Sudden death of ZX Spectrum 48K

edited December 2015 in Hardware
Hello

A friend gave me a 48K rubber keyed Spectrum. I brought it home, hooked it up to my little TV and powered it up.
I was very happy to see the familiar "(c) 1982 Sinclair research ltd" screen. I typed in a couple of short BASIC programs. They worked fine too. So I decided to load a game. I unplugged the adapter and plugged it back in to fully erase the contents of the RAM and to my surprise, the Spectrum that was working flawlessly a second ago was refusing to initialize. I tried to reset over and over again by unplugging the 9V jack, waiting a few seconds and plugging it back in. No difference. All I am getting is the screen in this photo: http://m.imgur.com/rU6IKwg
Does anybody know what might have caused such a sudden failure? I didn't attach any interfaces or even a cassette recorder to the Spectrum. It was just the 9V adapter and the TV, both of which I have been using with a bunch of other Spectrums without any issues in the past. It was sad to witness the death of yet another Spectrum.
Any ideas, opinions, suggestions would be appreciated.
Thank you :)
Post edited by memrah on

Comments

  • What test equipment do you have, and what test equipment do you have access to/know how to use? First rule, test voltages.
    Oh bugger!<br>
  • edited December 2015
    The problem might be the lower RAM or ULA judging by the I've heard that they can be particularly prone to failure. I'm not a hardware expert though so it could be something else. Test voltages as DEATH says. That should give you an indication of where the problem is.
    Post edited by PROSM on
    Host of CSSCGC 2020
    Currently working on Postie's Peril, a point-and-click adventure game for the 48K Spectrum made with my ZX Adventure Designer engine.
  • Guesses based on typical failures:-
    Transistor TR4 has died
    One of the 4116 DRAM chips has died
    One (or more) of the electrolytic capacitors that are in TR4's circuit has died, causing TR4 to die.

    There are not alway obvious reasons for failure. There could have been an existing defect and many years without power followed by being powered up have stressed it enough to cause failure.

    I once bought a good well known and well liked branded second hand CRT monitor. Used it for over an hour. Two hours later, it's line output transformer had died.

    Mark
  • edited December 2015
    I have a relatively good quality multimeter so I can test voltages. Alas no oscilloscope. I took the Spectrum apart and all the upper RAM chips are socketed. Also 4 of the logic chips near the CPU are also socketed. When I saw that, I assumed the unit was repaired in the past. But then I looked at the bottom of the PCB and the socketed chips show absolutely no sign of being soldered by hand unless whoever did the job was a super skilled expert at soldering. There is not a single solder point that looks different than the machine soldered ones. Anyway. What is also interesting is, one of the logic chips (IC 26) is double socketed. When I first saw that, I thought it was really weird. Upon closer inspection, one of the pins of that chip is cut off and the adjacent pin is bent to fit the missing one's hole on the socket. I am not sure what's going on here and but I thought I'd mention it to you guys before proceeding with voltage checks.

    Here's a photo of the strange mod:

    http://m.imgur.com/0geRCck

    Post edited by memrah on
  • edited December 2015
    @1024MAK is there a quick way of testing TR4 and the capacitors on its path without removing them from the PCB?
    I wish the 4116s were socketed too, like the upper RAM chips on this Spectrum..

    Is there a simple troubleshooting guide for the 48 motherboards? I will download the service manual.
    Post edited by memrah on
  • The quickest, easiest way to test if TR4 is okay, is to test the +12V and -5V supplies (they are generated from the nominal +9V input by the inverter/converter, of which TR4 is a major part). You can also perform some limited tests using the resistance and diode test ranges on you meter.

    There is no practical way of testing electrolytic capacitors while soldered in-circuit. Electrolytic capacitors are only rated for operation at their rated temperature (85C) for typically 2000 hours. They do last longer at lower temperaturse, but the inside of the Spectrum does run warm... As ZX Spectrums are over 30 years old, it is worthwhile replacing all of the electrolytic capacitors regardless.

    There are a couple of guides. The Spectrum fault finding guide (I will post the link later) and Zorn has done some you tube videos.

    Which issue board do you have? The modification on your board may well be to use DRAM chips that the board was not designed to be used with.

    Mark
  • memrah wrote: »
    Here's a photo of the strange mod:
    http://m.imgur.com/0geRCck

    I assume this is an issue 2 Spectrum, in which case the above is almost certainly how it left the factory. It's to allow OKI RAM to be used as upper RAM rather than the Texas Instruments types the Spectrum was designed for. Later versions have a wire link to select for TI or OKI upper RAM.

    Check for a short (or low resistance) between the two windings of the inductor coil. TR4 very often (though not always) fails with a short between its base and collector, which can be tested for as just described.

    I think it very likely the failure was due to an old electrolytic capacitor taking out the voltage converter circuit by some means. This may also have damaged one or more of the lower RAM ICs. It's a very common failure and is why I recommend replacing the original electrolytic capacitors in Spectrums, especially if they've been packed away for years. They're of an age now where more often than not one will give up within a few hours of being pressed back into service.

    You might find some useful information in some of the videos about ZX Spectrum hardware that I've published:
    https://youtube.com/playlist?list=PL4H3oOJzIQMkKiG94z2x0laC7gqqmwBE8.
  • edited December 2015
    Thank you all very much for all the responses. I will follow your instructions and try to diagnose the fault when i get home tonight. I'll let you know how it goes.
    Post edited by memrah on
  • edited December 2015
    Hello again
    @1024MAK Thank you very much for the links. I tested the +12 and -5 volt lines like you suggested and they seem to be carrying the correct voltages.
    I also watched Zorn's excellent "ZX Spectrum Initial Tests" video and my Spectrum passes all of those tests as well.
    @Zorn: Thank you very much for your comments and especially for the link to your great YouTube channel. Your ZX Spectrum videos are just excellent and I am planning to watch all of them soon.
    So at this point, since the machine passes all the tests you describe in your video, what should my next step be?

    Also, the upper RAM chips are indeed OKI chips but the Spectrum is actually not an issue 2 but an issue 3. I still don't know why the double socketed chip is there. Here you can see a few photos of the PCB, the upper RAM chips (all socketed) and a close-up of IC26 underneath the 7805 heatsink.
    http://m.imgur.com/a/VkbSG


    Post edited by memrah on
  • memrah wrote: »
    Also, the upper RAM chips are indeed OKI chips but the Spectrum is actually not an issue 2 but an issue 3. I still don't know why the double socketed chip is there.

    That's a bit strange, though not likely related to the failure. I assume the wire link (next to the EAR and MIC sockets) is set for TI RAM. OKI RAM is supported directly if the links are set appropriately.
  • edited December 2015
    Zorn wrote: »
    That's a bit strange, though not likely related to the failure. I assume the wire link (next to the EAR and MIC sockets) is set for TI RAM. OKI RAM is supported directly if the links are set appropriately.

    I wanted to look at the links and see how they were set and noticed a very strange metal bit attached to the end of the PCB. I have never seen anything like that on a Spectrum PCB but again I am not an expert so it may be a normal/common thing. It looks like it's bridging solder points on the links but honestly I have no idea what it is. Here is a photo of it: http://m.imgur.com/GBai8ac

    Also, if you look carefully at capacitors C5 to C8 in this photo, you'll notice that it looks like something leaked out of them but I thought those type of capacitors didn't leak. http://m.imgur.com/MGeC6xe Is that normal?
    Post edited by memrah on
  • memrah wrote: »
    I wanted to look at the links and see how they were set and noticed a very strange metal bit attached to the end of the PCB. I have never seen anything like that on a Spectrum PCB but again I am not an expert so it may be a normal/common thing. It looks like it's bridging solder points on the links but honestly I have no idea what it is. Here is a photo of it: http://m.imgur.com/GBai8ac

    I've seen that before. I think it was supplied with a RAM upgrade kit, intended for issue 2 Spectrums. It'll work fine in later versions or, instead, the link could be set for OKI RAM and the IC adaptor not used.
    memrah wrote: »
    Also, if you look carefully at capacitors C5 to C8 in this photo, you'll notice that it looks like something leaked out of them but I thought those type of capacitors didn't leak. http://m.imgur.com/MGeC6xe Is that normal?

    Ceramic capacitors shouldn't leak anything. I think it's just some general contamination and grime that has got in there over the years.

  • No. I was confused. It wouldn't have been intended for the issue 2, because that has a simpler high or low RAM type selector, not near the edge of the board and so that metal piece couldn't be clipped on. I still think it was supplied with a RAM upgrade kit, however. It seems bit odd, then, that the IC adaptor is used to re-wire the address connections for OKI RAM when the appropriate link could have been set instead. Anyway, I'd be inclined to leave it alone as I don't think it's related to the fault.
  • Zorn wrote: »
    No. I was confused. It wouldn't have been intended for the issue 2, because that has a simpler high or low RAM type selector, not near the edge of the board and so that metal piece couldn't be clipped on. I still think it was supplied with a RAM upgrade kit, however. It seems bit odd, then, that the IC adaptor is used to re-wire the address connections for OKI RAM when the appropriate link could have been set instead. Anyway, I'd be inclined to leave it alone as I don't think it's related to the fault.

    I agree that it's extremely unlikely to be related to the fault. So like you suggested, I'm leaving it alone.

    This may be a very bad idea but, would it hurt the Spectrum further if I remove the socketed upper RAM chips and power it up? If powering it up without the upper RAM chips in place wouldn't cause any further damage, perhaps I can rule out upper RAM failure if the Spectrum initializes and works as a 16K machine. But, again, like I said, I'm definitely not an expert, so this may be a really bad idea. :)
  • It'd certainly be reasonable and probably sensible to remove the upper RAM and concentrate on getting the Spectrum to work as a 16K model (as it should with no upper RAM).
  • edited December 2015
    Zorn wrote: »
    It'd certainly be reasonable and probably sensible to remove the upper RAM and concentrate on getting the Spectrum to work as a 16K model (as it should with no upper RAM).

    I removed the upper RAM and powered up the Spectrum. I get the exact same fault screen as I did before. So apparently the upper RAM has nothing to do with the problem. Since I don't own an oscilloscope to perform more in-depth tests, should I just replace all the electrolytic capacitors and see what happens? Or are there any other tests that I can perform with my multimeter? Like I wrote before, I performed all of the tests/measurements described in your "Initial Tests" video and the Spectrum passed all of those.
    And thanks again for your time. I really appreciate it.
    Post edited by memrah on
  • I don't think replacing the electrolytic capacitors will solve the problem, but I recommend doing so. However, given that probably isn't the cause of the fault, I might concentrate on trying to find that first. The symptoms make me suspect a faulty lower RAM IC or possibly a faulty CPU. Without further test equipment, it'll be difficult to determine which without simply replacing components until it works again.

    If you get stuck, I can offer a repair service (send me a message via http://zx.zigg.net) as can at least one or two other regulars in these forums.
  • Re: capacitors C5 to C8, some ceramic capacitors are coated with a wax like substance. I think what you are seeing is the difference between the remains and the worn off areas combined with general 30 years of grime and dirt.

    Do you have any expansions / interfaces that have a ROM in them? Or which can take a ROM cartridge? Or which have a reset button?

    Also what colour does the border come up with when you power up? Is it always white, or does it come on as a any other colour?

    Mark
  • I'd agree with Zorn, that fault is indicative of faulty lower RAM. To be honest, it could be anything though!

    Since you've pointed out that the upper RAM and that several of the 74LS chips are on sockets leads me to believe that your machine left the factory as a 16K machine, and to be honest, an issue 3 16K machine is quite rare, and far more common to be found on an Issue 2 board.

    As Zorn has said, test lower RAM, it's all you can really do for now, Did you test the output of the 7805 regulator? You mention that the -5 and +12V line is working, but no mention of the +5V
    Oh bugger!<br>
  • To be fair, he was never asked to test the +5V rail. As the +12V and the -5V are apparently good, I suspect the +5V is also good. But it is always wise to check.

    I also suspect one of more of the 4116 "lower" DRAM chips has gone faulty. If he has an expansion / interface that can take a ROM/EPROM, a ROM based RAM test program should confirm this.

    Mark
  • edited December 2015
    Zorn wrote: »
    If you get stuck, I can offer a repair service (send me a message via http://zx.zigg.net) as can at least one or two other regulars in these forums.

    Thank you. That would actually be great but I don't live in the UK and I'm afraid international postage to the UK and back plus the parts and labor costs would probably exceed what it would cost to buy a working Spectrum locally or even from Ebay. Let's see if I can manage to get this spectrum working again and if I get stuck, I'll see if I can find a cheap international shipping option.

    1024MAK wrote: »
    Do you have any expansions / interfaces that have a ROM in them? Or which can take a ROM cartridge? Or which have a reset button?
    Mark

    I do have a DivIDE+ but I am afraid to plug it into a non-working spectrum, in case it also kills the interface. If you think connecting it to the faulty Spectrum wouldn't damage it, I can go ahead and try that. What would that prove if it works? ROM fault?
    1024MAK wrote: »
    Also what colour does the border come up with when you power up? Is it always white, or does it come on as a any other colour?
    Mark

    9 out of 10 times it is white. I think I saw it come up Magenta just once.
    DEATH wrote: »
    As Zorn has said, test lower RAM, it's all you can really do for now, Did you test the output of the 7805 regulator? You mention that the -5 and +12V line is working, but no mention of the +5V

    I did check the output of the 7805 and the +5V line. They are fine. Actually, like I mentioned in my post above, I performed all the tests that Zorn mentions in this video:
    and the Spectrum passed them all.
    1024MAK wrote: »

    I also suspect one of more of the 4116 "lower" DRAM chips has gone faulty. If he has an expansion / interface that can take a ROM/EPROM, a ROM based RAM test program should confirm this.

    Mark

    Can a DivIDE+ be used for that purpose, with some sort of a diagnostic test firmware? Also, I have a RAM turbo interface which takes ROM cartridges but I unfortunately don't have an EPROM programmer. It may be a good idea to buy a diagnostic cartridge.
    Post edited by memrah on
  • edited December 2015
    It may be possible to use a DivIDE+. I have one, but have never tried to change the EPROM, it may also be possible to store a test program in the battery backed RAM, but again this is not something that I have tried.
    I think you’re better off buying a ZXC4 cartridge from Paul Farrow or a SMART Card from Phil.

    Paul can preprogram the ZXC4 with a ROM based RAM test. It's not cheap, but it can do far more than just run a test program.
    Links:- http://www.fruitcake.plus.com/Sinclair/Interface2/Cartridges/Interface2_RC_ZXC4.htm
    http://www.fruitcake.plus.com/Sinclair/Interface2/Cartridges/Interface2_RC_New_RAM_Tester.htm
    http://www.fruitcake.plus.com/

    Alternatively, the SMART Card from Phil can also be used to test RAM as well as be used as a mass storage device (using SD cards) http://www.retroleum.co.uk/smart-card-for-zx-spectrum/

    Balford has further developed a diagnostic program, details here https://github.com/brendanalford/zx-diagnostics
    That test program can run on any ROM/EPROM/EEPROM/E2PROM/Flash ROM that maps to the area of the BASIC ROM (0 to 16k) including the Interface 1 and compatible ROM cart interfaces (like the turbo), the SMART Card and the diagnostic board (currently not in production).
    A thread showing him repairing a ZX Spectrum is here http://www.worldofspectrum.org/forums/discussion/44400/the-spectrum-resuscitation-thread/p1

    At the time of posting, http://www.retroleum.co.uk was suffering from a server problem :-(
    If you are interested and the web site does not come back up, let me know and I will look for Phil's email address.

    Mark
    Post edited by 1024MAK on
  • 1024MAK wrote: »
    I think you’re better off buying a ZXC4 cartridge from Paul Farrow or a SMART Card from Phil.

    Mark

    I think I'll do that. Thank you Mark.
Sign In or Register to comment.