Jump to content


  • Posts

  • Joined

  • Last visited

  • Days Won


Everything posted by Womble

  1. Shinobi is System 16 pinout, not JAMMA. It looks like JAMMA, smells like JAMMA but you'll damage the board if you plug it into JAMMA.
  2. Cool, I did wonder as the photo you posted showed what looked like newly labelled 2732s which was probably wrong for a DK. In terms of likelihood, an actual bad EPROM at 5A is pretty low down the list, the fault may be located at 5A but I would double check the chip itself physically first (legs bent under rather than correctly sat in the socket) or tarnished legs. Also check it isn't in backwards, if it has been in backwards when the power was applied then the chip is toast. Also, the sockets on these boards only make contact on the outside of each leg, so I'd make sure the outer edges are clean, you can also very slightly bow the legs outwards a bit, so they press harder on the socket contacts. Tread carefully if doing that as it can make it harder to get the chip back into the socket correctly. Beyond that I'd be point a scope, or at minimum a logic probe at the control signals at that chip, /CE, /OE and /WR, as it is more likely that the board has a logic fault that isn't controlling that chip correctly. But it could be a bad EPROM, or be nothing at all to do with 5A. If was a betting man I'd put money on a logic fault.
  3. Assuming your voltage are ok, it's likely a contact issue, the ribbon cable and associated connectors are known to cause issues after nearly 40 years. Ideally you'd replace all board interconnect cables and sockets with new ones but its a sod of a job. They can drive you mad with intermittent flakyness that seems to come and go randomly. Single wipe sockets can cause issues too but the ones on DK are usually pretty robust. It could also be a logic fault, have seen both, but I'd start with the ribbon cables, if they aren't an issue today, they will be at some point. Btw - The black pins on the CPU isn't heat related, those pins were silver plated, the black is a reaction to sulphur dioxide in the atmosphere over many years. The dark spots under the RAM chips is heat related, as those RAMs run pretty hot, totally normal. Also - just check your mate didn't burn 2732s to replace 2532s, they aren't pinout compatible, DK can take both but you'd need to modify the PCB links (from memory) to change the type.
  4. Machine wise it would be the ZX Spectrum 48k, that was my only machine for about ten years, until I bought a second hand Atari ST. On the games side it would be Gunfright, one of Ultimate's isometric titles. Is probably the least rated of their games but I always used to play it as a kid, its still good for a quick spin.
  5. I had a QL for a while as part of the collection, a bit of a dull machine really as there is next to no software for it, and what there is looks very basic. I couldn't find anything that really taxed the machine. On the speccy it's worth checking out some of the games that use modern multicolour engines like Bifrost and Nirvana. Totally amazing what can be achieved with these on stock hardware, would have been game changers back in the day.
  6. Can I grab the LAI Dual Coin slot if it's still available? Sending PM too.
  7. My bets would be... Still bad RAM in the socket Good RAM in socket but track damage under the socket from when the original ram was removed. Logic fault somewhere nearby
  8. Usually the 63701 error means the 63701's RAM (IC22) is bad, I've never met a bad 63701 itself. Hard to tell from the angle but it looks like IC22 has already been socketed, although that might be a shadow in the image. If it is socketed might be an easy swap, or it could be a sign that someone has already tried replacing that chip.. Is IC22 in a socket?
  9. Actually, if you do get activity on pin 37, can you check for activity on pins 3 and 39 as well.
  10. Depending on levels you may or may not get any useful signs of life by probing the crystal's legs, but if you probe pin 37 on the CPU you should see extremely fast activity. You don't really need to know how fast its, you are only looking for activity, clock faults are almost always an "all or nothing" affair, if there is life you can assume its correct, but if you get zero activity then you have a clock problem somewhere upstream of that. While you are at the CPU, poke pin 40 as well.
  11. Probably normal, they're the big smoothing caps in the power supply feeding the amp. On the board as one will be wired across the negative power rail, and one across the positive so one will look backwards if they didn't run the traces in a more complicated way.
  12. Hmm, nothing obvious certainly, it looks like you get the SAII timeout error if the CPU isn't running and doesn't replying in time.. Basically I'd test 1) Is the crystal oscillating. 2) Clock signal at CPU 3) Dump and read all ROMs - Reading all the ROMs has the side benefit of reseating them all, and you can check for bent pins too. 4) Test 6264 RAM 5) Replace 6502 CPU - for some reason certain 6502s just die, I've seen quite a few over the years. Arcade faults are almost never caused by a bad CPU, unless its a 6502. Do you have a scope or logic probe? Looks like everything useful is socketed, so I'd swap in a new 6264 SRAM chip and test, then do the same with a borrowed 6502.
  13. Hmm, that error is for the microprocessor section of the sound PCB, so it could be anything up to and including the connection to the main board. Could you upload a photo of the sound PCB? All google finds are bad photos of links to dead eBay auction photos. I might be able to suggest tests once I can see what it is made up of.
  14. Large projects always go massively over budget, because the lowest bidder wins the tender, then they find out all the things they forgot to take into account and the price blows out. Holding them to the original quote just means that company goes bust and the project collapses, so you are better off paying the variations to keep it going. Any group that prices it up correctly is never going to win because they will be almost twice the price of the other mob who say they can do it, but probably actually know they can't.
  15. Ah sorry, no I didn't go that far, you would need a heat bar bending machine to get a uniform bend, trying it with a heatgun would get a very rippled and lumpy effect I expect. It should be quite easy to measure up the angles and the spacing from the cabinet itself, the fit isn't tight at all (probably on purpose as the cabinet has a bit of flex in it) and the metal brackets cover up any gaps.
  16. Thanks guys, am really pleased with how it turned out, and that I actually finished it in a reasonable time frame. @Shan - the canopy came from a local perspex supplier, its standard stuff (the original tint was "bronze" and seems to be a standard colour), it just needs to be cut and bent to the right angles. They used the smashed pieces from another cabinet's canopy as the template and made up two sets, one for that cabinet and one for me.
  17. I can help, will drop you a PM later today.
  18. Some people don't rate it but I actually think its is just as playable, plus it pushes the vector graphics harder than the original and has a tonne more speech going on, technically it's more impressive even if the game flow is weaker. In SW you end the level by destroying the Death Star, but the flagship level in ESB is in the middle with the AT-AT and the ST Walkers. In ESB the final stage which is a rather odd avoid-the-meteors stage before you see a tiny Millennium Falcon fly off before the game loops. Anyway, it's a few daughter-boards that go on in place of the original ROMs, and the original CPU slots into a socket on the daughter-board that sits in the CPU socket.. ... and finally I actually need some keys for the cabinet. I could replace the locks but the cost to do that is about the same as getting keys made for the locks it came with, or buying key blanks, because I had picked up a tubular key pick, decoder and cutter. No point doing things the easy way after all. Part of the reason for keeping the original locks is that the two doors on Atari over/under coin doors should have different keys, so the guy collecting the coins can't get into the top door where the service panel lives, and the service guy can't get into the bottom door to pinch some coins. As I'm keeping it original I might as well stick with the two different locks it came with, I just need to decode the locks to get the pin pattern and cut me some keys. Its relatively simple to decode them and I did that bit ages ago, once the pinning tool has formed itself into the pattern that unlocks the door you then read off the pin depths against the stepped key to get the numbers, which I'll put here for Google to find. Upper - Lock code M1222 = Key Code 1533731 Lower - Lock code M1224 = Key Code 1547135 After some fighting with the cutting tool I ended up with a working key for each, and do to mark the end of the restoration, in the same was I did Vindicators, I got these made up. The StarWars key for the top door... ...and the ESB key for the bottom door. The back door goes on for the last time... ...the back door key goes on the hook inside the service door... .. and with the coin doors finally locked up that really is it. Restoration Complete!!! All up it took me 9 months, far longer than I thought, but shorter than some of my grand plans. So how does it play? Really really well, the controller is smooth as silk, the sound is crystal clear and the vector monitor just looks stunning. The mini Wombles love it! The force is strong with this one... That's all folks, thanks for watching.
  19. It seemed simpler to cut that wedge free again and rotate it with the yoke rather than trying to unpick the glue that was on the inner face close to the fragile wiring. I'd previously glued the wedge back over the old marks on the glass face so maybe this tube was always slightly wonky, now it is bang on level. New silicone, masking tape and a Lego brick for leverage clamped it back in place to cure. The next bit to go back in is the cardboard bezel, which is in pretty good condition considering the rodent party in the cabinet. It had just sagged a bit over the years, which looks dreadful and lets light in around the bottom of the tube... ...and hangs down over the image at the top. It was also very dusty so it met the vacuum cleaner and go a wipe down with a damp cloth until clean. I need a way to reform this and give the horizontal surfaces enough strength to stay level, without adding much weight to it, or I'd just move the sagging to somewhere else. I opted for a balsa wood rod, strong and insanely light... ...glued on the top and bottom edges ... Probably enough clamps and pegs! The only remaining issue was the bowing lower left section giving it a rather lopsided look. Turns out you can iron cardboard flat again... ... and it works really well, but there is a fair chance it wouldn't stay like it for long under gravity, an icy-pole stick glued to the outside edge should prevent it reverting to form. Once that was dry the tube face got one last clean, the bezel slotted in, and got three staples shot in where the originals had been. With no need for any further access to the tube, the rear upper panel can go back on. Am very happy the tube is finally out of harms way, with kids in the house a bare tube sat on the floor is just asking for trouble. With the bezel in place I can correctly set the X and Y size of the image using the two piggy backed trim pots on the deflection board, and also go through the calibration steps in the manual to set up the vectors correctly. Ideally I'd have done this during the bench testing, as now I have to sit behind the tube to make the board adjustments, so I can't see whats happening on the screen. Some people use a mirror for this but I don't have a portable mirror that's big enough. I ended up finding an app that lets you mirror one iPhone's screen to an iPad on the same wifi network, so I pointed my phone at the screen on a tripod and had the iPad round the back where the adjustments were. The resolution wasn't great but it let me square things up well enough. The test mode seems to make setting up the monitor far harder than it should be, I found the easiest was to get the size and position right with the grid, then set attract mode running and dial in the rest when the Star Wars logo is on the screen, much easier. Now I'm not bothered by reflections during set up the glass comes out of storage for a good cleaning. Unfortunately it has some rather obvious scratches on the front, possibly from the same douche-bag who attacked the controller with something sharp... ...but amazingly you don't see them if the glass is reversed. So the mint inner face is now the outer face and all the blemishes vanish. Could pass as a new panel! The cockpit liner also comes out of hibernation and gets a good clean up... ...before slotting back in and securing with 4 painted bolts. This clamps the glass firmly in place and as the opening is slightly smaller than the open face of the bezel you don't see the sticks glued on the outer edge. The CP then bolts in with three bolts on the remade kick panel, and two larger ones in the upper face. The controller slots in next... ... with me sticking one arm through the coin door to put one nut on while I hold it in place with the other arm. The final harness connection is made inside hooking up the controller means this should be playable again. Can't fight the empire without a roof though, so the Perspex roofing sheets finally get unwrapped.... ...and installed with the two metal brackets to hold it all in place, including felt washers again. There's one final part that's been in storage since day one, the glass marquee from the rear light box. It's also the last item that's still as filthy as day one... ...even the inner face. The sections where the decal has worn away are where it has rubbed on the cabinet body, meaning those sections are hard up against the structure of the cabinet, so the gaps shouldn't show through, hopefully. The panel slots up into grooves in the bodywork but needs the two metal brackets to hold it safely in place, and those originally had some foam strips on the contact edge to stop the glass rattling about. I'd picked some up from Clarke Rubber when I bought the first mat... ...and it just sticks on the edges that contact the glass to provide a cushioned edge. Three more painted bolts go in to secure it in place with the lower bracket, it's one of those jobs where three hands would be useful as you need to hold the glass, get the lower bracket lined up and get at least one of the lower three bolts in place. Once that is secure the top plate goes on to finish the job. The result looks amazing, and really finishes off the cabinet... ... and it looks even better when back lit. That's the rebuild complete, or is it? Not quite. The short ends of the fat T-moulding just won't stay in the slot, its such thick stuff that it wants to stay straight. The bends at the back are so sharp and there isn't enough of a run after the bend to provide enough grip. At the factory Atari punched in staples at both ends of this stuff, I'd assumed that was just to speed up production as the front ones aren't needed, it seems the back ones are. So I borrowed and air stapler and compressor and fired some in before and after the bend to stop it bowing out. Once painted black these are next to invisible. Next is an Empire Strikes Back upgrade kit from Vector Labs which lets you swap between the Star Wars and the arcade sequel using a combination of button presses on the controller. ESB was actually the third Star Wars game released by Atari, with Return of the Jedi in between the original SW and ESB. The Star Wars game was released first in 1983, 6 years after the 1977 movie it was based on and it was a smash hit in the Arcades. Atari followed Star Wars with a Return of the Jedi game in 1984 using a raster monitor making it a totally different animal to the vector games, and incompatible with this cabinet. A year later in 1985 Atari released Empire Strikes Back as an upgrade kit for the original Star Wars cabinets as a ROM swap upgrade for the original Star Wars cabinets. Apparently it was a commercial failure with most operators ignoring it. Some say its just not as good a game, but the most plausible is that the original Star Wars game was still raking in coins and operators didn't want to risk disrupting it by replacing it with an untested game. They already had a very successful game so there was little point spending a tonne of money on an upgrade in the hopes they would still have a successful game. I remember seeing Star Wars cockpits in arcades around 1987 but rarely played them as they were the most expensive game in the place, at least twice the price of other games, maybe even four times. Clearly not a game that was going to be put out the back and set to 10p plays unless it gets an upgrade.
  20. The ARII was still bolted to the outer bracket from the bench testing and it makes sense to install it with the cage rather than trying to refit it inside the cabinet. The the whole assembly slides in on a lower rail, with two bolts on the upper edge. With that done the PCB can go home. The filter board goes back on the board stack and all the wiring gets hooked up. Once that's done it's easy to see where the remaining cable mounts are supposed to be screwed to the cabinet, so those go in too. Now the front end and back end are rebuilt I'm really only left with the middle, and that didn't quite go to plan. The powder coated seat-bracket comes out of storage and is fixed to the board with 3 large screws, the speaker harness plugs in and is tied down. All good except for the speaker contact that fractured and fell apart on me. . I'm hoping this is the last victim of the mice, with the connector just corroding enought to lose its springiness. So the harness comes off and a new spade connector goes on And then I find the panel doesn't fit. Basically I can't get the seat edge to engage with the seat surface panel because the speaker panel is rotating up at too tight an angle, and the lower metal lip is fouling on the edge. So this is why the original mats were smaller than the one I made a perfect fit, to give more wiggle room - ah crap! The only way I could get the panel to engage was to have the lower edge of the panel sat on the edge of the mat, but then the screw holes didn't line up. The mat also gripped the lower edge so well I couldn't knock it in to place. Even if I could it means this would never come off again as I'd need to be inside to kick it out. My options seemed to be limited to cutting the mat back a bit, trimming the lower edge of the panel, or removing the lower leading edge of the seat to give the bracket some rotation room. The mat is sacred, no one is going near that with a knife, even me. Plus it would mean pulling up the trimmed strip, which would probably pulling up more glue and base board, it gave me nightmares just thinking about it. Trimming the bottom of the panel would leave a gap and an unpainted edge which may be obvious when it is installed. Plus the process risks trashing the paint job, or scuffing the bolts and grilles. The safest option at this late stage was to take some meat off the lower edge of the seat board, the only draw back was that now isnt the time to be creating a load of sanding dust. I considered using the router, but as I'd need the guide bearing above the cutter I'd have to import a cutter a great expense. Or I could just use the sander and aim for a consistent cut. After raiding the laundry for sheets and blankets to mummify the cabinet... ... I sand back a 45 degree angle and make a filthy mess in the process. This doesn't weaken the seat as the bracket provides a lot of under-support, and the weakened face is well enclosed. Once the dust settled the speaker panel could finally rotate into place. The four screws finally go in. Next up was the last great unknown, the monitor - refurbished, minty fresh but totally untested. At this point the high voltage board was still missing the BU406D transistor that drives the line output transformer (LOPT - aka Flyback in the US) so it can't produce the HV to charge up the tube yet. That makes it safer to do one last preliminary check to verify if the +24 & -24 regulators, and the 555 timer circuit were doing their thing. The board hooks quite nicely onto the door sill woodwork giving me multi-meter and probe access to the underside. The voltages were in the acceptable range, apparently the 50W resistors across the regulators make the measurements a bit screwy, so onto the 555 timer, measuring the output at the collector pad for the missing transistor - signs of life! I had nothing to compare this to but I was really only looking for activity versus dead silence. One of the online documents said there should be a 20KHz signal present here, a more modern scope would tell me what frequency this was but mine is probably as old as this cabinet. I could have worked it out from the graicule but I was really only looking to see if it was doing something rather than nothing. So with those three things confirmed I powered it all down, dug out the transistor and slotted it in. The middle leg (emitter) gets snipped on these as it is common with the metal tab, which connects to the board through the metal heat-sink, which itself acts as a jumper between two tracks. I used heat-sink compound on the body section but left the tab clean for a good metal to metal contact, with the two getting screwed together tightly. The HV board does need calibrating to the new LOPT but those steps need it to be running at the time, so after connecting it all back up I flicked the main switch. I got the high pitched crackle of the tube energising, along with the short ringing noise of the degauss system demagnetising the tube. We have HV!! Expecting success I went round the front of the tube to see...bugger all! The first thought was that I just needed to turn up the screen brightness control on the LOPT but this did nothing. The game board was certainly running, and I could hear the faint chatter in the deflection coils (aka the yoke), the spot killer light wasn't lit, and I had heard the HV come up. Assuming I had all that, the only missing component was electron gun activity in the tube neck. These take the RGB signals from the deflection board so the only thing that could be missing was the heater voltage. You should be able to see a faint red glow in the neck when the cathode heaters are on, but there was no glow, it all looked very "off". There should be a 6.3VAC feed from the HV board to the tube neck for the heaters but a quick scout with the meter on the back of the HV board showed I had 0 volts coming off the HV board on the heater wires. Weird, this comes out of a dedicated winding in the LOPT and as it is just a transformer you get volts on all windings, or none, unless one of them is broken, which is a possibility but unlikely on a new untested LOPT. I need to see whats coming out of the other windings to rule the transformer in or out, and the simplest place to measure that is at the rubber anode cap on the back of the tube. Except you can't jam a meter in there as this should be running at around 20,000 volts. Enough to totally destroy a normal meter, vaporise the test leads and probably blow the shit out anyone holding it. You can get dedicated HV probes that you connect to a meter that basically steps the voltage down by a factor of a thousand or so, giving a reading of 20V when the probe tip sees 20,000. But my meter is a cheap and nasty Dick Smiths meter I invested $30 in decades ago, it is more than good enough for almost everything I do, but I don't have much confidence in using it for anything grunty. Plus a decent HV probe is about $200, a little steep considering I've never needed one before, and only need one for 60 seconds now. Thankfully the guy I bought the CRT tester from also threw in an untested HV probe that he got with it years ago, exactly what I needed knowing this was coming up I jumped at it. Despite looking like a sex toy for Daleks this is just the job. On arrival it was rattling badly as the giant resistors inside were loose. After some digging I found someone in India with the same one, surprising considering this is 1971 vintage, who could unscrew the red bit and send me a photo of what was inside. All I was missing was a fat spring that holds everything together and gives the probe a bit of bounce. One of Bunning's finest steel springs later and I had a working HV probe for $2. With that poked under the anode cap, connected to chassis ground, and carefully wedged in place. I flicked the switch again, got the HV crackle, the degaussing ringing noise and the needle swung violently up to 27,000 volts before dropping back to 0. So I am getting HV, but it's instantly cutting out, which as the voltage is going far too high means it's the saftey system kicking in. The doco that came with the LOPT contains the calibration steps to set this, using two trimmer pots on the HV board to set the running voltage as well as the threshold that the HV protection should kick in. The instructions contradict others online about how you do this (whether R17 needs to be fully clockwise, or anti-clockwise), but basically you turn R17 way up (clockwise by my obvservations) to push the HV shutdown voltage well out of the way, then with R7 you adjust the output voltage so you get 25,000 volts at the anode cap. Then you wind back R17 until the HV shuts off, so you set the HV shutdown to be 25,000v, which is below the 30,000v danger level, at which point this apparently turns into an x-ray machine. Once that is set you power off, to reset the HV cut out, wind R7 back down a bit and then power up for a second time, you then adjust R7 until you get 19.5KV at the anode cap and thats the working voltage dialed in. All good, except I couldn't get it to work, with R17 fully clockwise (which seemed to the only position that the protection mode wouldn't kick in) I could only get a stable voltage with R7 at the midpoint, and that was 27,000, far too high. I didn't dare leave this running to take photos as thats not good for the tube, the heaters, the LOPT or possibly me. Any adjustments to R7 in either direction caused the voltage to shoot up and the system to shut down. After some head scratching my only thought was the board is driving this LOPT far too hard, it's already up against the max cut-off possible, so any adjustment just trips it out. So out comes the HV board, and the superb schematics from the Dezbaz's site. It took a while to get my head round it but looks like the 555 timer is used to send a stream of pulses into a darlington pair of transistors that drive transformer T1, which drives the BU460D transistor which chops the LOPT primary coil in and out of circuit, I guess constantly collapsing and reforming magnetic field in the LOPT secondary coil gets dumped to the tube up the annode wire to the cap. The tube itself is a giant capacitor so it handles the smoothing itself. The duty cycle of these pulses is what sets the resulting voltage. The HV shutdown seems to be controlled by a comparitor circuit, with a voltage divider using the +24v as a reference voltage, kicking a thyristor in the circuit to bypass a small capacitor in the 555 timer circuit when a threshold is breached. With the capacitor bypassed the 555 oscillation stops and the HV is cut. The 555 timer timing is set up by a combination of R5, R6, the R7 trimpot and the mylar capacitor at C6. R5 and R6 measured correctly, and the trimpot tested OK the board, giving a smooth increase over the 2K ohm range, but C6 was just plain wrong. According to the schematic this should be a 0.0047uf (4.7nF) capacitor, but this was a 10nF cap, which actually measured as 11.57nF off board. This cap looked totally original, with the solder job looking as virgin as the other original joints on the board, but I guess it was part of the bodgy install of the TV LOPT, and the solder had had seen enough years to look original. Assuming my new LOPT was a better electrical replacement than the bodgy one this cap needed to be put back to what the design intended. Except it was 16:10 on a Sunday and Jaycar shut at 4pm. I went through all my scrap boards and old TV chassis but only found one 4.7F cap, which wasn't in great health. On the meter it started at 4.7nF and gradually increased over a few seconds passing 10nF within 30 seconds, it went in the bin. With nothing else to do until the next day I found a 555 timer simulator online that let me plug in the values in the schematic to see what impact this cap has on the output frequency of the 555. The circuit has a 2K pot in series with a 3.9K resistor, giving an effective range for R6/R7 of 3.9K to 5.9K, plugging those into the simulator with my 11nF capacitor gave me the following frequency ranga.e. 13,828 Hz when pot at 0 ohms 9,5734 Hz when pot at 2k ohms Whereas the correct capacitor would give me a range that was miles higher, not even a slight overlap. 4.7nF 34,042 Hz when pot at 0 23,567 Hz when pot at 2k I'm assumimg that the higher the frequency the more time per-second that the transistor is actually off, the less time the primary coil spends energising, so when it turns off there is less energy stored in the field to get dumped up the anode wire when the field collapses, giving a lower HV voltage. If the capacitor value is 0, which is what the thyristor bypass achieves when the shutdown thyristor is conducting then the output falls to 0Hz which stops the HV entirely, but in general the faster the pulses the lower the HV voltage. I've been burnt by Jaycar's capacitor ratings in the past so I took my capacitor meter in and went through their box of 4.7nF green caps, and only one was actually 4.7, the rest were up in the 5nFs. Based on my figures above a lower value is better than a high one, so I wanted one as close to the spec as possible. Thirty cents well spent, with it installed the HV was sat at a healthier 20kV, with the absolute top end achievable of 25kV. After setting R17 to cut off at 25kV I wound back the working voltage to 19.5kV and called it job done! As the monitor was staying on for more than a split second I also had the happy red glow of the cathode heaters in the neck again. Expecting good news I walking round to the front of the tube, and ah crap!. We have life, three colours, but the Y collapse fault is still there. I'd hoped I swept up the collapse fault during the refurb, but as I hadn't found a smoking gun I guess that was just too optimistic. At least I had red again, but again...wtf!? The Star Wars logo should be blue, not red, and the text should be red not blue! The only way to swap the colours over like this is to reprogram the game code or get the monitor wiring mixed up, but this wiring is all harnessed up with keyed connectors. The harness looked original, apart from the mouse damage, and there was no damage in the section that carries the colour signals from the game board to the deflection board. There is a connector mid path, but it is keyed and you can't plug it in the wrong way, smoke would be the result if you did anyway as there is 60VAC on three of the conductors that powers the monitor PCBs. On one side of the connector I have Red Green Blue, but on the other side I have Blue Green Red, so what should be drawn red is blue, and blue is red. The only conclusion is that someone's been here before! Getting the pins out of these connectors is not easy unless you have a de-pinning tool to do it, you can usually see when someone has tried to do this with out a de-pinning tool as it leaves marks and gouges on the plastic where they shoved needles or small screwdrivers down the side to try to bend the latches back, but this connector was mint. It won't have been like this since it left the factory in 1983 and it can't be an accident, plus whoever did it knew what they were doing and how to do it without trashing the connector. The only reason for doing this is when you are trying to track down a missing colour on a monitor. If you are missing Red you need to know if the monitor can't draw red (chassis fault, or bad tube), or if the board isn't giving out a red signal in the first place (board fault). Without a scope to sniff the signals a quick way to test is to swap the signals over, if you squirt the red signal into the blue gun, and you get blue stuff drawn on the screen, then you know the red signal was present. If you squirt the blue signal that was working before into the red gun and you get no red then the fault is in the monitor chassis or the tube, not the game board. As all three colours are back, and no faults were found on the monitor boards that would explain a missing red, the likely candidate for that fault was just dirty contacts, which I cleaned during the refurb. I suspect a previous owner had someone out to service this, who did a quick check in the field to suss out how bad the problem was before giving a quote. As it was never put back the right way around they clearly never got that guy back and it stayed faulty. Any way, am not fussed by that at the moment, running the monitor in this state is bad for the tube, so I'm more concerned by the negative Y collapse fault. The usual cause for this is a failure in one of the 4 bottle cap TO3 transistors, which drive the deflection coils, but I'd replaced all of those with new ones, as the originals were a mixed bag of manky looking ones. The other usual culprits are the MPU05 and MPU07 pre-amp transistors that feed the drive transistors, but I'd tested all of those before they went back on the board, and even tracked down an MPU07 to replace an old BD193 transistor that was sat in the X section. I guess I know the one I bought is fine and isn't a Chinese fake, as the X is working fine. When a vector monitor isn't getting any deflection the beam fires dead centre (the spot killer circuit is designed to catch this scenario and kill the beam, which it does as this will burn a hole in the phosphor coating), but the screen is basically divided into 4 quadrants with the X voltages going +ve to pull the beam right of centre, and negative to pull the beam left of centre. Same with the Y for above and below center. Upper half = [-X & +Y] [+X & +Y] Lower half = [-X & -Y] [+X & -Y] The upper half is fine, so I've got both positive and negative X signals, and the upper portion of the Y is fine as the image is drawn properly across the whole height of the upper half. I'm lacking any ability to drive the Y coil into the negative voltages, to cover the lower half. Going over the board again didn't throw up anything new, so I dragged the cabinet closer to the repair bench scope so I could have the deflection board out and running to see what was going on. Armed with the schematics from Dezbaz's site I started poking around the Y section where the signal comes in from the Vector PCB and hits the base of transistor Q1 and it looks healthy at the front door, with portions of the signal going higher and lower than the 0v midpoint... ...but at the emitter the signal is squashed, everything is cut off at the 0V level with nothing in the negative range, suspicious as that's what I'm seeing on the screen. Comparing this to the X deflection section, at Q11 which is the X equivalent of Q1 in Y, and the base and emitter pins both have the signal at full range. The question is whether the signals are being clamped by something, or whether the circuit that pulls the signal low towards the -24 volts line isn't working. It's simpler to start with the second option and check that the -24V rail is getting to where it needs to be in the negative side of the circuit. If that was missing then there would be nothing pulling the signal down so it would sit at 0v when not being pulled high. It was, but it's interesting that this is the part of the circuit with all prior damage, burnt resistors and dead diodes. Probably no a coincidence at all, so I figured I'd missed something, but nothing stuck out while poking around. I found a couple of the 0.1uF capacitors that looked a bit beatern up as they were right by the burnt up areas, so they got replaced with some salvaged from another monitor chassis, but no change. I ended up at the other side of the schematic chasing my tail for a while, pulling out parts to test and comparing with the X section when I could, but no leads. Everything seemed to be fine, which left the the usual suspects for deflection, the 3716 & 3792 bottle cap output transistors or the MSPU's that feed them, which for negative Y is Q5 and the bottle cap pair at Q6 and Q7. Q6 and Q7 were brand new, but Q5 was original and had lived through whatever fault did all the damage originally. So I desoldered it and put it on my component tester... ... yep, despite it testing fine back when I rebuilt this board in stage 2, it looks like it died as soon as it was put back into service, it now behaves like a diode. I can only guess that until the other parts were replaced it had no load on it, but when asked to work hard again it died. The only problem is these are hard to get, they are long discontinued and were used in some audiophile amplifiers in the 80s so they are very expensive, and anyone with a lot of shiny ones in stock are probably selling fakes. It's the same type I bought in stage 2 to for Q15 in the X deflection circuit, that was a good price but still silly for a single transistor, now I needed another one which would take weeks. I have a habit of keeping all the bad parts and bits that aren't being reused during a project so I still had the unknown equivalent transistor from Q15. It isn't a type referenced as a usable equivalent these days but it had clearly been on the board a very long time, and the X section had worked fine with it, so I cleaned it up, put some fresh heat-sink compound on the back and soldered it in... ...and fired it up. Negative Y is back and thats the last of the original faults tracked down!! That BD239 transistor is pretty ugly, it has the wrong pin-out for the board needing the legs to be twisted over each other, so I'll keep an eye out for another MPU07 at a sane price, but this lets me crack on with the project without parking it for weeks. Only two minor issues left, firstly the colour swap, easily fixed by swapping the pins back in the connector.... ... clearly they were swapped by someone as the signals should be lined up with the individual signal grounds, and they come out of the harness as a twisted pair. Blue is blue and red is red again. Secondly the whole image is at an annoying angle, down at the left side and too high on the right. Sometimes this is because the tube isn't level in the cabinet, a fact that may be less obvious if the edges are hidden by a bezel, but in this case the tube was straight and the image wasn't. This can be fixed by gently rotating the deflection yoke on the tube, the foam wedges that hold it can allow it to creep over time, but the top wedge was glued to the yoke with hot glue, and I'd siliconed the back of it to the tube.
  21. Re-assembly has to be the best bit of any restoration, everything is shiny and new again so it's like the best Lego set ever. There's just the nagging concern that it might not work. The last time any of this saw power was in the seller's garage on the day I bought it, and every single part since then has been stripped, scrubbed, dried, and refurbished, so there's nothing left that I could consider as "known-working". Testing as much as possible before installing it back into a cramped dark cabinet seemed like a good option, even if only to set the power supply voltages without having to lie on the floor to do it. The wiring harness on this is a bit of a tangle so I wanted to check it made sense and label up the connectors to make the re-installation a bit less of a plug and pray event. Although most of the connectors are unique, there are a couple that you can get swapped, with fairly impressive results I expect. Laying it all out on the bench makes it a lot easier to sanity check what is getting connected to what before the smoke test. The only part I can't test is the monitor as the tube is just a naked tube and I'd need to build a frame to hold it upright it outside the cabinet. The deflection and HV boards that make up the monitor are pretty accessible when in the cabinet anyway so I can test those far easier than the parts mounted in the dim dark depths. First job is to dig out the speakers, the ARII PCB that provides the main PCB voltages also contains the audio amplifiers so I can test the audio if I release the complete set of four speakers. I'd removed the pair below the seat earlier when I was stripping that panel and they were absolutely filthy. So much dust that it had totally filled the edge and been compacted into a thick ribbon from the speaker action over three decades, sucking in dusty, smoky air and compacting it as the cones moved in and out. Once that was chiselled out they were actually in were in great condition. They had sat in the parts corner for months now, while the other two stayed in the cabinet behind the seat. It must be much dustier at lower altitudes as these two were pretty clean, and only needed a dusting with a paintbrush. Bench building! After testing the power brick in isolation, I hooked up the ARII and checked it's outputs, before plugging in the PCB and flicking the switch again. Was met with a power-on pop from the speakers, the three indicator lights on the CPU PCB lit up and went out, and Obi Wan's voice boomed out the "The force will be with you". It works!! Slight adjustment needed on the ARII to get the 5V spot on. A critical test to perform on vector boards is the range of the XY signal voltages coming out of the vector board. A failure in the vector output section can dump higher voltages into the vector monitor than it is designed to cope with. So these need to be checked and OK before connecting it up, as a bad board can damage an otherwise healthy monitor. On a healthy board the DC voltage between the X and Y test points and ground should be less than 1 volt, and the AC should be no more than 5 volts. These signals are swinging all over the place as the board draws the vector lines by constantly changing the voltages going into the deflection coils, so you need watch it for a few seconds and note the maximums, which for me were. X DC 0.6v AC 2.5v Y DC 0.4v AC 2.2v The voltages are within spec, but it's no proof the board is drawing anything sensible, but as a CRT oscilloscope is also a vector monitor, most can be configured to display a monochrome image from these old games. Mine also has a Z input on the back that can the blanking signal provided by the board to turns off the beam while the focus point is moved to the start of the next line, otherwise everything is joined up like a spider web. Graphics look good, and both halves of the screen are being drawn, it seems to all work perfectly. After a quick session marking the connectors I'm ready to rebuild this X-Wing. First job is to run the long harness section down the central channel using the high tech method of taping the end to a stick and poking it through... ...and then go fishing under the seat to find the rear connector when it arrives. With that done the metal blast plate goes back in, this sits under the power brick and shields the wooden cabinet from anything exciting and unplanned happening on the underside of the power brick. Power brick goes home. Mains switch re-fitted. Safety shield over the top, to keep fingers off the mains terminals. Yep I need to clean that glue residue off. Mains cable installed, screwed down and connected to the brick. The rest of the harness remembers where it wants to go, despite being coiled up in a heap for 9 months. After a few checks with the old photos from day 1, the cable anchor points get screwed back down. The fan assembly slots in and gets hooked up to the AC and ground. The cabinet central grounding point is connected up to the harness and bolted down hard. So far so good, the rest of the connectors flapping around will have to wait for the ARII... ... but I'm off down the other end for a while. The rear section of the harness appears just behind the seat... ... and a short harness section plugs in under the seat to splits out two pairs for the seat speakers, and two longer pairs that go up behind the seat for the upper pair. Originally the rear lighting was set up like this... ...with 250VAC 50Hz magnetic ballast and starter. The plug-in jumper block on the power brick configures it to act as a step-down to provide 110V the fan and the fluoro tube, but at some point the original light must have died. Of all the original parts it probably struggled the most from finding itself in a 50Hz country rather than the 60Hz it was designed for. At 50Hz it would probably have worked fine, for a while, just hotter than it was designed for. When it finally died the owner resorted to splicing the fluro power wiring into the mains switch wire so they could install a ballast bought locally, which would need ~240V and expect 50Hz, which is why they hacked up the wiring to tap into th raw mains before the step-down. All good except I had reversed that hackery. In the intervening years someone invented the Internet, Amazon, and cheap global shipping, so I'm no longer limited to what I can find at Bunnings. There are no 110V 50Hz magnetic ballast options around, not surprising as there is no target country for such a combo. A better option these days is an electronic ballast, as the input frequency is irrelevant because they don't rely on lumps of iron and transformer resonance, they also don't get hot. Another option would be to replace the system with an LED tube, some of which takes mains direct to the tube as they have built in switch-mode power supplies. The issue there was finding one locally that was the right length and specified as compatible with 110V. I suspect most are, but the packaging is printed for the target market, so who knows what the true specs are. It could be an easy win, or an expensive fail if I find tube after tube won't work. Basically, I'd be back to importing one and hoping it survived the shipping. LED tubes also tend to be more directional, compared to a $12 fluro tube from Bunnings that gives out 360 degree light, so it's possible an LED tube would give a brightly lit stripe across the marquee glass and be dim elsewhere. So I picked up this, a 120V AC electronic ballast for $8 delivered... ...and it doesn't care whether it gets 50Hz or 60Hz, so I can keep the original fitting and get the original backlit effect. It's actually a known brand, with a detailed datasheet available too rather than a YumCha unknown. Just a case of splicing the wiring in, bolting it down, and dropping in a new tube. I don't need to import a tube as the tubes are globally universal, they run of high voltages created by the ballast so as long as that is correct for the region any tube should work. The speaker wires also get reconnected and the tie points screwed down to keep that from flapping around. Testing time, at this point I should get light at the back and fan-spin at the front. Success, so far so good. Back up the front I realise I'd forgotten one thing, refitting this. The serial number sticker from the power brick that I'd rescued before the sandblasting. It's a small detail but I'm already going well overboard on this project, so why not. It would have been a hell of a lot easier to refit it before I'd installed everything. Didn't seem worth ripping everything out again, but now's the time to do it, as the coin door hole is pretty handy access. I used some of the mat-fix-fail double-sided tape, sticking the label on one side, then razor triming it to size, and installing it through the door hole. Next up - Coin door, coin cage and service panel assembly. Major improvement over day 1. These Atari over-under coin doors clamp to the inside of the cabinet with small plates and bolts. Which had copped a lot of mouse damage, but came up a treat after blasting and plating. These clamp against the inner face... ... and the cage hooks over the middle bar. Coin box goes home. It's starting to fill up in there. Fitting the parts so far is much easier without the CRT hanging in the way, but the remaining parts would make fitting the tube harder and therefore more risky. So now is the time to refit the tube... ...another 2 man job, and much relief once the four nuts are back on. Keeping with the monitor theme, the HV board goes in... ...followed by the deflection board... ...and the golden PCB cage.
  22. Worth checking whether you have the correct power rails on the TMS5220 speech chip as it needs two power rails to function. Test the voltages at the socket, with the chip taken out, it's easier and removes the risk of slipping with the probe and killing the chip. 5v on pin 5 (relative to ground on pin 11) -5v on pin 4 (relative to ground on pin 11) If you don't have the original power supply and cab wiring then the -5v may be missing. The rest of the board doesn't seem to need the -5v at all, so it may well run perfectly, just with no speech with just the single 5v rail. Not sure what you are running it off, but there are a few cheap arcade PSUs out there that have a -5v terminal and labelling, but it's just a dummy and connects to fresh air inside the case. If you do have -5v present then ..... How long are you leaving it when you ground the self test pin. I'll have to check on mine when I get a chance but you may have to wait 10-20 seconds while it runs the self test before you get the menu. I know on Vindicators it sits there with a white screen for about 10 seconds before you see any signs of life when you switch to test mode. Also, just to check, are you grounding the test pin, and leaving it grounded? You need to keep it grounded, a momentary grounding may well freak the board ou
  23. Ouch! That's gotta hurt. Am pretty sure Logitech wanted to stay close to the original Acorn design, they are both fugly and pretty similar! Go for Twin Worlds when you write some disks for it!! Brings back some memories!
  24. It's not uncommon to find a dead TMS5220 voice synthesizer chip on these. I'd verify that the voltages at it's socket are ok and then swap in the one from the spare board.
  25. Same process, different glue. This stuff actually spreads well... ...I didn't get too close to the edges so it wouldn't bead out when attacked with the roller. First half down, same process on the second half. Done! Perfect With no more need for the elbow room I could flip the cab on its nose and re-fit the kick panel. Much easier in this position as you aren't fighting gravity while you try to fix it in place... ... and while the cabinet had it's rear in the air the rear panel screws went back in. Finally time for some t-moulding. This cabinet has an insane amount - upwards of 60 feet, and I had a mix of repaired slots and original slots that I hoped were still OK. If it goes on easily it'll be a quick and vast improvement, but if it fights me all the way I'll have a 60 foot battle on my hands - time to find out. My main concern was this area, where the two panels almost meet, the slots in the gap were pretty loose but there is no way I could get access to fix them without ripping one or both of the side panels off. Starting on the rear side, the repaired and routed slots on underneath aren't too tight... ... needed to snip out some of the tang at the corners so it would bend and not foul up... ... and snip the tang so it could splay out and follow the concave bends. I highly recommend using a roller for putting t-moulding in! I've used a rubber mallet in the past but it is so much easier to roll it on, plus you can feel the sections that need more pressure to get them fully in. Same deal for the front panel, I'd worked out on the last panel that it is easier to start with the trouble point and work away from it. Leaving it to the end means I'm fighting against the section I just installed when I get to the gap. Side one done. Onto side two... ... using the back of a chisel to get it to engage in the gap... ... and straight away hit a problem - a huge section of trashed slot that I had some how missed. The test before was if a wooden stick could easily go in then it was too loose and needed to be filled and re-cut, in this section I could almost get two in so no idea how I missed it. A quick session with the woodg lue and 3 icypole sticks filled it in, and after a couple of hours I ran out of patience, gave it a good going over with a hair dryer and fired up the router hoping the glue would hold up, it did. After that it was easy progress all the way round. Done! As I don't need to lie the cabinet on its side again I can safely refit the CRT brackets, which would otherwise get scratched up by the floor. There was a lot of slack in the bracket bolts but lining them up with the 35 year old outline imprinted on the vinyl helped. Paint job on the bolts looks great. Lots of parts can go back on now I'm no longer flinging the cabinet around. The panels and perspex go back on for a final (dusty) fit while I have it in the garage and in good light. These cabinets were never that well made and in their old age there is bit of slack in the panel spacing. Basically it's a case of finding the best looking fit and going with that. Pretty pleased that in the best-looking alignment the original screw holes lined up. When I got this it was held together with a random selection of screws which had chewed out the screw holes leaving them pretty loose, so they got a blast of wood glue to put a bit more meat back in them. Waaaay back when I first got the cabinet I got tired of constantly having to unscrew the back door, so I pinged a locksmith I'd used before and got a key cut for the original back door lock. So the lock goes back in, and the panel back on the cabinet to check all the alignments. Aside from all the dust, it looks great! Next up is the other t-moulding, to replace the 1.5" Atari custom stuff that is currently totally unobtainable. The only similarly patterned 1.5" stuff I could find anywhere was this from T-moulding.com The cross section isn't the same as the original but it should look pretty slick, better than bare wood, or the torn-up remains of the original stuff. The only issue was the tang is a lot thinner than the original stock, meaning the slot in the cabinet is about 3 times too wide. I could go down the well worn path of gluing wooden strips in and routing a fresh slot, except there's no way to get a router in to cut the slot at either ends of the run, as the side panels get in the way. Also knowing my luck, a week after I finish this cabinet someone will pop up somewhere on the planet and announce that they are re-making the official t-moulding again. Ideally I'd like to put the original stuff back on if it ever comes up so if I've filled in the slot I'd need to repeat the whole process and route again, a long-winded and messy job. So this t-moulding needs to be firmly fixed, but removable. I'd heard of folk doing this trick to avoid routing whole cabinets, so I thought I'd give it a go on these short sections. Strips of cardboard scored ... ...folded into a V shape ...that are pressed into the slot with the t-moulding to grip the tang. If the thickness is right the fit should be very tight, but also removable later if I ever need to. Main issue was finding card stock that was a suitable thickness, the stockroom at work came up trumps. This certainly isn't the easy option, it's fiddly, when it is thick enough to grip it gets hard to press in without the cardboard catching and tearing. Hopefully "hard to install" equals "wont fall out". To help it bend round the corners I trimmed the tang in those sections. The only tricky bit is the full 90 degree bend right under the marquee, this is thick stuff and doesnt want to bend that sharply, even with the tang snipped off. Originally Atari had pumped long staples in at both ends, I may have to fish a couple back out of the bin and hammer then in again. Side one looks good... ... but after fitting, it was clear how bad the inner side panel edges looked. Originally these were black but a combination of dust, dirt and general exposure to the elements had left them looking pale compared to the black t-moulding. The perspex roofing panels rest on these and the risk is they'd stand out as a light strip down each edge. So I used the last gasp of black paint in the tin to go along all the upper edges. The near side t-mould went on much the same, except whoever routed the slots on this side at the factory must have been drunk, the slot wanders all over the place and goes from very thin to really wide in places as the router drifted. No wonder the old t-moulding was covered in nail holes, probably from factory original nails, to make sure it stayed in. Looks good, but it's missing something. The edge protector strips to stop shoes trashing the edge. They are just bits of aluminium extrusion bent with a slight 90 degree overhang. I could have wasted a day driving all over the outer east to the various aluminium suppliers and still come up blank, so I picked up this polishing kit instead. The brown polish block and rough wheel are for cutting back the surface to get rid of the scratches. Which it does, after a while anyway. The white block and softer wheel are to put the final polish on, which also worked well. The one on the left has been cut and polished, the one on the right has just had the first half of the treatment. Both look good enough to go back on now, I suspect they will calm down a bit and return to a duller aluminium look over time, but that's fine, I'm not sure how shiny these would have been when new anyway. Fitting them was easy enough, just had to line up the lip with the marks on the side-art... ... and double check with a fragment of the old t-moulding from this area... ... before drilling some pilot holes. Four stainless steel screws later... ... and the same on the other side. Done! Not quite finished the t-moulding yet, there are three parallel runs of 1/4" on each side, under the roof arch, and on each side the outer of the three has pulled out at the bends. Not sure whats happened here as the run now seems to be too short for the cabinet, despite being nailed in at both ends at the factory, might be the violence of the nailing that did it. Either way the flexibility is gone anyway and it had snapped snapped at the staple bend. Out with the old. In with the new, this is actually the 5 year old left overs from Vindicators as the tang is slightly thicker on that old stock and gripped better than the newer stuff. T-moulding complete, makes a hell of a difference! There's just one last item on the list for this phase, a job I'd been kicking down the to-do list for weeks, due to the impact that stuffing it up would have on the project - applying this to the newly painted control panel. The glue on these is insane so there'd be no way to peel it off again without trashing the sticker and leaving the glue behind. Getting the glue off would trash the paintwork, so I'd be back to square one. The prospect of stripping, sanding, priming, painting, buying a new decal, waiting weeks or months for the next print run, and then for shipping didn't appeal, so this had to go well. First up I needed to cut out the center square, but the grey blank square... ... isn't actually where you cut. On the original the decal cut out was larger, extending beyond the bolt holes. Step 1 is getting it perfectly lined up at the top edge, using a clean smooth strip of wood to keep it flat across the whole width and to spread the load to avoid marking it. Getting it lined up along all edges is tricky due to all the bends in the CP, the decal is quite thick and stiff. It took a couple of weeks lying flat to lose the urge to curl up from it's time in the shipping tube, now I'm trying to bend it in ways it doesn't really want to go. Step 2 is to flip it over and mark out the square, and all the bolt holes on the backing paper with a thin texter. These lines weren't where I need to cut, but shows the positioning when it is back on, and makes marking out the true cut lines easily. Success! The decal is also slightly too wide, as the right arm is longer than the left arm by about 4mm, leaving a rather fragile overhang. Posisbly a bug in the artwork file. It's much easier to fix now with a ruler and sharp knife than later when it is stuck on. Running a knife along the edge after installation is a great way to scrape off the paint and let the rust back in. After a final dust and wipe down with iso alcohol I had to find the correct alignment again. Trickier now artwork was extra floppy with the middle cut out. With it clamped in position just before the first bend in the CP I peeled back the first inch of the backing paper, trimmed it out of the way, and stuck the top edge down. Then moved the clamp up onto the newly stuck down edge to make sure it stayed put... ...and slowly worked downwards, peeling the backing off about a centimeter at a time, making sure it bent into the corner all along the length. Done - and my heart rate slowly returned to normal. It can't actually go back onto the cabinet until the 3D cockpit liner goes in, which needs to go in after the cardboard inner bezel, which needs to wait for the CRT, which needs to go in later than a lot of the other stuff so it isn't in the way. I really don't need a cracked tube on the very last day of this project. In fact a lot of the parts in this phase need to stay off until they are needed during the rebuild, otherwise I'd be fitting them and instantly removing them again. So that's phase 3 complete! The to-do list is empty, nothing left to sand, paint, clean, repair, fix, stick, service or find. Time for the grand rebuild, and to see if it can fly. --- To Be Continued ---
  • Create New...