Fullscreen HTML5Med Res (??MB)HTML4

SERVICEMAN’S LOG Mixing it up a bit Dave Thompson It’s frustrating when there is a flawed product on the market, and instead of recalling or fixing it, the manufacturer blames the user instead. Your mobile phone has no reception? You must be holding it wrong! This time, it was our blender, and I had to turn to other users for a solution… It’s hard to be a serviceman these days without hearing ominous stories about the ‘right to repair’, especially regarding large US corporations. This has become a really hot-potato topic and indeed has been commented on with some insight by the Editor and other contributors to this magazine. One of the most serious concerns is the increasing use of subscription models for hardware, which is becoming more and more ‘de rigueur’. I think that is terrible news for consumers and repair people. Court cases and laws preventing monopolies and protecting the right of repair for consumers have driven some companies to introduce subscription models, so they are assured of continued income as well as protecting their ‘intellectual property’. One way they do that is by trying to maintain control of their products after they are sold. I suspect this will put many local repair people and servicemen out of work, often in very small communities, unless, by some miracle, they can score a maintenance contract with the vendors. No doubt they would demand exclusivity anyway. If the manufacturer designs products so that only they can reactivate them after a part is changed, how is anyone else supposed to fix them? The repair business isn’t what it used to be So, a bleak outlook, then. My own computer repair business, almost 30 years old now, has seen the wave and wane of the industry. Throw in a deep recession in the late 2000s and the city and my workshop being ruined by earthquakes in 2011, and it’s a wonder we still have a business at all. At one point in the mid-2000s, we were averaging 65 calls a day. I employed four guys and two vans on the road. These days, it is just me, and I’d be lucky to get 65 calls in six months. It turned out this way because computer service and repair have long been sunset industries. These days, it’s all mobile devices, and they are consumable, so if one is dropped and broken, insurance or savings pays for a new one. In some cases, data recovery may be required, but even that is moot as much of our stuff is backed up in ‘the cloud’ anyway. Every bread-and-butter job we had in the 2000s has long since disappeared, only to be done now by techsavvy householders or the advent of self-install plug-andplay internet. That’s OK with me, as I am nearing that age where I’ll hang up my floppy drive anyway. But for dozens of other companies and service guys, this really is the end of an era. Manufacturing for unrepairability It’s the same with just about everything these days. Most appliances, for example, are manufactured without repair in mind (or, if you are a bit cynical, with anti-repair in mind). If you can even source replacement logic or controller boards, they are usually hellishly expensive because they just aren’t made as available as they were in the past. The manufacturer wants you to buy a whole new unit, not fix the broken one. That’s one reason why repair being monopolised by manufacturers is so troubling. There is a conflict of interest, so they are more likely to quote you unreasonably high repair prices in an attempt to convince you to give up and buy a new one instead. My wife bought a high-end food blender/mixer type thing a while back, and overall, it works pretty well. It has this thing called “wireless detect”, which took us a while to figure out, but all we really use the thing for is blending and mixing, using the various controls on the front of the machine. One thing that always annoyed me is that it will only run with a jug or attachment sitting in it. Actually, many blenders have such a safety feature that prevent them from being used with nothing attached to them. In this particular 80 Silicon Chip Australia's electronics magazine siliconchip.com.au Items Covered This Month • • • • Overly complex food mixer ‘repair’ Tracking down interference using an SDR Three different antenna repairs Dual tracking power supply excessive ripple 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 model, this wireless detection feature allows the mixer to detect different-sized attachments and adjust pre-set programs automatically. In other words, it detects the particular attachment and adjusts things accordingly. As there are quite a few different-­sized glass grinder bowls and cups available as optional extras, it seemed like a handy feature to have. These attachments are basically glass bowls that screw into the base that has the blades built into it. The whole thing then mounts on the blender. The glass bowl part is removed from the hard-plastic blade-driver section for loading whatever you want to grind into it. Let’s say you want to make some powdered salt. You fill the glass bowl with the desired amount of granular salt, invert the blade-holder part and screw it blades-first to the glass bowl. You then flip the whole thing up the other way and plop it, blade side first, onto the top of the blender, engaging the splined drive socket. You can then use one of the pre-programmed routines or manually drive the mixer using the controls. An exercise in frustration It all seems simple enough. Except, on this model, with the wireless detect feature, you have to have the main mixing jug, one of these herb grinders or any of the other attachments installed on it to even power up. And it turned out to be so finicky that it made it almost unusable. The main jug – the one that comes with it – seems solid enough in operation. However, those optional glass bowl attachments have what I discovered are NFC (near-field communication) chips buried inside them that are read by electronics inside the device. In practice, though, the majority of the time we tried to use these attachments, the mixer would not detect that the bowl was in place, so it would not start. It was the same with all the attachments we bought for it. Obviously, this was not going to fly. The attachments have arrows moulded into them that show how much the two halves should be torqued for the thing to work, but even when the arrows are perfectly aligned, it just will not switch on most of the time. That certainly created a lot of blue language from the kitchen! Another problem is that, given the size of the smaller attachments, cranking them up to have the arrows aligned means crushing a large O-ring type seal between the two halves. Trying to undo them once torqued is nigh on impossible for my wife and almost impossible for me. The whole thing was starting to reek of poor design and implementation. siliconchip.com.au Of course, this whole idea is product protectionism cleverly disguised as a safety mechanism because thirdparty fittings and attachments that don’t have the correct NFC tag will not work. Only original attachments can be used on this mixer, and they’re not cheap. When it works, it works well, but getting it to work was often highly frustrating for us. Our first stop was the big-box store where we had bought the main unit, along with these extra fittings we thought we’d need. Of course, the guy there, while extremely knowledgeable when we were shopping for it in the first place, now seemed to be struck dumb and claimed he’d not heard anything from customers about it. Perhaps our one was faulty, and if we liked, he could feed it back through the warranty system and in just six short weeks, we could have it back. It was only a month old at this time, so I suggested that if he thought it was faulty, perhaps they could see their way clear to replace it under the Consumer Guarantees Act. Well, you’d think I’d suggested sending his grandmother on a one-way trip to Switzerland! That solution apparently wasn’t going to happen, for various reasons, first and foremost because we had used the mixer! I contemplated going through the finer points of finding faults without actually using an appliance. Still, this guy had obviously been down the annoyed-client road before and, like a debating team captain, had a pat answer prepared for everything and anything I could say. At this point, my serviceman’s lizard brain kicked in, and I thought I’d open it up, have a look and see if there was anything I could do. Perhaps the sensor had fallen off or had been glued in at an angle, or something silly that I’d be able to fix with my rudimentary knowledge of blender repairs. I mean, how hard could it be? We all know the answer to that, and you’d think that after all these years, I’d know too! Australia's electronics magazine December 2023  81 At least it was well made Pulling this thing apart was not that difficult. These are ‘proudly’ made in the USA and using American-made parts, or so the blurb states. That means no dumb security screws, just straight-forward, meat-and-three-veg screws that can be undone with a longish-reach Posidrive screwdriver. Everything came apart so easily. No breakaway clips, no hidden screws under mouldings. Very refreshing! I did have to pop out the rubber feet from the bottom to reveal some case screws, but I’ll give them that as a neat design. Once the screws were out, the two halves of the case came apart easily. There are no warranty-voiding stickers across the join or any of those breakable foil screw covers over anything. At least these appliances are designed to be repaired, and I like that a lot. Spares are apparently widely available from what the sales guy told me the first time we were at the shop looking into buying one. Inside is what you’d typically find in a blender. After removing a well-made protective metal cover, I could see the main space was taken up with a large brushed motor. It directly powers a splined drive socket at the top of the mixer via a square drive shaft at the end of the motor’s armature. The splined drive socket is easily removable by loosening a grub screw with an Allen wrench, if need be, and while the splined and square drive parts are cast from relatively heavy metal, the body of the drive socket is hard plastic. This is actually by design; if something in the jug or bowl fouls the blades and stalls the mixer, this plastic moulding will shear or crack, and the metal square-drive part of it will just spin harmlessly inside the moulding to protect the motor from stalling and potentially burning out motor windings and electronics. It is a relatively crude but very effective protection system. Replacement drive sockets are inexpensive and readily available. Another big plus for the repairability of this device. The mouldings and mounting for the motor and electronics are all super heavy-duty plastics, almost like Bakelite, especially given what I usually see in most cheaper modern appliances. This is definitely higher quality, and it is typical of the brand. The unit is certainly built to last, which you’d hope for, given the relatively high purchase price. The front panel controls – two toggle switches, a speed control pot and the LED display – are all directly mounted to a circuit on the inside front of the case. This PCB is populated with the usual mix of SMDs and discrete components, with heavy wiring to the motor and power switch on the right rear side of the mixer. It all looked pretty standard and what I would expect to see in any reasonably advanced blender. However, I could see nothing in or around the top of the unit that resembled an NFC reader, so I assumed it was mounted on the main circuit board instead. This ‘initiator’ side of the NFC system should throw out a magnetic field that would (hopefully) detect the passive NFC chip embedded in the attachments and then allow the mixer to be powered up, or not. The thing is that I could see the NFC chips embedded in the glass bowls on opposite sides, so why was this not detecting attachments 90% of the time? As the attachments have a spline-shaped base, they can sit at any angle in the drive system, but no matter where they sat, the blender 82 Silicon Chip would not detect them most of the time. There seemed to be no rhyme or reason. So, to my mind, there was either a fault with our unit or the NFC system is somewhat flawed. I reassembled the blender, as there was really nothing I could do, except feel a little better that I had at least tried to do something! An unexpected solution Next, I did what I always do and hit the web to see what was going on. Perhaps unsurprisingly, it turns out that this is a well-known problem with these blenders; a lot of people were moaning about it in online forums and videos. Nice one, big-box store guy; we won’t be shopping with you again! As is typical with the information available, there is a lot of it, and not much is helpful. Plenty of these slick kitchen-type presenters were talking as if we were imbeciles and saying all we have to do is align the arrows on the attachments, and it will work. Oh really? They are either completely ignorant or wilfully obtuse, and the comments sections usually refer to the former. While there are no instructions with the attachments, and the arrows are pretty hard to find unless you are looking for them, this resolution didn’t seem to help the majority of affected consumers. The party line from the manufacturer themselves was that a video would be ‘out soon’ to explain how to make this more reliable. To date, nothing has been posted, so, as is typical for a lot of technology, they leave it to end users to resolve their issues and find a workaround. As it turns out, there was only one video among hundreds where a home-chef type presenter found an almost foolproof way of making it so the attachments were detected and worked every time. She claimed she had just stumbled across it after spending many hours trying to get her (much more expensive model) mixer to work properly. Her method was to screw the two halves of the attachment together and line up the arrows. She would then place the attachment on the blender, and typically, it would not be detected. While it was in place, she cranked the glass part about 30° more and magically, the blender would see it. She could replicate this every single time, and of course, the arrows on the case of the attachment no longer lined up, but the appliance would detect it just fine. This, of course, would make it impossible to undo again due to being so tight. However, she then backed off the bowl in-place, using the grip of the blender to help her. She ensured she was still maintaining the seal – any contents would soon fall out if she undid it too far – and then removed it and flipped it upside down before completing the unscrewing and removing the blades part of the attachment. Of course, the first thing I did was try that method with ours, and it worked every time. The fact that the manufacturer hasn’t modified the attachments to show new arrow positions, or at least put out a workaround video of their own, is extremely disappointing. Sometimes, there is no electronic fix, just a clever end-user who figures out how to make it work. The source of the interference G. G., of Macleod, Vic thought he was solving one problem when he was actually creating a new one. He explains Australia's electronics magazine siliconchip.com.au how a software-defined radio (SDR) helped track down the source of the problems... I have a weather station to monitor the roof cavity temperature so I know when to turn on an extractor fan on a hot day. I realise that it could be thermostatically controlled, but I don’t want it running when we’re away. The roof cavity sensor/sender seemed to be chewing through batteries. Because of the nuisance value of getting into the ceiling, I decided to power it from a plugpack plugged into a ceiling power point. That worked OK for a few weeks, then the display stopped updating. Then I started noticing that the remote controls for our alarm system and garage doors had become less sensitive and we had to be much closer to their receivers to get operation. Next, a remotely-controlled ceiling fan refused to operate. At about this time, I had brought home a system for repair that included radio microphones and a mobile internet dongle. My wife was convinced it was causing the problems. I replaced the batteries in the alarm remotes, which gave a slight improvement. They had tuning capacitors, so I tried tweaking them and got a bit more range, but barely enough. Retraining and new batteries in the garage remotes seemed to gain a little more range. A web search told me that the alarm remotes were on 304MHz, so to check their outputs, I thought I’d install an SDR that had been given to me years ago but that I had never used. The software installation was tedious and even required manual installation of the drivers, but it eventually sprang to life. Stepping across that part of the spectrum, I couldn’t see any response to my button presses. Testing with the radio mic in the system in for repair confirmed that the SDR was working correctly. I then did a web search on the garage remotes. I found a very useful site (www.remotepro.com.au) that gives all manner of Australian wireless remotes and the programming of garage door openers, and even has full installation details for many garage openers. That site told me that my Merlin controllers were on 433.92MHz, so I tuned the SDR to that frequency to check the garage remotes. siliconchip.com.au Australia's electronics magazine December 2023  83 I found that there was already a very strong continuous signal, 30dB greater than the local FM radio stations. Rough direction finding with the whip antenna gave a null when pointed toward the weather station sender. Powering down the circuit going into the ceiling immediately stopped the rogue signal, and all remotes started operating perfectly. Reapplying the power even restored the temperature display, and on the SDR, I could now see a short update burst coming from the sender about once per minute. It seems that an occasional software glitch sent the weather station into a continual transmission mode and, as it was within a couple of meters of all the other devices’ receivers, it swamped their reception. Likely previous similar glitches had flattened the batteries before we’d noticed any effect, but my new power source was able to keep the rogue transmission going. After the reset, it has been performing normally for a few weeks; until I get around to replacing it, I at least know how to restart it. I later discovered that the alarm remotes also operate in the 433MHz band. A trio of antenna repairs Around fifty years ago, I. G., of Banyo, Qld was a Radio Trainee with the Department of Civil Aviation, field training at his home station, the Gold Coast Airport (Coolangatta)... I was lucky to have great mentors at the station, the supervisor and technician, who involved me in fault clearance and regular maintenance. Still, one time I was left to my own devices as they worked on a particularly troublesome fault with the non-directional beacon (NDB). The NDB was the most common and simplest navigation aid at the time. It is a low-frequency 200-400kHz AM transmitter, transmitting a short two- or three-letter identifier in Morse code a couple of times per minute. Before WW2, broadcast stations were used as navigation 84 Silicon Chip aids. With bearings from two stations, you could determine your position on a map or track towards a known location. That was not ideal as few broadcast stations transmitted 24 hours per day, and when broadcast networks became the norm, you could not be dead sure which station you were tuned to. NDBs were a more reliable alternative. The lower frequency gave a better ground wave, with no chance of skip. The Coolangatta beacon’s antenna was electrically short, a single vertical wire supported by several horizontal wires strung between two 22-metre tall towers. These horizontal wires formed a capacitive top-load to increase the antenna current and thus the antenna’s efficiency. The NDB transmitters were a pair of 100W vacuum tube units, providing operational redundancy. They were monitored by a receiver fed from a short whip antenna inside the NDB hut. If any of the monitored parameters fell below the Low-Performance Level, the monitor would change from the running transmitter to the standby. Frequent intermittent faults were causing changeovers. It was determined that the fault was causing a varying carrier level. After a lot of investigation, the fault appeared to be in the antenna itself. The DCA lines section was called in to lower the antenna and investigate its condition. This was reasonable because, being a coastal station, salt corrosion was a likely culprit. However, the antenna checked out OK and was hoisted back into position. The fault persisted. During this process, the trainee (me) was superfluous and left to his own devices. As I wandered about like a lost soul, in one of my walks around the hut, I noticed that the iron roof had a metal drainpipe down one corner that finished just above the ground and level with the ant cap on the building foundation. The drainpipe was not fully anchored and moved in the breeze, bumping into the ant cap. I wedged it back with a piece of timber and sought out the boss. We found that the fault could be induced by pushing the drainpipe against the ant cap. Grounding the roof and downpipe altered the signal strength at the monitor receiver. Problem solved! After completing my training, I was stationed at Charle­ ville in southwest Queensland and became the acting supervisor after a few years. This time, there was another very intermittent fault with the Charleville NDB. It only happened occasionally during wet weather. In this case, the fault kept recurring for a long time with an unknown cause; the short duration made it difficult to pin down. When the beacon was eventually updated, the new installation required re-siting the transmitter in the “transmitter hall” and the complete replacement of the antenna coaxial feeder (changed from 70W to 50W). When the old feeder was removed, they discovered a female-to-female connector under a little ‘sand dune’ in the building’s sub-floor cable duct that dated from the previous NDB upgrade. Heaven knows how long ago that was. It was not weatherproof in any way and showed signs of distress. In the words of Homer Simpson, “D’oh!” The last item is also from Charleville. In the 1970s, before the adoption of SSB high-frequency communications for air/ground communications, comms were amplitude modulated. To cover all of the Flight Information Zone with varying ionospheric conditions, three frequency ranges were used near 3MHz, 6MHz and 8MHz. Australia's electronics magazine siliconchip.com.au The fault this time was interference on one of the 3MHz channels. The interfering signal was the local radio station program (918kHz) mixed with the ident code from the local NDB (267kHz). This was determined by sitting down with a calculator and figuring out what combination of harmonics of these two transmitters fell on the problem receiver frequency. The receiver antenna system was three half-wave dipoles strung between two towers, with the lowest frequency at the top and the highest at the bottom, to maintain the same height relative to the wavelength. The feeders (shielded twin) were laced to a vertical guide wire at the centres of the dipoles. Since the problem manifested itself only in the local receiver, it was likely local. The immediate low-tech solution was to belt the receiver antenna feeders with a broom handle, which alleviated the fault! The fault was located in the supporting guide wire in the receiving antenna system. Initially, the eyes used in the mechanical structure of the supporting cables and other fittings were provided with small plastic sleeves to stop spurious rectifying joints from being formed by contact between the dissimilar metals and/or their oxides, making an unintended but efficient mixer. The plastic sleeves were long gone in the western sun. To clear the problem permanently, all these joints were eventually bonded. Fixing AC ripple in a dual-tracking power supply T. I., of Penguin, Tas had a trusty old power supply until it could no longer be trusted. Some gremlins were lurking within that would need to be dealt with... Following the completion of my electrical apprenticeship last century, I completed a course in Industrial Electronics, culminating in the construction of a Dual Tracking Power Supply kit, the details of which appeared as a project in Electronics Australia in March 1982. The power supply utilises LM317 and LM337 three-terminal regulators and provides ±1.5-22V DC at up to 2A. It has been my main DC source for experimentation in electronics over the years. “Tracking” refers to the magnitude of the negative rail voltage following the positive rail across the entire voltage range. It does this by measuring the positive regulator’s adjust/reference signal, inverting it and feeding it to the negative regulator’s adjust terminal. There is also a fixed 5V reference supplied by a separate regulator. The power supply has performed faultlessly over the years – until recently. Having built Nixie tube projects in the past, I am now in the process of building a VFD (vacuum fluorescent display) clock with a 32,768Hz crystal timing reference. I design, build and test the PCBs using the power supply mentioned above. I completed the crystal oscillator timing board and the divide-by-32,768 circuit to provide the 1Hz count for the clock timing. I connected my CRO lead to observe the 32,768Hz waveform, only to find significant noise on the trace. Although it was a definite sinusoidal waveform, I could not achieve a clean single trace and initially thought that the crystal was possibly being overdriven. However, after spending some unnecessary time changing components siliconchip.com.au around the crystal, I just could not get a clean signal. Instead of a clear trace, it appeared as a sinewave drawn by a 10mm-thick noisy trace. Somewhat frustrated and overdue for lunch, I switched off the AC supply to the power supply with the CRO still connected, and the signal instantly became a clean sinusoidal trace until the power supply’s onboard filter capacitors drained their charge away. That got me wondering whether the unit I’d built all those years ago was in trouble. I was able to prove things weren’t right by powering the crystal oscillator with an alternative DC supply; it produced a perfect trace on the CRO. I then put the CRO leads across my power supply’s output and could see significant AC ripple that obviously shouldn’t be there. My timing circuit was being modulated with AC ripple from the DC supply. I removed the four screws holding on the lid and slid the cover off. I could see four tantalum capacitors and several aluminium electrolytics. Given the age of the unit, I suspected that at least one was faulty. Looking at the circuit, I could see a 1μF tantalum at the input to each regulator, a 100μF electro across the output of each regulator, and a 10μF tantalum across the voltage adjustment potentiometer. I could also see some discolouration on one 120W resistor between the adjust and output terminals of the positive regulator. I clearly needed to remove the PCB and therefore took heaps of photos and marked the wires before going any further. I desoldered the main transformer AC connections plus the wiring to both the regulators, which are mounted on the side of the case for heatsinking. I then removed four other connections to various switches and indicator LEDs. I could then swing the PCB out far enough on the remaining wiring to enable component replacement. While the board was out, I checked the integrity of all the onboard diodes and any suspect dry joints. However, all was good and certainly acceptable, given my inexperience at the time I built it. I replaced the four tantalum capacitors, the two 100μF electros and the discoloured 120W resistor, then set about restoring all wiring connections. After checking and rechecking, I plugged the unit back in with the lid still removed and with fingers crossed, switched it on. Great – no smoke, so a good start. A test of the voltages proved that the unit was functional across the full range. Connecting the CRO leads showed a perfect, ripple-free DC supply. However, I then noticed a red LED fully illuminated. This was the dropout LED, which should only be illuminated if a fault draws too much current on the output so that the regulator drops out of regulation. What was going on here? I had no load connected, and the voltage tested perfectly across the entire range. Spending too much time measuring voltages around the components driving the LED, I finally realised that the sunshine coming through the window (yes, we do get sunshine in Tassie’s winter occasionally) was shining through the back of the LED, which was mounted on the front panel, giving the impression that it was illuminated. Shading the sunlight stopped the glow. No wonder the wrinkles on my forehead keep multiplying! With the lid back on, I connected the crystal oscillator to the power supply and tested the signal with the CRO, to see a perfectly clean trace. Hopefully the unit will serve me for many more years to come. SC Australia's electronics magazine December 2023  85