Fullscreen HTML5Med Res (??MB)HTML4

SERVICEMAN'S LOG A feline-themed cautionary tale Dave Thompson Cats can be quite difficult to manage, especially if they each need specialised food. With modern tech you now have devices like a microchip pet feeder, which allow only certain cats access to a particular food/medicine bowl. However, there’s always a caveat with allencompassing one-trick problem solvers. Items Covered This Month • • • Do (not) feed the cat Tektronix TDS744A oscilloscope repair Restoring an electronic organ *Dave Thompson runs PC Anytime in Christchurch, NZ. Website: www.pcanytime.co.nz Email: dave<at>pcanytime.co.nz It’s no secret we own a few cats (or should I say that we serve a few cats?). I have mentioned them before in this column. One of the problems with having multiple cats is that they are very much individual characters, with their own preferences for food and attention. So it is hard to implement a strategy for one cat without affecting the others. For example, one of our cats has allergies to something in the soil around our cat run. Every spring, these allergies flare up, but the other two cats aren’t affected at all. Having to dish out special food or medication to one cat and not the others can be a lesson in frustration, as they all graze on each others’ food. So when we saw an advertisement for “Microchip Pet Feeders”, we thought it might be the answer to our problems. Our cats are all microchipped within a few weeks of birth (as all pets should be). Having the ability to allow one cat to feed from a particular food/medicine bowl while the others have no access is very appealing. That is exactly what these devices claim in their advertising bumf; apparently, you can program access for up to 30 individual animals into each feeder using their unique microchip ID tag. In our situation, we only needed to register the one cat to it. To shop online or not We ordered one of these units online, and it duly arrived on the doorstep. I’m a modern guy; I like this ‘new’ way of buying products; not because I’m lazy, but these days one tries to avoid going out of the house (if one is even allowed!). Clearly, online shopping is boom46 Silicon Chip Australia’s electronics magazine ing at the moment. I have to wonder whether many people will go back to the ‘old ways’ even when restrictions are lifted. Who wants to deal with the hassle of driving to a shopping centre, finding a parking space and pushing your way through crowds to the shop, only to be disappointed that what you want isn’t even in stock? It’s quite convenient just to have stuff show up at your door a couple of days after clicking some buttons on the computer, and overall it doesn’t cost that much more since you are saving on petrol, stress and (perhaps more importantly) time. However, there are still problems with this approach. If you know what you want, or have used the products before, or they are the sort of items you can buy based on specifications (like a lot of electronic devices), it’s perfect because you know exactly what you are getting. But online shopping isn’t that suitable for buying the likes of clothes or shoes, for example. Unless you have gone to a bricksand-mortar store to try on the same product beforehand, and know it fits, buying these things over the Internet can be fraught with problems. What if it doesn’t fit? You are then in a position where you have to go through the hassle of contacting the vendor, returning the item, and possibly paying the return shipping costs. That can mean that your item ends up costing you more than if you had just gone down to the store and purchased it in the first place. At worst, you have no usable product and are out of pocket for all the related expenses. Defects and warranty returns can also be a can of worms. Some online sellers are great about returns, such as siliconchip.com.au most Amazon sellers and local computer stores. But I have experienced vendors who start ‘ducking and diving’ and trying to place the onus on the supplier or manufacturer, not wanting to accept the return themselves. High-street stores are bound by all manner of consumer laws to protect customers, but an online store might be located off-shore and so all that legal responsibility goes out the window. What happens then? Trying to outwit the cats Anyway, back to our new microchip feeder. We liked it so much we purchased another, and although we bought it from a different source, it is the same brand. The cats seem to love them, and weren’t as put off by the movement of the lid or the small amount of mechanical noise they make when opening and closing as we initially thought. However, cats are inherently very crafty and intelligent animals. The first feeder we purchased kept our alpha male out for a while, but he soon learned that in the few seconds it takes for the bowl door to close once the registered cat has left it, he can swoop in and hoover a large amount of any leftover food before the door slowly closes on him. Clever! The door-close-after-eating timing is variable, to a degree, via a three-position slide-switch (short, medium and long delay). But even on the shortest setting, there are a few seconds the contents of the bowl are vulnerable to pilfering. Our A1 male soon made good use of this potential design flaw by sitting very close nearby and swooping siliconchip.com.au when the registered cat leaves. I told you they were clever! Other than that, the device does what it says. There is a food bowl buried under a horizontally-closing bi-folded trap door. Once this door is closed, there is no way for the cat to actively lift the door to access the food or get to it in any way (and they try, believe me!). The door design is smart too; it cannot be opened by simply hooking a claw (or finger) under the split centre section and pulling upward, as the motor’s mechanical lever assembly and natural friction/resistance holds it firmly shut. It could probably be forced open with enough force, but cats aren’t strong enough (at least in the manner Australia’s electronics magazine required) to achieve that. It’s a good system; it works well and is quite robust. They have obviously put much thought into the design. There is a kind of plastic halo over the whole thing, and the cat must stick their head and neck through this to access the food (most microchips are inserted between the pets’ shoulderblades and are thus in the right place to trigger the device). That is probably the biggest aspect any pet would have to get used to. This halo obviously has the antenna for the microchip reader inside it. If the pet is not microchipped, a tag is supplied in the box, and this simple RFID disc can be attached to a pet’s collar and used instead of an embedded chip. As soon as the pet comes close to the February 2021  47 feeder, the sensor picks up the chip/ tag, identifies it and either allows or denies access. If the chip/tag ID is recognised, the door sequence starts. One LED flashes and the motor runs to open the door. Once the pet withdraws its head from the hoop, after the pre-set door-close time, the motor runs back the other way to close the hatch. While relatively simple, it’s a system that’s quite tricky to implement in a low-voltage (6V) battery-powered package. There is no provision for a mains supply, which in my opinion is a major design flaw. Batteries are expensive, wasteful and don’t typically last that long, and these units require pricey alkaline types for ‘optimal’ performance. An AC power option would have been a valuable addition. A turn for the worse However – there is always a however when a serviceman is involved – last week, the first feeder we purchased started misbehaving, with the door not operating correctly. According to the user manual, the indicator LED should flash red once every few seconds when the batteries get low, but that wasn’t happening, so I assumed it wasn’t a power problem. I changed the four C-sized cells for new alkaline ones to be sure, but the problems persisted. The door would not open fully, or then it would open fully but then not close properly. I removed the batteries and tested them with my multimeter, just to satisfy my nagging doubts. Of course, we all know this is not a complete indication of actual battery state anyway, but in lieu of a proper battery tester, a basic voltage test does tell me if they are getting past their best. All measured well over 1.5V, so I was reasonably confident the batteries were still OK. Things got worse over the next 24 hours, with the door often refusing to open at all. This feeder has several buttons that can be used to either program the pet ID, set times or open the door manually. Usually, a press of the manual door button would open the hatch straight away, but this became increasingly erratic. Something had to be done. And this is where the whole online shopping system can start to break down. I tried to find somebody to contact on the original purchase site for 48 Silicon Chip a warranty claim, and it turned out it is almost impossible to reach anyone there. The 0800 number (a free-calling toll number here in NZ) didn’t go anywhere, giving me only a pre-recorded message stating they ‘couldn’t take my call’, but at the same time providing no option for leaving a message. Great! There was also an e-mail form, complete with one of those annoying CAPTCHAs, but when I filled it in and clicked the Submit button, I got a page-not-found fault, along with the claim that the message could not be delivered. Excellent! This is increasingly the case with online sellers. They just don’t have any real customer service. Call me oldfashioned, but it makes much more sense to pay a little more and go to an actual store, where any potential problems can soon be ironed out without all this faffing about. Or maybe someone can open an online shop with slightly higher prices, where you can actually contact someone for help. I know it’s a crazy idea, but it just might work! But none of that helped me now. The only thing left to do was to open it up and have a look. And I also have another feeder to compare this one to, so what could possibly go wrong? Disclaimer: I am not a certified Microchip Pet Feeder serviceman, or even a non-certified feeder serviceman. Just so we’re clear. Four screws held the large moulded plastic back frame on. Once those were removed, the back split away cleanly, exposing the wrapped-wire chip sensor antenna, the underside of the door motor mechanism and a long, narrow, single-sided PCB absolutely stacked from one end to the other with surfacemounted components. I was surprised at the complexity of the circuitry and PCB, but then again, I guess there’s a lot going on in there. There was also a smaller ribbon-cable connected button-board PCB near the top rear of the unit, containing the manual door and programming buttons. Other than that, it was all fresh air inside. Fortunately, and against type, none of the component identification numbers had been obfuscated. I could see there was an ARM microprocessor mounted near the middle of the PCB and several other support chips for it surrounding that. At one end was the power supply section and at the other, Australia’s electronics magazine the motor driver, which included an array of what I assumed to be Mosfets or similar. The door motor appeared to be a simple 6V DC motor – nothing special. I unsoldered the two motor leads from the PCB (handily black and red) and used a benchtop power supply set to 6V to run it backwards and forwards. It operated perfectly and smoothly, so there was no mechanical reason it would be jamming or misbehaving. I wired it back in and used my power supply instead of batteries to power up the unit; the door opened to full travel and sat there trying to open further, with the motor ‘hunting’ slightly. Something was obviously not right. I was also not entirely sure how they were relating the position of the motor and door assembly to the driver circuit; how does it know when the door is fully open or closed? There are only two wires to the motor, so perhaps they are just sensing drive current in the line when the door won’t go any further and feed this information back to the micro to tell it to stop driving. I repositioned the door/motor manually back to closed, and powered on again. Once again, the door opened fully and tried to go further. It didn’t seem to matter where the motor was sitting; it just tried to open up and keep going. At this point, I realised there was little I could do. Without circuits, firmware or anything to work with, it was becoming a waste of my time. My cunning plan goes awry I know what you’re thinking: I have the other one! From a troubleshooting point of view, there is nothing like having another working unit to compare to a faulting one, so I went and brought the working one out to the workshop. Amazingly, though they looked very, very similar, they were actually completely different models. The PCB was very different; the case moulding was different, and even the door mechanism was different. So there was nothing I could use from the working one to relate to the non-working one. Awesome! All I could do at this point was button it all back up and go back to the seller’s website, and try to look for a way forward. After making many approaches with no luck, I ended up going directly to the feeder manufacturer, and they were siliconchip.com.au very happy to help me out. However, there are many hoops I now have to jump through, and I’ll likely end up having to ship this thing at my expense overseas. It isn’t a small package. This is not a great outcome, but better than nothing. In the meantime, we bought yet another feeder to replace this faulty model, and although we got it at a knock-down price (compared to the others), it does leave a bad taste when online vendors don’t play by the rules of civilised shopping. I’m fully prepared to write this one off, especially if it is going to cost too much to rectify. Cynically, I’m sure some of these online sellers take this into account, because at some point, it is just not worth the effort at the end of the day. It’d be a shame, as they are an excellent device and work very well. As an interesting aside, there is another aspect to this repair. The PCB inside the faulty feeder has a tiny screen-printed message visible on the top that reads: “My name is Ozymandias. King of Kings. Look upon my works oh ye mighty, and despair.” What a weird comment to add to your feeder’s PCB design! It is obviously meant to be read by somebody. At least it didn’t read “Help, I am being held inside an electronics manufacturing facility against my will!”. So should I despair? No, service work can be fascinating. Tektronix TDS744A oscilloscope repair A. L. S. of Turramurra, NSW, fixed his Tektronix TDS744A 500MHz 4-channel oscilloscope, but he had to totally disassemble it to get to the root of the fault, as he describes... I purchased this scope secondhand some years ago, and it worked perfectly until one day it refused to boot. It just flashed the LEDs on the front panel. Looking through the service manual, I could find no reference to this type of fault. This manual was obviously not designed for component-level repair because it had no circuit diagrams or even a block diagram! It had diagnostic procedures to isolate faults to a particular module, but it has to power up first, so that was no help. I immediately jumped to the conclusion that the power supply module was the culprit, so I started searching for a replacement or any information on it. siliconchip.com.au After trawling the net for some time, I discovered many other faults which are common to this model such as acquisition board failures, attenuator failures and poor SMD electrolytic capacitors, but nothing on power supply problems or any schematics. One guy on YouTube had the same flashing LEDs, but it was for an HP spectrum analyser, and it required a complex repair of the switchmode power supply. I also found some very good YouTube teardowns and repairs of this model, and some useful tips. They mentioned that there is an internal “protection” switch which can be accessed from the side panel through a hole. Sometimes, switching this can bring the scope to life. I switched it to protection mode, but nothing happened. So it was time to open it up and take a look. I thought there might be some visible evidence of burnt-out components, or perhaps it was just an internal fuse that had blown, or it just needed a reset. I always leave devices with highvoltage CRT power supplies alone for at least a couple of days to allow everything to discharge, especially when there is no available information on exactly where the high voltage is! After removing four screws at the back, the rear panel and outer case came apart with a light nudge from a rubber mallet. I removed two Tektronix “calibration is void if removed” stickers given that the warranty period had expired over a decade ago. This proved that the instrument had not been opened or messed about by the previous owner, at least since its last calibration. The internal layout was beautiful and well-designed for servicing. Each large PCB was easy to extract, starting with the one on top, which is the processor/display board with eight connectors. After these were removed, plus a small panel which has a Centronics connector and an RS-232 connector, the PCB slid out. Next, there is a large aluminium protector board with a high-voltage warning. I had to remove several screws to get that one out. I also had to gently lever it out of a slot, as it was very stiff. The power module was then exposed, so I removed it for a closer look. It was a pretty heavy board with a fairly standard switchmode architecture, capable of delivering all the low voltages for the scope. There were signs of overheating stress, but there were no immediately apparent shorts or problems, and all the electrolytics checked out fine incircuit. The inability to boot made it impossible to check the voltages given in the manual, so I had to look into other possibilities. I couldn’t find any power modules or identical ‘parts’ scopes for sale, except for a few scopes that probably had failed power supplies. But further research indicated that the TDS684A (1GHz, 5GS/s model) and the TDS784A (1GHz, 4GS/s model) have identical power modules and identical processor/display modules, so I took another look. I found one broken TDS684A for sale, which showed some activity on a very damaged and dull CRT screen. I therefore deduced it must have a functional power module, so I bought it for around $350. The bad screen wasn’t a huge worry because this model has a VGA output. My idea was to swap the modules to get my 500MHz scope working, and maybe even repair the 1GHz scope and get it working too! The non-working 1GHz scope arrived after a couple of weeks, and I hurriedly checked it out, but it was worse than the eBay seller’s photos led me to believe. The raster and graticule were folded over at the bottom of the screen. Was there something wrong with the power supply, or was the vertical amplifier or scan coil faulty? Would an external monitor even work? Also, the seller told me that he was pretty sure that it was showing all four traces, but there were no traces at all, Servicing Stories Wanted Do you have any good servicing stories that you would like to share in The Serviceman column? If so, why not send those stories in to us? 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. Australia’s electronics magazine February 2021  49 so the acquisition board was probably kaput too! This appears to be a common problem with this model. It’s possible that the seller assembled the whole scope out of junk modules from previous repair jobs. I decided to try swapping the power module anyway – after all, it did boot up. So I tore down the spare scope and extracted the ‘good’ power supply board. Because I was getting good at disassembly, I had it swapped in no time. Then, after checking all the connectors five times and cautiously plugged the thing in, I stood back in anticipation of sparks and blue flashes and pressed the “on” button at arm’s length. The LEDs were flashing again, and no boot up at all! The fault must be on one of the other modules; perhaps the processor/display module. So I swapped that too. Again, nothing, just those darn flashing LEDs! I noticed that on top of the processor module was a single 7-segment red LED display, and it was flashing “8” in time with the LEDs on the front panel. I could find no mention of what this means anywhere. So, I thought I had better change the acquisition board, which required delving deeper because the screws into the BNC attenuators were behind the front control panel. The whole instrument had to be inverted because this module is on the bottom. The front plastic panel had to be removed and this was supposed to “snap off” (according to the manual), but 20 years of grease, dust and grime acted like an excellent glue, so I had to use two screwdrivers in tyre-lever fashion to ever-so-gently prise it off. Of course, it cracked, but luckily the crack was on the bottom, and it was almost invisible. Then I had to remove the BNC cover along with the front control panel to expose four screws which finally released the acquisition board. I smartly swapped this, hoping it would cure the boot problem at last. But no! I was still getting the LED light show. What next? Deep inside the bowels of the instrument, there are several PCBs which are stacked like a house of cards, inside a three-sided metal box. From this emanated myriad wires connecting the cathode ray tube (CRT) and the EHT tripler, along with the scan coil drivers and output transistors and some ICs. There was also a large processor IC, and I was horrified to see that the EHT cable had touched this, leaving black soot on it, but it was all part of the design to squeeze everything into a tiny space. The CRT and its associated scan coils, rear PCB and correction magnets were also squeezed into this box. This was all that was left, so the fault must be there somewhere, mustn’t it? I was beginning to doubt my ability to fix the two scopes, but I had already committed enough cash to motivate me to continue. The next step was to inspect this daunting mess, so I disconnected everything, taking care to note where all the wires went. I disconnected all the cables and connectors except for the EHT lead, which was blocked by the thick aluminium chassis. So the CRT had to come out, but first, more stuff had to be removed such as the softkeys, which were mounted on a thin metal bezel, and also the floppy disk drive. Then, very gently, I pulled out the CRT to the limit of the high voltage cable, about 4cm out of its housing. I was only just able to remove it from the tube by lifting the rubber insulator on the side of the tube and lightly squeezing the prongs with a medical clamp to release it. I made sure it was fully discharged by shorting these metal prongs to the chassis. Fortunately, the two-day safety discharge period I had allowed had done its job. I have some old-world experience with TV repair, so I was extra careful not to knock the skinny end of the tube because that is the weakest part. I also wore safety glasses because if these things are broken, they can implode and fling glass everywhere. I held it with all the delicacy of a newborn baby and stacked it face down in a cardboard box and set it aside away from harm. Now the whole instrument was stripped down to the bone (as shown at lower left), and all that was left was the box of components with the highly suspect board. Several electrolytics had to be bent out of the way to access the screws which needed to come out. Some of the electrolytics completely blocked the screwdriver access. I don’t know why they didn’t measure the electrolytics before they allocated the screw positions. Despite having removed the screws, the assembly just wouldn’t come out. It would not fit through the gap left The Tektronix oscilloscope was taken apart to determine the suspect components/boards. ► Some burnt and shorted tracks were found on the EHT module around the TIP30C driver transistor and associated electros. 50 Silicon Chip Australia’s electronics magazine siliconchip.com.au by the CRT, but by twisting it on its long axis and angling it in a certain position, it finally came out like one of those kids’ puzzle games. On the EHT module, I found some burnt and shorted tracks around the TIP30C driver transistor and its associated electrolytic capacitor (shown opposite at lower right). This was not visible from the top, and the cause was not immediately evident until I removed a capacitor and a transistor. This short had obviously caused an overload, shutting down the whole instrument. I repeated the extraction of the EHT assembly on the very cannibalised TDS684 1GHz scope, leaving it as a bare chassis. By this time, I had become very proficient with these techniques, and the extraction took just five minutes. I mounted up the suspect “spare” EHT module in my scope and inserted the original CRT to see what would happen. Well, it finally booted but as you might have guessed, that wasn’t the end of my problems! The raster and colour display were unexpectedly good, but I could not get a trace, and that meant that the substitute acquisition board was bad. Now I had to retrace my steps and put the original acquisition board back in, which proved pretty easy because I had so much practice. After I restored this board I got a trace, but the gremlins had multiplied, and error messages plastered themselves over the screen. Retracing my footsteps, I restored the original processor/display module and waited for the boot-up, which takes about a minute. I thought I was dreaming as everything worked perfectly; the self-test passed the display was steady, and it even came up with my last settings! The only downside was that the brightness was a little lower than it had been; that can be improved with the internal brightness adjustment, but I had just about had it, so that would have to wait for another time. If I can get the 1GHz acquisition board going, I can extend the scope’s bandwidth from 500MHz to 1GHz, so I will have a go at that later. For now, I just want to enjoy my once again working scope! Restoring an electronic organ K. V. of Kallangur, Qld, has put some time and effort into restoring electronsiliconchip.com.au ic organs, which are sought after these days, and quite valuable... Some years ago, my wife and I were given an old Hammond “Grandee” electronic organ. My wife can play quite well, but I can only fiddle with something electrical. The organ was on the way to the dump, beyond repair – but I was given first choice! I eventually got it all working thanks to a gentleman in Sydney, who obtained a service manual for me. This was a big help because one circuit board was missing. Apart from that, the cabinet had been home to a family of mice for some years, leaving quite a mess! My son helped me make up a new circuit board, and after a big cleanup, it all came to life. One of the first things I had to do was to add a speaker switchon delay to alleviate the loud thumps. Electronics Australia published a letter I wrote about these repairs back in 1997. Most of the problems we have had over the last 20 odd years have been because of poor contact in the many plug-in connections. It is good to be able to remove a circuit board to check by unplugging it, but generally, there is nothing wrong with it. Clean the contacts, a little wipe with Vaseline, plug it in and it goes! One problem that took 20 years to track down was that the hum level was higher than it should be. I had changed all the electrolytics in the power supply with very little improvement. The hum level on the two amplifiers was below Hammond’s specified level, so I left it at that. The connections to the organ from the power supply and amplifiers were by two 15-pin plugs. I never did like them. If they were wriggled, they made scratchy noises in the speakers. Pluggable terminal blocks looked to be the answer. These plugs carried a mixture of various DC voltages, signal voltages and mains voltages. The modifications involved a fair bit of work, but it was worth it. While I had the chassis out, I thought I might as well change the Leslie speaker plug to a pluggable terminal too. The Hammond drawing on the Leslie speaker shows six wires, two for the speaker and two each for the highspeed and low-speed motors. But this organ had five wires, not six! When this organ was built, someone decided to save a bit of wire (about 800mm), because all the returns were Australia’s electronics magazine February 2021  51 terminated off the one Earth bar on the chassis – but they added another 6-pin plug and socket at the Leslie speaker end. I decided to wire it up as per Hammond’s drawings and delete the extra 6-pin plug. It was superfluous. When the organ was put back together, everything worked OK. My wife played a few tunes and was quite happy with it. That annoying hum level had gone! I sketched out the connections of the Leslie speaker showing how they were and how they should be. The tremolo speaker return connected to the fast motor return, which is always running, then through two doubtful plug contacts to the common Earth bar. The tremolo speaker had every chance of picking up some 50Hz current, producing the hum. 52 Silicon Chip Sometimes the designers’ plans are not always carried out on the workshop floor, but the organ worked – so out it goes to be sold! All the soldering in the Hammond was excellent and I never had any trouble with dry soldered joints. That reminds me of another organ I had to repair, a Baldwin “Fanfare” built in 1977. It had been sitting idle for years, and it too had become the residence of mice. That meant another big cleanup, checking and replacing corroded contacts. When it was ready, I connected it to a variac and slowly increased the voltage in stages. At full voltage, most of the organ worked. About this time, a service manual arrived from W. D. Greenhill & Co in England, so I was able to check the power supply. I found that only one section Australia’s electronics magazine was within tolerance. I removed the power supply to the workbench, replaced the electrolytic capacitors and one open-circuit transistor. All voltages were set within the ±3% as specified. When it was replaced in the organ, the -12V supply was down to -7.5V, and clearly overloaded. A bit of circuit tracing revealed that the -15V and -12V supplies were crossed over in a plug. The organ worked much better after correcting this error. This is another case of an original fault that was ‘allowed through’ because most of the instrument worked! I did eventually get this organ all working after tracking down numerous faults, corroded connectors, poor solder joints and some faulty ICs, as well as fitting a delay relay to the speakers. SC siliconchip.com.au