Fullscreen HTML5Med Res (??MB)HTML4

SERVICEMAN'S LOG Treadmill trials over trails Dave Thompson Being stuck inside for a long time, we’ve found that we (try to) use things that haven’t been touched in a while. Some of them have been sitting around for so long that they no longer work properly. In the case of our treadmill, the repair job provided more exercise than actually using it! At the moment, I’m only allowed to go out of my house to shop for essential supplies (though what constitutes essential is open for debate) or to walk or cycle for exercise. I have to say I’ve never seen so many people out and about; like us, they probably want to get out of the house to stave off ‘cabin fever’. It is sometimes so busy on the footpaths it is challenging to maintain the required 2m separation! Combine this with increasinglygrubby autumnal weather and walking has become a lot less appealing. Luckily, a few years ago we invested siliconchip.com.au in a good quality treadmill. However, like the vast majority of exercise equipment, after six months of solid use, we used it less frequently, and it now sits in the spare room gathering dust. To be fair, the treadmill isn’t totally unused; the wife uses the arms to hang washing on, and we store boxes of who-knows-what on the mat! Given the current situation, though, it seemed prudent to press it back into service. After a good clean, it looked brand new, even though it is going on for 10 years old. That’s the great thing about Australia’s electronics magazine equipment that typically doesn’t get much use. At least it stays in good condition! That said, we did do many kilometres on this one back in the day, though my motivation was more wanting to get my money’s worth out of it rather than personal fitness! This model is marketed under the name ProRunner; a brand likely dreamt up by the big-box company that sells these treadmills. It wasn’t inexpensive and is very well made, rock-solid and almost to the level of what you’d find in a fitness centre. It has done everything we’d asked of it, June 2020  61 so I considered it money well spent at the time. The treadmill stops running To prep it for use, I vacuumed all the dust off the frame and control panel and wiped down the belt surface. I also broke out the long-necked squeeze-bottle of silicone spray grease that came with the machine and as per the user manual, lubricated the deck and the underside of the mat. So far, so good; the machine was running as smoothly as ever, and the wife and I had several sessions over the following days. Then, a few days ago, as the wife was finishing her program and was in the cool-down phase, it shut down unexpectedly. The control panel flashed on and off about once a second, and each time it went dark, a beep sounded from the builtin piezo buzzer. Thinking it had simply ‘crashed’, she hopped off and turned the main switch off and on a few times, hoping this would reset it. There was no change; all she got from it was the rhythmic buzzing and blinking. She called me in, but there was little I could do. Full disclosure: I know as much about treadmills as I do about cardiothoracic surgery. That is, nothing. Well, I suppose that is not totally true; I know there’s a motor and a power supply in there, and likely some electronic jiggery-pokery going on up in the control panel and the two sections talk to each other, but that’s it. I’ve never seen inside one or viewed a circuit diagram. Like any serviceman though, I considered it my sacred duty to get in there and at least try to figure out what was going on! After a quick internet search, which Items Covered This Month • • • • Stuck in the house sans spare semis C-Bus home automation system repair Sharp R350Y microwave repair A Japanese fridge in Russia *Dave Thompson runs PC Anytime in Christchurch, NZ. Website: www.pcanytime.co.nz Email: dave<at>pcanytime.co.nz 62 Silicon Chip revealed little-to-no technical information about this make or model, I learned that some machines have an electronic reset somewhere. However, I didn’t recall reading about this in the user manual and a quick look over the panel and around the motor housing confirmed there was no breaker or pin-hole, or any other obvious reset mechanism. Opening it up There was nothing else for it but to open it up and see what I could find. Before doing that, however, I did my due diligence and looked further on the web. This proved frustrating; all I could find were outdated ads from the retailer or the odd expired listing for similar units on auction sites. An image search proved just as fruitless; there are many, many types of treadmill and all look much the same. So I had no choice but to break out the tools, get the covers off and see what I was dealing with. I did learn there are several components to consider; down at deck level, there will be a motor and a driver board for it. This sits (on our treadmill, at least) in the lower front section of the machine. Directly above that, in between the arms, is the control panel. Australia’s electronics magazine This is the part we mindlessly look at when we are slogging through the pre-set programs and it consists of a couple of displays, one LCD and another LED, and few rows of membrane-style buttons (one row for preset speeds and one row for degrees of incline). Similar controls are also on the ends of the support arms, with speed control on the right side and incline up and down at left. There are also exposed metal contacts on each handle that the user can hold onto, and the machine will display their heart rate, and from that, with speed and distance figures, the controller will estimate information like calories burned and an estimated time of when my heart will explode! Nothing extraordinary there, but relatively comprehensive compared to some treadmills. The problem could lie with the display/control panel up at the top, or it could be the motor and its associated driver down in the deck housing. I started with the deck housing simply because I thought it more likely to be something to do with the motor and power supply. The large, moulded-plastic housing looked as if it would just pop off, but was held up with something I couldn’t siliconchip.com.au sink. I thought that this was where the problem lay, because this part does the lion’s share of the work, and likely wears out first. Plus, I read a few forum posts where ‘experts’ postulated that failed motor driver boards are the cause of most powered treadmill failures. The sticker on the motor states it is rated for 90-180V DC. That’s some serious juice, and given it has to drive the belt with someone weighing up to 140kg standing on it, impressive in itself. My research on the web revealed some generic circuit diagrams, but nothing matched this setup. However, it appears that most powered treadmills use very similar technology to control the motors, and given there are only three types of motor typically used in treadmills, and two of those types are relatively rare, I could safely assume this controller is a PWM type. The PCB assembly certainly looked very similar to images of PWM boards I found online. Taking the easy pickings see. I could move the housing a little, and unclipped it from two locations in the very front, but something was holding it together further back. I couldn’t see any screw holes, and went so far as to lift the machine up so I could check underneath for fasteners, but nothing was visible. A viscous problem I assumed it to be just strong clips holding it together, so I worked my fingertips in the gaps between the housings and applied increasing pressure, expecting it to let go, but it stubbornly refused to give. What a great start to my treadmill adventure. I couldn’t even figure out how to get it open! After much huffing and puffing and purple language, I eventually gave it some real salt and pepper, and it started letting go. One by one, six turret clips finally popped loose. When I manoeuvred the housing away from the deck, I could see what had been holding it up: glue. Big, opaque blobs of glue. Great, it was going to be one of those jobs. Someone, somewhere, must have thought that clips alone just aren’t good enough for our treadmills, let’s smother everything with glue and make it almost impossible to service! siliconchip.com.au A quick look around the motor bay proved my hypothesis; anything exposed or connected was slathered with a generous dollop of the stuff. And it isn’t like hot-melt glues and silicones I’ve encountered before; in some places, it is very hard and brittle and breaks away, while in other areas it rolls and stretches, making it very difficult to remove. I was loath to get my heat gun anywhere too close in case I damaged any of the other parts, many of which are plastic, so when I had to remove glue from anything, I resorted to picking away at it with my fingers. I had to admire the motor and controller assembly. The DC motor might be long and relatively small in diameter, but it is exceedingly torquey. One end of the armature drives the mat via a toothed belt and the other end boasts a plastic, segmented sensor wheel. What I assume to be an optical sensor straddles the wheel, looking much like a disc-brake arrangement on a bike. This sensor monitors the motor speed and feeds data back to the control panel and/or motor driver. The motor driver PCB is bolted to the metal floor of the motor bay on a solid, right-angled aluminium heatAustralia’s electronics magazine As the motor can be tested using a car battery or bench supply, that’s where I started. I first had to pry the glue off the connections, then wired one of my workshop power supplies directly to the motor. I dialled in some current and then gradually raised the voltage until the motor started to spin. I got up to about 15V, and as the motor was humming along nicely, I considered it to be serviceable. Avoiding (for now) the glue-fest that is the motor driver assembly, I took the path of least resistance and removed the upper control panel, which is only held in with nine PK-type screws. These are easily accessible from the underside of the panel and once removed, the whole assembly lifts out from the top. And of course, because all the flying leads, sockets and plugs that connect the panel to the rest of the treadmill are dripping with glue, it took a lot longer than it should. Once free, though, I could at least take it to my workbench, making it much easier to work on. I’d singled out the data cable from the motor bay to the control panel from the loom going up the tubular frame and found it connected via a four-pin plug. Two of the wires were June 2020  63 black and red, so no prizes for guessing where to connect a power supply for testing! With 12V applied from my bench supply, the panel lit up, and all the familiar displays were working aside from the main display cycling through several errors codes (probably relating to the lack of sensor and data connections). I wasn’t too bothered about what these error codes meant… yet. For now, all I needed to know was the display was working, so I reassembled it temporarily back into the treadmill and moved on. Everything so far was pointing toward that motor control board. Removing it was as simple as taking out the two machine screws holding the heatsink to the deck, and then, of course, prying all the glue off everything connected to it. On the bench, I began by removing the single huge 500µF 450V electrolytic and testing it, simply because it was the easiest potentially-suspect component to get at. It measured 0.05W ESR and 490µF on my Peak ESR tester. So no problem there. I then removed the semiconductors from the heatsink so I could more easily reach and test them. All are stacked side-by-side and clamped to the heatsink with strips of metal and screws and lashings of thermal grease. There is a bridge rectifier in a SIL package, a dual diode array in a TO-220 package and an IGBT in TO3P format, all clearly identified. The bridge rectifier tested fine, as did the diode array, but I couldn’t test the IGBT with my Peak semiconductor tester. So I had to use the diode test function on my digital multimeter. IGBTs are quirky things to test; with the negative lead clipped to the emitter, and the positive lead on the collector, there should be nothing, until a brief touch to the gate with the positive lead turns the transistor on. Then a measurement can be made across the collector/emitter junction. If the gate and collector are then shorted with a fingertip, the junction should reset, and the meter measure open circuit again. In this case, the only measurement I could get on any pin combination, with any lead polarity, turned out to be the forward-bias of the fast-recovery diode. Also-called a ‘freewheeling diode’, according to the 18-page 64 Silicon Chip datasheet, it is connected internally in ‘anti-parallel’ across the collector/ emitter junction and provides both faster switching recovery and inductive reverse current protection. A potential fix At least now I knew what could be wrong. After yet more internet searching, I found replacement IGBTs readily available from local suppliers and AliExpress, with the usual crazy price disparity. I also found a vendor on AliExpress selling replacement boards, identical to this one. Interestingly, they were meant for Reebok-branded machines in the USA, so it seemed that there was some badge engineering going on. They were asking a couple of hundred dollars, which isn’t too bad considering. A dead IGBT might be just the tip of the fault iceberg, and I could be wasting my time sourcing and replacing it rather than just swapping out the whole board, which given my lack of treadmill servicing chops, might also not be the problem! But sourcing anything from overseas would take at least two weeks, and while the IGBTs were available locally, they cost six times as much as the parts on AliExpress. But nothing is being shipped until we are out of level four lockdown, which will be at least two weeks away, so we are hammered either way. And by then we’ll likely be done and dusted with lockdowns and can get back to real walking, with the poor old treadmill being relegated back to hanging clothes and junk storage. This is one of those rare cases where I know what the problem is, but there’s no straightforward way to resolve it. It’s a disappointing end to the tale, but fear not, I shall order a replacement part just as soon as I can and relate whether that did the trick. Fingers crossed, it will. Clipsal C-Bus home automation system repair About 18 months ago, D. S., of East Melbourne, Vic purchased a house with a fairly large Clipsal C-Bus installation controlling all of the lights, blinds and sundry other things in the house. It’s a good thing that he is a retired electronic engineer, as it wasn’t long before the system started to malfunction... Australia’s electronics magazine The home automation system in our house includes multiple touch panels, many wallplate buttons, motion sensors etc. When I saw how complex it was after moving in, I decided to do some research on how the system worked, just in case something went wrong. All of the electrical devices to be controlled are wired back to three cabinets located next to the switchboard. The cabinets contain a mixture of DIN-rail mounted main units: three 12-channel relay units and six 8-channel dimmer units. The C-Bus system consists of an Ethernet-like pink cable that connects in daisy-chain fashion to all of the control inputs – switches, motion sensors and touch panels. All of the input devices are connected in parallel and are powered from the C-Bus. Although the C-Bus cable looks like an Ethernet cable and uses RJ45 connectors, it is not at all compatible with Ethernet. It uses a single, duplicated pair and the signalling is superimposed on the DC supply. The C-Bus power (nominally 35V) is supplied by some of the units; they can optionally contain a 200mA power supply, at extra cost. The installer works out the total power requirements of all the input devices, then uses the required number of powered units to meet that requirement, when paralleled. Up until recently, the system performed flawlessly, and my tinkering has been limited to minor reprogramming of the touchscreens for new LED lights. However, we came home one rainy night to be greeted by darkness. Cursory checks showed there was still power to the house, but the C-Bus system was completely out of action. A check in one of the cabinets containing the C-Bus main units showed the C-Bus status lights were all off, indicating a problem with the bus itself. Two of the 12-channel relay units were chattering away with their lights blinking randomly. This should have been a major clue but, you know... I disconnected the C-Bus cable from the top of one of the relay units (which is also the connection to the upper floor of the house). The chattering stopped, and the C-Bus status lights flickered back on. So, my immediate thought was that there was possibly a short upstairs, perhaps caused by the rain. siliconchip.com.au One of the two 12-channel C-Bus relay units with its lids off. I chopped an old Ethernet cable in half and made up a test plug. With a meter connected across the data wires, I checked the C-Bus voltage with only half the network running (34V) and with the full house plugged in (10V). The minimum acceptable voltage is 20V, so, it seemed my hunch could be right. However, a resistance check of the upstairs section showed around 27kW, which seemed reasonable. Was it breaking down with voltage applied? I tried connecting a 27kW resistor across the bus to roughly simulate the additional DC power loading of the upstairs section, and the system continued to run OK. That eliminated the Above: the power supply board had failed with two ‘dried-up’ electrolytic capacitors. power supplies as the problem, or so I thought. Anyway, after a fruitless day fiddling with re-connecting parts of the bus in the upper storey and finding that the system just became less and less reliable, I went back to have a closer look at the power supplies. The three relay units and two of the dimmer units have an optional power supply, so I disconnected each one and tried a 180W test load on them individually. Three of the units in the lower cabinet held up well, only dropping a few volts under load. But the two 12-channel relay units that had initially been chattering went berserk when the load was connected, with lights flickering out and relays clicking. So, it seemed like the real problem was that the power supplies in those two units were faulty, leaving the input devices with only about 3/5 of their total power requirements. Looking online, the RRP of these units is over $1500, so merely replacing them was an expensive option. A manufacturer’s label showed they were barely nine years old, so should have life left in them. They were showing symptoms of dried-up electrolytic capacitors (a fault which will be familiar to readers of this column!). So a repair attempt seemed like a good option to me. Bench testing the repaired relay unit with a 180W load resistor. This time, the output voltage only dropped from 35V to 29V, as measured on the multimeter. siliconchip.com.au Australia’s electronics magazine June 2020  65 The first challenge was extracting them from the cabinet. Each unit was connected to 24 power wires, power wires for the unit itself plus two C-Bus cables. Fortunately, there is a circuit breaker at the end of the DIN rail that cuts power to the entire unit and its peripherals. I wondered how I would keep track of which wire went where, but the stiff wires remained in correct alignment even after they were disconnected from the unit. Finally, the first unit was out and on the bench. The next challenge was opening the case. The case is in two halves, split vertically along the middle. There are clips along the bottom of the case, with three blue covers clipped on the top that hold the two halves together. The blue covers proved to be a real battle. They slide into vertical channels and have lugs to hold them in place. However, some genius at the factory had decided to add dobs of plastic glue to make these covers almost unremovable. It took about an hour of levering and battling with various screwdrivers to finally crack the glue before I could get the covers off, with some battle scars to both the covers and the case. The innards are divided into three boards: a large relay board, the C-Bus controller board and a power supply board. The power supply turned out to be a simple switching supply with three output rails. Visual inspection didn’t show anything amiss, so I started by removing and testing the mains filter capacitors, which measured OK. Next, I removed and tested the 22µF 63V filter capacitors on each of the output rails. The first, which was a little raised off the board, measured 0nF. For the second, my Fluke meter read OL, which is not listed in the manual as a valid measurement (no, it wasn’t still charged, or shorted). Anyway, I assumed this capacitor was bad. The third capacitor measured OK. So, it seemed that I had found the problem. To be on the safe side, I also removed and tested the three other electrolytic capacitors on the boards, but they all tested OK. So, I ordered six Nichicon PW-series 105°C replacement capacitors (same as the originals) for overnight delivery, intending to replace all three capacitors in each unit, on the assumption that if one lot was bad, the other lot 66 Silicon Chip would be too. Indeed, after battling through the same difficulty getting into the second unit, the same capacitor measured 0nF, while the other two measured OK. The following morning, the replacement parts arrived, and I soldered them into the boards. With the units hooked up on the bench, I used my test cable and resistor to load them up. This time, the output voltage only dropped from 35V to 29V under load with no relay chatter. I hoped there weren’t any other hidden problems. I re-installed the units into the cabinets, re-connected everything and switched on with fingers crossed. With some relief, I saw the lights come back on, and everything was back to normal. The two units that failed are at the top of the upper-most cabinet, so probably had the highest heat loading. Nevertheless, there are three other units with power supplies which may also need repair in the future. I’ve ordered some additional capacitors, just in case. Sharp R350Y microwave repair R. S., of Fig Tree Pocket, Qld was not happy with the price he was quoted for a replacement module, so he decided to fix that module instead. That’s often the only economical option these days, as he explains... The inverter in our Sharp R350Y microwave failed. I looked for a replacement, but a new one costs more than $300. There are some reconditioned ones on eBay for around $100. I thought that was still too expensive so I thought I’d have a look at it, to see if it was repairable at the component level. I found a copy of the service manual online which contained the inverter circuit diagram, reproduced here. The control unit is shown as a black box. I’m not sure why since the control ICs consist of two LM339 quad comparators and one LM324 quad op amp; it’s not exactly high tech. The bridge rectifier tested OK. I applied power and checked the gate drive signal to the IGBT Q110, and it looked OK. I used a low-cost battery oscilloscope for this, as this circuit operates at a high voltage relative to Earth. This IGBT is a Toshiba GT40T321 rated at 1500V, 40A and is available on eBay in pairs, at around $3 each. The drive signal to the IGBT from the control circuit is buffered by a pair of complementary (NPN/PNP) transistors, not shown on this circuit because it’s part of the control system. To be safe, I replaced both gate drive transistors, the IGBT and the 10W IGBT gate series resistor. I also checked that varistor VRS110 (TVR10102) between collector and emitter of the IGBT was still connected. I found that a PCB track to one side of the varistor had burned off the board, so I repaired that. The other varistor, VRS111, is not fitted to the board, as indicated by brackets on the circuit diagram. I also checked for track damage on the gate connection to the IGBT. It is probably a good idea to leave the col- The circuit diagram for the inverter section of the R350Y microwave, the text in the diagram is so small it can’t be reproduced at a reasonable size, so check the manual online: www.manualslib.com/manual/677215/Sharp-R-350y.html Australia’s electronics magazine siliconchip.com.au lector of the IGBT disconnected until you check that the gate drive looks OK, with square wave pulses of about 15V peak. It seemed all right, so I reassembled the microwave, put a glass of water inside and heated it for a couple of minutes. The water started boiling, so that had obviously fixed it. You can find the manual for this microwave at: www.manualslib.com/ manual/677215/Sharp-R-350y.html Editor’s note: I paid less than $300 for a brand new 1200W microwave with inverter control. No wonder so many appliances wind up at the tip when replacement parts are so expensive. Fridge repair from Russia The “frost-free” fridge which had a broken thermostat. J. L., of Orange, NSW was visiting an Australian couple who live in the far east of Russia and they happened to mention that their son’s fridge was not working properly. Being an old fridge tech, he kindly offered to help... My friends’ son was expecting the first addition to his family, so a working fridge was a necessity in a Russian summer. Hence they were about to buy him a new fridge. But I said I would have a look at it first, to see if I could save them the expense. Their son only lived a few blocks away so it was convenient enough and we popped around. The fridge was a very old Japanese model which was powered using a step-up transformer – apparently, the fridge was made for the Japanese market but ended up in Russia, hence the different voltage requirements. The freezer compartment had some cooling, but the refrigerator compartment had none. The fridge was a “frostfree” design. A frost-free fridge has a fan which circulates air through a hidden cooling coil and discharges the cooled air into the freezer and refrigerator. Frost forms out of sight on the cooling coil, which is automatically defrosted several times each day, to keep the coil clear of ice, allowing the air to circulate. The defrosting process is initiated by a defrost time switch, typically every six hours. The defrost timer stops the compressor and initiates an electric heating element to melt the frost off the cooling coil. Heating continues until a small disc thermostat attached to the cooling coil senses a temperature high enough to indicates all the frost has been removed (typically around 6°C). The heating element then switches off and the fridge sits idle until the defrost timer runs out (typically after 30 minutes), allowing the compressor to start again, cooling the coil back down. I removed the back panel of the freezer compartment to check the coil. The coil was mostly clear of ice, indicating that the defrost system was working but the build-up of ice at the top of the coil suggested that the defrost thermostat was terminating the defrost action before all the ice was gone. With a little ice left over after defrost, the ice accumulated more each day and finally, the airflow became blocked and the fridge could not cool anymore. I tried a non-traditional fix, relocating the defrost termination thermostat to a higher location on the cooling coil, but after a week it was clear that the ice build-up problem was fast returning. Getting another defrost thermostat proved impossible in the far east of Russia – we just got that “idiot American” look from the servicemen to whom we spoke. Even if they had a thermostat to sell, I don’t think they would have sold it to us on principle. On the way home from searching service stores, the father said he had a couple of old fridges he was given to support the family’s work with orphans, but the fridges had died because of city power supply problems. Could one of these fridges have an equivalent part that I needed? After dismantling one old fridge, it siliconchip.com.au Australia’s electronics magazine turned out to be a frost-free style – the style I needed – and so I went searching for the defrost termination thermostat. The fridge was a Russian-made model but the principle of operation is the same everywhere with old fridges, so I removed the part and began testing to see if it would do the job. The test was to soak both thermostats in the freezer compartment of a working fridge, to simulate normal fridge conditions. I removed the thermostats to the kitchen table, to gradually warm up, and with an ohmmeter, I was able to determine that the Russian part needed a higher temperature to open the circuit (and end the defrost operation) than the original thermostat. The actual operating temperature was not important but the fact that it was a higher temperature than the original thermostat was a definite plus. The Russian defrost thermostat would mean the defrost element would operate longer than previously and should ensure all the ice is defrosted. So I fitted the Russian part into the old Japanese fridge. Fortunately, defrost termination thermostats are a fairly standard design, a pre-set bimetal disc around the diameter of a 10¢ coin. Fitting it was a breeze. I fired the fridge up and it seemed to work fine for the remaining week I was in Russia. Two years later and the old Japanese fridge has not missed a beat. It’s a great feeling to have beaten the odds with some thinking outside the square to produce a lasting, good result at no cost. I was the hero for a while and “the fridge job” still gets trotted out periodically to visitors! SC June 2020  67