Fullscreen HTML5Med Res (??MB)HTML4

SERVICEMAN'S LOG Rocking Raucous Retro Roland Repair Dave Thompson I’m trying to stay positive despite the world falling apart around my ears. Earthquakes, plagues, waves of misguided activists – they’re all conspiring to ruin what’s left of our idyllic way of life. At least customers still occasionally find their way to me, with devices that sometimes can still be repaired; in this case, a throwback to the 1980s. And that’s just fine with me, because they knew how to build fixable gear back then. Items Covered This Month • Rocking Roland repair • Samsung aircon repair • Fixing LED light fittings • Repairing a water heater and • • • key-fob Multiple LED downlight failures Repairing a TV with constantly decreasing audio levels Fixing outdoor lighting *Dave Thompson runs PC Anytime in Christchurch, NZ. Website: www.pcanytime.co.nz Email: dave<at>pcanytime.co.nz siliconchip.com.au W e live in ‘interesting times’. The pandemic is disrupting lives, if not directly through infections, then by hammering businesses through the collateral damage of lockdowns, and a drop in people being able to go out and buy goods and services. The economic toll is becoming increasingly apparent and harming us all. There is also significant lobbying going on from all manner of crackpot pressure groups trying to make everyone’s lives worse. They might not see it that way, but there’s no escaping the reality. I get the feeling Australia’s electronics magazine that so many policies these days are not being thought through, especially with politicians now taking advice from school kids rather than listening to the experts. In the electronics world, I’ve been worried about how getting older is affecting my ability to do fine repair work, but I also have all the above in the back of my mind, which doesn’t help my state of mind. Work, while sparse, is still coming in. Recently, a Roland Juno DX2 keyboard came into the shop accompanied by its owner. It had simply stopped August 2021  61 working. Not only were there no dulcet tones one usually associates with a Juno, there was nothing at all. No lights, no power. These vintage 80s-era keyboards are now quite sought after for their genuine ‘analog’ retro sound, so its owner very much wanted it to be fixed. I told him that, in theory, everything was fixable – if not by me then by someone with more experience – and it all really depended on how deep his pockets were. As with many musicians, it turns out his pockets were not very deep at all! Quite shallow, in fact. I advised him that I would assess it and then let him know what was ailing the machine, and it would be up to him as to whether he wanted to carry on. But it was also up to me as to when to pull the pin on any given job, which is part of our serviceman’s creed. He agreed. He had already done some of the work in taking all the screws out from the bottom and cracking the case open, so that saved me some time. He could get this thing apart so easily because he didn’t have any of those stupid security fasteners to deal with. Manufacturers back then were usually sensible, making products that were able and meant to be repaired, with any proprietary spares usually available from dealers (for a good while at least). The electronic components used throughout were often available from any good local electronics store. Access to the circuit boards was usually good as well, with no break-away plastic clips or similar impediments restricting any attempts at repairs. I liked that philosophy then, and I like it now. Dead on arrival Testing this thing was also simple. I plugged in the supplied power lead and hit the “On” button. Result: nada. Zilch. Bupkis. Nothing. The first thing to do was to test the power supply. Many a repair has come unstuck because the serviceman overlooks the patently obvious; that the lead or power supply has failed. It’s easy to do, and I’ve done it myself many times, being too keen to roll up my sleeves and get stuck into a job. One must temper one’s urge to get into it and test the obvious first. In this case, the power lead is likely as old as the machine, and without knowing its history, it could very well have fatigued and failed. The obvious test is to unplug it from the keyboard and stick a couple of multimeter probes into it in a way that I could wiggle it around and ring it out without electrocuting myself. This I did, and I got a healthy 240-odd volts AC no matter what I subjected the cable to. So I was happy that it was OK. The next step was to plug the cable back into the keyboard and test some points in the power supply circuit board. This was separate from the rest of the circuitry and easy to spot, given the transformer, capacitors, diodes and associated leads connecting to it. I love this older hardware; everything is so obvious as to what it does. A quick probe around the board showed that mains potential was going into the transformer but not coming out. This is quite unusual; transformers are one of the most basic of all electronic components and for one to fail is, in my experience, quite rare. I mean, if there’s a 62  Silicon Chip lightning strike or other huge power anomaly, then yes, a transformer can burn out, but in ‘normal’ use, it is unusual. A transformer is essentially just two coils of insulated wire wound onto a soft-iron core; how those coils are arranged determines what type of transformer it is. In the Dominions and other areas where 220-240V is the norm, the primary and secondary windings will be different than in the USA and other territories where the mains voltage is 110-115V. If a device is intended to be used in either location, it will often have two 110-120V primaries that can be connected in series for 220-240V operation or in parallel for 110-120V operation. This is one reason why many modern appliances (computers, printers etc) have a separate power supply; the basic machine runs on the same internal voltages, it is just the supply that differs. Of course, these days, most of those switchmode supplies can run off a wide range of voltages, like 90-250V AC, so they are suitable for use worldwide. Obviously, if there is a frequency-dependent component in the device (that is, it needs the 50Hz or 60Hz signal as a reference to operate), the internal power supply will vary between countries. In this case, however, the Juno was designed to be used in this part of the world, with 220-240V AC mains. I removed the power supply board, a simple operation with only four PK-style screws holding it down, and then desoldered the transformer leads from the board. With the board resting safely on the chassis, I used my non-Variacbranded Variac to slowly apply AC voltage to the circuit downstream from the (now removed) transformer. With my multimeter probes attached to the board’s outputs, I expected to see whatever DC voltages the board was designed to provide. Sadly, I got nothing; perhaps there was more to this than I thought. While I’d need the right transformer, I might also have to see what else was damaged before I could resurrect the Juno. I had hoped I might get lucky, but usually, by the time a transformer blows, there is a lot more collateral damage due to unusually high secondary voltages being produced as it fails. Even though there is a fuse, by the time that blows, a lot of harm can already be done. I might still have many more problems to sort out, but I would have to replace this transformer to find out. The appeals of retro Did I mention I love working on older stuff like this? Each section of the keyboard’s functions sits on a separate circuit board. The VCO (voltage controlled oscillator), VCF (voltage controller filter) and the VCA (voltage controlled amplifier) sections are all separate. The same goes for the keyboard processor, signal processing and audio amplifier sections. If one section fails, a new (or repaired) PCB can simply be installed, and regular operation resumes. This philosophy is unlike how modern instruments are produced, where everything is typically on one big circuit board with propriety COB (chip-on-board) ICs and no spares available from the manufacturer. Even worse, no circuit diagrams generally are provided, whether you are a repair agent or not. If something goes wrong, you usually have to chuck the whole thing away (into a landfill) and pay an exorbitant price for an entire new keyboard, as the cost of repair is Australia’s electronics magazine siliconchip.com.au so high. Nowadays, it seems to be all about IP (‘intellectual property’) protection and making hyper-consumable products with almost zero ability to repair. Back then, for better or worse, they used standard parts and standard (if increasingly clever) circuitry to achieve what they wanted to do. Foreign powers with commercial aspirations often hijacked these designs, calling the resulting device something else, but essentially cloning and copying the original company’s design. Affected manufacturers responded by making it increasingly difficult to reverse-engineer their products, usually by using proprietary parts and making spare parts or components and replacement circuit boards increasingly unavailable, meaning repair was simply not an option. No wonder people are up in arms about huge increases in e-waste and the rise of built-in obsolescence! I’m actually with them. This is just wrong, and while it might be great for IP protection, I don’t think it’s the best way forward. However, as our Juno 2 has discrete components on separate boards, it is a veritable dream for a serviceman like myself to fix. So it seemed that the problem with this machine was in the power supply board. I was hoping that if I could resolve this, the rest of it should still be OK. But like any good mystery, I wouldn’t know until I got the power supply board working. I do know that many of these older analog ICs were pretty hardy devices in their day, so the lack of any obvious burned-black spots on the other PCBs, holes in chips or that distinctive stink of burnt electrical components was a good sign. I was reasonably sure that once I got the power supply board working, the rest of it would start up again and start producing sounds. Fingers crossed! To be thorough, I should test every component on the power supply board. That’s not too onerous a task, to be frank, because there is not much on it. There is a diode rectifier array, two relays, a regulator, a few capacitors and a fuse. The fuse had not blown, so that usually indicates a lack of shorts. The regulator is a standard 78XX type, and the capacitors and diodes are all common components. All are clearly marked, as is the PCB assembly. To really save the planet, modern manufacturers should take a hint from the way this keyboard is manufactured. My first check was to measure the transformer’s output, and I got nothing, indicating that at least one of the windings was open-circuit. Fortunately, the transformer has a part number on it, but a quick Google search found nothing relating to it. The great news is I have a commercial transformer-winding machine. The bad news is that I would have to dig it out of storage in my garage to use it, and that idea wasn’t appealing at all. The good news is that I got about a hundred miscellaneous transformers when I bought the winding machine, but the bad news is I’d have to trawl through those transformers to find a suitable replacement for this one. The good news is that I didn’t have to! I had a Jaycar replacement in the drawer that would not only fit, it would also be suitable electrically. That really is excellent news... This was not so surprising, as all I really needed was a mains transformer with a 12V AC output with a reasonable current capacity, and those are a dollar a dozen. siliconchip.com.au Australia’s electronics magazine August 2021  63 Removing the old one was simple, and replacing it almost too easy. Surely this repair couldn’t be this straightforward. Once I had replaced that, I powered the machine on and... nothing. I knew this was too good to be true. I now had 12V AC, but nothing past the diode array. Out the board came again, and this time I replaced the existing diodes with four 1N4001s. Desoldering these old boards can be tricky, but I have to say I love the smell of that old solder; it brings back many memories. I replaced all the diodes, and for good measure, the regulator as well while everything was out. After reassembly, I hit the power button with expectations of it all working – and it did! The display and all the lights lit up as they should. With an amp plugged in, I hit a few notes and was rewarded with that warm, mellow, smooth, rich, laid-back, melodious, euphonious analog sound. Thank goodness for that! While I likely could have fixed other parts of the circuit if necessary, I was glad that I didn’t have to work my way through it all. The time involved alone would have deterred the owner, and he likely would have bailed on the project. At least now I could tell him that he would probably have many more years of use from this beautiful vintage keyboard. 64  Silicon Chip He was over the moon, and although he said he thought it would be an “easy fix”, he doesn’t know the half of it. Samsung air conditioner repair N. K., of Kedron, Qld likes to wield a soldering iron to repair written-off devices. It’s part of his hobby, and he enjoys the challenge of solving mysteries and saving a lot of money at the same time. In some cases, there is a lot to gain and little to lose... This one was brought to me by my son. His friends, a young couple with small children, had their air conditioner written off by the repairman in the middle of Brisbane’s hot and humid summer. It was an old Samsung split system, and apparently, replacement boards are no longer available. They could ill afford the $2700 quoted for a new system, so my son generously offered to take the boards so he (meaning me) could check them at the component level. So I could not test them in operation. The boards were the indoor and outdoor unit power supplies, the display board and the controller board, hosting the main microcontroller and several other surface-mounted ICs. The repairman said he found dead geckos on both power supply boards and blamed them for the failure. I found evidence of fried gecko on both Australia’s electronics magazine boards, but the marks were confined to the mains areas of the boards. So I doubted that was the real problem. They reported no lights on the indicator panel and the system was completely unresponsive. I figured that if the problem was with the outdoor power supply, or anything else outdoors, the indoor indicator lights should still come on. The outdoor unit is a linear supply with a conventional transformer followed by a four diode rectifier bridge. The fuse was intact. It’s only used to power relays anyway, and the diodes tested OK. There are no Mosfets, so I was confident the problem was elsewhere. I looked briefly at the indoor indicator board and saw nothing visually wrong. There was not much I could have fixed there anyway, so I put it aside. Likewise, the complex controller board looked intact. So I dismissed it as both unlikely and too hard. That left just the indoor power supply board. Its fuse was also intact. It is a switchmode power supply with 230V AC directly rectified to 325V DC. The rectifier bridge checked OK with my multimeter, as did the 400V electrolytic filter capacitor. Tracing the tracks on the board, this fed a TNY266PN offline switcher IC, with Mosfet switches to chop the DC into the primary of the step-down transformer. The secondary, low voltage side of the transformer went to a single half-wave schottky rectifier diode followed by filter capacitors and a small KA78L05AZ linear 5V regulator. There were other components, but they either checked out OK or did not look like suspects. Testing the schottky rectifier diode hit pay dirt. It was short circuit. However, without expensive test gear, I am always suspicious of the black magic lurking in switchmode supplies. You can never tell if something else failed first, damaging the diode, or if the diode failed and took other components with it. In any case, I never trust the Mosfets in a failed switchmode power supply. That TNY266 could have refused to power up due to a sensed low impedance on the primary of the transformer, caused by the shorted diode on the secondary. This, coupled with my overall suspicion about failing Mosfets, led me to replace the TNY266 as well. I could not tell if the KA78L05 5V DC regulator had been damaged by siliconchip.com.au the failed rectifier diode, so I decided to replace it too. The replacement components only cost a few dollars, plus $25 for shipping. I could not find an equivalent axial lead schottky diode, so I used a surface-mount equivalent, soldered to two posts on the board. It was all a gamble, but when my son reinstalled the boards and reconnected the power, the air conditioner sprang to life and started doing its job. So I had the thrill of the hunt, the satisfaction of success, and a suitably impressed son. This also resulted in a very grateful couple and a disappointed air conditioner salesman. Fixing LED light fittings J. N., of Mt Maunganui, New Zealand, had a go at fixing light fittings with failed individual LEDs... I had to replace an outdoor light because it had become too corroded and the housing was letting water in. Fortunately, I had a spare replacement Arlec ABL003 LED unit on hand. After installation, all went well for about three weeks until the light failed. As I am a retired technician and it was out of warranty, I decided to see if it could be repaired. After removing the cover, nothing seemed out of place, so I removed the unit to my workbench. I discovered that it had 18 LEDs in series, powered by an AC-to-DC converter. I applied power and verified that it was producing a reasonable DC voltage. Next, I tested the LEDs and found that two were faulty (marked with red arrows in the photo below). I couldn’t be bothered replacing them even if I could find replacements. Then I realised that I could simply short out the two faulty LEDs by applying solder siliconchip.com.au across them. Upon reapplying power, bingo, the light worked! To be on the safe side, I decided to add a 2.2kW resistor in series with the LEDs to reduce the current through the remaining LEDs to a similar level as it would have been with all of them installed. After a soak test of 48 hours, I reinstalled the light which is still working well after two months. Repairing a small water heater and a faulty key-fob K. D., of Chermside, Qld, had to make two repairs recently, both of which involved fabricating new parts. In both cases, those new parts are far superior to the failed ones that they replaced… I was asked to look at a small unit that heats water to about 40°C and circulates it through a mat. Made thirty years ago, the device had initially been used to keep premature babies warm. It had long been made obsolete from that job and repurposed for use in the laboratory. The complaint was that the unit wasn’t heating the mat. There were four likely points of failure: the element, the control electronics, the pump or the plumbing. Looking at my notes, I had previously replaced the cartridge heater in this unit. The original element was 3/8-inch (9.5mm) in diameter, and the only replacement I could get with suitable electrical ratings was 10mm in diameter. That necessitated the careful reaming of the thin-walled pocket the element fitted into with a chucking reamer. To quickly check the element and control electronics, I measured the power consumption of the unit. I found that the element was clearly Australia’s electronics magazine being cycled on and off by the control electronics. That meant that the failure was most likely in the pump or the plumbing. I disconnected the hoses, and the lack of flow or suction confirmed that the pumping system or plumbing was at fault. Water drained passively from the hose connections, though, indicating that there wasn’t a major blockage in the piping. That left the pump itself as the likely culprit. With the cover removed, I could feel that the rotor of the pump motor was turning, so the impeller had to be the source of the problem. Splitting the integrated pump/heater block required complete disassembly of the unit. All I found in the pump chamber was a protruding shaft. There was simply no impeller to be seen! I did find, however, lots of tiny pieces of Bakelite or phenolic material in the chamber and the water passages. Reassembling these like a jigsaw gave me a flat piece ~40mm x 4.5mm and about 0.8mm thick. It was a very simple impeller that must have been attached to a flat on the shaft with a couple of spots of glue. I thought of various ways to replace the impeller, such as making one from a brass shim or PCB material either glued or soldered to the shaft. Then I realised that I could 3D print a better impeller that would be a press fit and held in position by the flat section on the shaft. I quickly drew a simple design in a 3D modelling package and sent the file off to a colleague for printing. A few days later, I had an impeller ready to fit. It was a snug press fit onto the shaft and turned freely in the pump chamber. When reassembled, the unit pumped far better than anyone remembered. The next repair began when I watched a friend unlock her 2001 Toyota Camry with the key and not the key-fob remote control she usually used. Some questioning led to the explanation that a water bottle had leaked in her bag some days earlier, flooding the remote which had stopped working. It was apparent how the water had gotten in as the rubberised button had perished and fallen out several years ago, leaving a large opening through which the small PCB-mounted switch could be operated directly. Looking into the hole, I could see August 2021  65 several surface-mount components with white corrosion on their leads. I pressed the button a couple of dozen times, and the car responded twice, so I thought it would be worth attempting a repair. The remote was obviously never intended to be serviced, as the case was glued together. Some leverage split the case at the join, revealing an oval PCB containing all the components, including a soldered-in coin cell. Most of the corrosion was near the button and was easily removed with a fibreglass pencil. I then washed the board with isopropyl alcohol. After a couple of days drying in the sun, the car responded to every press of the button. I masked the switch with tape and gave the board a generous coat of Electrolube HPA conformal coating in case of future water ingress. Next, I had to make a new button. I covered the outside of the hole with tape and filled the recess from inside with Dow Corning 3140 conformal silicone. Once cured, and after some trimming with a scalpel, I had a pliable button thoroughly sealed to the case. That left re-joining the case itself. I chose not to glue it, in case I ever had to change the battery. Instead, I used Permatex Form-a-Gasket compound to adhere the case halves together with a watertight joint. The remote has worked for many months, with the homebrew button and case joint still in place. Multiple LED downlight failures R. H., of Ferntree Gully, Vic, must have been busy fixing LED downlights as he has had quite a few fail, as he relates... I was prompted to write this by the LED lamp repair story (Serviceman’s Log; May 2020, page 51). I replaced all our ceiling lights with 12W multi-LED lights of two different brands. Over the last two years we have had sweltering summers, and this appeared to precipitate failures in these lamps. Also, when I installed the first lot of three LED lights in the kitchen, I found that we could not watch TV due to interference. I put about six ferrite rings on each light power cable; that reduced the interference so we could at least watch most TV channels. I ended up shifting the antenna to another side of the house using RG6 quad-shield coax. I then installed another two lights in the dining room, and the second bedroom (my office). I put six or so snap-on ferrite rings on each of the mains power cables again, but still got interference! I can only watch TV with the lights off. As these two rooms were in line with my aerial and the Mt Dandenong transmitter, I had to do another antenna shift; this time positioned so the antenna points away from the house. I also added a masthead amplifier to improve the S/N ratio. We can now watch TV with the lights on. When the weather gets really hot (around 40°C), the roof cavity gets to nearly 60°C, and the LED lamps measure 40°C+ on their faces. Initially, one LED in the group of a dozen or so LEDs in each offending light will flicker annoyingly on/off. Fortunately, at the time of our LED light installation, I purchased an extra spare LED lamp per room. As all the new lights have been fitted with a GPO power point in the roof loft, it was easy enough for me to swap the failing lamp for one of my spares. With multiple lamp failures, rather than throw them away, I have been able to swap good LEDs from a failed unit onto another failed unit to make it work properly again. For our seven installed multi-LED lights, I have changed about 20 individual LEDs on the 120mm aluminium platter. After marking and disassembling the faulty lamp and identifying the LED polarity, I get out my mini gas flame torch. With the 2cm flame burning vertically, I hold the LED platter with pliers and place the faulty LED above the flame. After about five seconds, I can lift off the faulty LED with tweezers and repeat the same to retrieve a good LED from a spare (wrecked) LED light. Again noting the polarity of the replacement LED, I put it where the faulty LED was removed, heat it again with the gas flame (from the reverse side) and the LED will ‘magically slip’ into place using the existing solder. If you look carefully at the LED array photos (mine shown below, and the one published in the May issue), the faulty LED has a black spot. Pretty much every faulty LED I have found suffers from this black arcing spot. The string of series LEDs fails at the point of the weakest LED, and once it has gone open-circuit, the whole string won’t work. The only other fault I have encountered with these LED lights (and with CFLs) is the 2.2µF (sometimes 4.7µF) 400V electrolytic capacitor having a swollen top. Replacing that capacitor usually fixes it. Repairing a TV with persistent lowering of audio levels L. J. C., of Forest Hill, Vic, has a story about an electronic fault that had an unusual cause, leading to a very frustrating intermittent fault... In 1965 my father in law, who lived Burnt spot on faulty LED 66  Silicon Chip Australia’s electronics magazine siliconchip.com.au in a Victorian country town, bought a new TV. I don’t recall the brand, but it was Australian made. It worked well for a few weeks, then developed an annoying fault. While you were watching it, the audio level would slowly decrease, so you would have to get up and turn the volume up. This would continue, so you had to keep turning the volume up periodically until eventually, it was at full volume. After a while, presumably due to a power line spike when a motor turned on or off (eg, a fridge), the fault would disappear, so you had to jump up and turn the volume down! Then, after a few minutes, the cycle repeated; it was most annoying. He had the shop’s TV technician try to fix it a few times, but he never succeeded. Eventually, the set went out of warranty, so I asked him if he would like me to fix it. He agreed. I removed the rear cover; in those days TVs had a circuit diagram conveniently pasted inside. It was essentially a valve TV, but it had transistor audio IF and audio output amps. I was a telephone technician, and my boss had recovered an old TV chassis from the rubbish tip, so I inserted the audio IF and output amp valves in and connected a speaker to use it as a signal tracer. I connected the input of the signal tracer’s audio amp to the input to the set’s audio amp. I then switched the set on, and waited for the volume to decrease. I determined that the fault was in the audio IF stage since the level coming from both speakers decreased. But when I attempted to measure the collector voltage on the first IF transistor, the transient caused the volume to leap back to the original level. Frustrating! I reasoned that the fault might be temperature sensitive, so I put a radiator at the back of the TV to warm up the components (my mother-in-law was not impressed). When the volume eventually reduced, it remained low while I made the measurements. I connected the input of the signal tracer’s IF amp to the collector of the first IF transistor and found that the audio level coming from the signal tracer’s speaker was also low. Looking at the circuit, I noticed that the IF coil was tuned by a 560pF plastic film capacitor. In those days, plastic film capacitors were cylindrical. They siliconchip.com.au The internals of a typical garden LED light. were made by rolling plastic dielectric films with the conducting films, with a pigtail wire emerging from both ends. I concluded that the connection between one of the pigtail wires and the respective metal film was faulty, ie, there was a thin film of oxide between the wire and the metal. When the set was switched on, the transients broke down the insulating film, so the volume was normal. But, as the cap warmed, the insulating film started to reform; thus, the capacitance became smaller, hence gradually detuning the IF stage. But when a transient occurred, either from the mains or me attempting to make a measurement, the insulating film broke down. It became a good Ohmic connection for a while until the insulation started to reform. The gradual detuning by the IF amp reduced the signal level going into the next stage and the FM discriminator, thus reducing the volume. I replaced the capacitor with a new one and thus solved the problem. Fixing simple outdoor lighting F. F. C., of Sydney, NSW likes disassembling broken things and investigating the build quality, finding and fixing problems etc. The subject of this current letter is those cheap outdoor solar lights that are known to fail frequently... These lights are attractive for garden areas, outdoor steps etc because you don’t need to run any wiring to them, and of course, the low cost is the other attraction. The problem is that they never seem Australia’s electronics magazine to last very long. That low cost means that it’s tempting to throw them away when they stop working and buy another one. But often, the fix is quite simple. Opening them up is usually not too difficult, and all you will find inside is a solar panel, a battery, one or more LEDs and a small control board with a handful of components. The ‘battery’ is often a single 18650 Li-ion cell (nominally around 3.7V). If you need a circuit diagram, use your favourite search engine to look up the part code printed or etched into the main chip. As you can see from the photo above, there are only three parts on the tiny PCB. One of the components is a commonly found 4-pin part in a SIL package. If one of these lights fails within the first year or so, the most common cause is corrosion of the battery contacts. While the housing should theoretically be sealed, moisture might still make its way inside, and the contacts will quickly become rusty. That will prevent the battery from charging. Of course, the battery itself can fail over time, but it usually lasts a few years under normal conditions. Another possible failure point is in the wiring to the solar panel, which can be quite fragile. Keep that in mind when you disassemble and reassemble the light to fix it. You could fix the original problem and create a new one if you fracture those connections! The circuitry is so simple that it is unlikely to fail. If it does, you can generally swap the board from another light with a different failure. SC August 2021  67