Fullscreen HTML5Med Res (??MB)HTML4

SERVICEMAN'S LOG Travelling makes me go cuckoo Finally back from a long trip overseas, I had the expectation of a holiday from my holiday, but it wasn’t to be. One of the tacky souvenirs I brought back as a gift was faulty and of course it needed someone to fix it. While most people would throw it away, this was a gift and so I couldn’t help myself and went straight to work. On slow days, most of us day-dream of relaxing in some exotic location, with nothing better to do but to chill in the sun and sample the local delights. Unfortunately, modern travel has put a wet blanket on those dreams for me. After far too many hours standing in queues, lounging about in airports the size of small cities waiting for connecting flights and being crammed into aeroplanes packed to the winglets with irritable travellers, we couldn’t wait to get to where we were going – whether far away or back home. I’ve concluded that this baggageclass travel lark is for other people; next time it will be business class or bust! In theory, technology exists to make life better but I saw plenty of evidence to the contrary on my trip. For example, those body scanners at airports. Not only are they personally invasive but they are actually slower than the traditional pat down and metal-detector approach! On the way out, all the women passengers were diverted from the queue into and through the scanner, and on the way back, all the men were. For those who haven’t had the pleasure, you walk into a large, walk-in wardrobe-sized metal and glass booth, plant your feet on two painted footprints on the floor and hold your hands up as if surrendering – which of course, you are. A back-and-front scanner laterally rotates around 180° and back before an image is displayed for the perennially Dave Thompson* Items Covered This Month • • • Fixing a cuckoo clock Vintage army computer repair Westminster chimes in Oz *Dave Thompson runs PC Anytime in Christchurch, NZ. Website: www.pcanytime.co.nz Email: dave<at>pcanytime.co.nz grumpy operator to view. (Wouldn't you be grumpy too if your job was to stare at images of tired travellers' saggy appendages all day?) While there is a display outside the booth that the passenger can view on stepping out, the security person barked out orders for me to move forward so sharply that I didn’t have a chance to see what it looked like before I got a full pat-down anyway. So what’s the point of these scanners? For another example, smartphones are everywhere now. In many parts of Europe, you can pay for parking, petrol, souvenirs, groceries or pretty much anything else just by using an app, texting a number or holding your phone near a terminal. In the airport, you can use smartphones to display online boarding passes at express check-in terminals and to pass through the departure and boarding gates. The express check-in is great, and a real time-saver, unless (like us) you have bags you can barely lift that need to be checked in manually. However, using the phone for boarding takes longer than when the ground crew check each boarding pass the oldfashioned way, so where’s the benefit here to the weary traveller? On more than one occasion, a passenger couldn’t get the phone to wake up or the scanner to read it correctly, holding up those waiting to board even more. Progress? I’m not so sure. Enough grumbling. . . for now Anyway, after two gruelling days of travel, we were happy to be home, 58 Silicon Chip Australia’s electronics magazine siliconchip.com.au and then came all the unpacking. We’d brought a few souvenirs with us for friends and family, as one does, and we’d packed them very carefully to prevent them from being damaged. YouTube is full of videos of baggage-handling staff at airports around the globe casually kicking or dropping suitcases 15 metres to the ground, or chucking bags from the hold onto the trolleys – and sometimes missing. I have no doubt that most airport workers are diligent but even with our hard-shell cases, we suffered some damage. It’s annoying but there it is; we knew the risks. It is even more annoying when you unpack something you purchased for someone else, only to find it doesn’t actually work. We have an informal but long-standing competition with one couple we know to bring back the cheesiest souvenir for each other from whatever country either of us goes to. In this case, we brought back a small and very cheap and nasty souvenir cuckoo clock purchased from a tacky tourist shop at a famous beer hall in Munich. This is ostensibly a miniature representation of one of the many cathedrals dotted around southern Germany that boast a “glockenspiel”, a mechanical automaton-style display built into the clock tower that comes to life on the hour, every hour and performs sometimes-complex routines in time with pealing and tolling bells. We saw quite a few of these displays from tourist-packed town squares, but none we saw resembled this souvenir siliconchip.com.au version, which includes a tiny, watchsized working clock movement and a pendulum underneath that swings back and forward – or at least, is supposed to. It all looked fine from the outside, but when I opened the flimsy cardboard box and inserted the two hearing-aid style batteries that came with it, nothing much happened. The second hand did advance as expected and the clock ticked away as cheap movements often do, but after 10 minutes, the hour and minute hands hadn’t moved at all and the pendulum stayed stubbornly on one side, no matter how much I helped it to swing. We couldn’t give this thing away like it was; no matter how cheap and cheesy it is, it should at least work. I had to try to get it going. But how can one rationalise spending any real time on fixing a $10 trinket? The Serviceman’s Curse strikes again, of course! Delving into the clock Working on it was a bit of a challenge because it is small and oddlyshaped and there is no flat face at the front on which to lie it down, so I sat it on a sponge. The back half is just a plastic frame but the main body of it is sculpted, painted plaster with tiny figures inside it, making it relatively fragile. So I'd have to be careful handling it during the repair. There are four small neodymium magnets set into the rear moulding to hold it to a fridge. These are mounted Australia’s electronics magazine on the rear corners of the plastic housing. Inside this plastic frame, I could see the clear plastic case of the actual clock mechanism, a very typical cheap movement likely manufactured by the millions in some Chinese factory. Getting to it meant breaking the glue holding the magnet housing to the plaster body and this was achieved with the aid of a craft-knife blade and a little force. With that housing out of the way, I had access to the four tiny screws that held the clock movement together. The time-adjusting handle stuck out from the back of this housing and for those wondering, I’d already played around with that in order to get the hands moving. While I could manipulate the hands with the adjuster, they wouldn’t move under their own steam. It is one of those systems where you pull on the adjuster to engage it and twist it either way to move the hands forward or back, to the correct time. My thinking was that perhaps the adjustment mechanism wasn’t clearing the gears when pushed back in and thus preventing them from moving. No such luck; even after twiddling the adjuster through the entire range, there was no hand movement at all. The second hand still ticked away happily but the time never advanced. As I had to remove the plastic frame first, and this housed the pendulum assembly, I decided to check that next. The pendulum appeared to be moved by some type of electromagnetic system, an elaborate set-up for such a cheap device. The pendulum is simply a painted, heart-shaped plaster weight moulded to a short length of silver wire, pivoting at the very top of the plastic frame and running through a plastic “C” core which must house coils of wire used to create the alternating magnetic field. The problem was that the pendulum was very stiff, so it stayed where it was no matter where in the stroke I put it. I soon saw the problem; the injectionmoulded plastic ‘bearing’ the pendulum pivoted on had come out of its housing and was sitting slightly askew. I tried to pop it back in, but it kept falling back into the misaligned position. I used a bit of pressure to spread the plastic housing apart and removed the pendulum assembly entirely from its mounts and had a closer look at the pivots. December 2018  59 Either it hadn’t been made properly during manufacture, or it had suffered a catastrophic event in transit, because one of the tiny pivot pins had mashed to one side and when I attempted to straighten it, it broke off completely. Excellent! This plastic pin looked to be about half a millimetre in diameter and about 1.5mm long, so replacing it would be tricky. However, I’ve worked on smaller stuff before, so it was out with the microscope and dad’s old box of teenyweeny drills. I was fortunate to inherit these drills and blanks when dad broke down his workshop. Repairs in miniature He’d sourced them when he was making miniature jet engines for model aircraft, using modified car turbochargers for impellers because the bearings could cope with the expected 100,000 RPM shaft speeds. He’d needed to make tiny fuel tubes, mostly from (if memory serves) 1-2mm diameter brass or copper pipes, which I think he also made. He’d needed these drills to bore a series of holes along the sides of the tube; a tricky task for any engineer, but he managed to do it. As different sized holes would change the engine’s performance, he drilled many holes in many tubes and did a lot of experimenting. He’d needed many different-sized drills for this task and had kept a lot of the blanks from having the drills made. These drills were really tiny, some so small you couldn’t even make out the flutes until you got them under a good magnifying glass. They make my Jaycar set of PCB drills look like monsters! I broke out my micrometer and found one the same diameter as the remaining plastic pivot pin (0.45mm diameter) and after trimming off the remainder of the old, damaged pin and squaring off the surface with a craft knife, I used a pin vice with my smallest chuck to manually drill the hole where the old pin was. After going into the plastic block as far as I dared (probably only a couple of millimetres), I simply cut the drill off using a pair of old side-cutters, forming a new pivot pin. I used a Dremel and a small cutting disc to very carefully round off the sharp end of the cut drill, barely touching it to avoid heating it. When done, I re-assembled the pendulum into the housing and tried it; it 60 Silicon Chip now sat square and freely moved back and forth. Hopefully, the clock mechanism would be as easily fixed. Onto the next job I removed the four tiny screws that held the back of the clock on and it came off with the adjuster handle mounted in it. A simple spring arrangement holds the adjuster clear of the clock’s gears until pulled out to move the hands. As mentioned, while the hands do move when adjusted using this method, they just won’t move any other way. My guess is there must be something not making proper contact somewhere in the movement’s gearbox; a gear must have slipped out of position or something like that. The clock movement is a simple quartz type, with a tiny stepper motor and a small gear train that moves the hands. The gears appear to be injection-moulded Nylon, and reasonably well-made; that is, they are clean and clearly defined, unlike many cheap injection-moulded parts. Individually, they all seem to move without binding, as demonstrated by being able to adjust the hands manually, but the problem of why the hands didn’t move became evident when I dug in further. One gear near the start of the train had several teeth missing, perhaps faulty from manufacture or more likely eaten off due to the clock running with the hands stuck or the adjuster preventing gear movement. When I advanced the gear to where there were some teeth, the hands moved as expected, but soon stopped again when the gear came around again. This was the worst-case scenario, as while I have a parts bin full of gears and small cogs recovered from old clocks, printers, scanners, video recorders and various other contraptions over the years, I had nothing remotely like this gear in there. To repair this clock, I’d either need another suitable clock mechanism to replace this one, or a 3D printer and a plan of the gear; none of which I have. I hate being beaten by anything, let alone something as seemingly insignificant as this but it happens all the time, at least in my serviceman’s world; perhaps I should have paid more attention at school. There are always jobs where I discover there are no circuits or parts Australia’s electronics magazine available, or the manufacturer has intentionally obfuscated components, making them next-to-impossible to identify and replace, yet every time it happens it is still a bitter pill to swallow. There is nothing worse than a run of jobs that don’t have positive outcomes, and it transpires that this one will stay broken as well. It’s a shame that after all this we can’t give it to our friends, so after gluing it back together, it now hangs on our fridge. We had to give them another cheesy souvenir that we had (luckily) also purchased when overseas. At least the clock sounds like it is working and the pendulum goes back and forth. That is a fix that I am quite proud of. I’ll take the win no matter how ridiculous it was to do it. Even though the clock doesn’t work, at least it shows the correct time twice a day! Military computer repair These days, if you have a problem with your computer hardware, there are all sorts of diagnostic tools to help you figure out what is wrong. That wasn’t true back in the 70s though; most computers were too expensive and specialised. G. C., of Briar Hill, worked for the Australian Army when he ran into the dreaded intermittent fault with a computer they were evaluating... In the late 1960s, the Australian Army was investigating the possibility of using a computer system to quickly and accurately calculate the angles required to aim artillery guns. A “paper evaluation” concluded that a British Army computer had features more suitable for the Australian Army than those of a similar computer used by the American Army. So an arrangement was made for one of the British computers to be evaluated by the Australian Army. Rather than sending out a British Army technician to look after the computer while it was in Australia, it was cheaper to send an Australian Army technician to England, to be trained on the equipment. I believe the arrangement was between the Australian Government and the manufacturer, Elliott Automation; the system that came to Australia didn’t belong to the British Army. In 1969, I was selected to go to England to do the three-month course on siliconchip.com.au the maintenance of the Field Artillery Computer Equipment (FACE) at the British Army’s School of Electrical and Mechanical Engineering (SEME). The equipment, along with diagnostic equipment and many spare parts, arrived in Australia in 1970. I then became intimately associated with the system, working with it for more than a year. The system comprised six major pieces with many interconnecting cables. These pieces were: the operator’s console, the computer, a program loading unit, a teleprinter, a DC-to-AC inverter (to power the commercial teleprinter) and a power distribution module. Due to the short length of one specific cable, the computer was mounted upside-down on the trolley which was built to hold the lot. A team of Australian Army Artillery personnel had been trained in the use of the system and it was then taken all around Australia, to various Artillery units, to show it off and to have its usefulness evaluated. I went along with the system, to make sure it kept working. It worked flawlessly for about six months, then it developed an intermittent fault. The fault showed up as an error code displayed on the console and the code (9000 from memory) indicated that it was a fault in the computer, but not what the fault was. The computer was an Elliott 920B, which was a lighter weight but ruggedised version of their 920A computer. This was used, among other purposes, to control traffic lights. As I had been trained on the test equipment, I figured that I could easily find the fault. The main piece of diagnostic equipment was the computer test set. All I had to do was undo some of the cables going to the computer module, connect other cables to the computer test set and start the test. A slight hiccup: some of the points the test set needed to monitor didn’t appear on any of the pins of any of the external sockets of the computer, so it had to be opened up and two smaller cables then connected to the internal points. Simple, except that there was the main cover to be removed then an internal electromagnetic shield. The cover was no problem, only 20 large screws to undo. The shield, though, had 64 screws holding it in place. And this was in the days before we had electric screwdrivers. It took about half an hour just to get the test set connected. Once the cover and shield were removed, two printed electronic circuit (PEC) cards had to be pulled out and re-installed using extender PECs. The two smaller cables were then connected to sockets on the extender PECs. The testing with the computer test set was all logical; it tested computer functions (circuits), in a specific order, and then used the tested functions to extend the testing. It had many rotary switches and these had to be switched in specific sequences. At each step I compared the results, shown on nixie tubes, to values in a table in the repair manual. The complete test took about an hour. The first time I did this test to find what the 9000 error code was actually about, the test set indicated that no fault was found. I reckon that I repeated the test about six times and it didn’t find any problems. I disconnected the test set, put the shield and cover back on and re-connected the system. Everything worked correctly; no error code appeared on the console. The system worked for another month or so, then it did it again. I repeated the test and still, no fault showed up. This happened once or twice again and each time, some sequence in the testing seemed to clear the fault before it could be detected. Then the fault started to occur more regularly and I was getting a “bit of stick” from the operators for not being able to fix the equipment. I was beginning to think it was a heat related problem, and that by opening the computer up, the cooling cured the problem. To prove this, when the error code next appeared, I closed the system down and left it overnight to cool down. The photo above shows the teleprinter at left and operator’s console being used, with a labelled diagram at right. This computer used a ferrite core system for memory with a total capacity of 147,456-bits. Refer back to the article in Silicon Chip, March 2014, for an explanation of core memory (siliconchip.com.au/Article/6937). siliconchip.com.au Australia’s electronics magazine December 2018  61 The Field Artillery Computer Equipment, with the Elliott 920B in the foreground. The next morning, the error code showed up immediately the system was switched on. Only running the test sequence cleared the problem. So, I tried to overheat the system to get the unit to fail completely. With the computer opened and the test set hooked up, I had a vertical bank of two-bar electric radiators pouring heat into the computer; still, it didn’t miss a beat on the test set. Finally, the error stayed and going through the test sequences with the test set didn’t cure it. But worse, the test set didn’t identify what the problem was. I got permission from the Australian agents of Elliott Automation to contact their head office, in Britain, directly. The quickest way to make contact, in 1970, was to use the Defence messaging system. This was a teletype system. Elliott Automation had a British Army message centre on their premises, so I could compose a message directly to them. I would write out what the problem was and what I had done and submit the message to an Australian Army message centre. They typed it up on their teletype system and sent it over the submarine cable to Britain. Due to the time difference, I usu- ally had an answer back when I got to work the next day. This was kept up for about a week, with their engineer telling me what to try next. I’d write up the results I’d found in another message before leaving work for the day and have a reply by the next morning. Most of their suggestions involved exchanging various PECs in the computer with a spare. The people at Elliott Automation must have sensed that this slight problem may be about to put the kybosh on the sale of the FACE systems to the Australian Army. The engineer asked me, by message, did I have a home telephone that he could call me on? No, I didn’t, and the Army unit where I worked had closed for the day by the time the engineer was at work in England, so I couldn’t stay back to talk to him. What I did have, though, were parents-in-law who had a home telephone. I arranged for the engineer to call their number at about 8:00pm Australian time and I’d be there to talk to him. This we did for about three days and nothing he could think of worked. “I’m coming out,” he said, and he was there in Sydney in about three days. He observed the fault first hand and stepped through the test procedure, many times, getting the same result as I had. The test set was not finding the fault. His analysis was that it was a fault in the computer memory. This was the only item for which a spare hadn’t been sent out with the system. The computer memory was a ferrite core system, with 147,456 ferrite doughnuts being the storage medium. Its capacity was 8192 18-bit “words”. The engineer then brought out the “big gun” from his luggage, a computer programming keyboard. We hooked it up directly to the computer; the keyboard had its own display. Because I had been taught the computer “language”, its instruction set, on the course that I had attended, he told me to write a program to test each bit of each word of the memory. So, I tried writing all 1s to each bit Westminster chime clock repair 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. J. H., of Nathan, Qld, ran into his own clock problems, with a custom part being faulty. However, his story turned out better than expected... As children growing up in the 40s, both my wife and I lived in households which had a Westminster chimes mantle clock. So as a special gift for my wife’s birthday I presented her with a Napoleon’s Hat Westminster Chimes clock. The clock has given wonderful performances for twenty years. The quartz movement gains so little time that the clock does not need resetting between battery changes. However, just recently, the clock lost its chime function and because of the sentimental value attached to this clock, I thought I should try to repair it. The clock consists of two sections – the quartz movement powered by a 1.5V alkaline cell and the chimes section, independently powered by Australia’s electronics magazine siliconchip.com.au Servicing Stories Wanted 62 and reading the bit back straight away. All good. Then I tried all 0s, still good. Then he suggested a chequer-board pattern, writing “10101...” (18 bits) to a location and test it straight away, then, if good, write “01010...” to the same location, read it back and, if good, go on to the next location. Well, that did it! Finally, the fault showed up as one bit that didn’t change to the appropriate magnetic state when it was being programmed. The fast changing of the magnetic state of that one ferrite core with the chequer-board program identified the problem. The computer test set didn’t perform such a test. A hasty message was sent back to Elliott Automation and a new memory unit was dispatched and it was in Sydney within a week. “These memory units never fail, that’s why there was no spare sent out with the equipment,” the engineer told me, “they are ultra-reliable.” They were also very expensive. The cost of all the other spare parts sent with the computer was insignificant as compared to the cost of one ferrite core memory unit. The evaluation of the FACE system continued, once the new memory unit was installed, and the Army went on to buy many of these systems. A representative of Elliott Automation told me later that the ferrite core in the memory unit that failed had a microscopic crack through it. Silicon Chip two 1.5V alkaline cells in series. Two wires run from the movement to the chimes section and, on the hour, the movement shorts these two wires which then triggers the chimes section to start the hourly chiming and tolling sequence. The faulty chimes unit consists of an epoxy-encapsulated IC about the size of a 10¢ coin which drives a tiny 4cm speaker. The speaker tested OK but there was no way that the epoxy covered IC could be repaired. An internet search for a possible replacement revealed that they cost about $US30, with about as much again for postage. One clock company in the USA had a replacement for about $US8 but the postage was a secret. I emailed them three times for the total cost including postage but never once received a reply. Determined not to be beaten, I had to fall back on my own resources to effect a repair. Surely, I thought, it wouldn’t be too hard to get a microprocessor to play the simple Westminster Chimes tune and then add the appropriate number of tolls. There are only four or five notes involved. I had some older mark 1 Micromites on hand so I used one of the pulse width modulation (PWM) outputs to generate the required frequencies and experimented with the duration and pausing between notes until I had a respectable melody. I used the two wires from the movement signalling the hour to wake the microprocessor from sleep mode, whereupon it would play the chime and toll and then go back to sleep. For good battery life, not only did I use sleep mode but I also set the CPU to run at its slowest possible speed. Also, I set the chimes to cease at 10pm and resume at 7am – as much as I like a chiming clock, the friendship ceases at 10pm. The new circuit was able to fit (just) into the space vacated by the original IC and as I assembled the clock, I thought I had solved the problem. But it was not to be. After a few hours listening to the chimes, I realised they weren’t Westminster chimes at all! Where were the bells? The square wave PWM output was just so mechanical and un-musical. Well, I thought, maybe I could make the sound more interesting by adding a second PWM channel with a note siliconchip.com.au separation of about 8Hz from the first to create a vibrato effect on the note. This did make the sound more interesting but it still sounded like the Westminster Chimes played on bagpipes! So I put the clock back together again with its bagpipe sound but I knew this wasn’t the end. Maybe I would get a reply on that unit from the USA? It was sometime later that I came across the DFPlayer Mini. This device is a 16-pin miniature MP3 player module measuring about 20 x 20mm. With its own 3W audio output stage, it can be configured to play MP3 tracks stored on a microSD card either by a set of momentary contact switches or by commands sent from a microcontroller via a serial port (see the El Cheapo Modules article on page 74 for details on this module). I had already seen websites from which the full set of Westminster chimes could be downloaded in MP3 format. I had played some of these on my computer and they sounded impressive. So here was the solution to my problem. A ménage à trois of the DFPlayer, a set of MP3 chimes on an SD card and a Micromite. It took four weeks for the DFPlayer Mini to arrive from China but in that time, I was able to build a circuit on Veroboard, ready for the module to be dropped in. I also prepared and tested a suitable program for the Micromite. The original clock only chimed on the hour but not on the half or quarter hours. As mentioned previously, the hourly chime was synchronised by the two leads from the clock being shorted together. But now that I had a full set of chimes, I decided to make my program Australia’s electronics magazine incorporate the half and quarter hour chimes also. The hour chime is fully synchronised to the quartz clock but the half and quarter chimes would have to rely on the Micromite’s internal clock. As the Micromite’s internal clock is not very accurate over the long term, it is reset by the program to correct time on the hour as determined by the quartz clock. The DFPlayer finally arrived and all was ready to go when disaster struck! As I was unsoldering the previous set of leads from the clock’s tiny loudspeaker, to connect it to the DFPlayer outputs, the loudspeaker’s terminal connection pad completely separated from the frame. A quick examination of the fine leads going to the loudspeakers coil verified that repair would be impossible. I had some small 6cm speakers salvaged from old computers but they were too big for the allocated space in the clock. So to test the new chimes circuit, I pressed into service a larger 12cm speaker. This speaker is the type usually seen in a car’s audio system on the rear parcel shelf. And what a surprise – with this new bigger and better speaker the clock sounded like Big Ben itself! It wasn’t a disaster after all. I decided to mount the speaker on the rear of the clock – it just fitted neatly but I would now have to remove it every time the clock’s battery needed changing. Also, the power requirements meant that the chimes could no longer be powered from a 3V battery source. I had to use a 5V plugpack supply instead. But this was a small price to pay for such a fantastic outcome. Gone are the tinny sounding chimes and gone is the bagpipe wielding Scotsman. SC December 2018  63