Commodore PET 4016

[ortek_service]

Originally Posted by ScottishColin

PETTESTER works as-is for 8K, 16K and 32K PETs so that's not it.

I'll get the keyboard in a better state first and see where we are with other suggestions.

No RAM ICs have been changed in the upper bank yet - I'd quite like to leave it 'broken' in order to carry on testing Julie's code.

re CAS0 - see attached with another CAS1 screenshot which looks better. Verdigris on the pins is the claim I'm making for why it's different today, but it's also likely to be operator error.

Colin.

That is strange then, if PETTESTER is working OK on your other PET('s) that have 32K RAM.

I was thinking of how many 4116's you'd already changed in the Lower 16K bank - around 5 or 6 ? And if so, then it may not be unreasonable to believe there could be 6 failed DRAM's in the Upper 16K (Even though that is quite unlucky, if all the voltage rails have always been correct).

Thanks for the updated Column Address Strobes waveforms - These do look rather-more like I'd expect memory-logic ones to look-like and would seem to confirm that everthing hardware-wise is accessing the (Both Lower & Upper 16K banks) DRAM's OK.

[TonyDuell]

Probably irrelevant but I restored a Philips P2000C which had been badly stored a few years back. This machine has 32K of RAM using 4116's on the terminal PCB and I ended up changing the full set. Perhaps one or two were good but most had failed. I think those were the only chips I had to replace, certainly the 64K (in 4164s) on the system PCB was good.

[ScottishColin]

It's a bit random and depending on so many things with my PETs.

I've had some that have had no RAM chips needing replaced, and some where several do.

It is what it is I guess. Six out of eight in low memory for now with this one.

It has been stored under cover outside for fifteen years.

Colin.

Originally Posted by TonyDuell

Probably irrelevant but I restored a Philips P2000C which had been badly stored a few years back. This machine has 32K of RAM using 4116's on the terminal PCB and I ended up changing the full set. Perhaps one or two were good but most had failed. I think those were the only chips I had to replace, certainly the 64K (in 4164s) on the system PCB was good.

[ScottishColin]

I did the PEEKing and POKEing suggested here:

http://blog.tynemouthsoftware.co.uk/2015/05/commodore-pet-4032-repair-only-16k.html

When I POKEd 16384 with 0, a PEEK returned 65

When I POKEd 16384 with 255, a PEEK returned 65

So BASIC can see the upper RAM but there are problems there.

Colin.

[ortek_service]

Well if you are really reading '65' back (assuming that wasn't a copy & paste etc. typo?) when trying to write 0 OR 255, then that would seem to indicate that all 8 DRAM's in the Upper-Bank are faulty!
- Although in different ways, as 65 = 01000001 in binary
So bit(D)0 & bit(D)6 stuck High, and all other bits stuck Low.

[ScottishColin]

Well it definitely was those numbers. However after a reboot I get

POKE 16384,0 - PEEK(16384) 65
POKE 16384,255, PEEK(16384) 113

Colin.

[ortek_service]

Originally Posted by ScottishColin

Well it definitely was those numbers. However after a reboot I get

Very strange....

Originally Posted by ScottishColin

POKE 16384,0 - PEEK(16384) 65

= Binary: 01000001 = D6 & D0 stuck High

Originally Posted by ScottishColin

POKE 16384,255, PEEK(16384) 113

= Binary: 01110001 = D7, D3, D2 & D1 stuck Low


Although the results are maybe looking a bit more sensible now, with two datalines stuck high, and four stuck low
- So there appears to be six faulty 4116's.

However, this is just at the first location 16384.

It may be useful to run a short program with a FOR-loop all the way to 32767, and see if you get a different result at other locations.

And maybe do a loop-test at a single location and see if you get consistent bits in error
- if data outputs from any are floating then you might get different result each time, if they are going into CMOS inputs.
(Although I presume these are going through a 74LS244 TTL-input buffer between their data outputs and the 6502 CPU?)

[TonyDuell]

Originally Posted by ortek_service

(Although I presume these are going through a 74LS244 TTL-input buffer between their data outputs and the 6502 CPU?)

I was surprised, but looking at the schematics, the RAM data lines go straight to the 6502 processor.

[ScottishColin]

Well that ties in with the Tynemouth diagnsotics errors.

Although we know from the past that the testing there isn;t as thorough as other testers and I'm thinking thatJulie's will be more thorough as well.

Colin.

Originally Posted by ortek_service

Very strange....

= Binary: 01000001 = D6 & D0 stuck High

= Binary: 01110001 = D7, D3, D2 & D1 stuck Low


Although the results are maybe looking a bit more sensible now, with two datalines stuck high, and four stuck low
- So there appears to be six faulty 4116's.

However, this is just at the first location 16384.

It may be useful to run a short program with a FOR-loop all the way to 32767, and see if you get a different result at other locations.

And maybe do a loop-test at a single location and see if you get consistent bits in error
- if data outputs from any are floating then you might get different result each time, if they are going into CMOS inputs.
(Although I presume these are going through a 74LS244 TTL-input buffer between their data outputs and the 6502 CPU?)

[ScottishColin]

In other news, I've spent a therapeutic few hours today stripping down the keyboard, cleaning the keytops in my ultra-sound cleaner, washing the plastic chassis, cleaning the keyboard PCB with IPA, testing and cleaning every single plunger and putting it all back together again.

All keys on the nice shiny keyboard now work.

Colin.

[ortek_service]

That's good to hear, with no special conductive paint / foil discs etc. required.
I've normally found with IR remotes, doing similar does normally get them going quite-well again, but some keys do sometimes eventually require a bit of extra help on the conductive-rubber part as tend to need more force / are a bit intermittent.

[ScottishColin]

While we're on hold, I've been checking out the cassette ports and neither work on the 4016.

Pin C (or 3 depending on which way up the mother board is) - the motor pin - on my 4032 is c. +6.5v when Play is pressed.

On the 4016, it's c. +8.9v at all times.

Is this potentially an issue with the Q2T2222 in UE16 (page 3 of 11 of the schematics)?

EDIT - Ground is good and I do have +5v on pin 2 on the 4016.

Colin.

[ScottishColin]

Next reading - Q1 (TIP29 NPN) is reading +9.29v on all three pins on the 4016.

That doesn't seem quite right.....

Colin.

[ScottishColin]

Comparint the voltage drops between Q1 pins between the working 4032 and the failing 4016, there's a big difference so I think Q1 has failed in the 4016. Q2 measures ok though.

TIP29s ordered.

Colin.

[ortek_service]

I thought you may have had to change TIP29's in other PET's before, so might have already had some spares of these - But maybe not enough if both ports have gone!

I wouldn't have expected TIP29 transistors to have failed with age - certainly not both - so I guess someone might have accidentally overloaded these in the past from a faulty Datasette / cable with a short.

Although less-likely to fail, it's probably not a bad idea to have some spares of the 2N?2222, if not got some already - maybe in the kit of assorted popular transistors I recall you'd got?

[SiriusHardware]

The Q2T2222 device Colin is referring to is actually an IC, at least in outline (it's an IC package) but it is really just a 'box' containing four ordinary 2N2222 transistors.

In the event of failure of one or more elements of one of these ICs it would in theory be possible to remove and replace them with individual 2N2222 transistors with their leads suitably cranked. Each of the individual transistors in the package can also be tested 'as' a 2N2222 transistor if there is a transistor tester available, and I think Colin has one.

One possible reason for the TIP29 to have the same voltage on every pin is if they are all shorted together, so it may be worth checking to see if that is the case (with the power off, of course).

I've seen past examples where the cassette tape interface has been intentionally 'disabled' by physically damaging these transistors, to what end I wouldn't know - maybe to prevent 'classroom' users from bringing in their own software or games to play on the machine.

[ScottishColin]

I'll check for any shorts.

In the meantime (as I suspect you may have suspected) I have bought a couple of TIP29s and a couple of 2222s.

I too thought I had replaced one before but I had a quick squizz at my 3016 and can't see any replacements in there.

Colin.

Originally Posted by SiriusHardware

The Q2T2222 device Colin is referring to is actually an IC, at least in outline (it's an IC package) but it is really just a 'box' containing four ordinary 2N2222 transistors.

In the event of failure of one or more elements of one of these ICs it would in theory be possible to remove and replace them with individual 2N2222 transistors with their leads suitably cranked. Each of the individual transistors in the package can also be tested 'as' a 2N2222 transistor if there is a transistor tester available, and I think Colin has one.

One possible reason for the TIP29 to have the same voltage on every pin is if they are all shorted together, so it may be worth checking to see if that is the case (with the power off, of course).

I've seen past examples where the cassette tape interface has been intentionally 'disabled' by physically damaging these transistors, to what end I wouldn't know - maybe to prevent 'classroom' users from bringing in their own software or games to play on the machine.

[ortek_service]

I presume that once the TIP29's are removed, they measure short across all pins, so were definitely faulty and wasn't something else on the board?

Plus you managed to find Q2T2222 'IC's', rather than discrete 2N2222's?
Surprising, the Q2T2222 (or at least the Eastern-Europe Bnnn clones) seem to still be fairly-available (as New Old Stock at least) from various eBay etc sellers for not too-much.

I'd somehow not encountered these Q2T2222's before (or other Q2Tnnnn versions TI also made based on other popular 2Nnnnn transistors).
I presume it is actually a true 'IC' with just one die internally, rather than being a module containing discrete transistors? So to replace only one, you'd have to isolate the 'IC's' pins, fit a 2N2222 (or a plastic TO-92 package version, instead of a TO-18 can) etc into the three holes for that particular transistor.

It does seem strange that Commodore used these, as I doubt they were cheaper than 4off 2N2222's (even if a bit easier to insert and solder into the PCB).
And they didn't really need the better-matching / thermal-tracking that you often got from having transistors in the same package (even though these weren't explicitly sold as Matched array like MATnn ones that were)

Originally Posted by ScottishColin

I'll check for any shorts.

In the meantime (as I suspect you may have suspected) I have bought a couple of TIP29s and a couple of 2222s.

I too thought I had replaced one before but I had a quick squizz at my 3016 and can't see any replacements in there.

Colin.

[SiriusHardware]

If one of the 'internal' 2N2222 transistors in the package proves to have failed and there is a desire to keep it looking original, the package could be removed, the pins attached to the 'dead' internal device clipped short so they don't quite reach the PCB, the package refitted / resoldered, and then a 2N2222 could be soldered to the dead section's pads underneath the PCB.

However, all this is a bit premature as we don't know that the original Q2T2222 has even wholly or partly failed.

[TonyDuell]

When I had a similar transistor array fail -- actually mine was 2 NPN transistors and 2 PNP ones, I soldered the equivalent discrete transistors to a header plug and put in a socket in place of the DIL original.