Image Map Image Map
Page 3 of 10 FirstFirst 1234567 ... LastLast
Results 21 to 30 of 99

Thread: PET 2001-32N - Diag works in UD9, but with Kernal it fails to start

  1. #21

    Default

    Quote Originally Posted by iz8dwf View Post
    That one else is me
    The symptoms really smell like a data bus conflict probably dependent on A1 address line status. I'd pull the 6520s and the 6522 to see if things get better.

    Frank IZ8DWF
    Argh. First things first:

    Pulling the PIA and VIA chips makes no difference. I still see good voltages (-5.04, 12.01, 4.97) with a high speed (> 1 MHz I'd say.. it's difficult to measure since it's not a regular sinusoid) ripple that is a peak to peak of 300 to 500 millivolts. I actually think I should be looking at this as noise rather than ripple at this point. Exactly the same variation is on all three supply rails and present on the 22 volt as well... but it's super tough to tell if it's coming into the system from outside or just propagating up from somewhere in the system (like the regulators or something).

    I'm totally frustrated with these three supply DRAMs. Even though I built computers 30+ years ago, I've never worked with them before this. I have real work to do today, but later in the day I foresee an 8116 DRAM test circuit ending up on my smaller breadboard. Either my stock of DRAM was all questionable, the board is killing them, or something else is really wrong. I can't seem to find any stuck glue logic anywhere.

    I'm nearing the point where I decide to do some major reworking and scrap the DRAM completely and build a much simpler modern SRAM circuit and wire it into the board. I totally get why it was designed this way, but the 1 bit DRAM makes fault finding of transient problems really difficult... for me at least.

    If anyone has other ideas or pointers to an existing DRAM testing circuit appropriate for the 8116, let me know. I'm pretty sure I have a 7905 somewhere around here somewhere...

  2. #22
    Join Date
    Jun 2012
    Location
    UK - Worcester
    Posts
    2,978

    Default

    So, the high frequency stuff smacks of poor decoupling. Is it possible that some/most/all of the decoupling capacitors are faulty?

    I agree, you need to work out where the noise is coming from...

    If it is faulty DRAM, my test should find it. That is the purpose of the MARCH-B and MARCH-C memory tests. I will just need to write the documentation for you to understand how to interpret the results. I think the source code is fairly well described though (or am I thinking of the 8086 variant I created for work?).

    Dave

  3. #23

    Default

    Quote Originally Posted by daver2 View Post
    So, the high frequency stuff smacks of poor decoupling. Is it possible that some/most/all of the decoupling capacitors are faulty?

    I agree, you need to work out where the noise is coming from...

    If it is faulty DRAM, my test should find it. That is the purpose of the MARCH-B and MARCH-C memory tests. I will just need to write the documentation for you to understand how to interpret the results. I think the source code is fairly well described though (or am I thinking of the 8086 variant I created for work?).

    Dave
    Thanks Dave; the thing is that the system isn't even starting well enough now to run your tests! (which it was two days ago)

    I replaced all of the .1 uF caps on the DRAM last night. I'll pull out the two big disks today.

  4. #24

    Default

    Quote Originally Posted by daver2 View Post
    So, the high frequency stuff smacks of poor decoupling. Is it possible that some/most/all of the decoupling capacitors are faulty?

    I agree, you need to work out where the noise is coming from...

    If it is faulty DRAM, my test should find it. That is the purpose of the MARCH-B and MARCH-C memory tests. I will just need to write the documentation for you to understand how to interpret the results. I think the source code is fairly well described though (or am I thinking of the 8086 variant I created for work?).

    Dave
    I doubt any ceramic capacitor would fail in other ways than going to a dead short. And even that, it's one of the rarest things with those TH ceramic capacitors. 4116 DRAMs in my experience are very reliable if the supplies are good. I've repaired quite a lot of computers using them and the failed ICs among them are still < 5.
    One might get a bad batch, of course. The high ripple on all the supplies, if not a measurement error, in my opinion could happen with a data bus conflict. But removing all the I/O chips already eliminates lots of candidates.
    Are the DRAM data buffers (2 x 74LS244 on the 2001N) working well?

    Frank

  5. #25
    Join Date
    Jun 2012
    Location
    UK - Worcester
    Posts
    2,978

    Default

    I have seen 'teeth marks' on ceramic disc capacitors before now! Some insect or something clearly got hungry!

    Contrary to you, I have a lot of problems with 4116 DRAM. So it looks like a bit of "hit and miss".

    It could also be the refresh circuitry of course...

    Dave

  6. #26

    Default

    Quote Originally Posted by daver2 View Post
    I have seen 'teeth marks' on ceramic disc capacitors before now! Some insect or something clearly got hungry!

    Contrary to you, I have a lot of problems with 4116 DRAM. So it looks like a bit of "hit and miss".

    It could also be the refresh circuitry of course...

    Dave
    This is all 8116 DRAM.

    Teeth marks are definitely a possibility. Remember, this PET was a mouse clan's home in some guy's garage for a couple of decades. I've finally gotten the last of the mouse urine and carcass stains off of the case and board.

    I pulled the two caps. One was fine, the other was reading almost half of the rated capacitance. Being fed up at this point, I replaced them with some big X2 caps that are slightly higher than the originals. The ripple is down to 150-300 millivolts peak to peak. It's running the PET DIAG pretty reliably again, though that's reporting every other column as B (first column) and it has developed a bit of a random screen artifact situation.

    I've watch the 244s and they have seemed to be working. I've also tried piggybacking them, with no differences observed.

    I'm leaving this alone for the rest of the day and going back to my real job for at least a few hours. If anyone has other thoughts, let me know. In the mean time I have a +5 -5 +12 switching power supply on the way (man, I used to use these all the time! In the mid-80s I worked repairing arcade boards and games) for my DRAM tester project and, I think, I will ultimately bypass the entire power circuit and feed the board directly from that. This transformer periodically makes some ominous humming sounds. (I have checked the current draw and it doesn't increase during the humming, so it's not something on the board drawing more power.)

  7. #27
    Join Date
    Feb 2009
    Location
    Southern California, USA
    Posts
    2,662

    Default

    Quote Originally Posted by dhoelzer View Post
    This transformer periodically makes some ominous humming sounds. (I have checked the current draw and it doesn't increase during the humming, so it's not something on the board drawing more power.)
    Tighten the 4 screws holding the transformer to the standoffs.

  8. #28

    Default

    Quote Originally Posted by dhoelzer View Post
    This is all 8116 DRAM.
    MB8116 is how Fujitsu called its own variation of the "4116" family. In all respect they have all the same addressing, refreshing and power requirements. I still believe it can't be a power supply issue at this point. There must be something that reacts badly on the data bus (possibly A0-triggered, but that would mean one I/O chip or one ROM only).

    Frank

  9. #29

    Default

    Quote Originally Posted by iz8dwf View Post
    MB8116 is how Fujitsu called its own variation of the "4116" family. In all respect they have all the same addressing, refreshing and power requirements. I still believe it can't be a power supply issue at this point. There must be something that reacts badly on the data bus (possibly A0-triggered, but that would mean one I/O chip or one ROM only).

    Frank
    Thanks Frank. Yes, 4116 and 8116 are equivalent. Was just clarifying, sorry if it sounded otherwise.

    I'm pausing to regroup until tomorrow afternoon when the other power supply arrives. It will give me a really good opportunity to measure that one very precisely, solder leads onto the board in the +12, -5, +5, and ground and see how it changes both the behavior of the board and the noise/ripple on that supply.

    Thanks for your and everyone else's help! I am determined to get this thing working and to at least understand what's making it behave this way, even if I end up scrapping the on-board RAM as designed.

  10. #30
    Join Date
    Feb 2009
    Location
    Southern California, USA
    Posts
    2,662

    Default

    Quote Originally Posted by dhoelzer View Post
    More interesting... Powered on this morning after it completely failing last night... It shows DRAM problems again, but starts the tests in the KERNAL slot.
    If you are saying the kernel pettest runs, but the EDIT slot pettest EPROM does not, then I suspect one or both of the D8 and D9 ROM sockets or traces around them. Every time you swap a chip there, things change.

Tags for this Thread

Bookmarks

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •