Who has an Issue 1 or Issue 2 Spectrum??

edited November 2007 in Sinclair Miscellaneous
Do you have an Issue 1 or 2 Spectrum (ie machines with 5C102E and 5C112E ULA's) ???

You're help is needed! PLEASE SEE THIS THREAD:

http://www.worldofspectrum.org/forums/showpost.php?p=225399&postcount=1

Thanks!
Chris
Post edited by csmith on

Comments

  • edited November 2007
    csmith wrote: »
    Do you have an Issue 1 or 2 Spectrum (ie machines with 5C102E and 5C112E ULA's) ???

    You're help is needed! PLEASE SEE THIS THREAD:

    http://www.worldofspectrum.org/forums/showpost.php?p=225399&postcount=1

    Thanks!
    Chris

    Issue two's - I have lots - finding them at the moment will be a chore, but they are VERY common.

    Issue 1 is a rare beast tho!
    Oh bugger!<br>
  • edited November 2007
    DEATH wrote: »
    Issue two's - I have lots - finding them at the moment will be a chore, but they are VERY common.

    That's great news! Can you keep your eye out for them/one?

    Specifically I want to see if it's got R57 in place between the ULA RAS and the Z80 RFSH, and whether the ULA RAS ever goes 'high impedance' - I think it probably does not, and so even with R57 in place, the 4116 RAS will never 'see' the RFSH signal (which is why it is removed on later issues and advised to be removed in the service manual).....

    I just need proof for my book-thing.

    Cheers!
    Chris
  • edited November 2007
    I have an Issue 2 machine, but it may not have the original ULA (I'll have to open it and check inside). Unfortunately, I don't have a storage scope so I probably won't be able to tell whether there's a missing RFSH. (Hooking up the logic analyzer to it will be a bit hard too since there's no pin headers, but if you're stuck, I might be able to break out one of the 4116s onto breadboard to allow me to poke the logic analyzer in somewhere, won't be tonight though!)
  • edited November 2007
    Winston wrote: »
    I have an Issue 2 machine, but it may not have the original ULA (I'll have to open it and check inside). Unfortunately, I don't have a storage scope so I probably won't be able to tell whether there's a missing RFSH. (Hooking up the logic analyzer to it will be a bit hard too since there's no pin headers, but if you're stuck, I might be able to break out one of the 4116s onto breadboard to allow me to poke the logic analyzer in somewhere, won't be tonight though!)

    Ok, thanks Winston.
    The trick is to see if the ULA RAS ever goes high impedance. If not, then coupling RFSH to the 4116 RAS via R57 (330R I think) will do nothing....

    This might be a simple test: Could you try the following some time?

    Can you put your scope on Z80 RFSH and the 4116 RAS? If the RAS tracks RFSH, then the signal is getting through and the ULA RAS output must be going high impedance. This will be difficult to read though, as the ULA updating the display is going to get in the way.... Hmmm

    Better still, trigger your scope on ~INT and look at what RFSH is doing within 6 ms from the trigger (any more and you'll be about to hit the first display row). Then do the same for 4116 RAS instead of RFSH. If the traces are the same, then RFSH must be getting through to 4116 RAS, so again, the ULA RAS must be going HI Z.

    You might not have R57 though.... Again, that would be good to know. If you have a real 5Cxxx ULA, then can you tell me how long after ~INT the ULA RAS go low?

    Edit: Oh, make sure that the Z80 has disabled interrupts and is spinning on it's head in the upper 32K. Perhaps DI followed by HALT ?

    Answers to any / all of these will prove / disprove lots of theories....

    Thanks again!
    Chris
  • edited November 2007
    My issue 2 has the 5C112E ULA and does *not* have R57 fitted (the absence of flux stains on the PCB seems to suggest it was never fitted).

    Might be the weekend before I can hook it up to the scope, my table with the scope currently looks like an explosion in an electronics factory!
  • edited November 2007
    I'm pretty sure I have at least one Issue 2 and maybe even an Issue 1.
    Does 16K/48K RAM versions make any difference?

    Also, which Issue had the Spider bridge?

    I can have a look at the weekend. I can tell you what ICs are present etc. but I don't have the skills or equipment to "listen" in on rapidly oscillating voltages etc. Not sure if I can help?
  • edited November 2007
    BloodBaz wrote: »
    I'm pretty sure I have at least one Issue 2 and maybe even an Issue 1.
    Does 16K/48K RAM versions make any difference?

    Nope, doesn't matter about ram.
    Also, which Issue had the Spider bridge?

    Issue 2
    I can have a look at the weekend. I can tell you what ICs are present etc. but I don't have the skills or equipment to "listen" in on rapidly oscillating voltages etc. Not sure if I can help?

    Where are you?
    There are a couple of people helping with the issue2 machine, getting a look at the issue 1 is a bit like finding the holy grail....

    Chris
  • edited November 2007
    csmith wrote: »
    Nope, doesn't matter about ram.
    Where are you?
    There are a couple of people helping with the issue2 machine, getting a look at the issue 1 is a bit like finding the holy grail....

    Chris

    Derby, UK.
    I'll check tonight the Issue #.
  • edited November 2007
    Slightly off topic (sorry!) ... But i am sure i read that issue one machines have a hole in the underside for some kind of adjustment ? :-?

    Anyone confirm this (pontentiometer i expect) and whats its purpose is ? :)
  • edited November 2007
    My scope refused to trigger on the INT line (may need to be sent through an inverting buffer). However, it did trigger on /RAS on the 4116 memory. On the 2ms/cm sweep setting, it showed approximately >5ms gap in the /RAS line to the 4116 memory. If I then connected the other channel to /INT it showed that the interrupt happened about mid-way during the /RAS gap.

    The time measurement isn't exactly precise, my scope is purely analogue and I had to try and calibrate the sweep time by trying to make the sweep take 2 seconds to go across the screen on the .1s/div in 2 seconds, so I could be a couple of ms off :-) I suspect you were spot on at the meet - with interrupts disabled, the lower RAM will not get refreshed between the ULA finishing with the paper area and starting on the paper area of the next frame. The Z80's RFSH signal is very definitely getting nowhere near the 4116 memory.

    With interrupts enabled, there was /RAS activity in this gap - the system variables getting updated once per interrupt.

    Here's a photo anyway. /RAS is the top trace, /INT the bottom.

    http://www.alioth.net/tmp/SpecIssue2RAS.jpg

    Incidentally, the same thing is true of the Issue 4S Spectrum.
  • edited November 2007
    I've an issue one pcb right in front of me. Bought it a couple of years ago for a small fortune on Ebay.
    I don't remember if I've tried it to see it working.
    I'll give that a go first.

    :smile:
    Frans
  • edited November 2007
    Just tried to see if it works.
    Sadly I get black/white/black/white 4 bars across the screen. Sometimes a few flashing squares in there too.
    It's not booting up...

    Oh, it's also got a 32kb memory expansion, but with or without it it does not start up.

    :(
  • edited November 2007
    Cameraman wrote: »
    Just tried to see if it works.
    Sadly I get black/white/black/white 4 bars across the screen. Sometimes a few flashing squares in there too.
    It's not booting up...

    Oh, it's also got a 32kb memory expansion, but with or without it it does not start up.

    That'll be the lower 16K memory gone. As others will point out here, may have been caused by a flaky power circuit...

    Don't panic though, it's quite fixable.

    Chris
  • edited November 2007
    I wasn't panicking yet...
    Too bad the memory chips aren't socketed on this pcb...
    I'm going to have to desolder them all...
    Does one replace all memory chips or is it possible to figure out which exact ic's are faulty?
    I have the components to build Winstons' tester but didn't get around to install Linux to be able to run that PCB program he uses. Or maybe someone has recreated his pcb in Eagle?
  • edited November 2007
    First thing to check is -5V, +5V and +12V lines on the 4116s.
    If they're all good see if one of them is getting hotter than the rest, if they're malfunctioning they often get hot.
    If you have access to a EPROM burner and preferably a multiface you can load a RAM diagnosis program via the multiface button which'll diagnose which RAM chip is dead... otherwise you're probably stuck with a brute-force method (i.e. checking the chips one by one). As it's an issue 1 please be careful, there's not many left. :)
  • edited November 2007
    Cameraman wrote: »
    I wasn't panicking yet...
    Too bad the memory chips aren't socketed on this pcb...
    I'm going to have to desolder them all...
    Does one replace all memory chips or is it possible to figure out which exact ic's are faulty?
    I have the components to build Winstons' tester but didn't get around to install Linux to be able to run that PCB program he uses. Or maybe someone has recreated his pcb in Eagle?

    You don't need PCB or Eagle or any other PCB layout software unless you intend on modifying the layout.

    Get a Gerber viewer - there are loads of free ones, and use the gerber files, which are all available to download - then you can print the gerbers on toner transfer paper (for TT method) or transparency (for UV method). You can also use the PostScript files directly if you have a laser printer - use the 'mirrored' file for the component side, and the 'non-mirrored' side for the solder side. (The gerbers will be best since you can just print off the files you need).

    Alternately, use strip board and make it up from the schematics using point-to-point wiring, at least one person has done that. The schematics are on the site as plain PNG images.

    To desolder the chips, the fastest way is to use hot air. An inexpensive electric paint stripping gun works well. Practise first with a PCB you don't care too much about so you can find the distance at which to hold the gun - melting the solder but without scorching the PCB. Then ease the chips out with a small screwdriver or a pair of pliers. Don't hold the board upside down while you do this, unless you want to resolder all the passives too, since these will simply start falling out the board when the solder melts! Incidentally, four big wide lines indicates that *none* of lower RAM is working - check the +5, +12,. -5 and GND connections to the 4116 RAM first. It wouldn't be at all surprising if the internal switch mode supply on the Spectrum's PCB is defunct.

    You'll still need to remove the excess solder from the holes in the PCB - solder wick, a solder sucker, or 0.8mm drill works well, but there is A LOT of solder on a Spectrum PCB.
  • edited November 2007
    Thanks Winston for all the tips. I didn't know about the Gerber viewer.

    Also, I do have an Eprom burner and Multifaces, what program do I burn to the Multiface to check Ram?

    Thanks, Frans
Sign In or Register to comment.