which is related to a The MIDI Synthesiser fits in an instrument case measuring 150 x 100 x 40mm. A different case could be used as long as it’s bigger, the height of it depends on the heatsink you use. siliconchip.com.au Australia's electronics magazine June 2022 25 Copyright © 2022 SILICON CHIP Publications. Downloaded by Mike Blake (#19283)
It turns out that phase is not gen- Fig.2: approximations of the harmonic structure of different instruments. From erally important because our hearing left to right, the bars represent the sequential harmonics above the fundamental. disregards it. This makes sense when The harmonic structure is what defines the timbre of an instrument, while the pondering sound waves bouncing fundamental frequency is determined by the pitch of the note being played. about in a room; despite the phases of different frequencies getting mangled, a Sydney Harbour ferry) was a break- As computing power has steadily we generally do not perceive any tim- through in sound production through increased in recent years, we have bral difference. sampling. It revolutionised pop music seen growth in physically-modelled with genuinely new sounds. sound, such as in the popular “Piano- Another appeal is that sounds in tech” VST pianos based on the phys- nature are based on vibrations where The irony is that the inventors ics of real instruments, ancient and the timbral sinewaves have frequen- started by using additive synthesis, modern. Additive synthesis can also cies that are integer multiples, or har- according to co-founder Kim Ryrie be categorised as physical modelling monics, of the base ‘fundamental’ fre- (interestingly, also the founder of to a degree because of its timbre-based quency. This well-defined relation- ETI magazine): “We regarded using approach and dynamic nature. ship lends itself to computation. Musi- recorded real-life sounds as a compro- cal instruments can be recognised by mise – as cheating – and we didn’t feel Harmonics and the equal- their characteristic harmonic levels, particularly proud of it.” tempered scale with some examples shown in Fig.2. This Fairlight model was a ‘sam- Real instrument sounds are gener- The large evolutionary family tree of pler’, with the ability to record sound, ated through vibration, such as the electronic synthesisers includes prom- soon followed by cheaper ‘Romplers’ movement of air in a flute, the vibra- inent examples of additive synthesis. with recorded sound baked into ROM. tion of a guitar string or the oscillat- For example, the beloved Hammond These days we can have gigabytes of ing of the skin of a drum. The vibra- organ dating back to 1935 stacks tones samples on solid-state hard disks. tions create standing waves, having generated by pickups placed close to fixed nodes and moving antinodes. rotating mechanical ‘tonewheels’. It is undeniable that sample-based The nodes divide the length into equal synths can give amazing results, espe- divisions, leading to the integral har- Also, the early Fairlight Quasar cially with many nuanced ‘layers’ monics seen in the frequency spectrum synth of the 1980s was additive, as for parameters such as note velocity. of many instruments. were the Synclavier and a few Kawai However, they use masses of mem- keyboards. Loom, a modern VST ory instead of modelling anything on Fig.2 broadly shows how these har- instrument, is also an additive type. a physical basis. Still, from the early monics have characteristic levels in sampled tape loops of the Mellotron, different instruments, although it is With enough computing power, used on classics such as the pipe organ extremely generalised. additive synthesis makes the ‘morph- in the Beatles’ “Strawberry fields”, it ing’ of timbres possible by altering the is clear that samples are here to stay. An instrument plays at a pitch we set of sinewaves being summed over time – akin to what happens all around Australia's electronics magazine siliconchip.com.au us with natural sounds. Additive synthesis also has the great advantage of operating in the fre- quency domain rather than the time domain. This makes filtering a simple concept, where the filter contour sim- ply scales the levels of the base sine- waves. Brick-wall filtering is nothing more than including or excluding cer- tain sinewaves. This method of synthesis can cre- ate rich, stimulating and captivating sounds. But it has limitations when emulating real instruments compared to sample-based synthesis. The problem is that natural sound is far more complex than just harmonics; there are ‘in-harmonic’ frequencies in the spectrum, especially for percussive sounds. There is noise from blowing, scraping and scratching. The harmon- ics are not always exact integer mul- tiples etc. So, as a sonic tool, additive synthe- sis is great. But it cannot always emu- late natural sound easily. The Fairlight CMI synthesiser of the 1980s (which took its name from 26 Silicon Chip Copyright © 2022 SILICON CHIP Publications. Downloaded by Mike Blake (#19283)
► Fig.3: the equal-tempered scale has the advantage that music can be played in any key without retuning the instrument. But some note harmonics do not precisely match any note in the scale, with the worst being the 7th and 11th harmonics. Usually, though, such high harmonics are not especially loud, so this tends not to matter. Fig.4: the main tasks and calculations that are constantly being processed by the six Tone Processor chips that do most of the synthesis work. recognise as the fundamental fre- between any note and its neighbour. harmonics have an exact match in the quency, but the tone has a ‘colour’ To calculate the frequency of a note scale, but others don’t (although they dictated by the relative strength of the a semitone higher, we can multiply are often quite close). harmonics. The fundamental is known by this fixed factor. Since notes one as the first harmonic; the second har- octave apart have a ratio of 2:1, if each This reveals a degree of ‘in- monic is at double the fundamental semitone has a fixed geometric ratio, harmonicity’ in harmonics (ironic, frequency (an octave higher), the third that ratio must be the 12th root of two given the name). The analysis applies harmonic at three times the fundamen- (approximately 1.0595:1). to all notes; the 7th and 11th harmon- tal frequency and so on. ics deviate the most from recognisable Remember that this is a human pitches, and if audible, they sound But it is seldom realised that some invention to get a system of equal ratios ‘flat’ and dissonant. harmonic frequencies only roughly so that music can be transposed with- match the pitches that we recognise in out altering how it sounds. Although These imperfections are well- the chromatic (12-note) musical scale! the oddities of previous scales contrib- known to instrument makers. For Our brains have heard the pitches of uted to the richness of music diversity, example, pianos are designed to have notes from our earliest memories. Yet, and some bemoan their demise, the the hammer strike the strings at the the musical scale we use today is rela- equal-tempered scale makes a certain seventh vibrational node to suppress tively recent, and human beings have amount of sense. this ‘ugly’ seventh harmonic! The tried several alternatives, going right 11th harmonic is less noticeable, often back to Pythagoras. Fig.3 is a detailed analysis of musi- being naturally quieter. cal note A3 (440Hz), showing how the The scale we use today is called harmonics of this note do not always Tone processors the “equal-tempered” scale, where accurately align to the pitches of the ‘equal’ refers to a fixed frequency ratio equal-tempered scale. Power-of-two Fig.4 shows the heart of one of our siliconchip.com.au Tone Processor “Note Instances”, Australia's electronics magazine June 2022 27 Copyright © 2022 SILICON CHIP Publications. Downloaded by Mike Blake (#19283)
Fig.5: envelopes are time-based profiles that can be applied to various synthesis are swapped so that we never have parameters. The easiest to understand is the volume envelope, which varies the to copy data slowly and inefficiently loudness of the note from the time it is triggered until it is no longer audible. from one table to another. Fig.6: ‘3D timbre morphing’ is a solution to the problem that the harmonics The wavetable for a note instance of various instruments can vary depending on which note is being played, gets regularly refreshed during its how hard it is being played etc. This is especially obvious on instruments like active life cycle. The rate of refresh is pianos, where each key can have a unique sound, and louder notes can trigger not fixed but is approximately 50Hz. various resonances. As a side note, it is fascinating to read research where they have found showing how it generates sound. A factors such as the note velocity. This that the threshold where humans can note instance represents a note we involves looping through all active detect a change in timbre occurring play on the MIDI-connected keyboard. harmonics and adding corresponding is often a lower rate than this. Tim- The Master Controller ensures that the sinewaves together. bre detection is clearly a demanding, played notes are evenly spread across abstract recognition task in the brain. the available Tone Processors. Because harmonics are exactly inte- ger multiples of the fundamental, the During note generation, several ‘gain When a note instance is started on a wavetable holds exactly one cycle of envelopes’ can be applied to aspects of Tone Processor chip, the first thing that the summed periodic wave. Once that the sound. This is the sound’s ampli- happens is the calculation of the wave- has been calculated, this physical table tude envelope. Fig.5 indicates how the form to be used based on the ‘static’ in memory becomes ‘active’ for the system calculates envelopes. Both lin- patch settings, plus other ‘dynamic’ note instance. To do this, table pointers ear and exponential envelopes can be 28 Silicon Chip created, and each section of the ADSR Australia's electronics magazine (attack, decay, sustain, release) enve- lope has a ‘target’ value. During each section, the envelope’s current value moves towards the target. Exponential curves exist every- where in nature, relating to energy decay. So, that is a natural choice, and not just for amplitude. For exam- ple, when plucking a string, the high- frequency content decays first. The jos- tling of atoms in high-frequency vibra- tion uses up energy at a higher rate. A note instance also features three ‘Low-Frequency Oscillators’ or LFOs: Vibrato varies the pitch, Tremolo the amplitude and Timbre the harmonic levels. LFO modulation can add so much character to sound. There are also envelopes for the depth of this modulation, allowing, for example, a gradual onset of vibrato. Another interesting point is that whereas many synths would use Trem- olo across the total sound output of the synth, this module modulates per note, making the overall sound more complex and interesting. Timbre morphing Determining the harmonic levels to use when constructing a waveform is a complicated process. Fig.6 shows that a patch holds the harmonic data for 75 waveforms, with the waveform to synthesise depending on three param- eters. The “Note” parameter is the position of the note on the keyboard. The “Intensity” parameter most often means note velocity. The “Waveform” picks from five waveforms. Each point in this conceptual 3D space grid has a set of harmonic levels siliconchip.com.au Copyright © 2022 SILICON CHIP Publications. Downloaded by Mike Blake (#19283)
defining a waveform. The current parameter values define the required point internal to this space, and the harmonic levels to use are interpolated from the nearest defined grid points in this space. Taking this further by modulat- ing through this space with the Tim- bre LFO can give impressive realism compared to plain vibrato. Typical vibrato purely modulates the pitch of the whole waveform, whereas timbre modulation is harmonic-based and therefore, a far more complex modu- lation for our brains to perceive. It is thrilling to hear this difference and realise that our brains feed off the interest in sound. Perhaps, con- sidering the incredibly clever pro- cessing that our brains can perform with language, pattern detection and all aspects of sound, this realisation should not be too surprising. Towards greater realism Fig.7: an overview of all the tasks that the Tone Processor chip has to handle, in order of priority. The highest priority events are those that would cause the We have already mentioned that sound to break up or otherwise give unexpected results if delayed. natural sound includes in-harmonic elements, which do not fit the neat amount, giving a chorus-like effect, frequencies, raising the harmonic integer multiples of harmonic frequen- which tries to account for the fact that oscillation frequency. This effect cies. In an attempt to address this, this instruments like the piano use multi- applies to all stringed instruments, synthesiser includes some additional ple detuned strings per note. and is something our system can- features outside of the purely additive- not address because the model relies synthesis approach. A final feature is the ‘Body Reso- entirely on integral harmonics. nance Filter’. This attempts to emulate Firstly, a noise envelope is available the body resonance of a real instru- A real-time system to help simulate ‘blown’ instruments. ment by filtering the overall system This is a white noise generator with an sound. After specifying the filter con- The whole of the module is an exam- adjustable low-pass filter. tour in the app, it scales the harmonic ple of a ‘hard’ real-time system, where levels. Although this method has great the deadlines of sample production Secondly, you can add short in- theoretical appeal, it has mathemati- and processing are immovable. This harmonic samples. These are hard- cal limitations. presents considerable challenges, and coded clips of the sound of taps, the development of the module was a scratches, clicks and bonks. However, Despite all these extra features, there slow evolution of coding, measuring, the sample feature also includes an are still real-life complexities that the refining and sometimes redesigning. implementation of the well-known module just cannot tackle. For exam- ‘Karplus Strong’ delay line technique ple, a piano has peculiarities due to the The Tone Processors all run entirely of plucked string synthesis. stiffness of its strings, where the higher in parallel and are polled by the mixer harmonics get sharper compared to chip to provide samples at the audio The 1983 paper by Kevin Karplus the expected harmonics of the string. sampling rate of 41.7kHz. Inside each and Alex Strong entitled “Digital Syn- Tone Processor is a hierarchy of inter- thesis of Plucked-String and Drum This is because the stiffness effec- rupts, as shown in Fig.7, made possible Timbres” first described this tech- tively shortens the string for higher nique. It is a computationally simple but effective method of generating real- istic, decaying string sounds that start off life in the delay buffer as noise. It adds a powerful tool to this module, even though it does not have anything to do with additive synthesis! There are also settings to randomly or systematically detune the frequency of played notes, in an attempt to intro- duce the impurities of real instru- ments. Plus, there is an option of using two wavetable oscillators per note instance, detuned by a specified siliconchip.com.au Australia's electronics magazine June 2022 29 Copyright © 2022 SILICON CHIP Publications. Downloaded by Mike Blake (#19283)
30 Silicon Chip Australia's electronics magazine siliconchip.com.au Downloaded by Mike Blake (#19283) Copyright © 2022 SILICON CHIP Publications.
Fig.8: the entire Synthesiser circuit, which is somewhat unusual in that it mainly consists of eight PIC microcontrollers (of three varieties), all communicating via two separate SPI serial buses. The remainder of the circuit comprises the EEPROM (IC11) used to store patch and performance data, the power supply, MIDI input and audio output. siliconchip.com.au Australia's electronics magazine June 2022 31 Copyright © 2022 SILICON CHIP Publications. Downloaded by Mike Blake (#19283)
by the dsPIC’s ability to assign prior- The Spectral Sound Synthesiser PCB to these chips are that each Tone Pro- ities. is relatively easy to build. Although cessor’s CS2a-CS2f input (pin 4) con- with about a dozen ICs, many of them nects to a different select line on the The main routine of a Tone Proces- having 28 pins, there are lots of solder Mixer, IC3. sor, the centrepiece of the entire sys- tem, is just a simple loop that recal- joints to make. Be careful to make The Mixer is a different (but related) culates wavetables. This ‘background’ each joint properly or it might not type of dsPIC processor. Besides being task is unpredictable in duration, is not connected to this SPI bus and the six on a deadline, and can vary depending function correctly. chip select lines for the Tone Proces- on the interrupt activity and the com- sors, it also has two differential analog plexity of the waveform being built. The performance advantage of this outputs from an internal stereo DAC. method massively outweighs rare This means that the timbre refresh anomalies that are probably not even These signals are fed to op amp IC4, rate could slow down in certain cir- noticeable. which converts the differential signals cumstances – although, in use, per- to single-ended audio signals suitable formance is very acceptable, and tim- The software and hardware are for feeding to the CON2 output jack. bre changes are perceived as fluid and designed for speed. All the dsPICs Simultaneously, this circuitry filters smooth. are running at their fastest. All cal- out the DAC step artefacts using low- culations are integer-based, coupled pass filters built from added capacitors The processing ‘layers’ above this with extensive use of the on-chip and the existing gain-setting resistors. base main loop are concerned with hardware multiplier via the compil- calculating envelope steps, calculat- er’s “__builtin” commands. The code A virtual ‘half-supply’ rail is gen- ing the sample output and processing extensively uses shift operations for erated using zener diode ZD1 biased received data. fast multiplication and division, and from the +3.3V rail so that the audio numerous lookup tables are used, signals from IC3 remain within the A trick used to improve throughput including a detailed sine lookup. supply rails of the op amp. on the SPI bus between the Tone Pro- cessors and the Mixer is only send- Circuit details Mixer IC3 also connects to the ing the changes in sample values. The 24LC512 EEPROM (IC11) using a two- summing of sines in a Tone Processor The full circuit is shown in Fig.8. wire I2C serial interface (SDA & SCL). can result in a total value exceeding 16 The first thing to note is that the six That chip has its own bypass capacitor bits. The total on the Tone Processor Tone Processors (IC5-IC10) are identi- plus pull-up resistors for those serial is a 32-bit signed integer, but the Tone cal dsPICs configured in the same fash- lines, and that’s it. Processor only sends the change in this ion, each with just a handful of asso- total, capped at 16 bits, to the Mixer. ciated components: one Vdd bypass The last task for IC3 is to drive the capacitor, one Vcap capacitor (required Mixer Alert LED, LED2, from its RA0 This can cause signal distortion, but for the chip’s internal regulator) and output (pin 2). statistically, this will happen rarely. one 10kW MCLR pull-up resistor to 32 Silicon Chip prevent spurious resets. MIDI input, USB and other control tasks fall on the PIC18LF25K50, IC2. Besides the power supply, the only It monitors the presence of USB 5V connections to these chips are a com- at its RA0 digital input (pin 2) via a mon SPI bus, as they are pure num- 2.2kW/10kW ‘divider’, which mainly ber crunchers, and all commands and exists to limit the current into that pin data are sent on this bus. The only and ensure that it’s pulled to 0V when differences between the connections no USB connection is present. Australia's electronics magazine IC2 and IC3 communicate via a sec- ond separate SPI bus, with a dedicated chip select line, from pin 7 of IC2 to pin 22 of IC3. IC2 also drives LED1, the MIDI Alert LED, from its RB6 dig- ital output (pin 27). External potentiometer VR1 (the volume control) connects to CON4, placing it across the 3.3V supply. Its wiper goes to analog input AN11 of IC2 (pin 25). IC2 reads the voltage at the wiper using its analog-to-digital converter (ADC) and passes the digital value along to IC3, which then scales its output to provide the desired vol- ume level. The pot value or type isn’t criti- cal, but 100kW is reasonable. Scal- ing the audio sample values entering the DAC, rather than directly adjust- ing the op-amp gain, simplifies the PCB at the cost of reduced audio bit- resolution with the volume turned siliconchip.com.au Copyright © 2022 SILICON CHIP Publications. Downloaded by Mike Blake (#19283)
down. In practice, it’s hard to hear this Parts List – Spectral Sound MIDI Synthesiser degradation. 1 double-sided PCB coded 01106221, 145 x 94mm That just leaves the MIDI input, 1 instrument case [Takachi YM-150; RS Cat 373-2255] clock signal distribution and the 4 stick-on rubber feet power supply to describe. 1 front panel label, 145 x 37mm (see Fig.10) 1 lid label, 141 x 85mm (see Fig.11) The MIDI signal is applied to CON6, 1 5-6V DC 1A regulated plugpack • and it powers the IR LED within 1 16MHz crystal, HC-49 (regular or low-profile) (X1) FOD260L opto-isolator OPTO1. A 1 PCB-mount DC barrel socket, 2.1mm or 2.5mm ID to suit plugpack (CON1) 220W resistor provides current limit- 1 3.5mm stereo DPST switched jack socket (CON2) [Altronics P0094, RS ing, while diode D2 prevents the LED from being reverse-biased. It is essen- Cat 913-1021 or CUI SJ1-3555NG] tial to use the FOD260L opto-coupler 1 2-pin polarised header and matching plug (CON3, for power switch) as this is suitable for 3.3V operation 1 3-pin polarised header and matching plug (CON4, for volume control) – other varieties may well not work. 1 through-hole full-size type-B USB socket (CON5) [Jaycar PS0920, The output transistor in OPTO1 is Altronics P1304A/P1304B] operated in common-emitter mode 1 5-pin 180° DIN socket, right-angle PCB mount (CON6) [Jaycar PS0350, with a 470W pull-up resistor. The resulting signal goes to the RX input Altronics P1188B or RS Cat 491-087] (pin 18) of IC2. 1 SPST or SPDT panel-mount slide switch (S1, power) 1 100kW panel-mount linear potentiometer & knob (VR1, volume control) A single external oscillator is used 8 28-pin narrow DIL IC sockets (optional; for IC2, IC3 & IC5-IC10) because we have eight microcon- 3 8-pin DIL IC socket (optional; for IC4, IC11 & OPTO1) trollers that all need clock sources. 1 TO-220 heatsink (REG1) [maximum 40mm wide, 13mm deep from tab, This is built using crystal X1, its two 33pF load capacitors and unbuffered <18°C/W; RS Cat 263-251 used for prototype] inverter IC1a. The resulting 16MHz 4 M3-tapped 6.3mm spacers signal is inverted by IC1b and buff- 1 M3 x 10mm panhead machine screw, shakeproof washer and hex nut ered by IC1c and IC1d, then fed to all 8 M3 x 5mm panhead machine screws the microcontrollers’ clock input pins. 2 M2 x 10mm countersunk screws and nuts (for slide switch mounting) 1 100mm length of rainbow cable (for wiring to S1 & VR1) We don’t recommend using a buff- 1 small tube of thermal paste ered inverted in place of IC1, such as • up to 9V can be used, but 5-6V results in more reasonable dissipation the more common 74HC04, as it might not oscillate correctly. Semiconductors 1 74HCU04 unbuffered hex inverter, DIP-14 (IC1) The power supply is simple; the 1 PIC18LF25K50-I/SP 8-bit micro programmed with 0110622A.HEX (IC2) unit is powered with 5V DC from bar- 1 dsPIC33FJ128GP802-I/SP 16-bit microcontroller programmed with rel socket CON1, and this flows via reverse-polarity protection diode to 0110622B.HEX (IC3) the inputs of linear regulators REG1 1 MCP6022-I/P rail-to-rail op amp, DIP-8 (IC4) and REG2. REG1 powers all the dig- 6 dsPIC33EP512MC502-I/SP 16-bit microcontrollers programmed with ital circuitry while REG2 powers the analog circuitry, which is basically 0110622C.HEX (IC5-IC10) just op amp IC4 and the bias for zener 1 24LC512-I/P 64Kbyte I2C EEPROM, DIP-8 (IC11) diode ZD1. 1 FOD260L opto-coupler, DIP-8 (OPTO1) 2 LF33CV 3.3V low-dropout linear regulators (REG1, REG2) Increasing the signal-to-noise 2 3mm high-brightness green LEDs (LED1, LED2) ratio (SNR) 1 1.8V 250mW zener diode (ZD1) [eg, 1N4614] 1 1N4004 400V 1A diode (D1) A challenge with any system com- 1 1N4148 75V 150mA signal diode (D2) prising mixed digital and analog cir- cuitry is to stop the digital noise bleed- Capacitors ing through into the audio output. The 1 100μF 6.3V electrolytic module PCB takes the basic steps of 1 10μF 16V electrolytic separating audio and digital compo- 8 10μF 16V X7R ceramic nents as much as possible, with sepa- 2 1μF 63V MKT rate regulators and the use of a ground 4 100nF 63V MKT plane. However, additional measures 13 100nF 50V X7R ceramic have been taken to ensure a generally 2 33pF 50V ceramic quiet and acceptable audio system. Resistors (all 1/4W 1% metal film axial) One such measure is an audio lim- iter in the mixer audio code using 1 1MW 6 4.7kW 1 1kW advanced look-ahead AGC. A limiter squashes the dynamic range slightly by 1 100kW 4 3.3kW 2 470W attenuating peaks, thereby effectively boosting the quieter sounds and low- 10 10kW 4 2.2kW 1 220W ering the noise floor. Kit – SC6261 siliconchip.com.au An almost complete kit is available, which includes everything except the case, feet, labels and plugpack. It is priced at $200. Australia's electronics magazine June 2022 33 Copyright © 2022 SILICON CHIP Publications. Downloaded by Mike Blake (#19283)
Fig.9: like the circuit diagram, the eight PICs dominate the overlay, all in 28-pin DIL packages. Make sure to orientate those correctly and don’t get them mixed up. Fig.11: the lid panel ► artwork (shown at approximately 85% actual size) is a nice finishing touch to the project. It’s designed to be printed onto a transparent medium. Note that the two LED positions could vary somewhat, especially if you’re using a different case; you could simply cut that part of the decal off and position it separately. This does not seem natural when Construction a socket for IC1 as there’s little reason trying to emulate polyphonic instru- to use one there. ments; however, limiters and com- The Spectral Sound Synthesiser is pressors are commonplace in audio relatively straightforward to build, as Start with the resistors, checking reproduction, and it has a significant the use of numerous microcontrol- each lot of values with an ohmmeter beneficial effect in this system. lers minimises the number of separate before soldering them in place. Fol- components required. Most compo- low with the three diodes. Each is a We are also using a trick called nents mount on a double-sided PCB different type, so don’t get them mixed pre-emphasis and de-emphasis. The coded 01106221 that measures 145 x up, and ensure they are fitted with the digital audio generated has high- 94mm. The overlay diagram for this cathode stripes facing as shown. frequency boost applied, and the ana- PCB is shown in Fig.9. log signal processing circuitry has a Next come the IC and opto sockets matching high-frequency attenuation There is nothing remarkable about (or ICs and opto-coupler). Ensure they applied through a low-pass filter on the construction except that it requires all have pin 1 facing towards the top op amps. This way, the higher, more good soldering skills to solder 200+ of the board and if soldering the ICs noticeable element of circuit noise is pins accurately! We recommend using to the board, be very careful not to get suppressed. IC sockets throughout, including for the different 28-pin types mixed up. the opto-coupler; while sockets can The module actually ‘boosts’ the cause long-term reliability problems After that, mount all the non- higher harmonic levels by carefully due to oxidation of the contact points, polarised ceramic and MKT capaci- attenuating lower harmonic levels. It there is no real provision for in-circuit tors; there are 100nF ceramic and MKT is nice that complex digital filters are programming. capacitors, so make sure the MKTs go not needed to do this! in the positions shown in Fig.9. Still, since most constructors will be Finally, the Patch Editor applica- using pre-programmed micros (or pro- Now install the electrolytic capaci- tion automatically boosts harmonic gramming them before assembly), you tors with the longer positive leads to levels to the maximum, ensuring that could consider soldering them directly the pads marked + in Fig.9, followed the summed wavetable waveform is to the board as long as you are confi- by the polarised pin headers and jack across the full signed 16-bit range, dent they have been programmed cor- socket CON2. maximising the SNR. rectly. Note that we haven’t specified Next, solder the LEDs in place with the longer leads to the side marked A. Fig.10: the front panel artwork can be downloaded, printed, laminated (or protected in another manner) and then attached to the drilled panel. 34 Silicon Chip Australia's electronics magazine siliconchip.com.au Downloaded by Mike Blake (#19283) Copyright © 2022 SILICON CHIP Publications.
Fit these with sufficient lead length so long as it’s large enough to house the so that its top edge is right up against that they will reach the top panel of the PCB, since all the connectors and con- one side of the case (for an instrument case once the PCB has been installed trols (apart from the two which are case, it should be the front panel). After (see the section “Wiring it up” below panel-mounted) are along one edge that, mark and drill/cut holes in the for a discussion on case selection). of the PCB. adjacent panel for the DC power barrel Follow with the two regulators, first plug, MIDI input socket, USB socket attaching the heatsink to REG1 using With a 145 x 94mm PCB, most and audio output jack. the machine screw, nut and washer. cases measuring at least 165 x 100mm should be suitable. The height If you’re using the specified case, That just leaves crystal X1, DC required depends on the heatsink you you can use the drilling diagram, socket CON1, USB socket CON5 and are using for REG1. The specified heat- Fig.12, to assist you. It could also be MIDI socket CON6 on the PCB. Mount sink is only 20mm tall, so cases at least used on other cases, but you will need those in order of increasing height. 35mm tall should be fine. If you’re to adjust the placement on the panel Finally, if you’ve soldered sockets using a taller heatsink, add 10-15mm to match your PCB mounting location. to the board, plug in all the ICs and to its height to figure out what cases the opto-coupler now, paying careful will be suitable. Wiring it up attention to their pin 1 orientation and not getting the different 28-pin and Possible alternative instrument Once you’ve confirmed these are all 8-pin ICs mixed up. cases include Altronics Cat H0374 accessible through the panel, if you or Cat H0378 (with a short heatsink), haven’t already, drill holes for the Case selection Jaycar Cat HB5912 or the Hammond volume pot and power switch in con- RM2055M, which is available from venient locations. Then solder appro- The PCB is designed to fit into the Digi-Key and Mouser. It should also priate lengths of ribbon cable strips to case specified in the parts list, and the fit into a UB2 Jiffy box like Jaycar those parts and crimp/solder pins to front panel label (Fig.10), lid artwork Cat HB6012, Altronics Cat H0152 or the other ends that you then push into (Fig.11) and drilling template (Fig.12) H0202, but they don’t look as good as the plastic polarised header blocks (or all fit that case. These can also be instrument cases, and it will be a bit solder direct to the PCB). downloaded as PDFs and a PNG from harder to fit the board in. siliconchip.com.au/Shop/11/6416 You will also need to mark and drill Mount the board in the case using two 5mm holes in the lid for the LEDs You could use a different case as machine screws and tapped spacers to protrude through. Depending on their lead lengths, you might have a little bit of flexibility in where those LEDs are placed as you can bend the leads slightly. Keep in mind that if you are applying the lid panel label, it will have to line up with those holes. Now is a good time to adhere the front and/or lid panel labels (see below for hints on making them) and cut out the holes using a sharp hobby knife. Verify that S1 and VR1 are wired up correctly, mount them on the front panel and then plug them into headers CON3 & CON4. You can then ‘button up’ the board inside the case, power it up and check that it’s operational. To do that, you will need to plug it into a computer running Windows, download and install the software described in the following section, Fig.12: the positions of the holes to drill/cut in the front panel. The volume control and on/off switch are panel-mounted, so they could be moved, but these positions are designed to clear other nearby parts. You can use this for cases other than the recommended one, but you’ll need at least one reference point to position it correctly. siliconchip.com.au Australia's electronics magazine June 2022 35 Copyright © 2022 SILICON CHIP Publications. Downloaded by Mike Blake (#19283)
Fig.13: the MIDI Synthesiser was combined with a standard MIDI controller and verify that it can connect to the keyboard, amplifier and speaker to form this electronic clavichord. Spectral Sound Synthesiser. Programming the microcontrollers Making the labels This project uses eight microcontrollers of three different types. They are all I’ve found that just printing with an Microchip products (one PIC and seven dsPICs), so they can be programmed inkjet printer then spraying over with with a PICkit 3, PICkit 4, Snap programmer or similar. Or you can build it from art ‘fixing’ spray works well. our kit, which will come with all the micros pre-programmed. For the lid label, I used spe- Each different type of micro has its own software. In other words, there are cial decal sheets for my inkjet three sets of firmware. The codes are given in the parts list, and the download printer from eBay (www.ebay.com. package on our website includes the source code for all three, plus the three au/itm/181840164873), which works. HEX files you need to program them. They have a paper backing. The pro- cess is: If you want to rebuild the source code to produce new HEX files (eg, you want to make changes to the way it functions), you’ll need the Microchip XC16 Pro 1. Print just like you would to any compiler (which can be ordered from the Microchip website; there are also free sheet of paper (but with the shiny trial versions). Otherwise, optimisation level 3 will not be available, and the result- side up). ing firmware will not be fast enough to work correctly. 2. Spray with varnish/lacquer/fix- The PIC18 code is less critical, so you can probably get away with using the ative. free XC8 compiler to build that HEX file. 3. Submerse in water for 30 seconds 36 Silicon Chip Australia's electronics magazine to a minute. 4. Gently slide the decal off (the very thin decal detaches from the paper backing when wet) and onto the case. 5. You might want to varnish over the dried decal again, for added protection. The ‘Patch Editor’ software The module has an associated, pow- erful Windows program called the “Patch Editor”, written in C# Win- forms. A screengrab of this software is shown in Screens 1 & 2. This is a ‘Click-Once’ .NET appli- cation that I am hosting online at https://collectany.blob.core.windows. net/ssm/SpectralSoundModuleApp1/ setup.exe This is in Microsoft Azure ‘blob’ storage, which means that the user is notified of version enhancements if installed from this online location. A comprehensive user guide for this software is available. The app includes tools to help shape the timbre ‘landscape’, the envelopes, the filters and more. It includes ‘visu- alisers’ to view the timbres both in the time or frequency domain, and even includes a harmonic analyser, where you can grab the harmonic content from audio! The app also has its own unique programming language called ‘Spec- tral Definition Language’ (SDL) [not to be confused with Simple DirectMedia Layer – Editor]. You can write SDL code to finely tune the patch defini- tions and easily reuse chunks of code. The idea is to ‘abstract’ sound design to a higher level, simplifying all the complexities of detailed configuration. siliconchip.com.au Copyright © 2022 SILICON CHIP Publications. Downloaded by Mike Blake (#19283)
To this end, you can store your own Screens 1 & 2: sample screenshots from the powerful Windows-based Patch code snippets and execute them as Editor software designed to interface with the Spectral Sound Synthesiser. Its necessary via the app menu – a pow- source code is included in the download package. erful concept, with ‘out of the box’ default examples for setting things like fundamentals about the precision Our brains work on impression and a ‘Hammered String’ envelope! needed for harmonic levels. Since recognising overall characteristics, so humans perceive sound logarithmi- maybe there’s potential in focusing on Final thoughts cally, adequate level scaling might techniques that make huge computa- result from simple bit shifting. Can we tional savings by disregarding things This project has been a very intense really tell the difference in harmonic that just do not matter to perception. It but rewarding journey, often feeling levels to such a degree that it justifies seems like there is still much to think like it is ‘shooting for the moon’. It anything better? about regarding sound synthesis! SC shows that sound synthesis is still fertile soil for experimentation and Moreover, we need to think more invention. Fig.13 shows my DIY about how our brains perceive sound, ‘Electronic Clavichord’ containing a and less about the purity of math- standard MIDI controller keyboard ematical calculation. coupled with this module and a tiny amplifier and speaker. The finished Synthesiser One tantalising idea for the future has two LEDs could be to approach the problem of on the top of a timbral-based system from a more the panel to holistic angle. Rather than each played indicate when note having its own wavetable, think of it is receiving the required harmonics from all played MIDI messages notes as one giant pool of oscillators. and when it is We could then use the phenomenon communicating of ‘psychoacoustic masking’ to signifi- with a computer (eg, cantly ‘prune’ the actual harmonics loading patches). that require calculation. June 2022 37 This would require the ability to pri- oritise the harmonic importance and ignore the ones of least significance. An interesting aspect of this approach would be that the threshold could be based on system performance, always processing the maximum number of harmonics possible but degrading the sound quality in a controlled way if needed. This approach might also be able to deviate from the integer-based har- monic requirement, offering more realism. Another idea is to return to a more sample-based approach, but instead of storing samples in the conventional PCM way, keep them as timbres or even as timbre changes. This might provide significant savings. Other ideas start questioning Useful Links The biological bases of musical timbre perception: siliconchip. com.au/link/abdc Synthesising plucked strings: siliconchip.com.au/link/abdd Synthesising wind instruments: siliconchip.com.au/link/abde Sound quality or timbre: siliconchip.com.au/link/abdf Details on timbre: www.dspguide. com/ch22/2.htm siliconchip.com.au Copyright © 2022 SILICON CHIP Publications. Downloaded by Mike Blake (#19283)
Review by Allan Linton-Smith This handy little radar speed detector has enough sensitivity to detect the speed of tennis, cricket, baseball, softballs and footballs. It’s also useful for checking the speed of your golf swing, running or even your car. Radar Coach how fast can you run, bowl, serve, kick or drive? T he Radar Coach is available from like big 7-segment LEDs but, as men- this as the minimum acti- Tennis Warehouse Australia tioned, are made of standard 5mm vation speed. Usain Bolt for $249, including GST and LEDs. The result is pretty effective. tops out at 44.72km/h, so delivery (www.tenniswarehouse.com. The display is easily visible in sun- that’s a pretty safe assumption. au/radar-coach-speed-gun.html). It light and flashes the speed for a few comes with a small tripod and carry seconds. It gives you the option to The circuitry case and is specifically marketed display (and possibly also announce) toward tennis players, to help them the measured speed in km/h or mph. From the outside, I could not see improve their serving. But as men- an opening for the radar transmit/ tioned above, it will work well for all One possibly helpful application for receive pads or antennae, so I opened sorts of applications. the audible speed readings is for race it up to have a look. Inside, I found marshals, who can listen for vehicle a radar module labelled “MC420S-G We don’t usually review this kind speeds as they enter the pits without 10.525 GHz”, which looks a lot like of product, but we were surprised by taking their eyes off the track. the MDU2750 from Microwave Solu- how well it worked and thought some tions that I am familiar with. of our readers would be interested in it. The manufacturer recommends that the device be put behind a wire fence Unfortunately, I couldn’t find any The Radar Coach has a large display to prevent damage from fastballs, data on the MC420S-G, but it could made of 5mm LEDs behind a translu- because it is only housed in a plastic just be a re-badged MDU2750. cent housing. The little holes at the case that could crack if it’s hit hard, bottom are for the loudspeaker, which especially by a cricket ball. As fencing These operate from a 5V DC supply, can be set to announce the speed (in an may obscure the visual readout, the transmit a chopped 10dBm (10mW) American accent) in case the display voice message is again helpful. signal at 10.525GHz and are accurate is obscured, and it is remarkably loud. to within 0.03%. They use a dielec- It can record the last ten readings. tric resonator stabilised FET oscillator, It can also discriminate between the I like that it can ignore your running which provides stable operation over a ball speed and the running speed of speed (such as running up to the broad temperature range in either CW the bowler or pitcher. cricket pitch) and only detect the ball or low duty cycle pulsed mode, and speed. It does this by assuming you a balanced mixer for good sensitivity The display is 165 x 120mm and the can’t run more than 45km/h and sets and reliability. numerals are 60mm high. These look Features & Specifications ∎ Accurate readings of speeds up to 199km/h or 150mph ∎ Easy-to-read numbers ∎ Voice reading can be turned on or off ∎ Portable with free-standing or hand-held use ∎ Ideal for tracking tennis serve and ground-stroke speed ∎ Can measure ball speed or swing speed ∎ Set up on the ground behind a net (for protection) on in-built legs, or use your own tripod for more height ∎ Includes carry case ∎ Powered by 5 AA cells (not included) ∎ Record button repeats the last 10 readings 38 Silicon Chip Australia's electronics magazine siliconchip.com.au Copyright © 2022 SILICON CHIP Publications. Downloaded by Mike Blake (#19283)
They are a type of Doppler radar that The internals of the Radar Coach. The 10.525GHz radar module is at the top detects the frequency shift between the and has a four-patch antenna arrangement. It runs from 5V DC and transmits transmitted and reflected signal from at 10mW, sufficient to penetrate the plastic housing. Note the small 8W a moving object. The mixer produces loudspeaker at the bottom for announcing the speed. a low-level signal which contains a signal at the difference frequency range to be around 6m. It also depends doors, plus a very cost-effective pro- between the transmitted and received on the size and reflectivity of the target. cessor, display and audio amplifica- signals – see Fig.1. The frequency of tion design. this signal indicates the speed of the Note that it will be inaccurate if an The tripod mounting system is pre- moving body. object approaches the detector at an ferred over a radar gun arrangement angle. Imagine a ball going diagonal because it does not require a coach This low-frequency signal is filtered to the radar beam; the detector will to point, measure and call out each out, amplified and processed to pro- only calculate the approach speed, speed reading. It really does replace vide an audible and/or visual speed. which is considerably lower than the a coach in that sense (but won’t give actual speed; 30% low for a 45° angle. you any tips!). The 10.525GHz radar module So balls should be aimed directly at Overall, this device is fun to use draws 60mA from a 5V supply and the device without actually hitting it and is much cheaper than any direct produces a 700Hz output signal for (hence the recommendation to put it equivalents I can find. If you’re hav- an object approaching at 36km/h. behind a fence). ing trouble getting accurate readings, The MDU2750 has a frontal range of keep in mind that it’s critical that you approximately 50 metres and a rear Conclusion set it up properly. range of only 2-3 metres. This unit As for its build quality, I gave it to functions from -30°C to +70°C, but per- The Radar Coach is considerably my grandson and he has dropped it formance may be degraded above 55°C. less expensive than other similar several times, hit it many times with devices we’ve seen, which often cost balls and it still works a treat. He is a In the Radar Coach, the radar trans- several thousand dollars. Despite this, fast bowler and It has helped him enor- mitter is behind a plastic panel, sepa- it does not compromise accuracy. mously with his wicket-taking. He is rated by a 2mm-thick piece of rubber. It seems to achieve this by using a the fastest on his team now! This presumably attenuates the signal low-cost mass-produced radar mod- SC somewhat, and we estimate the usable ule originally designed for automatic The Radar Coach is ideally located Fig.1: the basic arrangement of behind a wire fence to protect it from the Doppler radar module. The being hit by the ball. Its large display oscillator generates a very high- uses 5mm LEDs behind a translucent frequency RF signal that’s sent housing which gives a readable out via the transmitting antenna. display in full sunlight. The little holes The frequency-shifted received at the bottom are for the loudspeaker. signal is mixed with the original siliconchip.com.au signal to produce a difference signal that becomes the IF (intermediate frequency) output. This is filtered, amplified and processed to generate a speed reading. Australia's electronics magazine June 2022 39 Copyright © 2022 SILICON CHIP Publications. Downloaded by Mike Blake (#19283)
High-Power F or under $20, you can buy some impressive LED panels from Ali- BLEuDckD-rBiovoesrt Express (eg, www.aliexpress. com/item/4001275542304.html). They Since we saw some ridiculously bright, low-cost measure about 22cm by 11cm with an LED panels for sale, we’ve been trying to figure active area of 20cm by 10cm. They’re out the best way to drive them. This Driver is the also available from other online sellers result; it is very flexible and useful for many other such as on eBay or Banggood. purposes, such as charging batteries from a DC source or converting between 12V DC and 24V DC. The panels are based on an alumin- ium PCB and have a silicone gel coat- By Tim Blythman Background Source: https:// ing over the LED array. They are spec- unsplash.com/photos/k4KZVfAXvSg ified as drawing 70W at 12V DC, and they simply expose two solder pads Features & Specifications for the power source. ∎ Switch-mode buck-boost current/voltage driver module There are several other modules ∎ Suitable for driving a variety of 12V LED panels with different sizes and power rat- ∎ Adjustable current and voltage settings using trimpots ings, although we haven’t tested any ∎ Alternative fixed voltage/current settings with fixed resistors of those alternatives. ∎ Lower-cost 5A option by omitting some parts ∎ Input voltage range: 11.3V-35V Having received some samples of ∎ Output voltage range: 7-34V these LED panels, we ran some tests ∎ Maximum output current: 8A using our 45V Linear Bench Supply ∎ Maximum input current: 10A (October-December 2019; siliconchip. ∎ Other uses include charging a 12V battery from another 12V battery com.au/Series/339) and produced the current/voltage curve seen in Fig.1. or other DC source This is consistent with four groups of LEDs arranged in series, each with ➿ ➿∎ Can also be used as a 12 24V DC or 24 12V DC converter a voltage drop of around 3V, giving a forward voltage of about 12V. 40 Silicon Chip Australia's electronics magazine Running the panel at 50W (close to 4A) for a while, it got pretty hot and was way too bright to look at directly. So we expect that these panels can be run at lower power levels than that and still be very useful. Running them cooler should also extend their working life. When supplied with a small amount of current, the individual LEDs can be seen, and there are 336 of them, arranged in 28 rows of 12 (see the photo at the end of the article). Each group of LEDs connected in parallel corresponds to seven rows. YouTuber Big Clive ran some tests on similar modules, and even tore back the gel coating to see what lies beneath. You can see his video at https://youtu. be/uIspnsBp3o4 He found that each group of LEDs is simply wired in parallel, meaning that the panel is mostly unaffected if one LED fails open-circuit. A short-circuit failure would tend to shunt the entire panel current through a single LED, quickly turning it into an open circuit! It also appears that the LEDs are actually blue, and the gel is a phos- phor coating. It’s an interesting con- struction that is quite robust, but sim- ple and clearly cheap to manufacture. As LEDs are often touted as being around eight times more efficient (in terms of lumens per watt) than siliconchip.com.au Copyright © 2022 SILICON CHIP Publications. Downloaded by Mike Blake (#19283)
incandescent globes, 70W of LED light The LM5118 handles the transition Fig.1: like any is equivalent to several hundred watts from boost to buck mode by using a semiconductor diode, the of incandescent light; easily enough to hybrid mode that is somewhere in current through these LED illuminate a large room very brightly. between at intermediate voltages, panels changes sharply with ensuring that the output remains sta- changes in voltage. As such, Limitations ble at all times. it’s not practical to regulate the panel brightness by It’s evident from the current/voltage It does provide current limiting, but controlling the voltage. We curve that applying much more than only to protect the inductor that is must instead control the 13V will put the panel over its nomi- used to store energy during the boost current, one of the features nal 70W limit. So directly connecting and buck phases. So we needed to of the LED Driver PCB. a 12V battery, which could supply as add some parts to the design to pro- when VIN is around 11.3V. This way, much as 14V or higher, is not a feasi- vide independent, adjustable output if a battery is used to feed the circuit, ble way to drive these panels. current limiting. it will be prevented from discharging below 11.3V, a fairly conservative level A 12V battery that’s nearly flat might Circuit details for most lead-acid batteries. only produce around 11.5V, so a resis- The 15kW resistor between pin 3 of tive voltage dropper is not suitable for Fig.2 shows the circuit that we have IC1 and ground sets the boost/buck powering these panels over a battery’s designed incorporating all these fea- oscillator frequency to around 400kHz, useful charge range. tures. Parts of it look similar to the which gives decent efficiency and low Hybrid Bench Supply because of the voltage ripple at the output. We also expect the current/voltage common external parts needed for the IC1’s pin 4 (EN) is pulled to ground curve to change depending on the LM5118 to operate. by a 100kW resistor, but can be pulled panel temperature. That will change up to VIN by shorting the pins of JP1. during operation as the panel self- Power comes in through a two-way Thus, JP1 can be closed with a jumper heats due to its own dissipation. barrier terminal, CON1, with the pos- to provide ‘always on’ operation, or itive supply passing through 10A Like most LEDs or LED arrays, a fuse F1. The 10A limit was chosen current-controlled or current-limited as a convenient level above the 7A supply is the best choice for driving limit of the LED panel. this one. While the voltage may drift slightly under constant current condi- A bank of paralleled ceramic tions, it’s a much more stable arrange- 10μF capacitors provides bulk ment. Thus our Driver incorporates supply bypassing to the power current-control circuitry. section of the circuit, while a 100nF capacitor is placed close The LED Driver to IC1, the LM5118, to stabilise its supply. Given that a common use case would be running these LED panels The VIN supply feeds into pin from a 12V battery or DC supply, we 1 of IC1 with grounds at pins 6 need a few specific features. The LED and 14. An 82kW/10kW divider panel operating point might be above across this supply to IC1’s pin 2 or below the battery voltage, so we UVLO (under-voltage lock-out) need to be able to increase or decrease exceeds its threshold of 1.23V the incoming supply. And to provide a consistent level of lighting, we also When the panel is off, you can need to regulate the output current. just make out the numerous For efficiency, we need to use a small LED chips that provide switchmode circuit. For this to both the light output under increase and decrease the voltage, it needs to be able to either buck (reduce) the phosphor gel coating or boost (increase) the incoming volt- (although they are a bit hard age. to see in this photo). Some circuits do this by having two separate stages; for example, first by decreasing the input voltage as needed and then using a second stage to boost the output from the first stage. The design of such circuits can be com- plex; more so when current limiting or regulation is needed. But chips exist that can work in boost or buck mode as needed. That includes the LM5118, a device we used in the Hybrid Bench Supply from April-June 2014 (siliconchip.com.au/ Series/241). siliconchip.com.au Copyright © 2022 SILICON CHIP Publications. Downloaded by Mike Blake (#19283)
connected to an external low-current IC1’s pins 12 and 13 connect across L2 and diodes D1-D4 are arranged in switch to give a simple on/off control. a pair of current measuring shunts to a bridge-like configuration that can be monitor the current through D3 and driven in either boost or buck switch- The capacitors on pins 5 and 7 D4, thus limiting the current through ing modes. Fig.3 shows how such a (RAMP and SS) set the ramp and soft- L1 and L2. This works whether the cir- bridge can work in both modes. start characteristics of IC1 to be suit- cuit is operating in boost or buck mode. able for our application. The circuit works as a buck switcher Pins 19 (HO) and 15 (LO) drive the for low output voltages (compared to IC1’s pin 8 FB (feedback) input is external high-side (Q1) and low-side the input voltage). When Q1 is on, cur- used to set the output voltage. The (Q2) Mosfets, respectively. Pin 16 is rent flows through L1 and L2 and then divider formed by potentiometer VR1 connected to an internal regulator that D1 and D2 towards the load. When Q1 and its two series ‘padder’ resistors provides around 7V with an external switches off, the current continues to feeds that pin with a fraction of the 1μF capacitor to stabilise this. circulate via D3 and D4. output voltage that is compared with a 1.23V reference within IC1. The 7V supply is used to drive the Above 75% duty cycle on Q1, IC1 Mosfet gates and is a good compro- operates in the hybrid boost-buck This adjustment gives a nominal mise between turning them on fully mode. Q2 starts to switch on with a output range between 6.8V and 34.7V. while maintaining fast switching. It duty cycle that overlaps with Q1’s The 34.7V upper limit is chosen to also powers shunt monitor IC2 which on-time. This increases the current stay well clear of the 60V Mosfet Vds we’ll get to shortly. through the inductors during the limit for Q2 while maintaining a useful on-time, and this extra energy gets range for 24V systems. The 1kW resis- Pins 18 (HB) and 20 (HS) are con- fed to the output during the Mosfet tor between the divider and the FB pin nected to either end of a 100nF capac- off-time, increasing the output voltage. reduces the interaction between the itor, which is charged and then used voltage control and current limiting, to drive the HO pin above the supply A simple implementation of the which we will explain shortly. voltage. This ‘floating’ gate supply boost mode would have Q1 on all the is needed to switch on the high-side time boost mode is active, but this is The 2.2nF capacitor, 4.7nF capac- N-channel Mosfet as its source termi- not possible with the LM5118, so it itor and 10kW resistor between pins nal can be at or near the supply volt- is switched on and off in synchrony 8 and 9 are a compensation network age when it is switched on. with Q2. that forms part of the feedback loop that controls IC1’s duty cycle. Mosfets Q1 & Q2, inductors L1 & This is necessary because the Fig.2: the circuit is based around IC1, an LM5118 buck-boost controller. It drives the H-bridge made from Mosfets Q1 & Q2, diodes D1-D4 and inductors L1 & L2. These allow it to step down the incoming voltage (by pulsing Q1 on) or step it up (by pulsing Q1 & Q2 on simultaneously). Varying the duty cycle/on-time allows it to change the output-to-input voltage ratio. We’ve added IC2 and some other components to provide an adjustable current limit. 42 Silicon Chip Australia's electronics magazine siliconchip.com.au Copyright © 2022 SILICON CHIP Publications. Downloaded by Mike Blake (#19283)
bootstrap capacitor needs to be peri- Fig.3: an illustration of how the LM5118 works, in buck mode (diagrams at left) odically refreshed to maintain the gate and boost mode (at right). The mode of operation is determined by whether S2 voltage, which can only happen while (actually a Mosfet) is switched with S1 or just left open (ie, off). In buck mode, Q1 is off. as the duty cycle approaches 100%, the output voltage approaches the input voltage while in buck/boost mode, a 50% duty cycle gives an output voltage equal All this is done transparently by the to the input with higher duty cycles boosting the output voltage above the input, controller inside the LM5118. approximately doubling it at 75% duty, quadrupling it at 87.5% and so on. Current limiting The LED Driver is designed to mount directly to the 70W LED panels, with just two flying leads between the two. As it has many other potential uses, you can The voltage at the cathodes of D1 mount it in just about any kind of box using tapped spacers. and D2 is smoothed by a bank of five 10μF capacitors accompanied by a Australia's electronics magazine June 2022 43 100nF capacitor. From there, it passes through another 15mW current sensing shunt, then through fuse F2 to output connector CON2. We can keep the grounds common between the input and output by plac- ing the current shunt in series with the positive output. This has several advantages, one of which is that you don’t need to have the ground cur- rent pass through this module; it can go straight from the load to the power source, possibly simplifying the wir- ing and reducing wire-related power loss. The voltage across the shunt is mea- sured by IC2’s pins 1 and 8 and ampli- fied with a gain of 50. IC2 is an INA282 current shunt monitor, and it takes its supply on pin 6 from IC1’s internal 7V regulator. It also has its own 100nF supply bypass capacitor. IC2’s pins 3 and 7 are both con- nected to ground, so the output volt- age from pin 5 is relative to ground. The voltage at pin 5 is divided and smoothed by the network consisting of the 100W resistor, 5kW trimpot VR2, 1kW resistor and 10μF capacitor. The smoothing is necessary to elim- inate instability which would cause LED flickering due to oscillations in the output voltage. The resulting voltage is fed into IC1’s FB pin via schottky diode D5. Thus, as the output current increases beyond a certain threshold, the voltage at the FB pin increases similarly to the situation where the output voltage is too high. IC1 attempts to control this by reducing its output voltage, thus reducing the current. The diode ensures that an output current below the limit does not drag down the reference. If the target cur- rent is not met, the control loop is based only on the output voltage. The result is not a ‘brick wall’ cur- rent limit; it allows higher currents at lower output voltages. This is because a higher voltage is needed at D5 to maintain balance at the FB pin as the siliconchip.com.au Copyright © 2022 SILICON CHIP Publications. Downloaded by Mike Blake (#19283)
output voltage drops further below Pairs of parts shown in the application notes for its setpoint. the LM5118, and are useful in certain You might notice from the sche- situations. The 1kW resistor between VR1 and matic that a few parts are duplicated the FB pin helps maintain this balance and paralleled. These include L1 & L2, We were initially unsure whether and limit the extent to which the two D1-D4 and the 15mW current shunts these parts were needed for stable parts of the circuit interact. connected to D3 & D4. The circuit has operation, but it turned out they been designed with these extra parts were not. Some enthusiastic readers With VR2 at its minimum setting, to handle up to 8A, by splitting the might be tempted to experiment with an output current of 1.8A will induce current between the pairs of compo- the design and use these component 27mV across the shunt or 1.35V at pin nents and thus moderating the heating locations, as shown in the LM5118 5, which corresponds to 1.23V at the of any single part. data sheet. divider output, meaning that this is the point that current limiting begins. For operation up to 5A, L2, D2, D4 The optional parts include an RC With VR2 set higher, a smaller frac- and one of the shunts can be omitted. snubber for the switching node and tion of the pin 5 voltage is sampled, The input and output fuses should also components to disable IC1’s internal and thus a higher output current is be changed to suit 5A operation. All regulator if the input supply voltage allowed. other components can work happily will always be within a suitable range up to the 8A limit. (about 5-15V). In practice, since IC2’s supply is around 7V, the maximum current set- While the shunt resistors do not Since the LM5118 can operate ting is around 8A. So setting VR2 above dissipate any significant amount of up to 76V (with some parts changes around 3/4 of its travel will effectively power, they are used by IC1 to moni- needed in our design to achieve that), disable the current limiting. tor the current through the inductors. this board would have many poten- Whether one inductor and one shunt tial applications. Some configura- Lower output current settings can or two inductors and two shunts are tions may not be as stable as the one be achieved by increasing the shunt present, the current limit through each presented here, so figuring out what resistance, although that would argu- inductor is the same. components are needed in different ably be a poor use of a circuit capa- use cases is left as an exercise for the ble of 8A. Extra parts reader. That the current limit tapers off is There are a few component loca- Options actually an advantage as it tends to put tions that are usually left empty. These the system closer to constant-power are shown in red on the circuit and R13, adjacent to VR2, is a different operation. For the LED panels, the PCB overlay diagram. We’ve incorpo- case. This fixed resistor is intended to operating voltage range will be quite rated these in the design as they are replace VR2 for a fixed setpoint. Alter- narrow in any case. natively, you can replace either VR1 or VR2 with a fixed resistor between Table 1: resistor values for fixed output voltages their two leftmost terminals, as they are simply wired as variable resistors Target voltage Calculated E24 resistor value Resulting voltage (rheostats). resistance Table 1 shows typical resistor val- 8V 210W 220W 8.05V ues for fixed output voltages, including the exact and nearest E24 series val- 10V 568W 560W 9.95V ues. The values are linear across the range, so you can interpolate them to 12V 926W 910W 11.91V find intermediate values if necessary. 14V 1284W 1300W 14.09V Table 2 does the same for current, with the listed values being at the 15V 1462W 1500W 15.21V point that current limiting first kicks in. Similarly, exact and nearby E24 20V 2357W 2400W 20.24V series values are given, and the cor- relation is relatively linear. 24V 3072W 3000W 23.59V Battery charging 28V 3788W 3900W 28.63V Although we have not done thor- 30V 4145W 4300W 30.86V ough testing with this configuration, the Driver is well-suited for charging Table 2: resistor values for fixed output currents a 12V battery from another 12V bat- tery. This might seem like an unusual Target current Calculated E24 resistor value Resulting current requirement, but it often crops up in resistance situations involving a caravan or sim- ilar that has a ‘house’ battery, usually 2A 119W 120W 1.98A a deep-cycle type. 3A 729W 680W 2.92A Such a battery is typically charged from the 12V system of a towing vehi- 4A 1339W 1300W 3.93A cle while the vehicle is charging its 5A 1949W 2000W 5.08A siliconchip.com.au 6A 2558W 2700W 6.23A 7A 3168W 3000W 6.72A 8A 3778W 3600W 7.71A 44 Silicon Chip Australia's electronics magazine Copyright © 2022 SILICON CHIP Publications. Downloaded by Mike Blake (#19283)
starter battery. Due to voltage drops Fig.4: most of the over long cables and the tendency of components on the modern vehicles not to fully charge board are SMDs, their starter battery, there may not be but only IC1 has enough volts available to fully charge closely-spaced such a house battery via a direct con- leads. Having said nection. that, some of the other components The Driver can overcome this and can be somewhat comfortably deal with batteries in all challenging simply charge states due to the current limit- due to the combined ing feature. The Driver is set to provide thermal mass of a voltage that suits the desired house those parts and the battery’s fully charged level, with the PCB copper. Most current limit set to a safe level for the components are not batteries and wiring. polarised or only fit one way; it’s A diode or VSR (voltage sensitive mainly the ICs and relay) on the Driver’s output may be trimpots that you necessary to prevent the house bat- have to be careful tery from draining through the Driv- orientating. er’s voltage sense divider. The Driver should be located close to the house Parts List – Buck-Boost LED Driver battery so that cable resistance does not affect sensing the house battery 1 double-sided PCB coded 16103221, 85mm x 80mm voltage. 2 2-way 10A barrier terminals, (CON1, CON2) [Altronics P2101] 1 2-way pin header, 2.54mm pitch, with jumper shunt (JP1) Construction 2 10A 10μH SMD inductors, 14 x 14mm (L1, L2) [SCIHP1367-100M] 4 M205 fuse clips (F1, F2) The LED Driver is built on a double- 2 10A M205 fast-blow fuses (F1, F2) sided PCB coded 16103221 that mea- 6 M3 x 10mm tapped spacers (to mount to LED panel) sures 85mm x 80mm. Fig.4 shows 10 M3 x 6mm panhead machine screws (to mount to LED panel) where all the parts go on the board. 2 5kW 25-turn vertical top-adjust trimpots (VR1, VR2) [Jaycar RT4648 or This design uses almost exclusively Altronics R2380A] surface-mounted parts of varying sizes, so you will need the usual set Semiconductors of surface mount gear. 1 LM5118MH buck-boost regulator, SSOP-20 (IC1) 1 INA282AIDR current shunt monitor, SOIC-8 (IC2) A temperature-adjustable iron will 4 SBRT15U50SP5 schottky diodes, POWERDI5 package (D1-D4) help greatly in dealing with the wide 2 PSMN4R0-60YS or BUK9Y4R8-60E N-channel Mosfets, LFPAK56/SOT669 range of part sizes that are used. Sev- eral of the components connect to solid (Q1, Q2) copper pours (for current and thermal 1 BAT54, BAT54S or BAT54C schottky diode, SOT-23 (D5) handling) and will likely require the iron to be turned up to a higher tem- Capacitors (SMD M3216/1206-size SMD X7R ceramics, 35V or higher rating) perature to make the joints. 16 10μF 1 1μF Tweezers, flux, solder wicking braid, 6 100nF magnifying lenses and fume extraction 1 4.7nF are all important requirements for 1 2.2nF assembly. Also, since you’ll need to 1 330pF keep the iron’s tip clean, have a tip cleaner on hand. Resistors (all SMD M3216/1206-size 1/8W 1% except as noted) 1 100kW Begin construction with the two ICs. 1 82kW IC1 has the finest-pitch leads, so start 1 15kW with it. Apply flux to its pads, then 2 10kW align the part with the pin 1 marker 3 1kW and tack one lead in place. 1 220W 1 100W Use a magnifier to confirm that the 3 15mW 3W M6332/2512 part is aligned with the pads and flat against the PCB, then tack the diago- A complete kit (Cat SC6292; siliconchip.com.au/Shop/20/6292) is available nally opposite lead and re-check its for $80. It includes everything in the parts list above. We can supply the LED position. panels, cool white (~6000K, SC6307) or warm white (~3000K, SC6308) for $19.50 each. Solder the remaining leads one at a time, or by gently dragging the iron tip Australia's electronics magazine June 2022 45 loaded with solder along the edges of the pins. These techniques depend on loading a small amount of solder onto siliconchip.com.au Copyright © 2022 SILICON CHIP Publications. Downloaded by Mike Blake (#19283)
the iron’s tip. Practice is the only way larger single lead is towards CON2. they are reversed, they will not oper- to get this right. The pad arrangement on the PCB ate correctly. should make this clear. Once finished, carefully inspect the Ensure that they are both wound leads for solder bridges. If you see any, Solder these like the passives, but to their minimums by turning their add some extra flux paste and then take extra care that the part is aligned adjustment screws anti-clockwise by use solder wick to gently remove the correctly so that the large tab that runs 25 turns or until you hear a clicking excess solder. under the part does not short onto the indicating that they have reached the smaller pads. end of their travel. Finally, clean away the flux residue with a flux cleaner (or pure alcohol While the packages used for Mosfets There are seven test points on the if you don’t have one) and a lint-free Q1 and Q2 may look unusual, they are board, but you do not need to fit PC cloth, then check again with a magni- actually much the same as an 8-pin pins; you can simply probe them with fier to ensure all the pins are correctly SOIC package IC, but with the leads a standard set of DMM test leads. soldered, and no bridges are left. along one side joined into one larger tab. This improves heat removal, low- Testing Use a similar technique to fit IC2 ers resistance and also makes correctly to the board. Then mount the smaller orientating them easier. You will need fuses installed for passive SMDs (except for the shunt testing, but since initial testing is done resistors) using a similar approach; Take care that the leads are aligned with a multimeter, you can fit lower- their larger pads are a bit more for- within their pads. The only real dif- rated (eg, 1A) fuses if you have them giving. Remember that some of these ference in soldering these compared on hand. If you have a current-limited parts are not needed (they’re labelled to SOIC-8 parts is due to the greater PSU, you can use that too. in red in Fig.4). thermal mass of the large metal tab and the copper areas on the PCB. Connect a voltmeter across CON2 The main trick here is to avoid and apply a power source of around touching the iron to one side of the Moving on to inductors L1 and L2, 12V DC (above 11.5V) to CON1. You part until you are sure the solder on the thermal effect will be even more should see about 6.6-7.0V at CON2. If the other side has solidified, or it might apparent here. They are not polarised, you get a reading near the supply volt- shift out of place. but you will need a good amount of age instead, you could have a short cir- heat to complete the soldering. cuit somewhere. In that case, switch The SMD capacitors are unmarked, off and check the PCB for faults before so be careful not to mix them up. It’s It’s best to lay down some flux paste proceeding. best to unpack and fit all the capaci- on one pad, add some solder to the tors of one value at a time. As some of other pad, slide and/or press down Slowly turn VR1’s screw clockwise. the capacitors (particularly the 10μF the part into place while heating that After the trimpot’s mechanism re- parts) are across ground planes, you solder, then add solder to the oppo- engages, you should see the voltage on might need to turn your iron up to site pad. Finally, refresh the first pad CON2 increase, rising to nearly 35V at make good joints. Ensure the solder you soldered. its maximum setting. If so, wind it back flows both onto the end of the part and down around 11V. If you can’t adjust onto the PCB pad below. Check that the solder fillets are the output voltage correctly, switch joined to both the inductor and PCB off and check for faults. The solitary SOT-23 part, D5, is pads before proceeding further. a BAT54 schottky diode. With one If you have used low-value fuses, lead on one side and two on the Now clean the PCB of excess flux change these now to your nomi- other, its orientation should be obvi- and thoroughly inspect all the parts nal value; for the LED panels we ous. Just make sure its leads are flat for bridges and dry joints; they will described earlier, 10A each is a good on the board, not sticking up in the be easier to see and fix after cleaning. choice. air, which would indicate that it’s upside-down. There are only a handful of through- You can also test that the current hole parts remaining. You can mount limiting works if you have a suitable Note that you can substitute a dual fuse holders F1 and F2 by installing a load such as a power resistor or test BAT54S (series) or BAT54C (common fuse and slotting the whole assembly load (like the one described starting cathode) diode as one of their two into the PCB. This ensures that the on page 48 of this issue). The mini- internal diodes connects between the tabs are aligned correctly and spaced mum current limit when VR2 is set same set of pads. The other diode in far enough apart to allow a fuse to be fully anti-clockwise is around 1.8A. the package will be unconnected and fitted. Like many of the parts, they may unused. need more heat to let the solder take You can easily monitor the out- to the large copper areas. put current at TP5 (near IC2) relative The remaining surface-mounting to TP3 (ground, at top left). This is parts are larger, so you might like to Next, mount the terminals for CON1 the raw output from IC2, and it gives raise your iron temperature before and CON2, ensuring that any con- 0.75V per amp. So 1.5V at TP5 corre- proceeding. Also, they are mostly nected wires can exit from the board sponds to 2A. arranged around the top half of the (most barrier terminals allow wires to PCB. be inserted from either side, but there Also, you can monitor the output are exceptions). JP1 and its jumper can voltage at TP6 (near CON2) relative Solder the three larger 15mW shunt then be installed near CON1. Leave the to ground. resistors, then the four power diodes. jumper in place for testing. The diodes have two small leads on Adjust your load until the current one end and a larger one on the other. Finally, fit the two multi-turn trim- limiting kicks in. Reducing the load In each case, the ends with two small pots, VR1 and VR2, near F2. Make resistance should let the output volt- leads go towards CON1 while the sure their screws are to the left, as age drop while the current stays mostly 46 Silicon Chip shown in the overlay and photos; if constant. Australia's electronics magazine siliconchip.com.au Copyright © 2022 SILICON CHIP Publications. Downloaded by Mike Blake (#19283)
LED panel mounting These panels are incredibly bright, and the two photographs above do not do them justice. They are too bright to look at directly when set to anything but the The Driver is designed to mount lowest setting (at left). on the back of the LED panel using the mounting holes near the power to keep the PCB from moving, flexing easiest way to guarantee they operate terminals, so you can use short flying and shorting against the aluminium at the same current. The main differ- leads to connect from CON2 to the back of the LED panel. See our photos ence is that the voltage needs to be set panel’s inputs. for details of this arrangement. to around 26V. This certainly seems to work fine, While your iron is on, you can Adjust VR2 to provide a suitable but the Driver is likely to be less effi- connect some leads to the LED pan- current and thus brightness. If you get cient in this mode unless the input els. As you will know by now, sol- much above 5A, you might find that voltage is raised to about 24V. dering inductors L1 & L2 to the PCB the current limiting no longer domi- You can change the UVLO thresh- requires much heat, but nowhere near nates, and the VR1 voltage setting may old to suit a 24V battery by chang- as much as is needed for soldering to need to be increased above 13V. ing the 82kW resistor to 160kW, and the aluminium-c ored PCB that forms 10kW resistor to 9.1kW. This will set the LED panel. Keep in mind that both the Driver the threshold to approximately 22.8V. and LED panel will get quite warm As noted in the Features panel, You might even find that you need during use, so they should be mounted you can also use the Driver as a DC- to preheat the panels with a hot air to allow free air circulation. powered battery charger, a 24V to 12V rework tool or similar before you can converter, or a 12V to 24V converter successfully solder those leads. We Suppose you see the LED panel rap- for many different purposes. also suggest that you pre-tin the leads idly flickering during operation. In that For the 24V to 12V arrangement, the and have a generous amount of solder case, the supply voltage is probably output limit can be set up to 8A, with on the iron’s tip (to accumulate some dropping below the UVLO threshold, a 10A fuse at F2, but with F1 reduced thermal mass). causing the Driver to cut out and then to 5A. In this case, you would also switch back on when the input voltage change the 82kW resistor to 180kW. To set up the Driver to work with recovers. Check your supply and that For a 12V to 24V arrangement, F1 LED panels, disconnect all loads, the connections to CON1 do not have should be 10A and F2 should be 5A, set the output voltage to around 13V too much resistance. with an appropriate current limit near and adjust the current limit fully 5A set using VR2. anti-clockwise to 2A. The 13V setting Driving two panels SC is simply a failsafe in case the current limiting stops working. We briefly experimented with run- ning two panels in series, as this is the Keep in mind that the LED panels are very bright; even at 2A, it will Australia's electronics magazine June 2022 47 likely be too bright to look at! We rested them on their edge during test- ing to aim them away from our faces. If you then connect the LED panel and power up the Driver, you should see the output voltage drop to approx- imately 12V as the Driver switches to its current-limited mode. If you don’t see the voltage drop, the current limiting may not be working. In that case, measure the voltage at TP7 neat VR1. This feedback voltage should always be around 1.23V when the Driver is operating correctly. Check that there is a slightly higher voltage at D5’s top right (anode) ter- minal; this means that the diode is feeding current into TP7 and con- trolling the output. If this is more than around 0.3V higher, D5 may be the wrong type or not injecting cur- rent correctly. If all is well, you can then perma- nently wire up CON2 to the LED panel and mount the Driver using tapped spacers. Use four tapped spacers with a screw at each end to mount the Driver PCB to the LED panel at its mount- ing holes. Then use two further tapped spacers mounted to the PCB only as standoffs siliconchip.com.au Copyright © 2022 SILICON CHIP Publications. Downloaded by Mike Blake (#19283)
Arduino Programmable Load Project by Tim Blythman To test devices like power supplies, driver circuits and current sources, you often need a particular or variable load resistance that can handle a bit of power. This Programmable Load is based on an Arduino shield that is easy to understand, build and use. It can be controlled manually or automated in a way that suits your application. D uring the design & testing But being connected to an Arduino or regulator is to see how it responds of our High Power Buck-Boost microcontroller means that it’s pos- to sudden changes in load resistance, LED Driver (starting on page sible to add some smarts. The circuit and it is capable of doing that. 40), we wanted to check how it han- also includes components to allow the dled various loads to test the robust- applied voltage and sunk current to Our sample code provides just the ness and versatility of the design. To do be measured. This means that it can basic features, including manual resis- that, we came up with this design, and calculate the power dissipated in the tance and current tracking modes, but it was so handy that we have turned it Load (P = V × I) too. it’s easy to modify the code to add cus- into a standalone project. tom features. Our sample code also Thus, you can program the Load displays all the data that is collected. Unlike the 50W DC Electronic Load to behave differently depending on (September 2002; siliconchip.com.au/ the application. Its functions include Circuit details Article/4029), the Programmable Load fixed resistance or current tracking is not infinitely adjustable and is not modes. It can even be programmed to The 50W DC Electronic Load from intended to sink a constant current. provide a dynamic load so that you 2002 uses a single Mosfet bolted to Instead, it uses switched resistance can test equipment under changing a large heatsink as the load element. elements that apply discrete load resis- conditions. That requires some careful circuit tance steps. design so that the Load can respond A typical test for a power supply to dynamic conditions. Features & Specifications On the other hand, our Programma- ble Load consists of 15 high-power ∎ Handles up to 70W continuous, at up to 15V and 4.7A resistors which have no trouble deal- ∎ Presents a load resistance between 3.1W and 47W in 15 steps, or 43kW ing with rapidly changing conditions. Crucially, there is no chance of them when ‘off’ presenting a short circuit as long as the ∎ Sinks 255mA to 3.83A in 255mA steps from a perfectly-regulated 12V circuit is operated within its working voltage range. source ∎ Manual control of unit loads or resistance The concept is simple. There are ∎ Software provides an approximately constant-current mode four groups of 5W 47W power resistors. ∎ Measures voltage up to 20V The groups consist of one, two, four ∎ Measures current up to 6.5A and eight resistors respectively, which ∎ Calculates power up to 130W can be switched into any combination from none to 15 resistors in parallel. 48 Silicon Chip Australia's electronics magazine The Load is optimised for use with siliconchip.com.au Copyright © 2022 SILICON CHIP Publications. Downloaded by Mike Blake (#19283)
Search
Read the Text Version
- 1
- 2
- 3
- 4
- 5
- 6
- 7
- 8
- 9
- 10
- 11
- 12
- 13
- 14
- 15
- 16
- 17
- 18
- 19
- 20
- 21
- 22
- 23
- 24
- 25
- 26
- 27
- 28
- 29
- 30
- 31
- 32
- 33
- 34
- 35
- 36
- 37
- 38
- 39
- 40
- 41
- 42
- 43
- 44
- 45
- 46
- 47
- 48
- 49
- 50
- 51
- 52
- 53
- 54
- 55
- 56
- 57
- 58
- 59
- 60
- 61
- 62
- 63
- 64
- 65
- 66
- 67
- 68
- 69
- 70
- 71
- 72
- 73
- 74
- 75
- 76
- 77
- 78
- 79
- 80
- 81
- 82
- 83
- 84
- 85
- 86
- 87
- 88
- 89
- 90
- 91
- 92
- 93
- 94
- 95
- 96
- 97
- 98
- 99
- 100
- 101
- 102
- 103
- 104
- 105
- 106
- 107
- 108
- 109
- 110
- 111
- 112
- 113
- 114
- 115
- 116
