Fullscreen HTML5Med Res (??MB)HTML4

SERVICEMAN’S LOG Another mixed bag of servicing stories Dave Thompson has returned from his arduous trek, which he made to pay respect to the most revered authority in New Zealand: the national Rugby Union team. We have some reader-contributed servicing stories while he is recovering. Regular service will resume next month! Common capacitor problems in appliances N. B., of Taylors Lakes, Vic repairs commercial laundry equipment and, given the constant use (and no doubt abuse) they receive, he is not short of work. Here are some of the more memorable repairs he’s made lately... I have repaired many Maytag coin-operated washers and dryers that use the power supply board shown in the accompanying photo, including models like the MHN33, MH30, MD20, MDA20 and the Neptune range (shown here). The fault is that the coin mechanism sometimes won’t count the coins. The display remains in the idle state. A significant ripple voltage is present on the +24V supply rail. Visually inspecting the board, it is apparent that the 5V rail filter capacitor has failed, but that was actually caused by the less obvious failure of the larger 24V rail filter capacitor. Generally, I replace them with a 2200µF 35V capacitor for the 24V rail and 470µF 35V for the 5V rail, and everything then operates correctly. Sometimes it doesn’t because significant ripple is still present on the +24V rail. The power supply arrangement in this machine is unusual. The board AC power is fed from a 110V AC 60Hz transformer. A 240V AC to 110V AC step-down transformer drives the primary of the isolating transformer that supplies power to the PCB. I don’t know why they didn’t use a dual primary transformer instead, with them in series for 230V AC countries and parallel for 110V AC. I thought the 2200μF capacitor might be drawing too high an impulse current, causing saturation of the magnetic circuit of one or both of the transformers, causing that ripple. 90 Silicon Chip To prove this, I disconnected the 2200μF capacitor and soldered two 1000μF capacitors in parallel in its place via flying leads. That worked, but I didn’t want capacitors hanging around off the PCB, so I decided instead to reconnect the 2200μF capacitor, adding two clip-on ferrite filters onto the wires from the PCB to the 20V AC output of the transformer. To my surprise, that worked too. Now the only problem is ensuring the technician installing the repaired PCB reads my notes and follows them! I also received a few Speed Queen (Alliance) Quantum dryer power supply PCBs (Alliance also makes Primus and Ipso brand machines). The machines were showing an “EHT” error on their displays, and the customer note said the dryer stopped working and only came good when the power was cycled off and on. Searching for that error code online told me the dryers were not reaching their drying temperature within the maximum allowed time. This happens if the flue is clogged with lint, the fan is going in the wrong direction, the heater has failed, or the dryer is too full of wet clothes. Finding no errors with the PCB, after finding out it was a very intermittent fault, I suggested that the customer should change the customer programmable “heat fault” setting to off. I would not suggest this if the dryers were gas-powered, but I knew all his machines were electrically heated, so the possibility of a fire is much lower than for gas machines, and the site is well supervised. Two other PCBs that came with that batch had the fault described as “no power”. Australia's electronics magazine siliconchip.com.au Items Covered This Month • Maytag coin-operated washers and more • Repairing a Seiko S451 watch pressure tester • Failed Li-ion battery packs in leaf blowers • Fixing the Silicon Chip 20W Class-A Amplifier • Poorly timed failure of a USB sound card • Microwave oven repairs Dave Thompson runs PC Anytime in Christchurch, NZ. Website: www.pcanytime.co.nz Email: dave<at>pcanytime.co.nz Cartoonist – Louis Decrevel Website: loueee.com I first looked up the chip descriptions on the internet for the ICs most likely to be power regulators. The first one closest to the rectifier and large 350V-rated electrolytic was a TOP256. With the 50-page data manual downloaded, I read the brief description and viewed the simplified circuit for that part. When I got to the paragraph titled “Soft start”, I found a brief description of the soft start circuit. The chip’s internal Mosfet is off at power-up; the rectified/filtered mains (now in the region of 350V DC) is applied to pin 4, where it is bled via an internal high-voltage current source to pin 2, the DC low-voltage input for the chip’s internal control and workings. On this pin is a 47μF 35V electrolytic filter capacitor (C9 in the sample circuit), which must be charged to 5.8V before the chip comes alive. A set of well-insulated flying leads to a voltmeter indicated that this voltage was never reached when the board was connected to a mains supply. A quick change of this capacitor got the supply running beautifully. Editor’s note: this is a fairly common fault in switchmode supplies where an electrolytic capacitor’s leakage increases to the point that the initial ‘trickle’ current is no longer enough to ‘bootstrap’ the circuit. The red arrow indicates the faulty capacitor, while the yellow arrow points to the TOP256GW IC. You can tell this PCB came from a dryer as it’s covered in lint! Note the conformal coating on the board that stops anything that might be conductive (eg, moist lint) from causing problems. The other PCB didn’t have this same capacitor problem; the starting voltage was correct, but the TOP256 IC had obviously failed, since it came alive after I replaced that. Seiko watch pressure tester repairs B. T., of Mudgeeraba, Qld writes: in the October 2023 Serviceman’s Log column (siliconchip.au/Article/15983), when recounting his adventures with his watch, Dave Thompson mentioned that he didn’t know how it was pressure tested. I may just be able to enlighten him! For many years, I repaired Seiko S451 watch pressure testers for jewellers and Seiko. The tester consists of a small pressure chamber surrounded by four PCBs and a separate air pump, similar to those used to inflate car tyres. The watch is placed face-up on a small holder inside the pressure chamber, and a very light lever rests in the middle of the watch’s glass face (crystal). A somewhat fiddly adjustment positions the watch until a front panel LED is illuminated. When the start switch is operated, the compressor pressurises the chamber to about three bar. This causes a good sealed watch to deform slightly; the crystal domes inward a little. A leaky watch does not deform; if it does, it will not maintain the deformation as the pressure inside the watch equalises with the pressure outside. The deformation is sensed by a phototransistor illuminated by a light-emitting diode. The lever resting on the crystal controls the amount of light the phototransistor receives. The electronics notes the position of the lever at the start of the test and compares this reading with that obtained when the chamber is pressurised and again after a delay of a minute or so. If it passes all the tests, the watch is declared “Acceptable”; otherwise, it is “Not Acceptable”. Most of the faults in these units were pressure leakage due to worn hinges or locks on the chamber door, poorly adjusted pressure switches that control the compressor, problems with the watch supporting platform etc. Occasionally, there was an electronic fault, but it was not common. One especially common fault was a blown-up compressor. This was caused by the fact that the unit had a power transformer that allowed it to operate from 100-110V AC (Japan/USA) or 220-240V AC (Europe/Australia). Unfortunately, the unit came with a US-type mains plug; most people used an adaptor to connect to our power. Left (p90): a PSU from a Maytag Neptune appliance. Left (p91): the faulty capacitor (red arrow) plus TOP256GW IC (yellow arrow) on the Speed Queen dryer power supply PCBs. Above: the Seiko S451 air pressure and water resistance tester. siliconchip.com.au Australia's electronics magazine June 2024  91 However, the compressor ran from 100V supplied by an internal transformer and fed to the compressor via a US-type three-pin socket on the back of the unit. It was therefore very easy to put the adaptor plug on the compressor cable and plug it into the 230-240V mains. The result was a spectacularly killed compressor. The compressor motor was a brush-type universal motor with a bridge rectifier in series with the 100V mains supply. That rectifier often saved the motor, as the bridge would rapidly spread itself over the inside of the case, and the motor often got off with just burn marks on the commutator and could be repaired. These units were very popular with jewellers for many years, long after Seiko stopped supporting them, but I haven’t heard of them for some time now. So I imagine the march of time has caught up with them. Repairing failed Li-ion battery packs B. P., of Dundathu, Qld writes: several years ago, my wife asked me to get her a battery-powered leaf blower to blow the leaves off the verandah instead of sweeping them. I purchased a 20V leaf blower on eBay for a reasonable price. It worked well for the purpose, but my wife asked me if I could get another battery for it so that she could use one while the other was on charge. A spare battery cost almost as much as the leaf blower, so I just bought another leaf blower. That way, once the original leaf blower reached the end of its life, we’d have another one to use in its place. That proved to be a good move, as some time later, the wire on the charger’s plug broke and I didn’t have a plug of the correct size to replace it. I ordered some plugs on eBay and we used the spare charger while waiting for it to arrive. Once the plugs arrived, I fitted one to the original charger and returned it to service. All went well for a few years until one of the batteries died. Removing four screws let me open it up. I checked the five 18650 cells and found that one was dead. I looked on eBay for a new battery, but they were no longer available. Replacement 18650 cells were very expensive, with five cells costing as much as the original price of the leaf blower. I remembered that I had a dead laptop battery that used 18650 cells, so I checked the cells in it, finding three that 92 Silicon Chip still had good voltages. I removed one of those cells and fitted it to the leaf blower battery. Once it had fully charged, my wife tried using the leaf blower with this battery, but she said it only lasted a minute and then stopped, so the replacement cell was no good. It was back to one battery again, but this situation only lasted a few months, until the second battery died. I dismantled the battery and found that one cell was dead again. Then I had an idea. I took one of the good cells from the first dead battery and fitted it to the second dead battery in place of the dead cell. This got the second battery working again, but it still meant we only had one good battery. I looked on AliExpress and found 18650 cells a lot cheaper than eBay, so I ordered 10. I also ordered some nickel strips on eBay so that I’d be able to fit the new cells when they arrived. The nickel strips arrived, but the cells did not. I followed the tracking for the cells, which showed they had been delivered in Sydney. How was that possible when we live in Queensland? My son said there is an almost identical address there, so someone couldn’t read the postcode! I got a refund, then ordered another 10 cells from a different seller at a slightly higher price. Unfortunately, in this case, the tracking number did nothing. I was getting concerned, but the cells arrived after 13 weeks. I used my 80W soldering iron to solder the new cells in place, then reassembled the battery and put it back into use. Unfortunately, the replacement cells were junk, and the leaf blower would only work for 20 seconds on high speed. It still worked on low speed, though. So we had one reasonably good battery and one that was of very limited use. Sometime later, another cell died in the ‘good’ battery, so I again replaced it with one of the leftover good cells from the second battery. Not long after, my wife said that both batteries were dead and the leaf blower no longer worked. I searched again for some decent 18650 cells and came across Tinker brand 3400mAh cells. I’d never heard of this brand, but the specifications suggested they should be good, as their weight was about the same as the original 2200mAh cells from the leaf blower battery. There were several five-star reviews on them, and some of the reviewers had done discharge tests and confirmed that the cells were what they claimed to be. Australia's electronics magazine siliconchip.com.au The leaf blower’s battery pack had died, so some replacements were sourced, which can be seen in the adjacent photo. They were rated higher than the originals (at 3400mAh) and worked well. These cells are available in Australia and come with a satisfaction guarantee or a refund. Had I finally found some good 18650 cells at a reasonable price? There was only one way to find out. They cost $7.55 each, with a 10% discount for buying 10 cells. I decided to order 10 cells, and they arrived quickly. I once again repacked the battery. Before charging it, I did a quick test by fitting it to the leaf blower to ensure the battery was in working order. I switched on the leaf blower and nothing happened. I got the other battery, and again, nothing happened. I got out the spare leaf blower, and both batteries worked in it. So now the original leaf blower no longer worked. I put the newly repacked battery on charge and, while it was charging, decided to dismantle the original leaf blower to see what was wrong with it and if it could be repaired. I removed the screws and split the case apart. It was apparent why the leaf blower no longer worked, as one of the wires had come off the switch. I got my 20W soldering iron out, soldered the wire back onto the switch and reassembled the leaf blower. Sometime later, the newly repacked battery was charged, and the repaired leaf blower was ready to use with the ‘new’ battery. I asked my wife to evaluate the performance of the leaf blower with the siliconchip.com.au replacement cells and see how it went and how long it lasted. She reported that the leaf blower now worked better than when it was new and the battery lasted at least 50% longer than it had done initially! That was an excellent result and it worked out at just under $37 per repaired battery. We also have a Hoover Linx vacuum cleaner. Last year, the battery died and I purchased a replacement battery on eBay for $57, but we still had the old battery. I took the old battery apart and, sure enough, it uses five 18650 cells. Unfortunately, I hadn’t thought of this old battery when I ordered the cells for the leaf blower, or I would have ordered 15 cells. But that is another job for another time. 20W Class-A Amplifier repair The Silicon Chip Class-A amplifier module first appeared in the July & August 1998 issues as a 15W module (siliconchip.au/Series/140). In May-August 2007, a 20W version was described (siliconchip.au/Series/58), and in September 2007, instructions for a complete stereo amplifier were published. J. G. of Bendigo, Vic built the 20W stereo amplifier from an Altronics kit... I modified it slightly in 2011 based on changes made in the later Ultra Low Distortion (Ultra-LD) Mk2 and Mk3 amplifiers, and it has performed outstandingly. However, when I powered it on recently, there was no sound from either speaker. The speaker protection relay did not appear to operate at power-on or power-off. The likely problem was a DC offset on the output of one channel of over 2V, triggering the DC offset protection. I removed the lid, powered it up and checked the module outputs with the power on. One channel settled quickly to around 40mV, while the other started at around +9V, dropping rapidly to +4V and slowly reduced to around +3V. I checked the DC voltages on the amplifier PCB against the published figures. The supply voltage is ±21V, not the ±22V of the original design, as a choke filter was installed in the power supply to reduce transformer buzz. This increases the time the diodes are conducting by storing energy in the choke, reducing the peak current drawn from the transformer. The downside is that the maximum output power is slightly decreased. Australia's electronics magazine June 2024  93 All the measurements I made were close to the published figures. The bad channel was amplifying a sinewave cleanly but clipping on the positive cycles. Given that the amplifier was working, the problem appeared to be a voltage mismatch in the input circuitry, with the output voltage offset developing to compensate for it. I disconnected the bad channel PCB to check the input transistors (Q1 & Q2) as I suspected a fault in those. Testing them out of circuit with a Peak DCA75, the Vbe figures were almost identical, and the gains were very close at 225 and 236. It was good to see they were closely matched after many years; however, they were clearly not the problem. Bizarrely, I found that the output voltage of the bad module was -0.45V with no power applied, while the good channel measured 0V as expected. The voltage was originating from the capacitor connecting the 510W resistor to the base of Q2, which ironically exists to reduce the amplifier’s DC offset. It is nominally 220µF but was replaced with a 1000µF capacitor as part of the 2011 changes. Compared with the same capacitor in the good channel, the top was raised slightly, a symptom of faulty capacitors manufactured from 1999-2003 (‘capacitor plague’). This capacitor was from my junk box and could have been manufactured at any time. It seemed to be suffering from a chemical reaction, causing pressure in the can and making it act as a battery. The voltage across it measures -0.5V with no load. With around -0.4V on the base of Q2, the output voltage had to go positive to compensate and drive the base voltage to +0.1V, to match the input signal. With the amplifier gain of around 20 times (20kW/510W), that -0.5V difference was amplified to about 10V. After reinstalling Q1 and Q2 and replacing the faulty 1000µF capacitor (as well as the same type in the other channel), both outputs were back to <50mV DC offset, and the amplifier is working well. USB sound card micro repair P. P., of Prospect, SA had to dive into a repair at precisely the wrong time, when he had lots of work to do, but couldn’t because his measurement device was broken... Isn’t it strange how things break exactly when you want to use them? It seems to be a rule of nature, similar to how, when you are searching for something, it is always in the last place you look. The logical inevitability of these sayings is of little comfort when you are in the middle of such a disaster. I was about to make a bunch of measurements using my audio test system and, well, nothing was working. This system uses the Silicon Chip USB SuperCodec (August-October 2020; siliconchip.au/Series/349), which has a tiny USB-toI2S (digital audio) converter embedded in it. The PC this plugs into was not finding the interface card, which foiled any hope of making the tests. I was in a hurry and had an extensive list of other tasks to get to, and here I was with the first task foiled. Because this was a PC-based test system, the logical assumption was that a Windows update had broken something, so I needed to reboot and check the drivers. One reboot later and the PC still sat there telling me that nothing was plugged in. After a few minutes of futile plugging and unplugging, I conceded that this laptop would never find the Codec. So I packed the whole lot up and moved to another computer, but it still wasn’t detected, confirming that the problem was the Codec. This was not good news, as I didn’t have a spare one; the I2S interface costs $140 and takes a week to arrive. My stress levels were increasing as I really wanted to get these measurements done. There was no option but to pull the thing out and look for obvious faults. The problem is that this card is tiny and loaded with M1608/0603 parts (1.6 × 0.8mm!) and a 0.5mm pitch IC with many pins. With repeated plugging and unplugging, I noticed one occasion where the PC complained that the USB device had failed. I took this as good news, as it meant that something was working sometimes. But what was causing this intermittent behaviour? As many of us have, I sat there looking disconsolately at a PCB loaded with hundreds of bits that I could only guess the purpose of, wondering where to start. I drank some coffee and had a think. My logic was that the computer only needs to talk to the processor (an XMOS IC) on this card for it to be registered in Windows, so I should check the USB cable, connector and any bits between that The audio interface board plus a close-up photo near the USB Type-B connector (marked with a red arrow on the lefthand photo). That marked transformer had a dodgy solder joint that was not clearly visible at a glance. 94 Silicon Chip Australia's electronics magazine siliconchip.com.au and the XMOS IC. The occasion where it almost worked convinced me that the fault was not catastrophic. Swapping USB cables took that as a cause off the table. Poking with a meter showed that the USB connector was fine, and I could get conductivity to the IC. I moved my attention to the soldering of the XMOS IC to the PCB, as some leads looked less than perfect. My usual check is to poke each lead with the tip of a sharp knife. Bad connections are really obvious as the leads bend very easily. While no leads appeared to have completely failed, some leads were clearly just soldered. So out with the iron, and with liberal amounts of flux, I reflowed all the pins on the XMOS IC. As a tip for those new to the service game, running a sharp knife along a row of SMD leads is a great way to find unsoldered/dry joints; the leads ‘jump’ as you go over them. I plugged the board back into the PC with its freshly soldered XMOS IC. The PC’s insistence that the board still didn’t exist increased my stress to the ‘muttering curses’ level. At this point, I purchased a new USB to I2S card, figuring that the sooner I ordered it, the sooner it would turn up. Just before I threw the presumably dead part in the bin, I took a peek through a microscope and noticed something a little less than perfect on the USB data line transformer (between the IC and USB socket). This is the only connection between the USB connector and the XMOS IC. I should have started there, as it is a really critical part of this device and not in a great spot for reflow soldering given that lumbering great connector near it. I gave it a squirt of freezer spray while the board was plugged in; nothing happened. I was one second from unplugging things and binning it, but as a last gesture, I poked the soldering iron on it (yes, while it was plugged in and powered on, which is bad form indeed). The PC found the card and a blue light came on! I sprayed it with freezer spray and it disappeared. Two minutes later, I had rather brutally reflowed the joints on that tiny transformer. Dodgy SMD joints can be really hard to find, not least because they are small but also because it is fiddly to rework them, and the actual fault can be underneath a component. I could not see the cracked joint, but I was able to demonstrate its presence, which was enough for me. Now I could freeze and heat the board, and it remained connected to my computer. So, a couple of hours later than planned, I had the Codec running again and was off to make the measurements I needed. My blood pressure was also coming down, and I was speaking English again. I also have a $140 spare card on its way as a lesson not to buy expensive spares until all reasonable courses of action have been taken! The internal temperature in the oven can be quite high due to heat from the magnetron (at 70% efficiency, 300W is dissipated). The capacitor also heats up due to its internal resistance. The capacitor is rated to 85°C; its polypropylene dielectric insulation resistance drops significantly with temperature. Measuring the voltage across the capacitor with an oscilloscope shows peak voltages exceeding 3000V during operation, so the capacitor is stressed by both voltage and temperature. The capacitors that failed were all made by BiCai in Ningbo, China. They use polypropylene insulation, and the volume price of the capacitors is about US$2 each. One solution is to limit the cooking time in summer. Alternatively, you can buy a 3000V AC capacitor at a higher price from the USA. Another microwave, a Sharp R395Y inverter oven, was tripping the mains supply circuit breaker during operation. I replaced the inverter’s insulated gate bipolar transistor (IGBT), type 40T321 (40A, 1500V), along with the protective gate-to-emitter zener diode and resistor. When the oven was tried again, the inverter failed again, indicating a faulty 2M368H(L) magnetron. With a new IGBT and another magnetron (I had a 2M319 on hand), the oven would still not heat. I then tried yet another magnetron (2M339) and finally achieved success. I measured the magnetron voltage at 6kV (magnetron disconnected) and 3.7kV with the oven at full power. So the faulty 2M368H(L) magnetron caused the inverter failure. I also had a faulty 2M319 magnetron. The mounting holes are different for the 2M368H(L) versus the other magnetrons, so I had to drill some new holes. I will now get a replacement magnetron of the right type for the oven. What is the difference between the magnetrons for inverter versus non-inverter ovens? Many magnetrons are similar. I tested the 2M386H(L) with a Megger and it broke down at 1000V. The cost of a new magnetron exceeds $300 and the inverter cost is similar. You can buy the magnetron on eBay for about $100 but not the inverter. Buying a new oven is cheaper than replacing both parts; an example of planned SC obsolescence. Microwave oven repairs R. S., of Fig Tree Pocket, Qld has repaired many microwaves and is familiar with many of the more common failure modes... This Breville BM0735 BSSANZ microwave oven (non-­ inverter type) has a voltage doubler circuit using a 1µF 2100V AC capacitor. If the oven is run for 10 minutes or more in summer (ambient temperature of at least 28°C), the capacitor can short-circuit, blowing the high-­voltage fuse. siliconchip.com.au During summer these HV 1μF capacitors were shorting in my microwave oven after extended use. Australia's electronics magazine June 2024  95