Stubborn 2001n/3032 with shorted 12V

[geekdot]

With this particular board, it is almost certain that at least one of the 2114 Video RAM IC's it defective, they have been on all versions of this board I have seen. You appear to have the original parts there, not good.

Mostly for these boards, the support IC's and buffers are generally good. The fault usually resides in te DRAM itself, one of the ROMs or the Video SRAM top culprit.

Thanks Hugo for joining the fight against this stubborn PET!
It's a long thread already and it's tedious to re-read everything Dave and I tried already. So here's aa quick recap:

• All RAMs and ROMs were removed, externally tested ("Retro Chip Tester" to the rescue)
• Dumped all ROMs and diff'ed them agains its zimmers.net equivalents
• Nevertheless did test with the EDITOR and BASIC ROMs replaced by the correct EPROMS (2726 and 2532) just to be sure
• Some 'misbehaving' or suspicious chips were replaced (UG5, the LS154 and the 555)
• PETTESTER runs like a charm. Many RAM test loops etc., also proving that the VRAM is OK (IMHO)

If it does not boot, then at least you know that you are defintely not dealing with a problem internal to DRAM or SRAM or ROM. So then you can get about checking interconnections and checking the support IC's for their logical operations with the scope, ROM select circuitry, buffers and so on and CPU signals and moving back to the Pettester if required.

That's where we're currently at, checking the ROM select lines which do seem to work fine with PETTESTER ROM but the original 901447-24.bin (as well as an EPROM) doesn't.

So what's the difference then? Like @daver2 pointed out, his PETTESTER is not checking for the /IRQ line... which is - when using the 901447-24 ROM - always high like a hippie
The only part I did not spend a whole day on is the 6522 VIA. Just digging through my basement if I can find another one (or at least socketed in some donor)...

 

[Hugo Holden]

Ok thanks for the recap and to help get more up to date: With all the known good ROM's in place (it could be worth checking them again) and known good DRAM IC's , and the SRAM probably ok and confirming the CPU power on reset pulse, what is the main issue when you boot the computer currently, is it just the apparently inactive /IRQ and failure to boot with a black screen or is it a garbage screen ? Is there any CPU activity; what is the sync pin doing? And what happens if the PIA ICs or VIA are removed or the upper bank of 8 DRAM IC's removed as an experiment, and boots are then re-attemped is there any change in the behavior with any of those tests ? And have you done the individual pin friction tests on all socketed IC's ?

I'm more or less suggesting, that since the board is not fixed yet, that something has been missed and to go back over some basic tests, without skipping any steps.

 

[daver2]

Agreed, we have clearly missed something obvious somewhere....

PETTESTER seems to run fine.

From what I can deduce, with the EDIT ROM in place, the screen is being cleared (signs of EDIT ROM activity) but the EDIT ROM appears to be the only thing that is then accessed - hence the suspicion regarding the /IRQ line.

For Hugo's benefit this PET was fitted with a separate graphics card, and A5 pin 9 was grounded, thus disabling the reading of the video RAM. We found (and removed) this modification. However, there may be other modifications that we have not found...

Dave

 

[Hugo Holden]

.......In looking for an explanation of why things seem fine with the PETTESTER ROM, but the computer won't run with the "apparently normal Edit rom".......another question: the photo that you posted on post #98, the Edit Rom you have programmed there in socket UD8, it looks like a 2732, from the photo (a little blurry). Is this the PETTESTER ROM, or the ROM you made with the correct Zimmers file in it for normal computer operations ?

The UD8 socket works with a 2716 if pin 21 is tied to +5V or a 2516 or a 2532, I have not confirmed if it will work with a 2732 myself, I have not tested that in my PET. It might depend on your board what was done with pin 21, whether it was connected to +5V directly or via a pullup resistor, which could affect the operation of a 2716 or 2732 in that socket. Also thinking that with the 2732, A11, pin 21 gets tied high with the standard pcb wiring, so the file inside it has to be in the upper half of the 2732's ROM memory space. But you might have only programmed the lower 1/2 of it from the Zimmers file ? (on the other hand, I think in the case of the 2532, the file would have to reside in the lower half of it)

 

[geekdot]

Yet another couple of "hope seconds"...
I saw that "x8xx" signal, which is generated by B2 (7425) by (N)ORing BA8-BA11. And that signal is used as CS0 of the PIA, CS0 of the IEEE VIA and OR'ed with BA6 as CS1 for the Keyboard VIA.
So it's not unimportant and what if the Editor ROM waits for one of those guys?

Clipped my probes onto the 7425 to see if that signal is generated by looking at the 4 NOR inputs. And because I had unused probes I've put one to BA11 before entering the inverter (A3) close by.

While the 7425 seemed to work correctly, something was strange in the input side... and lo and behold, input pin 5 should be an inverted BA11... but it was totally off what the probe on BA11 showed.
So all probes went onto the 74LS04 (A3): And sure enough, all its inputs we not correctly inverted on the outputs! Unsoldered and checked it externally: It's dead (as a Parrot).
Again: Solder a socket and in went a fresh & tested 74LS04... crossing fingers, power on: Same-Same

 

[daver2]

A3 is used all over the place... for example:

1. It drives the /RESET line.
2. It drives the /B R/W line.
3. It drives the RAM R/W line.

So how many of the six (6) inverter gates of A3 were dead?

I would have liked to have seen traces...

A3 pins 1-2, 3-4, 9-8 and 13-12 appear to be used (4 out of 6 gates).

A3 pin 8 drives the DRAM circuitry.
A3 pin 12 drives the video circuitry.

Dave

 

[geekdot]

A3 is used all over the place... for example:

In-system I checked on 3-4 and 11-10 (drives "SEL 8") as they are (literally) on the same page
They were both outputting nonsense... but given what you wrote I gave it another test, this time in my "Retro Chip Tester" which reports it as "OK" - that is strange.
Well, given /RESET as well as RAM worked fine with it, I guess that was a measuring f-up.

 

[Hugo Holden]

In-system I checked on 3-4 and 11-10 (drives "SEL 8") as they are (literally) on the same page
They were both outputting nonsense... but given what you wrote I gave it another test, this time in my "Retro Chip Tester" which reports it as "OK" - that is strange.
Well, given /RESET as well as RAM worked fine with it, I guess that was a measuring f-up.

74LS04's do fail from time to time, I once had one fail in use, with my SOL-20. In this case it behaved as though its input pin was O/C inside the IC, so that its output logic state went to a normal TTL low (O/C TTL inputs normally assume a high state)

One interesting thing, while that failure is easy to detect in an inverter gate because a signal goes missing at the output, if you get that on a multi-input gate, the gate's output can respond to a pattern on the remaining inputs when it should not. When you check with the scope, checking each pin separately, all looks normal with pulses, but the logic is fouled up. I have had NAND gates, supervising counter chain conditions do this sort of thing, makes for awkward faults. This is where it does pay to check an IC is obeying its truth table.

Did you look into that possible issue with the 2732 ?

 

[daver2]

So this was the machine with the disabled internal video circuitry (via the little link) and the additional video card.

With all of this chip changing, can we go back to PETTESTER and make sure that the video RAM, ROM checksums and DRAM is working correctly again...

Dave