Commodore PET working intermittently

[ajgriff]

Originally Posted by John Earland

I've tried this several times and it always does the same - it's almost like a reset but no cursor.

The absence of a cursor can be indicative of a problem associated with one of the PIAs (UC7 in this case).

Alan

[John Earland]

Thanks Alan,
Yes - I got the 6520 and 6522 round the wrong way! When the PET boots it now does so with a cursor and works well. It's just this intermittent screen dump of random characters! Mmmm. When I apply slight pressure to the board and switch it on it often boots then when I try without it it doesn't so it does feel like a bad connection. I've tested the new socket with my multi-meter and that seems fine.

[SiriusHardware]

Have you already tried methodically removing and reinserting every socketed IC? Normally this advice would be viewed with some trepidation because the original PET sockets are so notoriously bad that disturbing them can sometimes create problems which did not exist originally. However I remember, and you mentioned earlier, that all of the sockets bar UF10's were replaced by an experienced solderer during your first round of problems, so the advice to try reseating all the ICs in their modern sockets does not present the hazard that it might otherwise.

Needless to say, if you do this you do it with the power off.

[NivagSwerdna]

The reason I mentioned the finger trick is that an intermittent connection on nSELE, nIRQ etc can cause this behaviour... and it is sometime shown up by the finger trick... the probable root cause will be a dodgy socket either at the CPU socket or the VIA.

[John Earland]

Originally Posted by SiriusHardware

Have you already tried methodically removing and reinserting every socketed IC? Normally this advice would be viewed with some trepidation because the original PET sockets are so notoriously bad that disturbing them can sometimes create problems which did not exist originally. However I remember, and you mentioned earlier, that all of the sockets bar UF10's were replaced by an experienced solderer during your first round of problems, so the advice to try reseating all the ICs in their modern sockets does not present the hazard that it might otherwise.

Needless to say, if you do this you do it with the power off.

Hi, yes. I've taken each one out and abrased the pins slightly and sprayed the sockets with electronic cleaning solvent - same thing each time, the random character screen. If I apply slight pressure on the ROM/RAM board it will boot but it will now not recognise the Disc drive = ?Device not present error, which I wasn't getting before.

[John Earland]

Originally Posted by NivagSwerdna

The reason I mentioned the finger trick is that an intermittent connection on nSELE, nIRQ etc can cause this behaviour... and it is sometime shown up by the finger trick... the probable root cause will be a dodgy socket either at the CPU socket or the VIA.

I did just this with the CPU and yes it did work, though now it won't recognised the disc drive. It is progress I guess. I've not tried with the VIA

[John Earland]

So latest update is that. If I hold down the CPU more often than not the PET boots. It will load programs from the cassette and they will run but it no longer recognises the disc drive. Sometimes when I type CATALOG I get a short delay before it returns 'READY', sometimes I get the '?Device not present' message. I have tested the socket connections and they seem to be fine but there is obviously some poor connection somewhere.

[SiriusHardware]

I take it the CPU socket was one of those changed last time around. What sort of socket is it, conventional or turned-pin? What sort of pins does the ROM / RAM adaptor have?

[John Earland]

Originally Posted by SiriusHardware

I take it the CPU socket was one of those changed last time around. What sort of socket is it, conventional or turned-pin? What sort of pins does the ROM / RAM adaptor have?

Yes, it was. I could ask someone else to double check them for me i.e. the guy who did them originally. The sockets are conventional and the ROM/RAM board has the turned in pin types i.e. not flat.

[SiriusHardware]

OK, well, turned (round) pins don't always get on well with conventional sockets intended to receive flat (IC type) pins, they can end up forcing the socket contacts apart and making them a bit of a slack fit. The RAM / ROM adaptors should ideally have had thin, flat pins like a normal IC.

If your intention is always to use that ROM / RAM adaptor then you could try having your friendly expert remove the conventional CPU socket and fit a turned-pin socket on the mainboard as that would be a very snug fit indeed - almost too snug. Certainly wouldn't be a loose fit.

[John Earland]

Originally Posted by SiriusHardware

OK, well, turned (round) pins don't always get on well with conventional sockets intended to receive flat (IC type) pins, they can end up forcing the socket contacts apart and making them a bit of a slack fit. The RAM / ROM adaptors should ideally have had thin, flat pins like a normal IC.

If your intention is always to use that ROM / RAM adaptor then you could try having your friendly expert remove the conventional CPU socket and fit a turned-pin socket on the mainboard as that would be a very snug fit indeed - almost too snug. Certainly wouldn't be a loose fit.

That sounds like a good idea! I might have a go myself with my new found confidence! Just a thought though - I wonder if that would influence the disc drive no longer working?

[SiriusHardware]

The main IC involved with the disc drive, assuming it is connected via the IEEE-488 interface, is UC6. As all of those ICs are in more or less the same area, pressing on the RAM / ROM module might very well be causing pins on nearby UC6 to connect and disconnect. You're just going to have to try to narrow it down further with more stress-testing or make a decision about which to try first.

If you do decide to remove the conventional CPU socket yourself the suggested method is to cut the three or four plastic spars which tie the two sides of the socket together and then gently rock / wobble each side in turn from side to side, they will get easier and easier to move and then eventually the pins will all snap just above the top surface of the PCB.

Turn the PCB over and melt and remove each pin stump in turn with a fine pointed pair of metal tweezers, then clear the holes ready for the new socket.

Bear in mind that this socket has already been removed once, therefore the pads have already been subjected to some heat stress in the past.

[ajgriff]

I might be wrong or have misunderstood but I think the Tynemouth board has flat pins. See attached.

Alan

[SiriusHardware]

That's what I would have hoped, but John's statement seems counter to that. The frame in which those pins are held does look like the frame of a turned-pin socket and it's hard to see what shape those pins really are.

[John Earland]

Originally Posted by ajgriff

I might be wrong or have misunderstood but I think the Tynemouth board has flat pins. See attached.

Alan

No, the pins are definately round.

[SiriusHardware]

Here's another view for AJ.

[ortek_service]

It does look like tynemouth are keen on turned-pin sockets, as many on there, rather than using standard lower-cost ones.

But you do have to be very careful with the pins on turned-pin sockets (and many other round pin headers), as the pins are quite brittle and so can snap off it they get bent a bit (particularly when attempting to straighten them again).
Whereas the flat pins on most normal ones are a lot more forgiving.

And if one pin has broken off and stuck in the board's socket, then it could be making intermittent contact with the shoulder of the other half of it on the plug-in board assemblies.

I saw previous that it was mentioned that the main board socket was a 'turned-in' rather than flat type. So I wondered if this meant that these were actually those with the sloping sprung contact (rather than those with Y contact, covered mostly by top plastic)

[SiriusHardware]

This?

The sockets are conventional and the ROM/RAM board has the turned in pin types i.e. not flat

My interpretation of that was that all the sockets fitted on the mainboard are conventional types, and the ROM / RAM board has the 'turned pin types'. The extra 'in' just looks like a typo.

[ajgriff]

My mistake then, sorry. Hopefully you can see why I thought the pins might be flat from looking at the image I posted.

Alan

[John Earland]

Originally Posted by ajgriff

My mistake then, sorry. Hopefully you can see why I thought the pins might be flat from looking at the image I posted.

Alan

No worries. It's not obvious from the image at all but unfortunately (or fortunately) they are round.