Hi, new to the forum, fixing a Commodore PET

[GilesGoat]

Hi guys,

I hope I am not going to break some rule in starting a new thread about this, it could have gone maybe into that "Random characters on a PET screen" because that is the symphtom but let's go in order.

I joined this forum just today after suggestion cause "here seems the right place to be" if you are into putting back to work ( or modifying ) old computers.

But let's go with the main reason I joined this forum.

After many years me and my mate we finally managed to put hands on a Commodore PET 3032, it comes out from ebay from a guy that was treating it like "his best bet" really rarely seen such an old computer kept in such a mint condition it looks almost like "no human fingers ever touched it".

Only the motherboard had quite an amount of accumulated dust that been carefully removed by using a dry toothbrush and a vacuum cleaner centimeter by centimeter until everything was nice again.

I think this machine last time been turned on had to be around 1990 or such, it was working but the guy reported that "instead of 32K it was showing only 16K" but otherwise working fine.

It had a small incident the fuse holder got broken but I managed to fix that too.

However, once started the display comes out with a really nice ( perfect condition ) picture of random garbage.

Actually "not really random" precisely is ALWAYS THE SAME "garbage" of mixed chars on screen.

After quite reading this forum we tried various things and all I could say is :

- voltages seems to be correct in all places
- the CPU receives +5v
- there is a quite ok looking 5V clock about 1Mhz at pin 37
- reset line seems high
- RDY line seems high
- IRQ line is high too
- display is perfectly stable and nice looking

However .. all the address lines seems to be "stuck" a '1' it 'seems' like the CPU is kinda "frozen" and nothing is running, you can't see anything changing on those.

Tried to remove/reinsert chips, changed a few electrolitic capacitors, changed the crystal with a "fresh" 16 Mhz one.

No chip seems to be hotter than usual or such.

I am not sure what to think .. at some point we even tried removing ALL the ROMs to see if anything was happening, it looks like the CPU is not running at all.

I did try also to replace it with another ( it should be working ) 6502 I have spare .. nothing .. same thing.

I could/would post you ( tomorrow ) a link at some pics of it I took, you will see "everything looks now almost like new".

So it's quite puzzling .. thing is to begin to try to change some components here means "a bit of a mess" because everything is soldered directly on the PCB, I must say I never seen a CPU being "stuck" with AD lines all at 1 ..

Unless it would be waiting for some "BUSREQ" or such.

Any help would be appreciated, we really want to bring this machine back to life and be able to tell his previous owner that "the story had an happy ending".

About me huh .. there would be "even too much to say", really I got a point I actually spend more time "doing things" rather than trying to keep updated a website or blog or such about those because it would take longer than doing it.

I prefer to spend time trying to fix this poor PET

I will only briefly say "currently messing with VDHL, CPLDs, a Z8001 and some other stuff" :D

Cheers and thanks guys in advance for any help, and remember, it's not for me, it's for that poor PET CBM that we want to bring back to life

[edit] - a few pictures ( on the links ) added.

The machine as it arrived, it is in a really mint condition apart from the dust inside :

As it arrived, a bit dusty

After patient cleaning with a thootbrush and a vacuum.

Cleaned board

Finally after replacing the fuse holder ( that was broken ) and a bit more cleaning.

More clean

 

[carlsson]

Do you have access to a reasonably advanced EPROM programmer, one that recognizes the 2532 type of EPROM? Some of the newer, cheaper designs may not.

If you have a such programmer, you could lift all the ROM chips and one by one read them in the programmer as if they were 2532's. The second last ROM (2K Editor) may be read as a 2516/2716 chip though. Once you have read the contents, you could download known good ROM dumps from the Internet and using some suitable tool compare the two. Your programmer software may even have a verify mode built in. Just make sure you get the right ROM dumps as there are a couple different revisions. If all but one chip verify OK, that ROM might need to be replaced. If every single chip verifies OK, your problem is elsewhere, often the 2114 RAM chips if I understood correctly.

I am just about to check two sets of PET roms from our friends in France and Italy, lifted from two boards that have been tested otherwise OK.

 

[GilesGoat]

I have Wiley's Eprom Programmer, it does support 2716/32 but I don't think it can program the PET roms, I wonder if they can be read as that.

I could try anyway to make an "adapter" ( not sure if possible ) to replace those roms with a 27C64 ( EEPROM ) that are the ones I normally use.

2114 .. sdeng I have only TWO remaining "new", they are 2114 N-055 .. I think this means 55 ns access time but should be ok.

The only thing is, to make such a test it means to de-solder the original rams ( that could end up in destroying them ) and put some new socket and try the new rams in ..

Could do .. I hope it's worth doing .. is lightly 'risky' ..

 

[carlsson]

Well, the experts on this topic will soon chime in with the piggy-backing method, in which I believe you temporarily may place a good RAM onto a suspicious RAM. I'll let someone else explain the details though.

 

[GilesGoat]

I have Wiley's Eprom Programmer, it does support 2716/32 but I don't think it can program the PET roms, I wonder if they can be read as that.

I could try anyway to make an "adapter" ( not sure if possible ) to replace those roms with a 27C64 ( EEPROM ) that are the ones I normally use.

2114 .. sdeng I have only TWO remaining "new", they are 2114 N-055 .. I think this means 55 ns access time but should be ok.

The only thing is, to make such a test it means to de-solder the original rams ( that could end up in destroying them ) and put some new socket and try the new rams in ..

Could do .. I hope it's worth doing .. is lightly 'risky' ..

 

[Chuckster_in_Jax]

Well, the experts on this topic will soon chime in with the piggy-backing method, in which I believe you temporarily may place a good RAM onto a suspicious RAM. I'll let someone else explain the details though.

Not too long ago I was repairing a Compaq Portable and tried the piggy back method for checking for bad RAM. I was skeptical, but it actually worked!

I also desoldered all the RAM chips on a PET 8032 while back and installed sockets on it. If you take that route you need to be aware that it is tedious and there is a chance of damaging the solder pads on the motherboard. I had a real bad time desoldering the ground pins. The traces for ground are large and acted like a heat sink.

Chuck

 

[MikeS]

I'd certainly explore a little further before unsoldering RAMs & ROMs; let's see if anybody else has any suggestions. I take it you have a 'scope?

This where a NOP generator is handy; it disconnects the data lines (and therefore the memory) from the CPU and just lets it continuously step through all the addresses.

 

[dave_m]

- voltages seems to be correct in all places
- the CPU receives +5v
- there is a quite ok looking 5V clock about 1Mhz at pin 37
- reset line seems high
- RDY line seems high
- IRQ line is high too
- display is perfectly stable and nice looking

However .. all the address lines seems to be "stuck" a '1' it 'seems' like the CPU is kinda "frozen" and nothing is running, you can't see anything changing on those.

Make sure there is a good 1 MHz clock on the Phase 2 output (pin 39).

Check to see that Sync (pin 7) is pulsing. It should pulse at every instruction fetch.

If there are no low address lines pulsing (AB0, AB1, AB2, etc), this may indicate one or more of the big 40 pin chips are bad.

On a Reset, the CPU should load the program counter from ROM location FFFC and FFFD. Can you force a reset by momentarily grounding RST or check to see that you are getting a reset pulse of about 0.5 sec on power up?

Look at the zimmer's site to find the right schematic and board assembly drawing for your PET as there are many variations.

http://www.zimmers.net/anonftp/pub/cbm/schematics/computers/pet/index.html

I see, from the photos you added, that the big chips are all on sockets. That makes things a lot easier.

Originally there was 16K RAM reported so there is a bad memory chip(16K X1) in one of the eight upper 16K chips. The garbage characters on the screen means that the CPU is not running and/or the two video static RAM chips (2114) are not working.

Your model PET is the non CRT controller (6545) version which means the video timing will run properly even without a running CPU which is needed to initialize the 6545 video chip versions of the PET.
-Dave

 

[GilesGoat]

It took me a good 2 hours of sweat and prayers to arrive to this :

Yes I had a little problem with one track but I corrected it with some wire.

Checked all, etc. should be ok ..

However .. yes there is a change .. no it's not fixed yet .. what happens now
contrarely to before where we were always getting the SAME identical pattern now
the patter is random and sometime the 6th ? 7th char flickers with another one.

Every time you switch it on/off now the pattern is different around like this :

But is random now, it changes at every restart.

By the way I hope I haven't put too big images, if that the case shout me and
I'll replace them with just the links !

 

[GilesGoat]

Updarte ..

I tried to remove first the two 6520 chips nothing was happening .. just the same ..

However after I removed the 6255 now the screen boots in random garbage but after about 1 second ( probably a bit less ) turns into this :

Am I getting somewhere ? Could it be the 6522 is busted ?

Eh .. if that is the case .. where could I find one as replacement ? ..

 

[dave_m]

However after I removed the 6255 now the screen boots in random garbage but after about 1 second ( probably a bit less ) turns into this :

Am I getting somewhere ? Could it be the 6522 is busted ?

Pulling the 6522 VIA chip seems to allow the system to start up better. Normally on power up, the garbage is seen for a second and then is cleared and replaced with the Commodore message. So yes it looks to me like the 6522 is one of the things that may be bad. However note that the signal 'Graphic' is output from the VIA so that when was pulled, the open line sets the character generator into graphic mode and so who knows what the screen would look like.

 

[GilesGoat]

Still no joy ..

We did try to replace the 6522 with one temporary took out from a Commodore 1541 drive that actually has 2 of them in.

.. nothing .. now we are again stuck in that 'random garbage screen' and looks like the CPU is not running at all ..

Maybe later on I'll try to check again with the oscilloscope and such, last time I was watching clock and all was ok, but the waveform coming out from "Phase2" was quite a bit odd, despite it should a NC pin I mean in the diagram looks like "Phase2" is not being used anywhere.

The clock was looking quite distorted, instead of being a nice square wave it was like there was a strong capacitor attached to it I don't say it was like a sawthoot but you know that kind of "banana shape pulse" ?

 

[dave_m]

but the waveform coming out from "Phase2" was quite a bit odd, despite it should a NC pin I mean in the diagram looks like "Phase2" is not being used anywhere.

The clock was looking quite distorted, instead of being a nice square wave it was like there was a strong capacitor attached to it I don't say it was like a sawthoot but you know that kind of "banana shape pulse" ?

The 6502 output clock "phase2" (pin 39) is an important timing signal. Double check the schematic (which one are you using?). Also check to see that your scope probe has been "compensated" (there is a tweaker adjustment in the probe). Hook the probe to the 1KHz calibrate output on the scope, and tweak the probe screw adjust until you get a nice square wave. Then you can believe the signals you see on the scope.

 

[GilesGoat]

The probe was still calibrated, however I re-calibrated it again 'at best', I am using it in X1 mode, DC.

I made some captures, the settings are DC, 2V/div, 0.5Us/div.

I am actually getting a ground with a wire inserted into pin 12 of an empty ROM slot so there's a light path.

This is the clock that enters at PIN 37 :

This is the Phase 2 clock at PIN 39

Which I think it's quite strange .. ( I did check also that pin 21 and 1 are both ground, they are ).

This is the Phase 1 clock at pin 3

I should ( but I can't hold them ) use 2 probes to capture both of them and see how the phase relationship is.

Finally this is the SYNCH pin at 7, for the first time ever "seeing something" yet after a couple of reset this is normally stuck to '1', sometime at '0'.

Now .. checking the diagram, the clock seems to enter in my case UA0 ( I think it should be '10' but is really a '0' ) pin 9, which is a 7417.

There was some "rot" around that chip but I cleared it well before to start all this.

Bit odd that phase 1 and phase 2 clock is it ?

 

[dave_m]

Now .. checking the diagram, the clock seems to enter in my case UA0 ( I think it should be '10' but is really a '0' ) pin 9, which is a 7417.

There was some "rot" around that chip but I cleared it well before to start all this.

Bit odd that phase 1 and phase 2 clock is it ?

Very odd waveforms because Phase 1 is not connected to anything and Phase 2 is buffered by A10 with no other loads. I would think the 6502 CPU is bad except I know you tried another one. I wonder if there is some strange capacitive loading going on here??

 

[Pet Rescue]

Another newbie with the same faults!

Another newbie with the same faults!

Hi I am new to the forum too and just like GilesGoat I have a 3032 that is poorly too. It is a non crt chip type!

So far it has similar symptoms to GilesGoat's Pet.

When I very first bought my pet it just displayed a screen full of number 7's.
Then after checking a few voltage lines this then changed to a screen full of garbage that alters each time it's switched on.

I have replaced the 2114 static video ram chips. This made no difference just garbage again, I then piggy backed the 4116 ram with known good chips. No Change.

I then put the 6502 Cpu and 6522 Via into my trusty old Vic 20 and they work fine also checked the 2114 video rams.
The original ones were not working but the new ones are. I have swapped the 6520 chips and even left them out.

The clock is pulsing but I only have a 15Mhz scope so cannot see the waveform very well.
I have checked each Rom against known good ones on zimmers web site and found the last Rom in the slots the basic 2 Rom to be faulty.
I have made an adapter and put a 2732 in its place programmed up to the correct bin file.

The Pet now powers up with no display, once it has been on a while when you turn it off the on you get momentry garbage on the screen but then it blanks.

I grounded pin 22 on the expansion bus to test the reset on the cpu but this blew the 74ls04 hex inverter.
I replaced that and scoped the reset pulse and it worked normally, momentry low then high.

Now I have plugged a recently aquired cassette deck into the cn2 port(but no motors runs) and now it has gone back to a garbage screen on startup and when I turn off and unplug the cassette deck it does not go blank anymore when turned back on.

One thing I have noticed is that A1 pin 7 on the 4116 is held low while the others are pulsing a waveform.

Has anyone got any tips, I am waiting to get a better scope (100Mhz) to give me better waveforms.

 

[GilesGoat]

As soon as I'll have some time I'll try to put up a "CPU tester", well literally just a socket with a 1Mhz ( MOS compatible ) clock and power to a CPU.

Those shapes in that 2 clocks output look very odd I want to understand if that is "normal" or what I think they should look lot more square, I have somewhere ( in all my chaos around ) a Tektronix probe I'll try to measure again but "if the first clock looks right, so it should look the second".

I don't have any 7416 around to try to replace that but yes this been one of my second toughts too, maybe the buffer is faulty ? Mah ..

I'll have to do an adapter as well cause my programmer can't read those roms, I was thinking in making some adapter anyway to use 28C64 in place of the standard ones.

 

[dave_m]

One thing I have noticed is that A1 pin 7 on the 4116 is held low while the others are pulsing a waveform.

That is not good as the address lines are bussed to all the RAM chips and if A1 is shorted it will be hard to tell which chip is the culprit. There is a resistor R32 that may provide a clue between a shorted RAM or if the multiplexer (E3)that switches between the refresh addresses and the CPU address is bad. If both sides of the resistor are completely dead then maybe the problem is only a bad E3 mux. If the mux side of the resistor show an effort to pulse then one of the many RAM chips is shorted and a long trial & error may await you.
-Dave

 

[Pet Rescue]

GilesGoat will your programmer do 2716 and 2732 roms?

Thanks for the pointer dave_m, just as a coincidence my 16 DIP sockets have just come in the post.
I will take a look at that resistor R32 and the E3 multiplexer before I take out the ram en masse and
test/replace the 4116 rams.

Thanks again for the reply

 

[GilesGoat]

GilesGoat will your programmer do 2716 and 2732 roms?

Thanks for the pointer dave_m, just as a coincidence my 16 DIP sockets have just come in the post.
I will take a look at that resistor R32 and the E3 multiplexer before I take out the ram en masse and
test/replace the 4116 rams.

Thanks again for the reply

I am not sure .. it's based on the "Willem Eprom" 097ja , I think it requires a "PCB3" version ( mine isn't ) for that type of Eprom .. I got in on Ebay long ago from somewhere in China ...