SuperPET / MMF9000 with strange video issues

[Gabriele72]

I'm back to vintage computing after a while, and decided to take a look at a MMF9000 SuperPET that's been sitting in a corner reclaiming attention for years.
The SuperPET is a heavily expanded 8032, and (to start with) the mainboard was completely dead.
After fixing some bad RAMs I was welcomed with the beep and a screen full of blocks. I noted that /CS2 on the char EPROM was being held high. Since it's driven driven directly by the 6545 I desoldered and replaced the CRTC.
Great improvement, but my joy was quickly shattered as I was left with this screen:




The boot message is doubled (or tripled) and misplaced, and enlarged ghost chars appear as you type.

Tried 3 different 6545s so the problem should not be the controller.
I tend not to blame the video RAM either, as I can poke and peek correctly from location 32768 to 34768.
That should be the end of the video memory (2000 locations) right? But if I go on pokeing beyond that address I can see that line of weird characters at bottom that gets overwritten.

Beyond that, evertything seems fully functional.

I don't have a scope or a logic analyzer, only a multimeter and a couple of logic probes.
What to check next...?

 

[dave_m]

Tried 3 different 6545s so the problem should not be the controller.
I tend not to blame the video RAM either, as I can poke and peek correctly from location 32768 to 34768.

At first glance it appears to me that the refresh address counters are getting mixed up, but that function is in the 6545 and you have replaced it. To account for the repeated Commodore message, perhaps two of the TA address lines or SA address lines are shorted together or one is shorted to ground due to solder splash, etc? With power off, you may have to do some ohmmeter checking and/or physical inspection of those lines.

Is the SuperPET board connected at this time?

 

[Gabriele72]

Is the SuperPET board connected at this time?

No, I'm troubleshooting just the 8032 mainboard.
However, I found it: it was the flip-flop at C1 that wasn't working right. It is the first time in my life that piggybacking helps

Anyway, happiness didn't last long and the machine developed an even more elusive fault: when the computer is cold, it boots up OK. After a minute or so, the chars start turning into solid blocks, and also the screen area outside.
Then the screen starts to shrink vertically more and more, as if the vertical drive is going crazy. Sometimes the monitor buzzes and whistles
This PET is going to drive me mad...

 

[RobF]

Anyway, happiness didn't last long and the machine developed an even more elusive fault: when the computer is cold, it boots up OK. After a minute or so, the chars start turning into solid blocks, and also the screen area outside.
Then the screen starts to shrink vertically more and more, as if the vertical drive is going crazy. Sometimes the monitor buzzes and whistles
This PET is going to drive me mad...

Check the voltage regulators on the main board; the video board (well, more accurately, the CRT controller) is very sensitive to fluctuations in the supply voltage -- and I've seen the SuperPET voltage regulators fail as they warm up.

The buzzing and whistling is the flyback transformer trying to cope with signals it can't handle.

- Rob

 

[Gabriele72]

Took a while to swap some ICs but the machine eventually reached a stable state. Now it seems to boot OK only to freeze when I enter a line number.
Once this (hopefully last) issue is resolved, I'll be ready to fit the daughterboards. I wonder how could I try them, though... do I need special SW or maybe the right SYS could start the fireworks? :D

 

[dave_m]

Now it seems to boot OK only to freeze when I enter a line number.

After the boot, does the cursor blink properly indicating the 60 Hz interrupt is running? Without this, the keyboard is not scanned.

 

[daver2]

Gabriele72,

Do you have the Waterloo disk set for the SuperPET?

If not, a set of '6702 dongle removed' d64 images can be found at http://mikenaberezny.com/hardware/superpet/waterloo-languages/ along with PDF file as to how to test all of the Waterloo languages for the SuperPET.

I got the original .d64 images from somewhere - but I was working on the SuperPET emulation for VICE and we needed to initially remove the dongle protection (so I don't quite remember where I got the original images from). I may have a set of 'unhacked' image files if I have a look around on my hard disk if you can't find them though.

Dave

 

[billdeg]

http://vintagecomputer.net/commodore/superpet/
a few disk images I made a while ago, if you have the means to build a disk from a .D80 file.

I could be convinced to post the language disks, assuming they're copy-able.

 

[Gabriele72]

@dave_m

yes, cursor is OK, I can type and enter BASIC commands. Some direct commands are executed, others result in a syntax error, and entering a line number freezes the machine.
If I plug in Nicolas Welte's 6502-RAMROM expansion and bypass the ROMS, the machine seems to work OK, that's why I tend to blame the ROMs themselves or some support logic IC (decoding logic?).
Unfortunately that expansion interferes physically with the daughterboards, so it's not a viable easy solution.

@daver2 and billdeg

No original disks unfortunately, just the bare machine but thank you both for the links.
I'll try the unprotected .d64 first (once I get SuperPET running, and I don't even know the status of the 6702) and then the rest.
I cannot write .d80s directly, I have an 8050 drive and several 3040s though, so if it's just a set of readable files I could use the d64 format as a bridge, more or less like this:

Emulated d80s -> Emulated d64s -> 1541 -> real 170k disks -> 3040+8032 -> 8032+8050 -> real 500k disks

 

[billdeg]

search ebay for TPUG Library CD, I believe the files are there. not sure the format.

UPDATE: This CD can be purchased from TPUG who appears to be selling a lot of them on Ebay (there is an auction selling CDs for $20). There is a directory on the CD PET-->SuperPet containing .d64 images.

Bill

 

[MikeS]

UPDATE: This CD can be purchased from TPUG who appears to be selling a lot of them on Ebay (there is an auction selling CDs for $20). There is a directory on the CD PET-->SuperPet containing .d64 images.

Bill

Thanks for the plug! Your commission check is in the mail ;-)

 

[dave_m]

@dave_m

If I plug in Nicolas Welte's 6502-RAMROM expansion and bypass the ROMS, the machine seems to work OK, that's why I tend to blame the ROMs themselves or some support logic IC (decoding logic?).

Does the machine language monitor work. Enter sys1024. If so, we can display some ROM data and see if anything looks suspicious. Does the board contain 24 pin ROMs like a normal 8032? Are they on sockets or soldered?

 

[Gabriele72]

Does the machine language monitor work. Enter sys1024. If so, we can display some ROM data and see if anything looks suspicious. Does the board contain 24 pin ROMs like a normal 8032? Are they on sockets or soldered?

Unfortunately SYS is one of the commands that returns a syntax error...
The ROMs are on sockets (another variable in the equation) and it looks like a standard romset to me.
Perhaps I can read them as 2532 in an eprom programmer ?

 

[dave_m]

Perhaps I can read them as 2532 in an eprom programmer ?

Yes, but unless you have an ancient EPROM programmer, it may not handle the 2532 type EPROM. But most can handle a 2732, but you would need to make an adapter out of a 24 pin socket to change the connections on a couple of pins. Do you have a programmer that can read the 2532 or the 2732?
-Dave

 

[Gabriele72]

Do you have a programmer that can read the 2532 or the 2732?
-Dave

Yes, I have a Needhams EMP-20, a HiLo ALL-11 and a Leap Leaper-10, all should be able to deal with the 2532.
I could just burn a whole new romset but I don't like wasting my few remaining 2532s... so I'll try to read & compare the original ROMs first.

 

[Gabriele72]

Guys, I'm lost...
I've dumped and compared the whole romset with the files available online and they're identical.
The romset is a standard 8032 set:

901465-20
901465-21
901465-22
901465-23

screen editor 901474-04

A brief summary:
The 6502 RAM/ROM expansion allows me to bypass the whole romset. I cannot choose which rom to bank out: it's all or nothing.
With it, the machine works. Without, BASIC is acting weird:
The only commands that work seem to be PRINT, END and NEXT.

/ <RETURN> gives me "press play on tape #1"
- <RETURN> gives me "press play on tape #1"
* <RETURN> gives me "press record&play on tape 1"
= <RETURN> gives me "Can't continue error" :shock::shock:
0..9 <RETURN> lands me in monitor but it doesn't work.
any line number <RETURN> freezes the machine.

There might be something bad in decoding logic or the sockets etc... but before blindly desoldering everything, does anyone have a clue which of the ROMs could be involved?

 

[dave_m]

There might be something bad in decoding logic or the sockets etc... but before blindly desoldering everything, does anyone have a clue which of the ROMs could be involved?

If you have a scope, you might use the simple NOP Generator to check all the address lines and address line decoder signals (chip selects) at the ROMs. Or you could install the Welte board and run short BASIC programs that loops one ROM address space at a time say from B000 to BFFF and probe the address lines on the B socket looking for anomalies and then go on to the C area, etc.
something like this to exercise the B address lines:
10 for i = 45056 to 49152
20 peek (i)
30 next i
40 goto 10 or stop, etc

If no scope, you might consider reflowing all the solder joints on the ROM sockets.

I took a quick look at the BASIC 4 disassembly listing and just got confused. To me it looks like the problem could be in the B000 (UD10), C000 (UD9) or D000 (UD8 ) ROM sockets.

 

[MikeS]

I assume you're just replacing the ROMs, and not the RAM as well? Also, presumably you've checked all the sockets?

Interesting that the ROMs are even in sockets; does it look like it came from the factory that way, or has someone perhaps already been trouble-shooting?

 

[Gabriele72]

If you have a scope, you might use the simple NOP Generator to check all the address lines and address line decoder signals (chip selects) at the ROMs.

No scope unfortunately, only logic probes. I could still build the NOP generator but I doubt it could be of any help.

Or you could install the Welte board and run short BASIC programs that loops one ROM address space at a time say from B000 to BFFF and probe the address lines on the B socket looking for anomalies and then go on to the C area, etc.

But then I would access the Flash EPROM on Welte's board, I think the board disconnects the original ROMS from the bus... or not?

If no scope, you might consider reflowing all the solder joints on the ROM sockets.

Just back from doing it... no improvements.

This board has issues beyond my understanding. Probably the best tool to isolate the problem would be the PETVet.
I think I'll go the easy route: as soon as Sinchai.de is back online I'll get another 6502 expansion and fit it in permanently. I measured the spacers and there should be enough room (I'm unsure whether I lose access to the option ROMs though).

 

[Gabriele72]

I assume you're just replacing the ROMs, and not the RAM as well? Also, presumably you've checked all the sockets?

Interesting that the ROMs are even in sockets; does it look like it came from the factory that way, or has someone perhaps already been trouble-shooting?

RAM seems to be OK, at least the first 16K. I had to replace some of the 4116s at the beginning, but the upper half still has one (or more) faulty chip.
Concerning ROMs, it looks as if someone installed sockets at a certain point, it's a rework for sure.

BTW I found 2 option ROMs I had forgotten about, I did remove them some time ago. UD11 is a 2532 and UD12 is MOS 901465-02.
But this is BASIC 2.0, isn't it?
Was the european SuperPET given the ability to run BASIC 2.0 or request? They are connected to a select switch which looks original (not a hack) but it's not mentioned in the assembly schematics.