› ›

48k spectrum restarts by itself

hedgehedge
edited January 2010 in Hardware
First of all hello from a newb and thank you for your help. This site has already offered me a lot of insight.

I have a 48k spectrum that will boot to the 1982 sinclair screen and will stay working for random times say 10secs to 20-30 mins and then it will restart as if I have pressed NEW - enter. Sometimes only once and often multiple re-starts and a crash.

I This page diagnoses the problem as "Check the TR6". Now on loking at it and http://www.worldofspectrum.org/SpectrumRepairGuide/ here I believe the voltages maybe wrong. But I am not sure, I have also looked here http://www.kpsec.freeuk.com/components/tran.htm and here ftp://ftp.worldofspectrum.org/pub/sinclair/technical-docs/ZXSpectrumIssue3b-Schematics.gif and would like to know if the BE BC and CE tests should be made with the spectrum powered down - unplugged. and indeed if any one can let me know which leg is which looking down from the top with the power facing away that would be great.

I am a humble hobbyist and have OK soldering skills but electrical diagnosis is not good although I can just about weild a multimeter.

So as well as re-setting, the system also sometimes beeps and re-sets and also occasionally the screen fills with ???? marks and crashes. I am thinking these other problems are symtoms of the TR6 resetting the machine and once its fixed the other issues will stop?

I have ordered new 2N2369 to replace the ZTX313 (TR6) however while I wait for these parts to arrive is there anything else I should be checking?

Looking at the description of how to test the 4164s I dont really understand where the -ve multimeter lead should go when looking at the voltages at various legs. However because my system starts I was able to do teh POKE - PEEK command to check the memory and it all seemed correct.

Finally - sorry for such a long first post - I ahve removed ands re-seated my ULA a couple of times. It is the only chip that is not soldered on my3B board. I have also change the UHF output to be composite video following the excellent guide here http://womblesretrorepairshack.blogspot.com/2008/11/zx-spectrum-composite-video-mod.html.

Thanks for looking

Paul
Post edited by hedge on

Comments

  • polomintpolomint
    edited January 2010
    Well... I must say that that is an excellent first post. I can't be of any help though, but there are many others here that surely will be.. (Winston seems to know the electrical side of the speccy, along with others).
    Won't be long before they reply, :)


    Damn, nearly forgot... WELCOME TO THE BEST FORUM ON THE NET, :D
    So far, so meh :)
  • guesserguesser
    edited January 2010
    assuming it's a digital multimeter, rather than an old moving coil one, to measure any of the voltages you just need to connect the -ve lead (usually black) to any ground connection on the board. I usually clip it to the metal can of the modulator if you haven't removed it entirely. otherwise you'll have to find some other suitable place to connect to. It's easy to find something connected to ground as there are huge areas of copper on the underside of the board around the modulator area etc.
    My rubbish website including the redrawn Amstrad schematics and the Sinclair Wiki.
  • rokoroko
    edited January 2010
    hedge wrote: »
    ....would like to know if the BE BC and CE tests should be made with the spectrum powered down - unplugged. and indeed if any one can let me know which leg is which looking down from the top with the power facing away that would be great.
    As It happens that I have an open 48K in front of me and the (A.Dickens) Hardware Manual (p.41) at hand: The rounded side of ZTX313 should be on the righthand side then, and (top view) from front to back: C=+5V, B=A0, E=/IORQGE.
  • hedgehedge
    edited January 2010
    Thanks for the replies. I am going to replace a number of the transistors in the hope this solves the restarting problem.
  • Jose_LeandroJose_Leandro
    edited January 2010
    Hi
    hedge wrote: »

    Looking at the description of how to test the 4164s I dont really understand where the -ve multimeter lead should go when looking at the voltages at various legs. However because my system starts I was able to do teh POKE - PEEK command to check the memory and it all seemed correct.

    You can check if spectrum detect all ram. Write

    PRINT PEEK 23732 + PEEK 23733 * 256

    The value printed should be 65535

    bye

    Jos? Leandro
  • nuggetreggaenuggetreggae
    edited January 2010
    if we look at basics.....

    The most common fault for otherwise-working-perfectly but randomly-restarting speecys it was 90% of the time due to the power cable having been pulled out and pushed in so many times a fault develops either within the power plug itself or the solder connections on the socket on the board. Ist things first try replacing the plug and/or resoldering the socket to the board. Very quickly removing and applying power will also cause the other crashes you were referring to.

    Rgds
  • WinstonWinston
    edited January 2010
    Yes - I'd go for "find dodgy connections" before anything else. As nuggetreggae says, check the power connector is good.
  • hedgehedge
    edited January 2010
    Before I remove the plug when I play around with the power lead i.e. try and move it left and righ or up and down when plugged into the spectrum at no point will the system re-start or at least if it does it could just be it was about to as I cannot repeat the re-start by moving the power cable.

    Could that plug still be a source for failure even under those conditions - excuse the naiveness just not used to electronics fault finding?
  • rokoroko
    edited January 2010
    I think you sufficiently tested the plug in that way. You might perform a thorough visual inspection of the PCB looking for dodgy situations, specially when there have been earlier attempts to repair. And check the voltages.
  • hedgehedge
    edited January 2010
    Wow, just replaced my first ever transistor. Thanks to roko for pointing me in the correct direction for which leg is which. I bought the PH2369's of ebay to replace the "possibly" defective ZTX313. Regardless of wether its in error its a real buzz actuallly replacing the transistor and then switching the spectrum back on and seeing the 1982 Sinclair Spectrum :)

    Lets see if this solves it.

    to be continued.
  • hedgehedge
    edited January 2010
    Well after hours of use without a re-start the issue seems to have been solved. I could not tell you which solved it but here goes.

    Note when initial testing was done it was without a keyboard attached, just booted up to home screen:

    - Replaced TR6 system worked for approx 45 mins before first re-start
    - Replaced TR4 system restarted within a couple of mins
    - Removed power port, re-tinned all the legs cleaned off all old solder and re-soldered making sure I had a good wet joint. System re-started after about 20-30 mins
    - Removed the - 7805 regulator (I think) - 3 pin component that is attached to the heatsink. Re-tinned its legs and cleaned all the old solder off the board then cleaned the area that contacts the heatsink and indeed the area around the board. Re-soldered making sure good wet solder joints.

    The machine has now run for many hours with no re-starts. Thank you for the help pointing me at the power area.

    Paul
  • WinstonWinston
    edited January 2010
    Of course the 7805 is a part that gets quite warm - if any problem has a bad solder joint and is going to cause problems, it's probably this :-) Good job finding the problem!
  • rokoroko
    edited January 2010
    I one time did a test run of 48K, without cooling the 7805. After a while some overheating protection(?) of 7805 jumped in. Restarting with cooling worked OK, but the 7805 remained unstable when getting warm. I never trusted them again....
  • PilsenerPilsener
    edited January 2010
    Besides the possibility that the 7805 may fail due to high temp, as mentioned, there's also a chance that small and almost invisible cracks in the solder joints expands due to heating, causing intermittent connection to the PCB.
  • rokoroko
    edited January 2010
    Pilsener wrote: »
    Besides the possibility that the 7805 may fail due to high temp, as mentioned, there's also a chance that small and almost invisible cracks in the solder joints expands due to heating, causing intermittent connection to the PCB.
    Of that I have seen a number of very bad situations inside Opus Discovery Interfaces, including my own....
  • PilsenerPilsener
    edited January 2010
    I once got a problem with a TV, which failed every 2 weeks.
    I replaced the faulty parts, and it failed again after 2 weeks.
    Turned out to be faulty solder joints on a small heatsink that also worked as a connection for grounding.

    I hate such faults. They're very hard to solve.
Sign In or Register to comment.