Silicon Chip500W Class-D Mono Amplifier - April 2023 SILICON CHIP
  1. Outer Front Cover
  2. Contents
  3. Publisher's Letter: Renewable energy costs are seriously understated by the media
  4. Feature: Underground Communications by Dr David Maddison
  5. Project: 500W Class-D Mono Amplifier by Phil Prosser
  6. Project: Wideband Fuel Mixture Display Part 1 by John Clarke
  7. Subscriptions
  8. Review: T48 Universal Programmer by Tim Blythman
  9. Project: Automated Test Bench by Richard Palmer
  10. Project: Silicon Chirp – the pet cricket by John Clarke
  11. Feature: Using TestController by Richard Palmer
  12. Serviceman's Log: Tips on kit and bits by Dave Thompson
  13. Product Showcase
  14. Vintage Radio: Browning-Drake 6A by Dennis Jackson
  15. PartShop
  16. Market Centre
  17. Advertising Index
  18. Notes & Errata: Secure Remote Mains Switch, July & August 2022; Capacitor Discharge Welder, March & April 2022; Programmable Hybrid Lab Supply with WiFi, May & June 2021
  19. Outer Back Cover

This is only a preview of the April 2023 issue of Silicon Chip.

You can view 39 of the 112 pages in the full issue, including the advertisments.

For full access, purchase the issue for $10.00 or subscribe for access to the latest issues.

Articles in this series:
  • Underwater Communication (March 2023)
  • Underwater Communication (March 2023)
  • Underground Communications (April 2023)
  • Underground Communications (April 2023)
Items relevant to "500W Class-D Mono Amplifier":
  • 500W Class-D Amplifier cutting diagrams (PDF download) (Panel Artwork, Free)
Items relevant to "Wideband Fuel Mixture Display Part 1":
  • Wideband Fuel Mixture Display PCB [05104231] (AUD $10.00)
  • PIC16F18877-E/PT programmed for the Wideband Fuel Mixture Display (WFMD) [0510423A.HEX] (Programmed Microcontroller, AUD $15.00)
  • Wideband Fuel Mixture Display short-form kit (Component, AUD $120.00)
  • Firmware for the Wideband Fuel Mixture Display (WFMD) [0510423A.HEX] (Software, Free)
  • Wideband Fuel Mixture Display PCB pattern (PDF download) [05104231] (Free)
Articles in this series:
  • Wideband Fuel Mixture Display Part 1 (April 2023)
  • Wideband Fuel Mixture Display Part 1 (April 2023)
  • Wideband Fuel Mixture Display, Pt2 (May 2023)
  • Wideband Fuel Mixture Display, Pt2 (May 2023)
  • Wideband Fuel Mixture Display, Pt3 (June 2023)
  • Wideband Fuel Mixture Display, Pt3 (June 2023)
Items relevant to "Automated Test Bench":
  • Automated Test Bench Swiss Army Knife PCB [04110221] (AUD $10.00)
  • 5V to ±15V boost module (Component, AUD $5.00)
  • ESP32 DevKitC microcontroller module with WiFi and Bluetooth (Component, AUD $25.00)
  • Testbench Swiss Army Knife short form kit (Component, AUD $50.00)
  • Laser-cut UB1 Jiffy box lid for the Test Bench Swiss Army Knife (PCB, AUD $10.00)
  • Firmware and laser cutting files for the Automated Test Bench Swiss Army Knife (Software, Free)
  • Automated Test Bench Swiss Army Knife PCB pattern (PDF download) [04110221] (Free)
Items relevant to "Silicon Chirp – the pet cricket":
  • Silicon Chirp cricket PCB [08101231] (AUD $2.50)
  • PIC16F15214-I/SN programmed for Silicon Chirp [01810123A.hex] (Programmed Microcontroller, AUD $10.00)
  • Silicon Chirp kit (Component, AUD $25.00)
  • Firmware for Silicon Chirp [01810123A.HEX] (Software, Free)
  • Silicon Chirp cricket PCB pattern (PDF download) [08101231] (Free)

Purchase a printed copy of this issue for $11.50.

DESIGN BY PHIL PROSSER 485W into 4Ω (single channel) Operates with loads between 4-8Ω Very high efficiency (typically >80% at moderate power levels) Very low in cost and easy-to-build with minimal soldering required Typically, 0.02% distortion over most power levels at 1kHz Frequency response from <5Hz to 20kHz, +0,-1.5dB Built-in speaker protection 5 MONOBLOCK 0 class-d amplifier 0 If you need a serious amount of audio power, are on a budget, and are not after ‘high fidelity’, this is for you! It uses two prebuilt modules and not much else, mounted in a compact metal chassis, WATT to deliver heaps of audio power all day long. B uilding a 500W+ amplifier is a serious undertaking. To make sense, a high-power Class-D amplifier would need a switch-mode power supply. After all, why bother with a Class-D amplifier if you need a 1kVA transformer and bank of capacitors, making the thing half the weight of a VW Beetle? DANGER – LIVE COMPONENTS Do not consider touching the heatsinks or anything on the PCBs when the amplifier is powered or for several minutes afterwards. Assume that contact will be lethal! Never, ever touch the PCBs if the amplifier is even plugged in. If you want to measure the heatsink temperature, use a noncontact IR thermometer. 26 Silicon Chip So we came up with the idea of using some of the relatively cheap modules available on sites like eBay and AliExpress. There were three questions on our minds: were they safe, would they even work, and would the performance be acceptable? So we started surfing online shops and came across two promising modules (see the adjacent panel). Deciding on the modules All the modules we purchased for evaluation have reasonably good availability and have been on sale for many months. Some things that drove us to choose them were: Table 1 – Measured performance into a 4W W resistive load Voltage Freq (RMS) Load Power THD+N Notes 8V 1kHz 4W 16W 0.026% Warm up test, heatsinks 36°C 20V 1kHz 4W 100W 0.017% Heatsinks 40°C after a few minutes 30V 1kHz 4W 225W 0.019% Heatsinks 49°C after a few minutes 40V 1kHz 4W 400W 0.03% 44V 400Hz 4W 484W Output started clipping 44V 1kHz 484W Output started clipping 4W Australia's electronics magazine siliconchip.com.au ● The power supply modules have decent mains-to-secondary isolation. ● They have decent heatsinking and quality capacitors. ● They are common/available parts sold in a range of voltages, ie, a moderately mature and supported design. ● The prices are neither too cheap to be true nor overly expensive. So we placed orders for one of each to test out (plus the modules listed in the panel overleaf that we didn’t end up using). The cost of each module was in the region of $100. $200 for a power supply and amplifier module is bonkers for this sort of power level. If you have built a 500W amplifier using discrete parts and a linear power supply, you will know that this would barely pay for the transformer, let alone the rest. So is this still too cheap to be true? Our greatest concern with purchasing this sort of equipment online is safety and electrical standards. In choosing these modules, we spent a lot of time downloading photos and trying to see how they were laid out, if there were slots milled between feedback opto-isolators and suchlike. Once we had received them, we inspected them to see if they matched the pictures – they did. We then tested them to the best of our ability using our old-school megger (500V) and found no measurable leakage from primary to secondary on both power supplies during a 60-second test. We are not promoting these power supplies as being compliant with any standard, mind you! But there is visible isolation built into the design and measurable isolation on test, which was enough for us to work with them. The power supplies purchased both claim to be capable of “1000W”, although the smaller of the two owns up to being more like a 500W continuous unit. We think it reasonable to rate both power supplies under 1kW continuous, given the parts used, especially the smaller one. Of the two sets of modules, we chose to proceed with the larger, black modules. We have provided some information on the ‘also-ran’ modules for interest but recommend that you stick with the two shown opposite. Performance Table 1 shows some spot measurements of distortion at various power levels. These agree with the claims siliconchip.com.au #1 Large Class-D Amplifier IRS2092S 1000W Class-D Mono Amplifier (see Photo 1): siliconchip.au/link/abic siliconchip.au/link/abid siliconchip.au/link/abie Claims Speaker protection operating from an independent power supply Supply voltage: ±65V to ±80V Photo 1: this Maximum output power: 1000W 500W+ Class-D amplifier module Efficiency: ≥90% was under $100 and includes a speaker Signal-to-noise ratio (SNR): 90dB protection relay. The control circuitry is Dimensions: 157 × 101 × 44mm mounted on a vertical sub-PCB. Net weight: 0.45kg This has four onboard 1000µF 100V supply bypass capacitors per rail labelled Nichicon HE(M), 18mm in diameter and 42mm high. The Nichicon data sheet we found did not list a 1000µF, 100V cap in this range, 820µF being the largest. The size of this capacitor is consistent with the ratings. The main switching transistors are both labelled IRFP4227. The output bobbin is wound on a substantial toroid (35mm diameter) using 1.2mm enamelled copper wire. This amplifier incorporates a speaker protection circuit with a substantial relay. It is more of a high-power AC relay, rated at 30A, but the DC voltage rating is only 30V. Still, we would rather have this in the circuit than not! #2 Large Switch-Mode Supply 1000W LLC Soft-Switching Power Supply (see Photo 2): siliconchip.au/link/abif siliconchip.au/link/abig Claims Output power: 1000W Input voltage: 220V AC (nominal) Output voltage options: ±24V, ±36V, Photo 2: this inexpensive “1000W” ±48V, ±60V, ±70V or ±80V (±70V switch-mode power supply seems to use in our case) reasonable quality components and, as far Efficiency: 88-93.7% as we can tell, is sufficiently safe. We were Standby power: 2W pleased that it passed a 500V insulation Size: 156 × 100 × 50mm breakdown test. Net weight: 350g It has four input filter capacitors rated at 180μF and 400V, which should provide sufficient headroom at 220-240V AC. The CapXon brand capacitors have a ripple current rating of 700mA each. When delivering 1kW, the ripple current will be just over that. So their ratings are marginal if we use this to its full rated capacity. The output capacitors are labelled Nichicon 1000μF 80V. These are low-­ impedance capacitors made for switch-mode power supplies that are the right size and look OK. The mains rectifier is a GBK2510, rated at 1000V & 25A. The output diodes are MURF2040CT 20A ultrafast rectifiers. The mains-side switching transistors have their part numbers ground off! As shown in Photo 3, the clearance on this module between Neutral and the mounting screw (which will be Earthed via the chassis) is just over the minimum allowable. However, it is better than the other one we bought and considered (see “The also-ran modules” panel overleaf), so it is OK. Photo 3: the distance between this component lead that connects to the incoming mains Neutral and the mounting hole is smaller than we would prefer, but is just enough to meet separation standards if Neutral & Active are swapped. That is more common than you might think, especially in old houses. Australia's electronics magazine April 2023  27 Fig.1: the frequency response of the 500W Class-D module is very flat, dropping by only 0.4dB at 10kHz and 1.4dB by 20kHz. It’s definitely suitable for driving an LFE (low-frequency effects) channel, given that there is no such roll-off at the low end. made by the module suppliers. A distortion level of around 0.02% at 1kHz is not exactly hifi, but it isn’t terrible either. It is certainly acceptable for many tasks, especially PA, sound reinforcement, or driving a subwoofer in a hifi or home theatre system. The frequency response into a 4W load is shown in Fig.1. There is a bit of a drop-off at the upper end, but it isn’t terrible. It is, however, totally flat down to 10Hz, making it perfect for driving a subwoofer. The slight rise at 5Hz is irrelevant as it is minimal. LFE (low-frequency effects) channel content might go down to 3Hz, at which point it will still be very close to 0dB. Maximum power testing Scope 1: the amplifier output (yellow) into a 4W load near clipping, close to 500W. As it approaches clipping, the Class-D switching frequency drops from 225kHz to about 56kHz, allowing it to deliver a lot of power with some distortion. The ‘choppy’ appearance of the waveform is normal for Class-D. Scope 2: the amplifier pulsed output at around 1kW peak into a 2W load. There’s something nasty going on near the zero-crossings that would lead to very high distortion (if you can see it on a ‘scope, it’s bad!). Still, it is capable of driving 2W as long as the signal dynamic range is high enough. 28 Silicon Chip Australia's electronics magazine Using a 1kHz waveform, the amplifier ran for an extended period delivering 500W into a 4W resistive load. When loaded, the 15V rail voltage increases, almost certainly a result of this rail being an unregulated winding on the switch-mode transformer. The dummy load was a set of 1W resistors made from very heavy duty Nichrome wire. At full load, they were just short of red hot, and the heat generated was enough to make it uncomfortable to hold your hand 20cm above the dummy load. The amplifier sustained this on a continuous basis throughout a 20 minute test – see Scope 1. Reducing the output to about 30V RMS and the load to 2W, the protection relay immediately switched off. Assuming this was overload protection, we switched to using a pulsed signal that is more typical of music, with six cycles at 1000Hz followed by 100 cycles of silence and then it repeats. The amplifier was able to generate this waveform at clipping into 2W. The output voltage was about 60V peak, consistent with a claim of close to 1kW – noting that they specify 10% distortion and the tests here were below clipping. The fact that the amplifier shut down for continuous duty but was capable of brief bursts of output is important. We doubt this amplifier would drive a 2W subwoofer with modern music, which can have significant content at low frequencies. The amplifier was happy with a continuous waveform into 4W, though. The distortion into 2W was visible on the scope (see Scope 2), so we would dread to think of the actual distortion level. siliconchip.com.au The ‘also-ran’ modules___________________________________________________ We considered other amplifier & power supply modules when designing this amplifier. The following modules looked OK, but we decided they were not as good as the ones we went with. Some readers might still be interested in using them in different scenarios, although note that the safety of the alternative switch-mode supply is concerning. #3 Small Class-D amplifier IRS2092S 1000W Mono Digital Amplifier (see Photo 4) siliconchip.au/link/abih Claims Supply voltage: ±58V to ±70V Output power: 1000W (±70V power supply, 2Ω load, 10% THD) Efficiency: ≥90% SNR: 90dB THD+N (±70V, 2Ω): 1% <at> 900W, 0.1% <at> 750W Frequency response: 20Hz ~ 20KHz Speaker load impedance: 2-8Ω Voltage gain: 36 times Input Sensitivity: 1.5V RMS Protection: output short circuit, speaker protection (no relay, though!), over-temperature Dimensions: 132 × 68 × 45mm Weight: 260g The output filter capacitors are two 470μF 100V units per rail, labelled Fulkon CD288H. Data sheets were not obvious on the internet, but they look about the right size for the job. The main switching transistors are both labelled IRFP4227, but the labelling is quite different between them. The output bobbin is wound on an E-core using Litz wire, which is reassuring. Photo 4: we also tested this Class-D amplifier module which could deliver a similar amount of power. We didn’t choose this one because we’d be running it right at the upper limit of its specified voltage range, whereas the other module has another 10V of headroom and also seems a bit better designed. #4 Small Switch-Mode Power Supply LLC Soft-Switching 1000W Power Supply (see Photo 5) http://siliconchip.au/link/abii Claims Input voltage: 200-240V AC Output voltage: ±35 to ±80V (±70V in our case) Other output voltages: independent 12V, auxiliary ±12V Voltage regulation: main ±3% with no load or ±10% with load; independent, ±15% with no load Output current/power: 880W for main, 0.5A each for independent and auxiliary Continuous power: 500W <at> 25°C Rated power: 880W for about 5 minutes at 25°C. A cooling fan should be added for long-term operation. Peak power: 1200W (less than 100ms) Efficiency: up to 95% Weight: 400g There are four input filter capacitors rated at 120μF and 400V, sufficient for running this from 220-240V AC with headroom. The input capacitors are smaller both physically and in capacitance than the preferred unit. At 1kW, their ripple current will be more than 800mA. The data sheet on the installed parts does not specify this parameter, but looking at similar parts, this will likely exceed their rating. The output capacitors are labelled SLF 1000μF, 100V in the CD288H range, specified for high-frequency and low-impedance. These look right for the job. The mains rectifier is a KBL608 unit rated at 800V, 6A unit. That is marginal. Somewhat disconcertingly, the clearance from the mounting hole (to an Earthed standoff) and Neutral on this PCB is closer than desirable – see Photo 6. With a shakeproof washer, it is a touch over 2.5mm, right on the edge of acceptability. A solution might be to use no washer or a smaller washer. Photo 5: the alternative power supply. It can’t deliver quite as much continuous power as the one we ended up using and seemed to use inferior components that are operated too close to their ratings for our liking (in some cases, beyond!). Photo 6: the power supply shown in Photo 5 also has too little clearance between the Earthed mounting hole and the nearest Neutral conductor. siliconchip.com.au Australia's electronics magazine April 2023  29 So in summary, the amplifier ‘does what it says on the box’ aside from delivering that kilowatt into 2W. Design So, let’s look at what it takes to turn these into a very powerful amplifier. The basic arrangement is shown in Fig.2. It is very much about the appropriate connection of the modules and the provision of some cooling. This is a ‘monoblock’ amplifier with no volume control. We expect you would feed it from a preamplifier that provides volume control, input switching etc. For stereo use, you would need to build two of these, although if you want to power a subwoofer, one should be fine by itself. In terms of a preamp as part of a stereo system, you could use our Digital Preamp with Tone Controls from September & October 2021 (siliconchip. au/Series/370) or our Ultra Low Distortion Preamplifier with Tone Controls from March & April 2019 (siliconchip. au/Series/333). You could, in theory, add a volume/ level control pot on the front panel and route the signal wiring to the amplifier module via that pot. We’ll leave that as an exercise for our readers as we expect most constructors will use a separate preamp. Build and testing We first had to work out how to house this safely and at a reasonable cost. We chose the Jaycar HB5556 chassis as it is just right in size, of good Fig.2: thanks to the prebuilt modules, the ‘circuit’ of this amplifier is dead simple. The power supply generates three rails: -70V, +70V and +15V, which are fed to the amplifier module. The 15V rail also powers the 12V fan via a 39W 1W dropper resistor. build quality and at a great price. This case also lent itself to us implementing some forced air cooling. There are three main baffles to keep things cool, as shown in Fig.3. We are striving to achieve forced airflow over the heatsinks for the power supply and Class-D amplifier. Even though these are better than 90% efficient, if you are driving 1000W into Speaker power handling Speaker power ratings are a bit of a vexing topic. Those who were around in the 1980s and 1990s will have seen the outlandish Peak Music Power Output or “PMPO” numbers that ran into the thousands of watts, often from a 10W IC amplifier chip! At a more pragmatic level, the power rating of a loudspeaker is primarily defined by the capacity of the voice coil to dissipate energy and, at a mechanical level, the excursion limit of the cone. For example, a tweeter typically has a 25mm coil weighing a small fraction of a gram. Many are rated at 100W or more, but the actual continuous power they can handle is only a couple of watts. They rely on the crossover and the nature of music signals to reduce “100W” to only a few watts seen by the tweeter. Woofers have a much tougher life. AES2-1984 defines the power handling test. Power handling is measured with pink noise with a 6dB peak-to-RMS ratio. For example, the BEYMA 21LEX1600Nd driver has a 3200W “program power” rating and a 1600W continuous power rating, equating to a 400W RMS sinewave power rating. Be warned that this amplifier could be very bad for the health of your domestic speakers! We have not recently produced a speaker design that can handle 500W continuously, although the Majestic Loudspeakers (June & September 2014 issues; siliconchip.au/Series/275) are somewhat close, at 300W (tested). 30 Silicon Chip Australia's electronics magazine a load, that is 50-100W being dissipated in each module, mainly via their heatsinks. They will get very hot running this way without air moving over them. Of course, this will not normally be the case. Typical music has a crest factor over 10dB (depending heavily on the type of music), which means that on average, with full-range music not being driven heavily into clipping, the output power would rarely be over 100W for very long. But consider the realistic use case for a 500W (or 1000W) amplifier; its niche is in subwoofer duty, where, with modern music, all bets are off. Modern music has periods of close-to-continuous bass output. So keeping everything cool is essential. With modest output, say, averaging up to 100W or so, these amplifier modules are fine in a case with passive cooling. If that is your application, you can avoid manufacturing the plenum presented here. If you intend to play loud music for extended periods, you need to bolster the cooling. Our plenum is made from three folded sheets of aluminium and uses the case’s lid as the top. This allows us to add a fan and force air over the heatsinks, increasing their efficiency. siliconchip.com.au Fig.3: the case is reasonably compact yet more than large enough to fit the two modules. A series of baffles direct air sucked in through the rear panel (by an 80mm fan) across the heatsinks of the amplifier module and power supply, then out through vents on the left side. The top vents are blocked off to prevent air from escaping before it has completed this route. Without getting too much into the details of heat removal, consider that heatsinks dissipate energy through convection (hot air rising from the heatsink being replaced by cooler air), radiation (mainly IR energy being emitted) and conduction from the heatsink into thermally connected materials. Without running our amp so hot that it’s about to melt, radiation is not a significant factor. Convection is an important means of heat removal, but the case stifles this somewhat, and even in free air, heat will only be removed by convection siliconchip.com.au so fast. By forcing air from outside through the case, over the heatsinks and then exhausting it from the case, we can increase the transfer rate between the heatsinks and the air, picking the heat up off the heatsink and dumping it outside the case. Making the baffles We folded aluminium sheets to form a labyrinth, with a fan forcing air in from the rear of the enclosure and using the perforations along the sides of the case for exhaust. The panels are all securely Earthed for safety. The Australia's electronics magazine cutting and folding details are shown in Figs.4-6, with instructions to follow. We made ours from three sheets of 1.2mm-thick aluminium, although a thickness between 1.0mm and 1.5mm will be fine. You could alternatively use polycarbonate sheets and glue or tap and screw them, or if you have the gear, 3D print it. Use our plan as a guide and follow the principles of forcing air across the heatsinks and out of the box. Assuming you’re making the panels as we did, first cut the metal sheets to size. We used a jigsaw. An angle April 2023  31 Fig.4: the plenum baseplate is bent up on either side to form the ends of the chamber. The cut-out in the upper left corner is for air to exit into the left-hand side of the case, where it escapes via side vents. Fig.5: this panel, also made from a bent aluminium plate, seals off the section of the plenum chamber closest to the case’s front panel. grinder with 1.6mm metal cutting discs also works but requires caution. Drill the holes as shown before bending. If you do not have a pan brake, 1.2mm aluminium can be successfully bent by clamping it to a workbench with a tight 90° edge and using a hammer and piece of timber to ‘panel beat’ the corners into the metal sheet. Go slowly and gently. Make sure the end panel is a good fit for the base. We achieved this by 32 Silicon Chip making the base piece first, then, once it was folded, adjusting the folded ends of the rear panel to achieve an acceptable fit. This does not need to be perfect; there will be a fair bit of airflow, so a leak here and there really does not matter. If you choose to paint your metalwork, make sure to mask off around the Earth lug, as you need a good electrical connection there. With the baffles made, cut the holes in the rear panel for the fan, Australia's electronics magazine input, output and power connectors, as shown in Fig.7. This is an inside view, so if you are cutting from the outside, make sure to mirror it. The final result from the outside (once all the components are mounted) is shown in Fig.8. Cutting the fan hole is a bit fiddly. We used the ‘drill and file’ method, in which you drill many 4-5mm holes around the inside of the final cut line to remove the bulk of the material, then use a file to smooth the edges. An siliconchip.com.au Fig.6: this baffle divides the plenum chamber into two halves, one side for the power supply and one for the amplifier module. The rectangular cut-out allows air to pass from one side to the other. Fig.7: this shows the cut-outs needed in the rear panel but note that the large hole at bottom centre, with two smaller holes near it, is for the Speakon terminal that constructors might opt to leave out. The RCA socket hole has been moved since we built the prototype, as it interfered with the fan. Fig.8: this shows how the rear panel should look once completed. The Speakon terminal is wired in parallel with the binding posts; only one is required, depending on the speaker connector you plan to use. alternative method is to use a jigsaw with a metal cutting blade. To make the holes in the base of the case, present the plenum base to the rear panel with the rear panel in the case, then mark the mounting holes. These are shown marked on the siliconchip.com.au drawing; there are six of them between the folds. This will ensure these are in exactly the right spot. Once marked, drill these, then the mounting holes for the PSU and amplifier modules. These holes need to be countersunk on the underside. Photo Australia's electronics magazine 7 shows how we aligned the plenum in the case to drill the mounting holes. Mount the amplifier and PSU modules now, as shown in Photo 9. Use countersunk M3 machine screws to secure the eight 15mm threaded PCB standoffs to the base. This will allow April 2023  33 Photo 7: once you’ve made the plenum base, you can fiddle with the baffle separating the two halves, so it’s a good fit and not too much air will leak past. Fig.9: cut a sheet of Presspahn or similar insulating material (thick cardboard will do) and mount it on the power supply to ensure sufficient airflow over both the heatsinks and transformer. the plenum assembly to sit flat in the case when assembled. Then use 6mm M3 machine screws and star shakeproof washers to secure the boards. Optimising the airflow We made an extra baffle for the power supply module to force more air over the heatsinks, shown in Fig.9, made from Presspahn. Unfortunately, Presspahn insulating card is becoming hard to get, although we did find an equivalent material (see the parts list). If you can’t get that, use thick cardboard, as we are not relying on its insulating properties too heavily here. Under no circumstances use metal. This is secured with two M3 machine screws and star shakeproof washers to the tapped holes in the top of the heatsink. Use Loctite to ensure these screws do not come loose over time. You should also stick a piece of card to the inside the top panel to cover the vent holes over the plenum. This way, the air does not escape through there and has to flow past all the heatsinks on the way out. Once that’s in place, cut and stick lengths of weather-stripping foam along all the top edges of the plenum chamber and baffles, as shown in the photos. This will make a seal with the case’s lid so that too much air doesn’t flow over the panels and mess up the airflow pattern. Wiring it up With the modules installed, mount the internal baffle. This is important as it controls airflow, as shown in Fig.3. You can see how this sits in Photo 8. Photo 8: the rear view of the 500W Class-D Amplifier’s chassis. 34 Silicon Chip Australia's electronics magazine siliconchip.com.au SILICONE SEALANT OVER EXPOSED METAL HEATSHRINK SLEEVES OVER SPADE LUGS & CONNECTORS N L +15V GND CABLE TIES 12V FAN V+ GND V– OUT GND 39W 1W RESISTOR POWER SUPPLY MODULE AMPLIFIER MODULE IN GND +15V GND V– GND V+ PRESSPAHN BAFFLE (NOT FULL HEIGHT) HEATSHRINK SLEEVES OVER ALL SPADE LUGS & CONNECTORS Fig.10: all the wiring for the amplifier is shown here, except that the Speakon connector has been left off. If fitting it, wire it in parallel with the binding posts. You could use the spare output terminals on the amplifier module for that if you wanted to. Don’t leave off the insulation or cable ties for the mains wiring (also see the photos) and ensure the Earth lug makes good contact with the chassis base. Once it is screwed in, install 10A mains-rated red, green and black wire between the ±70V outputs from the PSU to the amplifier module’s power inputs, referring to Fig.10. This rating is essential as there is 140V DC between these conductors and they can carry significant current. Use medium-duty hookup wire to connect the independent 15V power rail to the amplifier module. Add lengths of 6mm heatshrink siliconchip.com.au tubing over much of these two sets of wires because we will run these cables through the hole in the internal baffle, and we will be tying these to the very top of this opening. This will control where these cables sit, and the heatshrink adds a level of protection and ruggedness to this cabling. Fan connection Next, connect the power to the fan. The fan is a 12V type, but the closest Australia's electronics magazine rail we have is 15V DC, so a 39W 1W resistor connected in series with the fan drops about 3V. The fan is wired to the 15V connector on the amplifier board. Use light-duty or medium-duty hookup wire. Mains wiring The mains wiring is also shown in Fig.10. There is not a lot of it; however, you must take caution with all wiring as most is either mains potential or April 2023  35 Photo 9: this shows how the two modules fit inside the plenum chamber within the case. The wiring between the two modules has been run along with the input and fan wiring, but the output and mains connections have not been made yet. high voltage DC or AC (the output). Ensure all wiring is secured with zip ties to keep it tidy and controlled if anything comes loose. Install the power switch on the front panel as shown in Fig.11. Then take two lengths of brown and blue mainsrated 10A wire and connect from the IEC mains connector to the switch as shown. Connect the topmost terminals on the switch to the IEC mains input, and then run a second pair of wires from the central switch terminals back to 36 Silicon Chip the mains input on the power supply. We used insulated crimp connectors on the IEC connector and switch. If you wish to solder these connections instead, insulate the joints with 10mm diameter heatshrink tubing. Keep these wires twisted and tidy, and zip-tie them such that they cannot come loose in the case. We found it handy to label the unswitched and switched input wires. Using a length of yellow/green striped 10A mains-rated wire, connect the Earth pin on the IEC connector to Australia's electronics magazine the M3 Earth screw that runs through the case and plenum metalwork. Before assembling this, take a utility knife and scrape the paint from the case around this bolt. Use a star shakeproof washer on the bottom and top of the case and attach a 3.2mm solder lug to this. Connect the Earth wiring and check continuity with a multimeter. Install an 8A or 10A ceramic fuse in the IEC mains input/fuseholder assembly. Remember to insulate the exposed metal strip on the back of this siliconchip.com.au Fig.11: just one hole is needed in the front panel for the power toggle switch. That is unless you elect to add a volume control pot or a power-on indicator (an illuminated switch could be used instead). connector with neutral-cure silicone sealant, as it will otherwise be live whenever the mains cord is plugged in. this to the top of the plenum with a cable tie and want this as extra abrasion protection. Output wiring Input wiring Use mains-rated 10A rated wire for the amplifier output connections. We used 400mm of green and red wire twisted together from the amplifier output to the output connectors. We included both Speakon and binding posts outputs; you may only need the binding posts. We sleeved the output wiring in a 250mm length of 6mm diameter heatshrink tubing. We did this firstly to ensure there could be no confusion between this and the power wiring and also because we will be securing Take 300mm of shielded cable and connect the RCA connector on the rear panel to the screw terminal header on the amplifier board. Use a short length of sleeving to insulate the exposed ground braid and 20mm of 3mm diameter heatshrink to form nice terminations. Caution At this point, you should have a standalone chassis with the amplifier modules installed and wired up. First and foremost: safety. If you are not totally comfortable working with high voltages then do not proceed without help. It’s also safest to do the first power up with the lid secured. This amplifier can generate a lot of power. To do this, it uses high supply rails of ±70V DC. It could easily stop your heart if you make contact with these two rails. Also, the switch-mode power supply operates from the mains and has close to 400V DC in parts of the circuit. This is also lethal. Second: danger to your possessions. The amplifier generates 44V RMS continuously into a 4W load. This is close to 500W. If you feed this into your speakers as a sinewave, we can guarantee you will destroy them. See the panel on “Speaker power handling”. This view shows how we wired up the output connectors and gives you a good view of the Presspahn baffle that optimises airflow over the power supply module. You can also see how the mains input wiring has been insulated. Also note how the output wiring and ±70V rail wiring is cable tied to the top of the plenum, just behind the Presspahn baffle. siliconchip.com.au Australia's electronics magazine April 2023  37 Similar cautions apply for test equipment; make sure that if you connect this to a distortion analyser, it is on a 50V or 100V RMS range. Testing First, check that the mains power switch is on, then with it unplugged, do a final check with a DVM on its 20MW range (or similar) and check for any measurable resistance between the Active and Neutral inputs and the output ground connector. If there is, then you need to stop and find the problem. Also perform a final check of your Parts List – 500W Monoblock Amplifier 1 1000W Class-D amplifier module (see links at the start of the article) 1 1000W 70V split-rail switch-mode power supply (see above) 1 vented metal bench enclosure, 304 × 279 × 88mm [Jaycar HB5556] 1 dual binding post for speakers [Altronics P9257A] 1 panel-mount insulated RCA socket [Altronics P0220] 1 fused IEC mains input socket [Altronics P8324] 1 10A+ mains-rated chassis-mount DPST/DPDT toggle switch [Altronics S1052 or Jaycar ST0585] 1 8-10A fast blow sand-filled or ceramic M205 fuse [Altronics S5934] 1 Speakon chassis-mount speaker connector (optional) [Altronics P0792] 1 quiet 80mm 12V fan [Altronics F1150] 1 80mm fan guard [Altronics F1022] 1 2-way 2.54mm-pitch vertical polarised header [Altronics P5472, Jaycar HM3412] 1 2-way 2.54mm-pitch polarised header plug [Altronics P5492 + 2 × P5470A, Jaycar HM3402] 1 39W 1W resistor Hardware 1 428 × 225 × 1.0-1.5mm aluminium sheet (for base) 1 225 × 103 × 1.0-1.5mm aluminium sheet (for baffle) 1 259 × 100 × 1.0-1.5mm aluminium sheet (for plenum end) 1 115 × 65mm sheet of Presspahn or similar insulating card [www.ebay.com.au/itm/293254125529] 9 M3 × 16mm panhead machine screws 12 M3 × 10mm countersunk head machine screws 16 M3 × 6mm panhead machine screws 12 M3 hex nuts 32 M3 star shakeproof washers 8 15mm M3-tapped spacers 1 3.2mm solder lug [Altronics H1503] 9 blue insulated 6.3mm female spade crimp lugs for 1.5-2.5mm2 wire [Altronics H2006B] 1 1.2m length of 9-10mm wide adhesive foam weather stripping [Bunnings 3970353] 1 1.5m length of 5-10mm wide adhesive foam weather stripping [Bunnings 3970353] 1 pack of small Nylon cable ties Wire & cable 1 1.5m length of brown mains-rated 10A hookup wire 1 1.5m length of blue mains-rated 10A hookup wire 1 0.5m length of green/yellow striped mains-rated 10A hookup wire (eg, stripped from a length of 10A three-wire mains flex) 1 1m length of red mains-rated 10A hookup wire 1 1m length of green mains-rated 10A hookup wire 1 1m length of black mains-rated 10A hookup wire 1 0.5m length of red medium-duty hookup wire 1 0.5m length of black medium-duty hookup wire 1 300mm length of single-core shielded audio cable [Altronics W3010] 1 1m length of 6mm diameter clear heatshrink tubing 1 200mm length of 3mm diameter clear heatshrink tubing 38 Silicon Chip Australia's electronics magazine wiring. A fault here will be both spectacular and dangerous. Plug the amplifier in, switch it on and listen for the speaker protection relay switching in after a couple of seconds. Carefully measure the voltage between ground, V+ and V− on the power supply output using some properly insulated DMM probes and a suitably rated meter. The rails should both be within 5V of 70V but with different polarities. Carefully measure the voltage on the +15V input to the amplifier and ensure it is close to expected. If any of the above fails, unplug the amplifier and leave it off for 10 minutes. After verifying that the mains plug is still out, disconnect the power amplifier from the power supply so you can check the PSU by itself. If you can’t see the right voltages at its outputs with no load, you have a faulty PSU. If the PSU measures OK, rebuild it and check your wiring carefully. Now plug in a signal generator to the input and a CRO with a 10:1 probe set to measure up to 70V peak to the output. Power up and look for the sinewave on the output. Increase the signal level until you see clipping; check that this is about 40-44V RMS. Connect a load and start the input signal at a low volume level, increasing to a manageable level. Only use a loudspeaker for this if you have no other choice and are happy to test at moderate levels only. If you have a dummy load, run the amplifier at as high a power as is safe for your load for 5-10 minutes. If you are using a speaker for the test, play some moderately loud music. At this point, we are really just checking that nothing goes wrong – no puff of magic smoke etc. After testing as hard as you feel safe, unplug everything and open the amplifier. Use an IR thermometer to measure the temperature of the PSU heatsinks, the E-core transformer on the PSU (in our tests, this was the hottest part) and the amplifier heatsink. If these are all below 65°C, everything is fine and you are all set! Otherwise, check the airflow management components (baffles, seals etc) to verify that there are no massive air leaks and confirm that you haven’t skipped any of the steps listed above. SC siliconchip.com.au