Commodore PET 2001 voltage regulators - Page 14 - UK Vintage Radio Repair and Restoration Discussion Forum

ajgriff · 1st Nov 2021, 11:37 pm

Quote:

Originally Posted by SiriusHardware

My Bad, I jinxed it.

Actually I think there's a piece of code inside every PET which returns the statement: "Ha! Ha! Think you've fixed it? Well, we'll see about that!"

Alan

retromit · 2nd Nov 2021, 5:48 pm

Quote:

Originally Posted by SiriusHardware

My Bad, I jinxed it.

Before you get too mired in all the possibilities, just lift UC6 and make sure there aren't any pins which got folded over the last time it was inserted, in particular pins 9 and 17.

If they look OK, do this statement again:

Quote:

POKE 59426,255

And after doing that, measure the voltages on the following chip pins:-

UA8, pins

2
6
10
14

and UA7 pins

2
6
10
14

The voltages on all 8 pins should be about the same.

Are they?

I checked UC7 and UC6 and they both looked pin - no bent pins. I also swapped them over and issues the POKE/PEEKs in order to see if there was any difference; again, the same values were reported (zero and 127).

I then checked the voltages for UA8 and UA7 after issuing the POKE 59426,255:

UA8, pins

2 = 3.21V
6 = 3.18V
10 = 3.16V
14 = 3.2V

UA7 pins

2 = 3.22V
6 = 3.2V
10 = 3.18V
14 = 3.2V

So they all seem to be in the same range.

I'm just wondering with replacing the processor, and with the flexibility of the board, if anything has been damaged from this?

Tim.

SiriusHardware · 2nd Nov 2021, 6:19 pm

I honestly don't think so at this point. By all accounts every other part of the machine bar the shift key is working, so it is very unlikely that you have a processor fault or PCB fault in that area.

Next set of tests:

As before,

Quote:

POKE 59426,255

And this time measure the voltages on these four pins of UA7:-

9
11
15
13

This is to compare the operation of the MC3446 buffer on bit 6, where it appears to be working, and bit 7, where it apparently is not.

SiriusHardware · 2nd Nov 2021, 7:37 pm

Incidentally, I took one of the Hitachi HN462532Gs (bought from AJ's link earlier in the thread) to work, and it programs fine using the specific setting for those chips on the old Needhams EMP20 there. I should have access to that for the next few years at least so if anyone needs those ICs programmed, I'm happy to do that.

For anyone looking to try to program them as though they are 2732s by using an adaptor to rewire the pins to the pinout of a 2732, I noticed that the EMP20 used a slow-but-sure 50mS programming pulse, 25V VPP and 5V Vcc so that's what you need to aim for if using any other programmer.

retromit · 2nd Nov 2021, 10:24 pm

Quote:

Originally Posted by SiriusHardware

I honestly don't think so at this point. By all accounts every other part of the machine bar the shift key is working, so it is very unlikely that you have a processor fault or PCB fault in that area.

Next set of tests:

As before,

Quote:

POKE 59426,255

And this time measure the voltages on these four pins of UA7:-

9
11
15
13

This is to compare the operation of the MC3446 buffer on bit 6, where it appears to be working, and bit 7, where it apparently is not.

That’s no problem. I was just trying to rule out anything which had potentially changed.

Okay, the POKE has been run, and we get the following for UA7:

9 = 4.88V
11 = 4.88V
13 = 4.88V
15 = 0.13V

So looks like there is a discrepancy on pin 15 (assuming that all four should match as per the previous tests).

Tim.

SiriusHardware · 2nd Nov 2021, 11:24 pm

Yes, I'm afraid it looks like the 'bit 7 input buffer' inside UA7 has failed.

If you wanted to be doubly sure you could lift UC6, gently bend pin 9 out to the side so that it does not go into the socket when UC6 is reinserted, and repeat the previous test again. This would be to check that it is not a damaged input on UC6 pin 9 which is holding the output from UA7 pin 15 low.

Or, if you are confident enough to try to remove UA7 and UA8 intact without damaging the PCB, remove them, fit sockets and swap over UA7 and UA8. If you do that, you should find that the test program reports bit 3 faulty rather than bit 7, thus confirming that the IC originally fitted in UA7 position was faulty.

I have to tell you that so far, this problem has always been due to a failure in the MC3446 IC so the above checks are almost hardly worth doing.

These ICs (MC3446) are not run of the mill ICs but ScottishColin managed to buy replacements from Cricklewood Electronics who seemed generally rather good at finding parts needed for PETs - not too badly priced at the time, maybe they still have some.

The ones they did have had a slightly different part number (in terms of the letters after the number) to the ones you'll have in the machine but as Colin can hopefully confirm, they are 100% compatible and work fine as replacements.

If you do buy from there or some other source I suggest you buy at least two or even a full set of three because there is no telling how much longer they will be available for.

SiriusHardware · 2nd Nov 2021, 11:31 pm

Here are the ones from Cricklewood. They've gone up a bit so maybe only order one, just to fix the fault, and then possibly order spares later, when funds allow.

https://www.cricklewoodelectronics.com/MC3446AP.html

retromit · 3rd Nov 2021, 12:06 pm

Quote:

Originally Posted by SiriusHardware

Yes, I'm afraid it looks like the 'bit 7 input buffer' inside UA7 has failed.

If you wanted to be doubly sure you could lift UC6, gently bend pin 9 out to the side so that it does not go into the socket when UC6 is reinserted, and repeat the previous test again. This would be to check that it is not a damaged input on UC6 pin 9 which is holding the output from UA7 pin 15 low.

Or, if you are confident enough to try to remove UA7 and UA8 intact without damaging the PCB, remove them, fit sockets and swap over UA7 and UA8. If you do that, you should find that the test program reports bit 3 faulty rather than bit 7, thus confirming that the IC originally fitted in UA7 position was faulty.

I have to tell you that so far, this problem has always been due to a failure in the MC3446 IC so the above checks are almost hardly worth doing.

These ICs (MC3446) are not run of the mill ICs but ScottishColin managed to buy replacements from Cricklewood Electronics who seemed generally rather good at finding parts needed for PETs - not too badly priced at the time, maybe they still have some.

The ones they did have had a slightly different part number (in terms of the letters after the number) to the ones you'll have in the machine but as Colin can hopefully confirm, they are 100% compatible and work fine as replacements.

If you do buy from there or some other source I suggest you buy at least two or even a full set of three because there is no telling how much longer they will be available for.

I'm happy to try the work above, but If the other checks are hardly worth doing, then I can just order the ICs. I'll need to desolder one IC and solder in a new socket, but as the issue seems to point to this chip, then it may be easier to just buy them (and potential replacements).

Thanks for the link to Cricklewood Electronics. If these ICs are all okay, then I'll get the order in.

Thanks again,
Tim.

ajgriff · 3rd Nov 2021, 12:59 pm

I've not used my PET's IEEE port although it passes the simple peek/poke test and my ROM/RAM emulator's VOSSI tester doesn't detect any issues. However, in view of the apparent unreliability of increasingly scarce/expensive MC3446 chips I got to wondering about buying in a spare or two and came across these:

https://www.ebay.co.uk/itm/403185609...IAAOSwn~dhTWAS

https://www.ebay.co.uk/itm/403185577...wAAOSwqedhTVnQ

Without going into specifics there are several good reasons for not buying these and I'm NOT suggesting that Tim or anyone else should take a chance either. I might take a punt though but won't be testing them if they eventually arrive as I'm not going to unsolder apparently good chips for no valid reason.

Alan

SiriusHardware · 3rd Nov 2021, 1:06 pm

Well, maybe if you do order some you might consider passing them to ScottishColin or Tim for testing (if they are willing, of course), as they will both have at least one buffer socketed by that point. It would obviously be useful for you to know that your spares are viable before you ever get to the point where you need to use them.

The price is obviously good, but as usual for stuff from China there always has to be a healthy dose of scepticism and no great surprise if they are found not to work.

ScottishColin · 3rd Nov 2021, 1:49 pm

Here's what's in my UA7/UA8 for reference. Let me know if Cricklewood don't have any as I have one spare.

I only paid £4 - they seem to be £6.50 now....

Marked as MC3446AP in case the photo isn't clear.

Colin.

ScottishColin · 3rd Nov 2021, 1:51 pm

Happy to test - I've got two sockets available now....

Colin.

Quote:

Originally Posted by ajgriff

I've not used my PET's IEEE port although it passes the simple peek/poke test and my ROM/RAM emulator's VOSSI tester doesn't detect any issues. However, in view of the apparent unreliability of increasingly scarce/expensive MC3446 chips I got to wondering about buying in a spare or two and came across these:

https://www.ebay.co.uk/itm/403185609...IAAOSwn~dhTWAS

https://www.ebay.co.uk/itm/403185577...wAAOSwqedhTVnQ

Without going into specifics there are several good reasons for not buying these and I'm NOT suggesting that Tim or anyone else should take a chance either. I might take a punt though but won't be testing them if they eventually arrive as I'm not going to unsolder apparently good chips for no valid reason.

Alan

retromit · 3rd Nov 2021, 1:54 pm

Quote:

Originally Posted by ScottishColin

Here's what's in my UA7/UA8 for reference. Let me know if Cricklewood don't have any as I have one spare.

I only paid £4 - they seem to be £6.50 now....

Marked as MC3446AP in case the photo isn't clear.

Colin.

Thanks Colin. I've managed to get an order with Cricklewood for the ICs so hopefully they'll be here soon. Thanks again for the offer, though.

Like Colin, happy to test any ICs if they are needed.

Tim.

ajgriff · 3rd Nov 2021, 2:30 pm

My thanks to Colin and Tim for offering to test MC3446s. I've just come across a recent post on another forum by someone in the UK who bought the ICs (5 Pack) from China. Took a month to arrive but all worked perfectly apparently. Think I'll go for it. Will report back around Christmas time!

Alan

Slothie · 3rd Nov 2021, 2:43 pm

I've bought chips from china and generally got good chips. I think the problem comes mostly with expensive or popular chips. But caveat emptor, as they say.

ajgriff · 3rd Nov 2021, 4:05 pm

Ok I've placed the order. Tossed a coin and went for the ones ostensibly branded as Motorola.

Alan

SiriusHardware · 3rd Nov 2021, 4:28 pm

Good man, taking a bullet for the team...

ajgriff · 3rd Nov 2021, 6:15 pm

Hoping to dodge the bullet Sirius! There's always PayPal Protection for my £8.38 gamble if the worst happens.

I'm sure a Cricklewood MC3446 will bring Tim's machine back to full strength and complete another successful PET repair.

Alan

SiriusHardware · 3rd Nov 2021, 6:32 pm

Mysterious though, wasn't it, the way the IEEE buffers appeared to be working and only just this minute suddenly decided not to? As you've said yourself, never let a PET know that you think you have fixed it.

The best theory I can come up with is that the mere process of running the test, which excercises all the buffers in the chips, made a buffer which was about to fail anyway, fail. This was a discovery which needed to be made and fixed, though, before attempting to connect something like an SD2PET to the port.

Slothie · 3rd Nov 2021, 8:04 pm

It could have been as simple as a cracked bond wire or similar, that worked (just) until the machine was powered up for long enough to warm the chip. expand the die and open up the gap. Unfortunately equipment stored for a long while can often work for a while and then deterioration of many years, or moisture ingress can put the final straw on the camel's back, so to speak.

2nd Nov 2021, 7:37 pm	#264
SiriusHardware Dekatron Join Date: Aug 2011 Location: Newcastle, Tyne and Wear, UK. Posts: 11,560	Re: Commodore PET 2001 voltage regulators Incidentally, I took one of the Hitachi HN462532Gs (bought from AJ's link earlier in the thread) to work, and it programs fine using the specific setting for those chips on the old Needhams EMP20 there. I should have access to that for the next few years at least so if anyone needs those ICs programmed, I'm happy to do that. For anyone looking to try to program them as though they are 2732s by using an adaptor to rewire the pins to the pinout of a 2732, I noticed that the EMP20 used a slow-but-sure 50mS programming pulse, 25V VPP and 5V Vcc so that's what you need to aim for if using any other programmer.

2nd Nov 2021, 11:24 pm	#266
SiriusHardware Dekatron Join Date: Aug 2011 Location: Newcastle, Tyne and Wear, UK. Posts: 11,560	Re: Commodore PET 2001 voltage regulators Yes, I'm afraid it looks like the 'bit 7 input buffer' inside UA7 has failed. If you wanted to be doubly sure you could lift UC6, gently bend pin 9 out to the side so that it does not go into the socket when UC6 is reinserted, and repeat the previous test again. This would be to check that it is not a damaged input on UC6 pin 9 which is holding the output from UA7 pin 15 low. Or, if you are confident enough to try to remove UA7 and UA8 intact without damaging the PCB, remove them, fit sockets and swap over UA7 and UA8. If you do that, you should find that the test program reports bit 3 faulty rather than bit 7, thus confirming that the IC originally fitted in UA7 position was faulty. I have to tell you that so far, this problem has always been due to a failure in the MC3446 IC so the above checks are almost hardly worth doing. These ICs (MC3446) are not run of the mill ICs but ScottishColin managed to buy replacements from Cricklewood Electronics who seemed generally rather good at finding parts needed for PETs - not too badly priced at the time, maybe they still have some. The ones they did have had a slightly different part number (in terms of the letters after the number) to the ones you'll have in the machine but as Colin can hopefully confirm, they are 100% compatible and work fine as replacements. If you do buy from there or some other source I suggest you buy at least two or even a full set of three because there is no telling how much longer they will be available for.

2nd Nov 2021, 11:31 pm	#267
SiriusHardware Dekatron Join Date: Aug 2011 Location: Newcastle, Tyne and Wear, UK. Posts: 11,560	Re: Commodore PET 2001 voltage regulators Here are the ones from Cricklewood. They've gone up a bit so maybe only order one, just to fix the fault, and then possibly order spares later, when funds allow. https://www.cricklewoodelectronics.com/MC3446AP.html

3rd Nov 2021, 12:59 pm	#269
ajgriff Nonode Join Date: May 2004 Location: Halifax, West Yorkshire, UK. Posts: 2,587	Re: Commodore PET 2001 voltage regulators I've not used my PET's IEEE port although it passes the simple peek/poke test and my ROM/RAM emulator's VOSSI tester doesn't detect any issues. However, in view of the apparent unreliability of increasingly scarce/expensive MC3446 chips I got to wondering about buying in a spare or two and came across these: https://www.ebay.co.uk/itm/403185609...IAAOSwn~dhTWAS https://www.ebay.co.uk/itm/403185577...wAAOSwqedhTVnQ Without going into specifics there are several good reasons for not buying these and I'm NOT suggesting that Tim or anyone else should take a chance either. I might take a punt though but won't be testing them if they eventually arrive as I'm not going to unsolder apparently good chips for no valid reason. Alan

3rd Nov 2021, 1:06 pm	#270
SiriusHardware Dekatron Join Date: Aug 2011 Location: Newcastle, Tyne and Wear, UK. Posts: 11,560	Re: Commodore PET 2001 voltage regulators Well, maybe if you do order some you might consider passing them to ScottishColin or Tim for testing (if they are willing, of course), as they will both have at least one buffer socketed by that point. It would obviously be useful for you to know that your spares are viable before you ever get to the point where you need to use them. The price is obviously good, but as usual for stuff from China there always has to be a healthy dose of scepticism and no great surprise if they are found not to work.

3rd Nov 2021, 1:49 pm	#271
ScottishColin Octode Join Date: May 2012 Location: Perth, Scotland Posts: 1,807	Re: Commodore PET 2001 voltage regulators Here's what's in my UA7/UA8 for reference. Let me know if Cricklewood don't have any as I have one spare. I only paid £4 - they seem to be £6.50 now.... Marked as MC3446AP in case the photo isn't clear. Colin. Attached Thumbnails

3rd Nov 2021, 2:30 pm	#274
ajgriff Nonode Join Date: May 2004 Location: Halifax, West Yorkshire, UK. Posts: 2,587	Re: Commodore PET 2001 voltage regulators My thanks to Colin and Tim for offering to test MC3446s. I've just come across a recent post on another forum by someone in the UK who bought the ICs (5 Pack) from China. Took a month to arrive but all worked perfectly apparently. Think I'll go for it. Will report back around Christmas time! Alan

3rd Nov 2021, 2:43 pm	#275
Slothie Octode Join Date: Apr 2018 Location: Newbury, Berkshire, UK. Posts: 1,287	Re: Commodore PET 2001 voltage regulators I've bought chips from china and generally got good chips. I think the problem comes mostly with expensive or popular chips. But caveat emptor, as they say.

3rd Nov 2021, 4:05 pm	#276
ajgriff Nonode Join Date: May 2004 Location: Halifax, West Yorkshire, UK. Posts: 2,587	Re: Commodore PET 2001 voltage regulators Ok I've placed the order. Tossed a coin and went for the ones ostensibly branded as Motorola. Alan

3rd Nov 2021, 4:28 pm	#277
SiriusHardware Dekatron Join Date: Aug 2011 Location: Newcastle, Tyne and Wear, UK. Posts: 11,560	Re: Commodore PET 2001 voltage regulators Good man, taking a bullet for the team...

3rd Nov 2021, 6:15 pm	#278
ajgriff Nonode Join Date: May 2004 Location: Halifax, West Yorkshire, UK. Posts: 2,587	Re: Commodore PET 2001 voltage regulators Hoping to dodge the bullet Sirius! There's always PayPal Protection for my £8.38 gamble if the worst happens. I'm sure a Cricklewood MC3446 will bring Tim's machine back to full strength and complete another successful PET repair. Alan

3rd Nov 2021, 6:32 pm	#279
SiriusHardware Dekatron Join Date: Aug 2011 Location: Newcastle, Tyne and Wear, UK. Posts: 11,560	Re: Commodore PET 2001 voltage regulators Mysterious though, wasn't it, the way the IEEE buffers appeared to be working and only just this minute suddenly decided not to? As you've said yourself, never let a PET know that you think you have fixed it. The best theory I can come up with is that the mere process of running the test, which excercises all the buffers in the chips, made a buffer which was about to fail anyway, fail. This was a discovery which needed to be made and fixed, though, before attempting to connect something like an SD2PET to the port.

3rd Nov 2021, 8:04 pm	#280
Slothie Octode Join Date: Apr 2018 Location: Newbury, Berkshire, UK. Posts: 1,287	Re: Commodore PET 2001 voltage regulators It could have been as simple as a cracked bond wire or similar, that worked (just) until the machine was powered up for long enough to warm the chip. expand the die and open up the gap. Unfortunately equipment stored for a long while can often work for a while and then deterioration of many years, or moisture ingress can put the final straw on the camel's back, so to speak.