Fullscreen HTML5Med Res (??MB)HTML4

SERVICEMAN’S LOG Where there’s a weld, there’s a way Dave Thompson It has always been my dream to build my own car. I worked on aeroplanes for many years, and if I could do that, surely a more terrestrial vehicle would be a doddle! Still, such an undertaking is a major project, which is why I have been working on it for around 15 years and still haven’t finished... It was either that or build an aeroplane; kit planes exist, but they are pretty expensive, and my garage isn’t exactly hangar-sized, so building a car is a somewhat more realistic goal. When I started those 15 years ago, my circumstances allowed me to indulge in this dream. It was all triggered when I came across a book on making a Lotus 7 replica using standard Ford parts for systems like steering, suspension and drivetrain, as manufacturing these critical parts is tricky for the home builder. There were a few problems, of course. Firstly, I’d need to find those parts – or suitable equivalents. Secondly, I’d need many tools I didn’t already have. First and foremost among those, I’d need a decent welder and some skills to go along with it. Dad was a pretty good welder – he wasn’t qualified, but learned by doing, and over the years, I spent many hours watching him use his trusty arc welder (called a ‘stick welder’ in some parts) to fuse metals together. I inherited his welder and accessories, which now sit under my bench. But I’ve never used them. MIG vs TIG The Lotus 7 replica was based on a tubular steel spaceframe chassis. To put it together and have it certified, I’d need to use either a MIG (metal inert gas) or TIG (tungsten inert gas) welder. Back when I started all this, TIG welders were expensive, and from the research I’d done, it was a much harder skill to acquire, so I decided to go with MIG. The principle of welding is simple; heat the metal joint (and filler rod) enough and, under the right circumstances, it will literally fuse together. In contrast, soldering ‘glues’ components electrically but gives no real strength, which is why solder alone should never be used for joints where physical strength is required. In electrical engineering terms, an arc welder is the simplest way to fuse metal. All you have to do is pass a huge alternating or direct current (AC/DC – rock on!) through the metal to be joined to heat it up. One of the electrodes is a flux-coated rod to assist sweating everything into a nice seam. While simple in theory, in practice, it takes a lot of skill and knowledge to know which rods to use, how much current to apply, how fast to move the rod along the seam, how fast to feed it in, and many other variables that only experience and practice can teach. A MIG welder is theoretically a lot easier to use for beginners. Instead of a solid flux, an inert gas (usually Argon, CO2 or a mixture of both) is used to isolate the weld as it happens. This prevents air from oxidising the joint at the high working temperatures, which would otherwise make it messy and not structurally sound. This all happens at the nozzle end of the welding torch. It is hollow and has an aperture for the gas to flow through, while a wire is power-fed to the joint down the centre. Pressing the trigger on the torch does three things. Firstly, a valve opens so gas can flow out the end of the torch. Secondly, a motor starts feeding wire out of the nozzle at a pre-determined rate and lastly, lots of current is applied to that wire. The circuit is completed by clipping a heavy-duty Earth Items Covered This Month • • • • • When there’s a weld, there’s a way Magnifying viewer repair Smeg dishwasher repair Troubleshooting a BWD 525 oscilloscope Fixing a pool chlorinator 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 siliconchip.com.au Australia's electronics magazine May 2022  85 clamp to the work to be welded. Wherever the wire from the torch touches the metal, the circuit is completed, and welding occurs. As you can imagine, there is a lot going on, but the variables are all easily adjusted on the welder itself. Gas flow can be changed by tweaking the regulator, wire speed is controlled by a potentiometer and the output current by either a pre-set switching arrangement or a continuously variable current regulator. The performance between different welding rigs varies enormously, as does the price. Hobby welders are notoriously cheap and often not designed for any serious work. All welders have a stated duty cycle, and this is usually part of the numbers one looks at when buying a welder, along with the maximum output current. Welders can only be used for so long at full noise before having to ‘rest’ and cool down. The standard ‘period’ is 10-minute units, so if the duty cycle on a 100A welder is 30%, it can theoretically be run at 100A for three minutes before either shutting down due to overheating, or the operator stops welding and waits out the remaining seven minutes. Obviously, there are variables in this; if you make a weld and then stop for a while before making another one, you can go for longer as it’s only the on-time that matters. Also, running it at a lower current will usually allow you to have a higher duty cycle. But that number does provide a good indication of the practical use of the product, and should be taken into account when shopping. Another consideration is the device’s build quality; many inexpensive machines use aluminium windings in the main transformer, usually one of the most critical components of any welder. Aluminium is cheaper than copper, so cheaper machines tend to use transformers wound with it. Much internet argument rages over the pros and cons of either material, and whether square or round-wound coils on the transformer core are better. Still, in practice, most serious welding machines use very beefy, copper-wound, iron-cored transformers. I mention this backstory because recently, a neighbour brought in a dead MIG welder to my workshop, asking if I could repair it. When plugged in and powered on, the cooling fan ran, gas flowed and the wire was fed at a pull of the torch’s trigger. However, there was no output voltage 86 Silicon Chip (typically 23-26V DC on a smaller MIG like this), and it wouldn’t weld when the circuit was completed. This didn’t bode well; I suspected several possible reasons. On many such welders, there is a massive full-bridge rectifier mounted to the case, while in others, a ‘driver’ PCB controls the current delivery. This rectifier (or any of the components on a driver board) could have failed. Many welders also use the same PCB (or sometimes separate smaller PCBs) to hold components for controlling the fans, electronic gas switching and wire-feed speeds. Still, as these features all appeared to be working, failure here was unlikely (though possible, of course). Depending on the type of thermal cut-out device employed, this may have also failed, preventing power output. While some machines use bi-metal thermal switches, others use simple single-use thermal ‘fuses’. Either can kill power to the whole machine, or only prevent the high-­ current side of things operating and keep the fans running to assist cooling. And if none of those things turns out to be the problem, it might be the transformer itself, which would put a whole different light on things. Either way, I’d have to open it up and take a look. I could see the bottom of a PCB through the vented case, so I would start by looking at that. This welder is a 180A ‘prosumer’ level gas/gasless machine with a claimed duty cycle of 60%; not too shabby, considering it was purchased many years ago. It can also weld aluminium (with the right welding wire fitted and the polarity to the torch reversed). Interestingly, the owner uses a large SodaStream CO2 gas bottle mounted to it for the inert gas supply, through a converter valve commercially made for that purpose. I wish I’d known about this when I got my MIG, as it is substantially cheaper to swap these bottles out than rent even the smallest one and get it filled from the local industrial gas suppliers. It is also much more portable than having a large gas cylinder to tote around. Opening it up There are few jobs easier than disassembling a welder. There is usually a side panel that can be unscrewed or simply unlatched to change wire spools and access the power leads to the torch and other interior components. Chunky PK-style screws hold the rest of the metal and plastic bits together, and it takes literally five minutes to strip the whole caboodle down to spare parts. The main transformer is the star of the show and takes up a good amount of space inside the box. It also makes up the vast majority of the weight of the machine. A large 100mm cooling fan sits near the back of the compartment, and the spool and wire-feeder mounts at the front, behind the control panel (such that it is). More modern ‘inverter’ type welders get away with a lot less electrical mass. While they typically do an excellent job, they tend to cost a lot more. The PCB I saw earlier was easy enough to remove, and as far as I could ascertain, there was nothing untoward with it. No electrical smell or signs the ‘magic smoke’ had escaped. The pot that controlled wire speed felt smooth, and a meter across it showed no signs of worn-out tracks when I slowly rotated the pot through its range. There is a fuse mounted on the PCB, and that tested OK. There is also a smaller mains to 12V transformer mounted on Australia's electronics magazine siliconchip.com.au this board; I tested it for continuity, and both primary and secondary looked good, with no shorts to ground anywhere. 12V DC applied briefly to the relay coil saw it pulling in and letting go properly, and the contacts also rang out OK. None of this was much of a revelation as this board controls the fan, gas valve and wire feeder – all of which I knew still worked. Moving on then. A bi-metal type thermal switch was mounted to a bracket that pressed the face of the switch to the coils of the transformer. If the coils got too hot, the switch would trip and interrupt power, preventing welding until it cooled again. Testing the switch was straightforward; after removing it from the bracket, I used a multimeter to measure the resistance across the terminals (with the leads disconnected from the rest of the circuit). The reading was almost 0W. I then used my hot air gun to carefully apply heat to the face of the switch, and it opened at around 50°C, or as near as I could measure it anyway. That seemed about right; if the outside of the coil were at around 50°C, the centre would be hotter, and that’s as hot as I’d want it to get. The final discrete component was the large industrial-­ sized bridge rectifier mounted to a metal block, which was then mounted to the steel case (for better cooling, I assume). Measuring across all points with my diode tester showed there were no shorted or open-circuit diodes. That left the main transformer. While it was possible the switch in the torch handle was failing, the rest of it was working when the trigger was pulled, so I suspected it was not the problem. Preliminary measurements across the primary of the transformer were encouraging. However, after further testing, I discovered that one of the two secondary windings was open-circuit. After disconnecting all the wires, I pulled the transformer out for better access, noting carefully where everything went so I could put it back together later. It certainly wouldn’t help if I wired it back up incorrectly! My fears were confirmed with one of the secondary windings appearing open, and it was the one that went off to the rectifier. That explains the lack of output to the wire. It was also possible that this welder was branded and marketed under one name by one company but sold by other companies (even in the same country) under another name, with the same (or very similar) hardware. A quick look on the Interweb brought up literally hundreds of very similar welders, but very little information on the parts inside or even who made them. Besides, I didn’t exactly have any part numbers emblazoned over anything in this machine either. I don’t relish making phone calls like this to clients, but sometimes these things don’t work out. But in this case, the client mentioned that when he bought his welder, an old friend of his had also purchased the same one. That one had fallen off the back of a ute at a job site years ago and no longer worked. The client reckoned his mate might still have it lying around (like many of us, he didn’t throw anything away either!), and if so, perhaps he could acquire it and I could burgle it for parts. Even better, I said, it might be easier to repair than this one! Sure enough, a few days later, the client turned up with his mate and his mate’s dead welder in tow. One look at the wreck told me that it wasn’t going to be repairable! It looked like it had been run over; I guess when heavy objects fall onto hard ground, they don’t usually fare well! However, transformers don’t bend easily, and as it is mounted in the dead centre of the case (to balance the weight and make it easier to move about, I suppose), it is about as protected as something could be in a relatively flimsy stamped metal case. In-situ measurement (once I’d bent a few things out of the way) proved it was still alive, so after some serious panel beating to get stuff out of the way, I was able to extract the transformer from the dead machine. Reassembling it into the original chassis was as straightforward as wheelbarrow mechanics. Once I made sure that Bringing it back to life This caused somewhat of a quandary; buying a new transformer, or having one made specially if we couldn’t find a replacement, would likely cost the lion’s share of a new, more modern welder. I’m in the very fortunate position to have a commercial-grade transformer-winding machine and ample copper wire stocks, but I’d have to face a couple of problems before I could re-wind it. For one, I’d have to break down the old, dead transformer to salvage the E and I iron core laminations from it – I have some NOS (new old stock) cores in stock but nothing that large. And for two, at the moment, that machine is buried under a household’s worth of junk in storage. Getting it out (very much a two-person job) and setting it up to re-wind one transformer (even a monster one) wasn’t going to fly. That was a shame, really, as it would have been a very interesting project for my machine. Oh well, such is life. A quick call-around for potential replacement transformers came up empty. This brand of welder was no longer made or sold, so it meant finding another one from another manufacturer – that is, if the customer wanted to go ahead with a repair. siliconchip.com.au Australia's electronics magazine May 2022  87 everything was in its proper place and wired in correctly, I held my breath and plugged it in. A quick brush of the wire to a scrap of metal held into the earth clamp proved that we now had plenty of juice at the torch. The moral of the story? It’s always good to have a spare! “Magnifying viewer” repair B. G., of St Helens, Tasmania has a short story about repairing a somewhat unusual device... A friend called me seeking help to repair a “magnifying viewer” for a vision-impaired friend. I duly picked up the unit and was told that it failed to switch on and ruined the RCD in his switchboard, which had to be replaced. I gathered from the weight that it contained an old CRT. The item to be viewed was placed in a tray under the tube, then adjusting the magnification and focus knobs provided a clear and magnified display of the object on the screen. I cautiously plugged the power plug into my test outlet, which has an incandescent globe in series with the Active line. The globe pulsed for some seconds, then tripped my circuit breaker. The unit was made by Telesensory Systems, a US company that now appears to be non-existent. So I had to trace the circuit. There was a nice toroidal power transformer with no markings, a regulator board with +16V DC and +12V DC outputs, CRT drive circuits, a small ‘vidicon’ camera underneath and two of the smallest fluorescent tubes I have ever seen. I discovered that the two 12V regulators had failed and replaced them. I couldn’t make much sense of the transformer; a mate suggested that I temporarily try one he had, to no avail. I checked for shorts on the mains side, but it seemed all right. Undoing an insulated cover on the left side, I found another board labelled “fluoro lamp driver” with a 4060 IC, some large capacitors, relays and transistors and a large black inductor. The inductor and the board were connected to the mains Active input and were easily unplugged. The unit powered up now; this time, a dim raster was visible, so perhaps the original power transformer was OK. I then realised that the large inductor was the ballast for the fluoro tubes. It measured open-circuit, and I bet it was breaking down with voltage applied. An internet search failed to find anything suitable like a 4W ballast, and given that it was not producing a dull picture, I fitted a string of white LEDs under the CRT. This allowed it to produce a very reasonable magnified display. I left it like that, and my friend was delighted to have it returned in working order. Smeg off and buy a new capacitor R. W., of Hadspen, Tas managed to repair a dishwasher for a grand total of $6. That’s less than 1% of what he was quoted for a new control board without installation... When we moved to Tasmania, our new house had a Smeg DWA U214X dishwasher installed, matching the kitchen cabinetry. It was about three years old, appeared to be in good condition, and worked reliably until one day, a year and a half later, it refused to start. This “magnifying viewer” utilised a large CRT display. It was made by a company called Telesensory Systems who specialised in making devices to help visually impaired people. 88 Silicon Chip Australia's electronics magazine siliconchip.com.au This unit has a large pushbutton switch that controls the mains supply. Upon switching it on, the machine gave a beep, but none of the LEDs illuminated. It would generally flash the two right-most LEDs to indicate completion of the previous cycle. Selecting a program would typically show the corresponding LED, but nothing happened. Cycling the power gave the same initial beep but no further activity. I trawled the internet and found that this was a common problem with Smeg dishwashers of this age, but no one had documented a repair. Some of the suggestions were entirely unhelpful, stating things like “you need a new keyboard for it”. I was able to find an assembly diagram but sadly, not a schematic. The next day, I again tried to operate the machine and was greeted with the same result. I was called away for an hour or so and, after returning, I realised that I had left it switched on and now the end-of-cycle LEDs were flashing. After selecting a program, it operated normally. The next day, the fault returned, but it worked after being left on for an hour. I contacted a local supplier of appliance spares, and they were able to find a replacement board, but it was over $650. Even second-hand items on auction sites weren’t cheap and certainly not guaranteed to work. This effectively wrote off the dishwasher, but I decided to attempt a repair as I had nothing to lose. I thought that faulty capacitors were the likely culprits, possibly not resetting the microprocessor, causing the switchmode supply not to start or limiting the available current. I retrieved the control board, and there were no tell-tale signs of failure or bulging electrolytics. There was a 22nF X2 capacitor to drop the mains voltage, and I remembered reading in a past Serviceman’s Log column that these had caused some problems in ageing equipment. I decided to replace it and all of the accessible electros too. A quick trip to Jaycar, and I had five capacitors for about $6. Some like the X2 were an exact fit, while others were larger, and I used a leaded 100μF electro to replace an SMD type. However, when I desoldered one leg of the SMD capacitor, it took part of the PCB track with it! I was able to delicately solder the lead to the remaining piece of track. Not ideal, but it worked. The PCB is sandwiched in two half-shells that mount in the dishwasher and guide the edge connectors. I had to make a hole in one side to accommodate the 450V electrolytic, as the original 400V unit was smaller. Smoke test time – it worked faultlessly. The LEDs appeared brighter than I recall, indicating that the X2 capacitor was indeed not passing sufficient current for the power supply to start. So a dishwasher was saved from the junk heap for just $6, one hundred times cheaper than a new board and many hundreds cheaper than a new dishwasher. Troubleshooting a BWD 525 oscilloscope J. D., of Crows Nest, NSW has an electrical engineering degree but wound up working in IT instead. He has kept his workbench going with the odd repair and project, but mainly in the digital electronics, low voltage space... High voltage for me meant mains power, and even then, it was only to step it down. But then, I got the opportunity siliconchip.com.au Australia's electronics magazine May 2022  89 to purchase an Australian-made analog BWD 525 cathode ray oscilloscope (CRO). It was already close to 40 years old by then. It worked great until one day, my single trace became multiplied, roughly 10 scan lines high. I thought I’d have a go at diagnosing the problem. I started by checking the various dials. The focus dial ‘worked’, meaning the multiple scan lines did all go in and out of focus but remained 10 high. Next, I fed in a 1kHz 4V peakto-peak square wave, and the 10-high scan lines remained, but the Y deflection seemed to be working. The X&Y controls moved my waveform as expected. Luckily, I had the service manual, and it included the complete circuit, with expected waveforms and voltages at various points. Vaguely remembering how to discharge a CRT, I opened the CRO, revealing discrete components – including capacitors as big as cans. I started looking at the focus circuit, which has -1450V DC applied to a series resistor string of 1.5MW, a 2.5MW pot and two 3.9MW to ground. The pot’s wiper went into the CRT terminal marked “focus”. This was somewhere to start. There’s also a bypass capacitor between the pot and the CRT. I measured the resistor values and disconnected the bypass cap; they seemed all good. One problem was that my multimeter had a maximum rating of 1000V DC, so I couldn’t directly measure the -1450V rail. There were even higher voltages marked on the circuit. I creatively measured the focus voltage by adding a resistor in series with my DMM’s input resistance and calculated that the focus voltage was somewhere between 800V and 1200V. The horizontal amplifier looked like it was working too; there was a 7400 TTL NAND gate doing some tricky switching, but roughly measuring the voltage and waveforms, it seemed to all be correct. The horizontal amp circuit also does the blanking among some other functions such as “alt” and “add” (dual-display/add the signals together) – I couldn’t find any faults there. Looking at the capacitors, nothing seemed strange; there were no bulges or anything like that. Also, the diodes and the bipolar transistors all seemed to be conducting correctly. 90 Silicon Chip The repaired BWD 525 oscilloscope displaying a nondistorted trace. Working around 8600V made me a little nervous. Next, I thought I’d check the power supplies. The 135V rail showed a whopping 9V ripple before getting clipped by a zener diode – which reduced it to 25mV. So I ordered and then installed some new unique-valued capacitors because they didn’t seem to be filtering the supply rails very effectively any more. I then calculated that I could directly measure the voltage at the CRT focus pin with the potentiometer set halfway; it should just be within range of my multimeter. Doing this, the scan lines decreased to 5 high. So I was definitely in the right area. Eagerly, I continued to measure. I measured across one of the 3.9MW series resistors, expecting to get a reading around 480V. Instead, it went to 1000V+, then the multimeter promptly failed. I confirmed that the resistor was open-circuit, replaced it and the original, crisp traces returned. But wait, didn’t I measure the resistor values before? That’s a real headscratcher. With newfound confidence (and a new multimeter), I wanted to tackle another problem that I’ve always had with this scope. The horizontal trace never made it all the way to the right edge of the screen. It would cut off about 2.5cm before reaching the right edge. Looking at the circuit, the horizontal deflection amp was made of pair of matching high-voltage BD115 transistors with their collectors connected to the left and right horizontal deflection plates on the CRT. Their bases were connected to identical pairs of emitter followers; that made sense. As they are mirror images, I compared and measured them. All the voltages for those transistors matched perfectly and were as shown in the manual. I then checked the emitter-followers, and I got twice the voltages specified in the manual. I measured -3.5V to -4.15V Australia's electronics magazine siliconchip.com.au instead of -2V to -2.8V. However, the voltage at the BD115s still measured perfectly. This was a mystery. The coarse & fine horizontal adjustments worked – scrolling left, it scrolls off the left side, as you would expect, but when scrolling to the right, it always clipped 2.5cm before hitting the right-hand edge. I started to look at the trace as it reached its end before the right edge and couldn’t see any distortion in the trace. It was as if the blanking/retrace circuit kicked in too early. So I swap the connections to the BD115s. My theory was that if it was the blanking circuit, the ‘cut off’ would also switch to the left side. It didn’t! It was still cut off on the right side, as if there was an invisible wall. While off, I carefully placed the CRO on its side, the right side down, thinking that whatever is obstructing is probably extending too far left from the right side. After all, I was running out of ideas. I banged the table, switched it on, and the trace moved closer to the right side! I banged some more, and it moved some more until it was back to normal – happy days. So now I have a fully working 45-year-old Australian-­ made CRO and a new multimeter. Pool chlorinator problems C. F., of Duncraig, WA had a problem with an AstralPool Viron eQuilibrium pool chlorinator. Luckily, the control board had relatively few parts and identifying the one which was not doing its job was not overly tricky... One day, I noticed the chlorinator pump was not operating at the usual time. We had some wild weather with occasional blackouts, so I thought the timer had been reset. I checked it, and the time displayed was 00:00. I set the time, checked the timer setting (which was fine) and put it in automatic mode. I was expecting the pump to start, but nothing happened. The display cycles through screens showing pool chemistry, chlorine production, the current time and timer status. When the clock display appeared again, the time was 00:00. My first thought was that I had stuffed up when setting the time, so I tried again, with the same result. I reset the system, but that didn’t help either. Each time after cycling back to the clock display, the time showed midnight and did not advance with passing minutes. I called the manufacturer support line but they couldn’t help me; all the person could tell me was how to set the clock, which was not the problem. As the controller is out of warranty, I decided to have a look inside. I disconnected the controller and brought it to my workbench. Removing four screws opened it up. Inside is a large PCB with a couple of transformers, a few relays and power transistors. A ribbon cable connects it to another PCB at the front panel. I took out four more screws to remove this PCB. It has the display, three ICs – one square with 64 pins and two eight-pin types, all SMD. The board is coated in a protective lacquer. This is good as the controller lives near the pool, potentially exposed to the elements and pool chemicals. However, it makes readings the IC markings a bit challenging. Eventually, I got the details by using a magnifying glass and illuminating the board with a torch from different angles. One of the eight-pin ICs is a 5V regulator, the 64-pin is the PIC microcontroller (a PIC18F6XK22) and the siliconchip.com.au 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. other 8-pin IC is a Microchip MCP7940N real-time clock. I checked its supply voltage, and it was correct. The clock chip is equipped with an I2C serial interface and the PIC microcontroller also has I2C lines. Following the tracks, I could see the connections between them. So it looks like the PIC microcontroller ‘outsources’ timekeeping to the clock chip. I used an oscilloscope to check what was happening on the I2C serial clock and serial data pins. As expected, there was activity on setting the time, and each time the display entered the clock and timer status display. I checked the external oscillator pin and the signal looked OK. Since, apart from the clock, all other functions seemed to be working, I thought the clock chip was not doing its job. If it were the PIC microcontroller, I would have no hope as I don’t have the software to program a new one, but I thought that replacing the clock chip would fix the problem. I ordered a compatible MCP79400 in the SOIC package. After replacing the chip on the PCB, I connected the ribbon cable to the power board and plugged in the unit. I entered the time and waited for the display to circle to show the clock and timer. The clock did not return to 00:00, the time was now correct, and after a minute, it advanced. So the clock was now working. All that was left was to apply some protective lacquer over the new chip and put the controller back together, which was the reverse of the disassembly procedure. I was pleased that I saved the controller from ending up a junk SC pile in this ‘throw-away society’. The AstraPool pool chlorinator is now keeping the correct time after replacing the MCP7940N/MCP79400 real-time clock/calendar (RTCC) IC. Australia's electronics magazine May 2022  91