Fullscreen HTML5Med Res (??MB)HTML4

SERVICEMAN’S LOG Relating a range of rambling repairs Dave Thompson Dave has been recruited by a shadowy organisation currently attempting to master the art of underwater sheep herding. While he is on an intensive four-week course learning to speak dolphin, we have a few stories from readers. Regular service resumes next month. My work laptop is connected via gigabit LAN. Unfortunately, there is only one spare LAN port in the rumpus room, so if I need to use my private laptop, it has to rely on WiFi. We have two access points that are reasonably centrally located on the ceilings of both floors of the house. When all is well, we get usable transfer rates of 300Mb/s. Recently, I was using my personal laptop to run a Microsoft Teams session to communicate with my coworkers on Brisbane’s cross-river rail project, located in the Brisbane CBD. I found that my laptop could not connect to the WiFi, so I had to resort to using a mobile phone instead. After the session finished, I set about determining the cause of the problem. Initially, I suspected the laptop because the WiFi driver had been reinstalled recently, but I noticed that my phone was not connected to WiFi either. I went to the downstairs access point and saw that none of its three indication LEDs were lit. The hardwood floor and a few plasterboard walls do a really good job of blocking both the 2.4-2.5GHz and 5.25.9GHz WiFi signals. Our network switch powers the access point 90 Silicon Chip via power-over-Ethernet (POE). Disconnecting and reconnecting the network cable to force a reboot did not produce any joy. The network switch is a second-hand enterprise-grade item (Cisco C3560X-24P), capable of supplying 30W from all 24 ports simultaneously. I tried another port on the switch, in case its POE hardware had failed on that port, but that also failed to make a difference. A final check was to put a basic continuity tester on the ends of the patch lead to the switch and the patch lead to the access point. This proved the patch leads and the house’s fixed wiring were good. Alexandra Hills is less than 4km from Moreton Bay as the crow flies, and we are on reasonably high ground, which results in salt corrosion. We had to replace some of the RJ45 sockets that were installed in the early 2000s, before WiFi was affordable. By now, it was reasonably clear that the access point had failed. The access points require 15W (17W peak from Cisco’s data sheet), so they run reasonably hot. The oncewhite plastic housing is now very yellowed and, in places, verging on brown. My initial thoughts were that I might get lucky and that failed electrolytic capacitors could be the cause of the problem. I opened the case by removing the four Phillips head screws concealed by rubber feet. This revealed a roughly square printed circuit board with five pressed metal antennas attached to the case. There were four aluminium electrolytic capacitors, with at least one showing signs of distress (a slightly convex end). The access points can be powered from a 12V DC adaptor, which had to be purchased separately. Because it was intended to use POE, no approved adaptor was available. After a quick look around the house, I found a potentially suitable adaptor. Using the adaptor with the access point connected to a switch without POE capability, it booted up displaying an amber power LED and two flashing green LEDs (LAN and WiFi). Checking the installation guide confirmed that the power LED is supposed to be green for POE and amber for 12V DC power. I forced my phone to connect to the access point by turning its WiFi feature off and on again while in close proximity to it. Using the access point’s web interface, I verified that the phone was connected to that access point. Now there was a realistic prospect that the access point was repairable. Australia's electronics magazine siliconchip.com.au Items Covered This Month • Repairing a Cisco WAP371 access point • A recurring fault • Failures in bench grinders • ... and another problem to grind • Fixing a Ryobi electric lawn mower 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 The electrolytic capacitors are all through-hole components, so I needed access to the other side of the PCB. There were no retaining screws for the PCB. It was located in the dished top of the case (when ceiling mounted) by bosses that prevented contact with the screws retaining the relatively flat lid. The holes in the PCB were visually larger than the bosses, so it was reasonable to expect that the board could be removed without any significant force. Before proceeding, I disconnected the three black coax cables to the antennas from the PCB. The remaining two grey coax cables were directly soldered to the board, but it looked like it would be possible to flip the PCB over without disconnecting the cables; this was a big mistake. The board proved to be a very tight fit on the bosses and required some leverage to release it. It came free with a jerk that broke one of the antennas off the tiny plastic spigots retaining it. I desoldered the other antenna, freeing the board from the case. The next mistake I made was not immediately desoldering the antenna that was still attached to the board. The coax braid was severed during subsequent testing, and the repair required the cable to be shortened and stripped for re-termination. Given that the coax has an outer diameter of less than 3mm, it was a challenging task. Examining the board, the circuitry associated with the LAN side of the power and communication circuitry could be clearly identified due to a several-­ millimetre-wide band siliconchip.com.au of translucent board substrate separating it from the rest of the circuitry. The band was bridged by the switch-mode power supply transformer, an opto-isolator (for voltage regulation feedback), the LAN transformer and several very chunky surface-mounted ceramic capacitors. Only the electrolytic capacitor associated with the LAN side of the power supply tested good in-circuit. The capacitor that looked likely to be the filter capacitor on the secondary side of the power transformer (CP9) measured as a short circuit. I recorded the capacity and voltage ratings of the capacitors in preparation for their removal. Removing the three suspect capacitors was not particularly easy, even with a professional vacuum desoldering tool. The use of lead-free solder, large ground planes and possibly a multi-layer board meant that a lot of heat and time was required to melt the solder. The process was aided by applying some additional lead/tin solder to improve the heat transfer. After removal, all the capacitors failed outof-circuit testing. The bad news was that there was still a short across CP9’s pads, even without the capacitor fitted. There were several reasonably large surface-mounting diodes near the secondary side of the transformer, all of which passed basic diode tests. Closer still to the transformer was an 8-pin package (QP3) labelled 9476GM, which looked like it should be an IC. A web search found a data sheet for a 60V 7.8A Mosfet in an 8-pin SOIC package. There was a very low resistance between its source and drain connections and the pads of CP9. At this point, the penny dropped; the power supply was using synchronous rectification to improve efficiency. Removing QP3 using a hot air rework tool eliminated the short across CP9’s pads. Out-of-circuit testing of the Mosfet indicated a high-quality source-to-drain short circuit. An internet search for a supplier of a direct replacement proved fruitless but a filtered search on element14’s website for the package and Vds rating came up with the SQ4850CEY as a potential substitute (rated at 60V, 12A). Additional checks on its Vgs threshold, on-resistance and maximum permissible gate-source voltage confirmed it as a viable substitute. I ordered that Mosfet plus some replacement capacitors, all low-ESR, 105°C rated parts from the Panasonic FN series. The rest of the repair was reasonably painless. I used hot-melt glue to retain the antenna that had broken free during dismantling. The repaired access point appeared to work normally. The only peculiarity was that when the access point was returned to its normal location, it would not work. The switch diagnostics claimed that the switch was working normally. However, the switch’s log file revealed that the relevant port had detected a current overload on many occasions prior to the access point being removed for repair. After rebooting the switch, the access point worked on the port to which it was originally connected. It is possible that the switch has an undocumented feature that causes it to give up trying to supply power after a large number of overcurrent events. Australia's electronics magazine January 2025  91 As a precaution, I have replaced the capacitors in the upstairs access point. This was an interesting learning experience and helped justify acquiring quality soldering tools when I retired. Replacing them with comparable WiFi 6 access points would cost around $600. We don’t currently have any devices that would benefit from WiFi 6 (802.11ax). D. H., Alexandra Hills, Qld. Intercom woes and a recurring test equipment fault I used to work as an RF technician for a commercial TV station in Brisbane, before and during the transition from analog to digital terrestrial TV. One day, the chief engineer asked me to fix the intercom on the transmission tower. It was an Aiphone brand installed by a separate company several years ago, before I commenced working there. There were handsets in master control, the base of the tower and several platforms up the tower. Even though we had VHF radios, and ‘phones for comms, it was needed as a backup. Since its installation, it had been very noisy and basically unusable. That was put down to the interference from all the RF floating around on the tower. There was the main VHF TV transmitter, various radio base stations, microwave links etc. Intermodulation products could also be present from various RF sources mixing together on the large metal tower. There was no documentation available for the installation, just a basic Aiphone user manual that was a couple of pages, with some basic wiring, showing connection with an AC adaptor for power. I just had my trusty Fluke multimeter, so I thought I would start at the handset in Master Control, as it was inside, out of the weather. As with other fault-finding, I decided to check the power supply first. When I opened the cover, there was a terminal strip with several unlabelled white wires. There were also two white wires connected to the only marked terminals, identified as + and −. When checking power supplies, it’s good to take a reading with both the DC and AC ranges to see what is going on. The result was a surprise; I measured 13V AC and basically no DC, when it was clearly labelled DC! I was expecting DC with maybe some AC ripple. Now the problem was: where was the power supply? Luckily, the station electrician remembered that it might be in the switchboard at the base of the tower. With his help, we removed the cover panel and found a Bell transformer that was the power supply we were chasing. The wiring matched, and it was definitely putting out 13V Servicing Stories Wanted Do you have any good servicing stories that you would like to share in The Serviceman column in SILICON CHIP? If so, why not send those stories in to us? It doesn’t matter what the story is about as long as it’s in some way related to the electronics or electrical industries, to computers or even to cars and similar. We pay for all contributions published but please note that your material must be original. Send your contribution by email to: editor<at>siliconchip.com.au Please be sure to include your full name and address details. 92 Silicon Chip AC – which did seem to match that Aiphone diagram, but I think there was confusion about the designation. Because of its location, rather than replace the transformer, I installed a bridge rectifier and a couple of big electros in a Jiffy box. The system performed perfectly now on a DC supply – there was no RF interference! The chief engineer was happy, and I earned a pay rise over it! I have a second repair story. Silicon Chip or EA published a couple of component checker adaptors for CROs. They were basically a low-voltage AC plugpack with a resistive voltage divider to deliver 1V AC to the probes of the oscilloscope in X/Y mode. They were really useful for testing components in unpowered equipment without removing them. It would quickly show on the screen if a component was OK. For example, a diode would give a hockey stick shape, capacitors would be ovals, resistors a diagonal line etc. That was great if you had a known-good board and a bad board; you could quickly compare the waveforms between the same component on the two boards. We used a Hewlett Packard 5342A microwave frequency counter for testing TV microwave links. It was used in conjunction with N-Type 20W dummy loads, N-Type pads and a DC blocker. Every couple of years, it would stop working and have to be sent out for repairs. The boss blamed us for not being careful enough when using it. When it died the last time, the boss decided the company had spent enough money on repairs and ordered a new model. As the older unit was now destined to gather dust on a shelf, I thought I might as well have a look at it. I got the repair information from the last time it was sent away – it stated: replaced 12V regulator and performed calibration – $5,000! It had several power supplies, including 12V DC and 5V DC outputs. I removed the 12V regulator and it was indeed dead. Before replacing it with a new one, I checked the load with an ohmmeter with the unit unpowered. The 12V rail seemed to have a very low resistance to Earth. The unit had about eight PCBs plugged into a motherboard, with a diagonal black line across the top so you can see instantly if they are in the correct order. I unseated them one by one and found one board that was the culprit. I found the track leading from the 12V rail on the edge connector. It branched off into several directions, and I was without any circuit diagram etc. I wouldn’t recommend it, but I made one or two cuts in the track with a Stanley knife to isolate the problem. There was a tantalum capacitor not far from the edge connector. Upon removing it, I found it was nearly a short circuit. I replaced with a good one, replaced the regulator and repaired the tracks. The unit powered up fine; I didn’t do a “calibration”, but I compared with the new machine, and it measured within a fraction of a dB. It was still very useful for vehicles and choppers etc in the field. I’m retired now, but the unit was still going when I left. I don’t know if that capacitor was the problem all along; I have found faulty tantalum capacitors before in other equipment, especially after a power surge. I would always be wary of using them, especially when they are close to a power input. A. G., Jindalee, Qld. Australia's electronics magazine siliconchip.com.au Component lead failures in bench grinders I recently received two small, identical 3-inch (76mm) bench grinders after they stopped working. They were a generic brand out of China meant for hobby use, for grinding and polishing. Someone with a mechanical bent had opened them up and pronounced they had black spots on the circuit boards; something that was beyond his skills to repair. They were passed to me as someone who knows about such things! Inspection of the internals showed them to be quite wellmade with appropriately wired and insulated connections. Apart from an On/Off switch, there was a starter capacitor and a small speed controller PCB. The board had a few components around a variable resistor and a three-legged semiconductor. Fortunately, no attempt had been made to remove component numbers during assembly, as is often the case, so I could see the three-pin device was a BT137 Triac. All fairly standard stuff, I thought. Close inspection of the boards showed the central or anode pin of the Triac was damaged on both boards. In the first instance, the lead was open-circuit where the right-­ angle bend had been made in the lead to allow the Triac to lie flat on the board after soldering. No obvious cause for this was evident, other than perhaps damage caused when it was bent. I soldered a short piece of wire from the stump of the lead to the board. On application of power, the grinder worked again. The second grinder had a slightly different fault. The anode lead was intact at the bend, but where it entered the board, the hole was blackened and the lead had broken where it made contact with the solder in the hole. The broken lead was still embedded in the solder on the reverse of the board, with no sign of any soldering defects, such as a dry joint. The blackening was probably the result of arcing after the lead had broken. I cleaned up the solder, remade the connection and tested it out. This time, the grinder ran, but there was no speed control. siliconchip.com.au I replaced the Triac and, to be safe, I also replaced the Diac. The grinder then successfully ran with speed control. Why these leads broke is a mystery. My thought is that they may have fatigued due to vibration. The grinders are not that well-balanced and, at high speed, they vibrate noticeably. The Triac is not secured to the board other than by the leads, which may have put stress on the shortest lead as it shook while operating. Unfortunately, I could not drill the board and secure the Triac by its mounting hole, as there were tracks on the reverse side of the board in that area. Time will tell if my theory on vibration is correct. N. D., Ocean Beach, WA. Another problematic grinder! I was using my Ferrex 125mm angle grinder with a 1mm cutting disc to cut some roofing sheets when it suddenly stopped. I’d had this grinder for a year and I hadn’t had any problems with it until then. Had the power or extension lead failed? I plugged the grinder directly into a working power point, but it still didn’t work. I removed the brush cover and then the side cover. I could see the brushes were still in good order with plenty left on them. I used my multimeter to test and there was no problem there. Next, I tested the switch. While holding the power button in, I checked for continuity between both sides of the switch and there was none, so that was the problem, the switch had failed. This was an Australia's electronics magazine January 2025  93 unusual type of switch, an SPST momentary rocker. I did not think I would have one in stock, but I checked anyway. I went through my box of switches and I had many different types, but nothing remotely resembling this one. I wondered if I could repair the switch, so I took it apart. The fault was obvious. The tiny contact had burnt. This switch is rated at 30A 32V DC and 16(14)A 250V DC. That rating is a figment of someone’s imagination because there is no way that tiny contact could carry that much current. No wonder it had failed after just a year of occasional use. A fine file quickly restored the contact, but there was no way I could reassemble the switch. It was obviously assembled by a robot because there is no way a human could put it together with all the small parts in it. So it was time to find a replacement switch. I thought I would ring the service centre number listed on the grinder. The person I spoke to said he doubted they would have internal parts for the grinder, but he would check and get back to me, so I left my email address for him to contact me. He said that they don’t have the switch. No surprise there, as so many things these days are designed to be thrown away and not repaired when they break. While searching with Google, I spotted the exact same switch from Altronics for $3.35. That was better, but the postage was between $10 and $13, so that killed that idea. However, my son mentioned that he would be going to Brisbane that day, and it just happened that he would be driving right past the Virginia store. He said he could pick up my order, so I ordered three switches (so I would have two spares). The way it was originally put together, it had crimp 94 Silicon Chip terminals to join the Active wire from the switch to the power cable, the switch to the motor and another crimp terminal from the Neutral wire to the motor. I didn’t like the idea of these crimp terminals. As the cable had some minor damage (not affecting its safety), I decided to replace it along with the switch. There is no actual cable clamp, as the cable is held in place by the moulded cable flexible strain relief. I started by pulling out the wires, then I used a drill bit (by hand) to remove the outer section of the original cable. I forced the strain relief over the new cable and used superglue to secure it. That works really well. Next, I checked if the replacement switch would fit. Luckily, it fitted easily with no modification needed. I connected terminals to the wires to avoid soldering the switch, as I was not sure if the plastic would melt if I soldered the wires to it. I did away with the crimp wire joiners and instead soldered the wires and covered the joints with heatshrink tubing. This is the only power tool I have come across with this type of joiner, and it’s a reflection of the quality of the grinder. All my other power tools have wires long enough to connect directly to the switch, and I think they all use DPST switches as well. The accompanying photo shows the inside of the switch area of the angle grinder after replacing the faulty switch. The broken switch and the crimp connectors can be seen above the motor. Like many power tools these days, this grinder is double-­ insulated and so has no Earth wire connection. The replacement cable I used was three-core flex rated at 10A, the same as the original cable. As the Earth wire was not used, I cut it off. This was a spare cable I had saved from something no longer in service. I reassembled the grinder and tested it, and it was once again working. I put it back into service and I’ve been using it for several days now. Even though this was just a cheap angle grinder, it was worth repairing it, as it was only the switch that needed replacing. There is some degree of satisfaction in being able to repair something that is unrepairable because spare parts are not available for it. Of course, a balance has to be struck in that it can’t cost more to repair something than what it’s worth. Otherwise, it’s better to just replace it. In this case, I spent $3.35, a bit of time and a bit of heatshrink tubing to repair a $30 tool. This is not the first time I’ve repaired a power tool when spare parts were not available for it. I have an XU1 angle grinder that wore out a brush in the motor and I could not get a spare part for it. However, I managed to track down a replacement brush on eBay in England and repaired the grinder and after several years; it’s still being used. I also find replacement brushes on eBay when spare brushes are not available. I have lost count of the number of devices I’ve been able to repair and get back into working order at minimal cost. B. P., Dundathu, Qld. Ryobi lawn mower repair I’ve fixed a lot of petrol-driven garden products in my time. When petrol engines are running, they’re great. However, they can be painfully difficult to start, especially if you don’t run them often. Australia's electronics magazine siliconchip.com.au Years ago, I had a petrol chainsaw that I rarely used and it would always take ages to get going. For that reason, I used it less and less, so in the end I never really used it even when I really needed it. On impulse one day I bought a mains-powered chainsaw at auction and have never regretted it – you take it out of the cupboard, make sure there’s chain oil in it, plug it in and start sawing. Electric lawn mowers are nothing new – those Flymo mains powered mowers were around when I was a kid, but I always wondered how many minutes it would be before I ran over the power cord. Battery mowers have come a long way. A friend raved about his 36V Ryobi mower when it came up in conversation, so when I drove past a Ryobi battery mower in a council clean-up, I immediately pulled over and had a look to see if was worth taking. It all looked pretty complete except for the key, so I threw it in the back of the car and took it home. I wanted to try it out. I own a few Ryobi 18V power tools; being able to swap batteries between many different tools makes battery management much easier. This mower turned out to be an 18V product, which was perfect for me, even if it wasn’t a 36V one like my friend’s. The first thing I did was bypass the ‘key’. Battery mowers all seem to have a removable key that allows you to disable the motor – I expect this it so that toddlers can’t put anyone in danger, including themselves. Luckily for me, there are no toddlers living at my house, just the cat, and he doesn’t like mowers at all. The key just consists of a removable short circuit on a couple of 6.3mm QC spades – it’s probably a blade fuse in a special moulding. I used a fuel pump relay bypass switch I made when I was trying to get my classic car engine to start. With my switch on and a battery in the socket, it was no surprise that the motor wouldn’t run when I pulled the run lever on the end of the handle bar. The lever felt pretty floppy, and I didn’t think it was doing anything. However, I decided to open the motor section and have a look at what was under the cover. It wasn’t too hard to open; just half a dozen or so Torx screws, all the same size. It took a few minutes to find the two underneath. Once I had them all out, the lid came off and I could see a motor and a separate electronic controller. The wiring was pretty straightforward, with a pair of small gauge wires from the controller running up to the switch. Apart from a few blades of grass and some dirt, it all looked good. I disconnected the plug to the run switch and was very happy that when I shorted out the connector pins on the controller with a piece of wire, the motor started. So the problem was in the handlebar switch, or the wires to it. The wires looked OK and, as I mentioned before, the switch lever felt a bit floppy, so I took to the switch mechanism with the same Torx driver. Like the base, it came apart pretty easily and I could see how it worked. Two hands are required to operate the switch – there is a switch plunger pushed by the lever, plus a button you have to press to enable the plunger to move. I found I could run the motor by manually activating the button and plunger directly on the switch. So what was wrong? The button was releasing the plunger to move OK, but for some reason the lever wasn’t pushing on the plunger. You have to be a bit patient with these mechanisms because you can never see them operating when they’re assembled. I thought perhaps the switch mount had broken and the switch had moved back, or something had broken off the lever. It took a bit of looking, but eventually, I found a threaded hole in the end of the lever. I think there had originally been an adjustment screw that has fallen out at some stage. I found a self-tapping round-headed screw about the same size in my scratch box, and without much trouble, soon had the mechanism operating properly. That was it. After reassembling the switch and putting the cover back on the motor, I mowed until the battery went flat, with no problems at all. Going forward, I just need to figure out a better key/fuse arrangement. It works really well. The battery doesn’t last too long, and at 37cm, the cut is a bit narrower than the 46cm cut my petrol mower has, so it takes a few more passes. However, it’s really quick to get out and start mowing, and lightweight, so easy to push around. It would be great on a yard about half the size of mine. I think a 46cm/36V version might be the go, if I can find one... SC D. T., Sylvania, NSW. Left: the workaround to the missing ‘key’. Right: the internals of the Ryobi lawn mower. siliconchip.com.au Australia's electronics magazine January 2025  95