Fullscreen HTML5Med Res (??MB)HTML4

Last month, we looked at some ways to improve amplifier cooling, either in an amplifier you are building or an existing one that is running too hot. This month, we go into the details of modifying a specific amplifier to improve its fan cooling. Part 2 by Julian Edgar Cooling Audio Amplifiers A fter ‘cooking’ two hard-working amplifiers in a hot roof space, I resolved that any further amplifiers put to this torture test would need to be commercial (rather than domestic) designs – and preferably fan-cooled. My budget didn’t extend to new amplifiers, so I looked for second-hand ones. After an extensive search, I found two LD Systems amplifiers – the XS-400 and XS-700. The XS-400 has an output of 2 × 200W into 4W, while the XS-700 develops 2 × 350W into 4W. Both are Class-D amplifiers that have a maximum distortion of less than 0.1%. Not hifi, but good enough for a wholeof-house sound system. I bought the XS-700 first and tested it extensively, using it to power two 15-inch (380mm) subwoofers, also located in the roof space. The testing showed two things. First, the amplifier worked well, and second, despite the fan cooling, certain internal components ran quite hot. I’ll concentrate on the XS-700 in this article, but I modified both amplifiers in the same way. Airflow will take the path of least resistance, and the inner surface of the top amplifier panel is often the smoothest, least obstructed path. Therefore, with air inlets in the front panel and an outlet fan in the back panel, unless it is prevented from doing so, a lot of air will flow along the underside of the top panel, completely missing all the components it is meant to cool! these initial temperatures were measured in 20°C ambient conditions). The heatsink in the audio section of the amplifier was noticeably hotter – about 45°C. What really concerned me were two voltage regulators positioned in the middle of the PCB. These were running at 60°C – and in hot ambient conditions, I saw 75°C! See Photos 2 & 3. The data sheets for these KA7815 and KA7915 regulators showed a specified operating range of 0-125°C. However, that’s the junction temperature, which is likely to be a fair bit higher than the external temperature (to calculate how much, we’d need to know their dissipation and multiply it by the junction-to-case figure in the data sheet). Still, they are likely well within their specifications. However, running 40°C above ambient seems pretty darn hot to me! Perhaps more worryingly, they’re located very close to two large electrolytic capacitors, which are known for not liking heat. Australia's electronics magazine siliconchip.com.au Initial temperature testing The amplifier uses two major heatsinks: one located in the audio amplification section, with the other for the switch-mode power supply (see Photo 1). Measurements from an infrared thermometer showed that the power supply heatsink was typically running relatively cool, for example, 37°C (all Air can be sneaky sometimes 62 Silicon Chip Photos 1-3: this LD Systems 350W × 2 Class-D amplifier has a single rear fan that draws air through two grilles in the front panel. The large heatsink on the right is for the power supply; the one on the left is for the audio amplifier. The thermal camera view inside the amplifier shows the hottest parts to be two voltage regulators – they’re nearly 39°C in 20°C ambient conditions after only a few minutes. Once more time has passed, those two regulators (circled) are over 60°C. While within their specifications, they are next to two large electrolytic capacitors. Such capacitors don’t like heat. For cooling, the amplifier uses two front grilles and a 35mm fan located more or less centrally on the rear panel (Photos 4 & 5). The two front grilles are internally covered with a dust filter (Photo 6). The fan operates at two speeds; it appears the increased speed is triggered when the audio heatsink is above 55°C. An airflow baffle made of PCB laminate is positioned transversely near the front of the amplifier, between the two main heatsinks, with some small holes in it. No airflow baffles are provided outboard of the two major heatsinks. So where was the air going inside the amplifier? I removed the upper panel of the enclosure and temporarily replaced it with a sheet of clear acrylic. I then used the smoke from an incense stick to carefully observe It is difficult to concentrate when an amplifier is belting out at full volume, so it’s best to use a dummy load when doing high-load testing. People in your household (and possibly your neighbours, and their neighbours) will thank you. The load comprises resistors of an appropriate value to emulate the speakers you are using – for example, 4W or 8W. Very high power resistors are expensive, but there’s a cheap and easy way to create your own load. Two approaches can be taken. In the first, buy two electric jug elements of the sort that have an exposed winding on a ceramic base. Unwind sufficient length from each so that you create a load with the appropriate resistance. For example, configure each as an 8W load and wire them in parallel to give a 4W load (see Photo 7). Or, since this type of jug element is now becoming more expensive Photo 4: a standard baffle is located between the two main heatsinks to prevent air flowing directly from the front vents to the rear extracting fan. However, testing with smoke showed quite a lot of air passed straight over the top! Photo 5: the rear-mounted fan has two speeds, with the slower of the two being inaudible. siliconchip.com.au the pattern of the airflow within the working amplifier. As always, when doing this type of flow testing, things were not as expected! There were three main paths that the air took between the inlet grills and the outlet fan – bypassing the audio heatsink to the left, bypassing the power supply heatsink to the right, and flowing over the top of the central baffle in the gap between the baffle and the lid! That is, none of the heatsinks had much airflow passing along their fins, and the two very hot voltage regulators were largely in static air, although they got a small amount of flow. Before doing any further testing, I decided to connect a dummy load. Dummy loads Australia's electronics magazine September 2025  63 and harder to find, buy some 5W, 1W wire-wound resistors and wire them in series to get the required resistance. Use thick cable to connect the loads to the amplifier’s speaker terminals – one load for each channel. Then fill a Pyrex (or ceramic) container with water and place the loads in it. Ensure that the resistors and connecting cables cannot short out and be aware that the water can become hot enough to burn. Make sure that neither you nor anyone else can come into contact with the water. I used eight 1W 5W resistors, wired in series to form two 4W loads, placed each side in a double ceramic cooking dish. The dish contained about one litre of water (Photo 8). It took about an hour of testing for the water to get really hot. One problem with using a dummy load for an extended period of testing is that, should your input signal fail, you may be unaware of that. To overcome this, wire a speaker to one channel of the amplifier through a 150W 5W series resistor. This will allow you to hear the input signal at a low volume, even when the amplifier is working hard. If the speaker is still too loud, increase the resistance. The monitoring speaker will also let you know if you have cranked up the amplifier high enough that it clips (the sound will distort), so you can turn it down a bit. While most amplifier testing uses a sinewave input, I suggest that for this testing, you use normal music of the This will allow the heatsinks to heatsoak and so be forced to work as heat exchangers. This test also allows you to monitor your dummy load, to ensure that the water doesn’t become too hot. If it does, switch the amplifier off and then carefully replace the water at appropriate intervals, or use a larger container. If you are unsure whether the amplifier has an automatic temperature-­based shutdown, monitor internal temperatures during this initial run-in period. Testing with the dummy load Photo 6: the front air inlet grilles had this filter placed over them. I removed it to achieve better flow. sort you listen to. A sinewave input will work the amplifier extremely hard, and unless you habitually listen to sinewaves for recreation, it’s also not indicative of the conditions under which the amplifier will actually be working. To set the input level correctly, take note of the volume control’s position when your normal speakers and source are connected and you are playing music as loudly as you ever will. Then, with the dummy load and monitoring speaker connected, replicate that level on the control. When testing, start by running the amplifier at full power (below clipping, remember) for 15-20 minutes. With the dummy load connected and the clear acrylic lid in place, I could fully test the XS-700 amplifier. My first concern was with the very hot voltage regulators. Their heatsinks were small, had vertical fins (whereas the airflow through the amplifier enclosure is horizontal) and furthermore, the two heatsinks were positioned at right-angles to each other. Editor’s note: those small blocky heatsinks are better than no heatsink but otherwise are mostly useless. Even a small flag heatsink will generally outperform them. Flag heatsinks have gaps in the fins, so airflow in virtually any direction will help them dissipate heat. Replacing these heatsinks with a much larger, horizontally aligned design seemed to be a good first step – but there was a snag. To remove the existing heatsinks would be very difficult; the main PCB would need to be removed from the case, and even then, Photos 7 & 8: a dummy load can be made by rewiring electric jug elements or using series wire-wound resistors. In both cases, match the impedance of the speakers you are using (eg, 4W). The load is then placed in a ceramic (or Pyrex) dish that has been filled with water. Warning: the water can become hot enough to scald; and both resistive loads for each channel should be kept separate as contact between them could damage the amplifier. 64 Silicon Chip Australia's electronics magazine siliconchip.com.au gaining access to the screws that held the heatsinks to the regulators would be difficult. Obviously, these components were installed early in the build process. Could the heatsinks be retained and airflow better directed at them? I created a smooth channel between the fan and the two regulators from two thin strips of cardboard. In effect, nearly all the fan’s air was then being channelled through the voltage regulators’ heatsinks. Doing this showed a dramatic drop in the regulator temperatures – from running at 60°C to 49°C. However, as you would then expect, the airflow pattern within the enclosure was altered – testing with smoke showed that the audio amplifier heatsink was getting much less airflow past it, and the infrared thermometer showed a commensurate increase in heatsink temperature. I then cut a small opening in the wall of the baffle closest to the audio heatsink, allowing the fan to draw some air from that direction. Smoke testing showed this was indeed happening, and the audio heatsink dropped in temperature (see Photo 10). But what about the other end of the amplifier – the power supply section? That heatsink had never run particularly warm, and yet a lot of airflow was passing around it – a waste of flow, if you like. I then extended the standard central baffle in that direction, reducing the flow around this heatsink. As expected, the heatsink’s temperature then rose a little – but it was Measuring temperatures For reasons of safety, convenience and speed, infrared temperature sensing is the best way to check the amplifier’s temperature during testing. An infrared thermometer measures the amount of infrared energy given off by an object. The amount of infrared energy coming from an object depends on its temperature and emissivity. The emissivity of a perfect radiator of infrared energy, called a blackbody, is 1. However, many objects have emissivities that are less than 1, and if a correction isn’t made for this, the temperature measurement will be wrong. If the object either reflects or transmits infrared energy, the emissivity value will be less than 1. Shiny polished surfaces, such as aluminium, are so reflective of infrared energy that accurate temperature measurements of those surfaces may not be possible without modifying them. Some infrared thermometers can be programmed for the emissivity of the surface you are measuring, but many just use a default value of 0.95 – the emissivity of lamp black or candle soot. If you are making only comparative measurements (has the temperature gone up or down with your modifications?), the emissivity won’t matter much, but if you want accurate values and you are measuring a shiny surface, you may want to colour it black with a marker, or on a large shiny heatsink, stick a thin piece of black electrical tape onto it. A thermal camera, while more expensive than a digital infrared thermometer, can also be very useful. Like an infrared thermometer, thermal imaging cameras (sometimes also called thermographic cameras) measure infrared radiation. However, unlike the thermometer, they then render that as a visible light image on a colour LCD. Typically, the ‘hotter’ the colour on the display colour, the higher the temperature of that area. The biggest advantage of a thermal camera over an infrared thermometer is that you can quickly scan whole areas – just point the camera at the open amplifier and you can immediately see the hot spots. Another advantage is that thermal imaging cameras automatically adjust the scale that they are using, depending on the variation in temperature. Therefore, quite subtle variations in temperature, that you would take a long time to find with the infrared thermometer, are immediately visible. However, unless you have other uses for a thermal camera (I have found that there are plenty), the infrared thermometer should be good enough for amplifier temperature measurement. Photo 9: it doesn’t photograph well, but it’s easy to see the smoke flow from an incense stick being drawn through the case. The top cover has been replaced by a sheet of clear acrylic. A temporary cardboard baffle (under the brown wiring) is reducing the flow that bypasses the power supply heatsink. Photo 10: a close-view of the temporary cardboard baffles. The cutout in the baffle nearest the camera allows airflow from the front inlets past the audio heatsink (left, out of view). This tiny cutout made a dramatic change to the measured flow past that heatsink. siliconchip.com.au Australia's electronics magazine September 2025  65 2 3 1 Photo 11: the temperature and flow testing setup. (1) Temporary baffles linking the voltage regulators to the fan. (2) Strip prevent air flowing over the top of the standard baffle. (3) Baffle to prevent flow bypassing the lower power supply heatsink. Table 1 – amp modifications Heatsink Standard Modified Power supply 37°C 39°C Voltage regulators 60°C 49°C Audio 55°C 52°C Photo 12: the final airflow baffles and guides can be made from insulating paper such Presspahn or this fibroid fish paper. The baffles and guides can be held in place with small dabs of silicone sealant. 66 Silicon Chip still the coolest major heatsink in the amplifier. Time for some more smoke testing. With the voltage regulator cooling tunnel in place, complete with the cutout in the wall to promote some flow around the audio heatsink, and the baffle preventing a lot of wasted airflow past the power supply heatsink, the interior airflow pattern of the amplifier had greatly changed. With some of the previous free-flow channels now blocked, a lot of airflow was passing over the top of the standard front baffle. I then added a cardboard strip to block this flow (Photo 11). Interestingly, the fan could now be heard working harder – it was drawing air past the components it was meant to cool, rather than happily bypassing most of them! Table 1 shows the results. They were measured just below clipping on music material, working as a subwoofer amplifier crossed over at 90Hz, in a 20°C ambient environment, with the fan operating at a low speed. As can be seen, at full load, the altered airflow has caused a slight increase in the power supply heatsink temperature, a reduction in the audio heatsink temperature and a major reduction in the voltage regulator temperature. In fact, many hours of testing showed that the voltage regulator Australia's electronics magazine temperatures were reduced by as much as 25°C in some conditions! Installing the baffles Rather than use cardboard to form baffles and guides, it is better to use an insulating product such as Presspahn. However, I found it difficult to get cheaply in small quantities, so I used fibroid fish paper, which is available from Rockby Electronics. It comes in a tight roll and needs to be flattened before it can be used. This can be achieved by rolling it in the other direction and/or using an iron. The paper can then be cut to size and inserted where the cardboard trial baffle and guides were. A few dabs of silicone sealant hold them in place. To seal the baffle (the one that had plenty of airflow over the top), I used a strip of soft foam rubber cut from a larger sheet. Again, this was held in place with some silicone. When the lid is replaced, it seals against this foam. Conclusion Whether it’s thermally connecting panels to act as heatsinks, re-­orientating heatsinks to allow better convectional flow, adding fans or altering airflow patterns within the enclosure by using guides and baffles, improving amplifier cooling can make a major difference to SC internal temperatures. siliconchip.com.au Photo 13: the finished modifications. They cost very little but give major reductions in the temperature of the hottest components. Versatile Battery Checker This tool lets you check the condition of most common batteries, such as Li-ion, LiPo, SLA, 9V batteries, AA, AAA, C & D cells; the list goes on. It’s simple to use – just connect the battery to the terminals and its details will be displayed on the OLED readout. Versatile Battery Checker Complete Kit (SC7465, $65+post) Includes all parts and the case required to build the Versatile Battery Checker, except the optional programming header, batteries and glue See the article in the May 2025 issue for more details: siliconchip.au/Article/18121 siliconchip.com.au Australia's electronics magazine September 2025  67