UK Vintage Radio Repair and Restoration Powered By Google Custom Search Vintage Radio and TV Service Data

Go Back   UK Vintage Radio Repair and Restoration Discussion Forum > Specific Vintage Equipment > Vintage Computers

Notices

Vintage Computers Any vintage computer systems, calculators, video games etc., but with an emphasis on 1980s and earlier equipment.

Closed Thread
 
Thread Tools
Old 24th Feb 2021, 12:14 am   #641
SiriusHardware
Dekatron
 
Join Date: Aug 2011
Location: Newcastle, Tyne and Wear, UK.
Posts: 11,482
Default Re: Non-working Commodore PET 3016

Assuming that goes OK, let's go on to a NOP test.

This is basically the procedure described in the document in your post #359, but in that version the CPU data pins are connected directly to 0V and 5V respectively. We don't like that because those pins are outputs as well as inputs and if they somehow go into output mode, they won't appreciate being welded to the supply lines.

So our recommendation (as per Mark) is that you make those connections from the data pins to 0V and 5V via resistors of perhaps 4K7 value rather than wires.

Physically this may involve stacking two or more sockets in order to allow you to make the necessary connections. Maybe you can do it with a conventional socket with all of its pins fitted but with the pins of the data pin receptacles bent out sideways, flat, so you can solder the resistors to those. If you do adopt this method bear in mind that the bent-out-flat pins with the resistors soldered to them must not make contact with the contacts in the socket below.

However you decide to do it, do not solder to the CPU pins because if you get any solder on the thin lower portion of the IC pins they may never fit into a turned-pin socket again.

Just to recap, the reason for doing this is to feed the CPU with a known harmless, predictable instruction to execute over and over again. Each time it reads a NOP instruction, it will advance the address by one step and then read the next instruction (which will also be a NOP) and so on. The end result is that the CPU runs smoothly through all the addresses trying to read from each one as it goes and selecting all the system devices one after the other in a nice predictable way, all of which can be observed on a scope. This is especially useful for spotting address line problems and address decoder problems but it should make it easier to observe data line activity as well.

Last edited by SiriusHardware; 24th Feb 2021 at 12:29 am.
SiriusHardware is online now  
Old 24th Feb 2021, 11:37 am   #642
ortek_service
Octode
 
ortek_service's Avatar
 
Join Date: May 2018
Location: Northampton, Northamptonshire, UK.
Posts: 1,394
Default Re: Non-working Commodore PET 3016

dave cox
Re: Non-working Commodore PET 3016
A little OT, but I managed to remove the top from a TMS25P32. It certainly looks like an EPROM inside.

Was it really cheaper to make with a lid than with a window or is it just an anti-recycling / anti re-flashing measure ?

SiriusHardware
Re: Non-working Commodore PET 3016
Maybe those windows were a bit more expensive than the plain lid (the material had to be UV-transparent, I guess, so possibly not ordinary glass).

Quote:
Originally Posted by dave cox View Post
Indeed, usually silica glass is used if you need UV transparent. I don't recall it being super expensive, unless you are comparing it to window glass!

dc

I thought they had to use Quartz, for UV transparency, and this did make it more expensive - like EPROM-Eraser tubes, that were like cheaper fluorescent ones without the coating and with special Mica? glass.

The ceramic package, needed for compitability with glass windows, also made these more expensive than plastic-package types of later OTP ones.
Although even if still in a ceramic package, but with non-transparent window, these OTP ones were probably a bit cheaper - the main reason why OTP types would have been used.
But maybe there hadn't been enough demand / alternative packaging process developed yet, for OTP versions of the 2732 so they just swapped the window-cover. The covers on this type does sometimes fall-off, with glue failure, and later ones tend to have a circular windowed embedded in the hole, flush with the surface.

It seems the very first non fusible-link 1601 etc. PROM's used the same silicon die as the slightly-later 1701 etc. EPROM's. But had a soldered-on metal lid, before they developed UV-transparent ones. Although there is rumour you might be able to erase the metal-lid ones with the right x-ray source (if it doesn't damage the IC)
ortek_service is offline  
Old 24th Feb 2021, 2:06 pm   #643
ScottishColin
Octode
 
Join Date: May 2012
Location: Perth, Scotland
Posts: 1,762
Default Re: Non-working Commodore PET 3016

Confirmed that they all have the same shape and do not overlap with each other.

Colin.


Quote:
Originally Posted by SiriusHardware View Post
They certainly look like chip select pulses so it would seem that even in its semi disabled state the system is managing to sweep through the address ranges of the PROMs. Just one other thing to try while you are in the area: Scope any two of the PROM pin 20s at the same time. You should see that they are selected at the same rate / frequency but at different times, so the chip select pulses will be staggered, one before the other, showing that only one PROM is selected at a time. If any chip select pulses are directly one above the other that would show two devices being selected at the same time (which should not happen).
ScottishColin is online now  
Old 24th Feb 2021, 2:19 pm   #644
ScottishColin
Octode
 
Join Date: May 2012
Location: Perth, Scotland
Posts: 1,762
Default Re: Non-working Commodore PET 3016

OK - 40 pin socket with pins 36-33 folded out sideways. Solder 4.7Ω resistors to each of pins 26-33.

Pins 29, 31 and 33 through individual 4.7Ω resistors to pin 1

Pins 26,27,28,30 and 32 through individual 4.7Ω resistors to pin 8

Then insert 6502 and check pin 9.

Have I got that right?

if so, how should I connect 5 individual resistors to pin 8? Just twist their ends together and insert 1 of the resistors to pin 8?

Colin.




Quote:
Originally Posted by SiriusHardware View Post
Assuming that goes OK, let's go on to a NOP test.

This is basically the procedure described in the document in your post #359, but in that version the CPU data pins are connected directly to 0V and 5V respectively. We don't like that because those pins are outputs as well as inputs and if they somehow go into output mode, they won't appreciate being welded to the supply lines.

So our recommendation (as per Mark) is that you make those connections from the data pins to 0V and 5V via resistors of perhaps 4K7 value rather than wires.

Physically this may involve stacking two or more sockets in order to allow you to make the necessary connections. Maybe you can do it with a conventional socket with all of its pins fitted but with the pins of the data pin receptacles bent out sideways, flat, so you can solder the resistors to those. If you do adopt this method bear in mind that the bent-out-flat pins with the resistors soldered to them must not make contact with the contacts in the socket below.

However you decide to do it, do not solder to the CPU pins because if you get any solder on the thin lower portion of the IC pins they may never fit into a turned-pin socket again.

Just to recap, the reason for doing this is to feed the CPU with a known harmless, predictable instruction to execute over and over again. Each time it reads a NOP instruction, it will advance the address by one step and then read the next instruction (which will also be a NOP) and so on. The end result is that the CPU runs smoothly through all the addresses trying to read from each one as it goes and selecting all the system devices one after the other in a nice predictable way, all of which can be observed on a scope. This is especially useful for spotting address line problems and address decoder problems but it should make it easier to observe data line activity as well.
ScottishColin is online now  
Old 24th Feb 2021, 2:30 pm   #645
ortek_service
Octode
 
ortek_service's Avatar
 
Join Date: May 2018
Location: Northampton, Northamptonshire, UK.
Posts: 1,394
Default Re: Non-working Commodore PET 3016

[QUOTE=ScottishColin;1345717]OK - 40 pin socket with pins 36-33 folded out sideways. Solder 4.7Ω resistors to each of pins 26-33.

Pins 29, 31 and 33 through individual 4.7Ω resistors to pin 1

Pins 26,27,28,30 and 32 through individual 4.7Ω resistors to pin 8

Then insert 6502 and check pin 9.

Have I got that right?

if so, how should I connect 5 individual resistors to pin 8? Just twist their ends together and insert 1 of the resistors to pin 8?

Colin.

/QUOTE]


Unless you've made some typos, I think they recommended 4k7 = 4.7kΩ resistors rather than 4.7Ω (which wouldn't really serve much purpose being that low).
Although this may not make too much difference, if just going to supply / ground and there's nothing driving to drive out from pins you're trying to pull up / down.
ortek_service is offline  
Old 24th Feb 2021, 2:42 pm   #646
ScottishColin
Octode
 
Join Date: May 2012
Location: Perth, Scotland
Posts: 1,762
Default Re: Non-working Commodore PET 3016

My bad. I meant 4.7kΩ

My question is really whether I can twist all the resistors together after soldering them to the pins to get them into pins 1 and 8?

Colin.


[QUOTE=ortek_service;1345726]
Quote:
Originally Posted by ScottishColin View Post
OK - 40 pin socket with pins 36-33 folded out sideways. Solder 4.7Ω resistors to each of pins 26-33.

Pins 29, 31 and 33 through individual 4.7Ω resistors to pin 1

Pins 26,27,28,30 and 32 through individual 4.7Ω resistors to pin 8

Then insert 6502 and check pin 9.

Have I got that right?

if so, how should I connect 5 individual resistors to pin 8? Just twist their ends together and insert 1 of the resistors to pin 8?

Colin.

/QUOTE]


Unless you've made some typos, I think they recommended 4k7 = 4.7kΩ resistors rather than 4.7Ω (which wouldn't really serve much purpose being that low).
Although this may not make too much difference, if just going to supply / ground and there's nothing driving to drive out from pins you're trying to pull up / down.
ScottishColin is online now  
Old 24th Feb 2021, 2:46 pm   #647
SiriusHardware
Dekatron
 
Join Date: Aug 2011
Location: Newcastle, Tyne and Wear, UK.
Posts: 11,482
Default Re: Non-working Commodore PET 3016

Twist their ends together, solder a wire to those ends and solder the other end of the wire to your +5V / 0V points (as applicable) on the UG4 pads.

4.7 ohms is quite a bit lower than we had in mind...

edit: Just saw your edit. 4K7 it is.
SiriusHardware is online now  
Old 24th Feb 2021, 4:38 pm   #648
ScottishColin
Octode
 
Join Date: May 2012
Location: Perth, Scotland
Posts: 1,762
Default Re: Non-working Commodore PET 3016

Done.

Not that I know what I'm looking at, but at least this is consistent on Pin 9. See attached file.

https://drive.google.com/file/d/18Fc...ew?usp=sharing

What further tests would be useful?

Colin.
ScottishColin is online now  
Old 24th Feb 2021, 5:26 pm   #649
SiriusHardware
Dekatron
 
Join Date: Aug 2011
Location: Newcastle, Tyne and Wear, UK.
Posts: 11,482
Default Re: Non-working Commodore PET 3016

Lovely. Now, look at the rest of the CPU address lines A1-A15 going upwards in order. I'm not going to do pin numbers (and mess them up) any more, you are obviously able to work these out for yourself. (Where there is scope for confusion I will still state pin numbers and hopefully get them right as well).

What you should see is the same kind of waveform but with the frequency halving / length of the high pulse / length of the low gap between pulses doubling each time you move to a higher address line. You'll probably have to wind down the horizontal speed each time you move up a couple of lines as the length of a pulse would soon be wider than your trace otherwise.

Once you've verified that all the CPU address lines have activity a quick verification check would be to measure the frequency on each of A0-A15 and you should find that the frequency halves each time you move up an address line. (Bear in mind that quirk of your frequency meter, where it needs a half-supply ground, if you try this).
SiriusHardware is online now  
Old 24th Feb 2021, 6:04 pm   #650
ScottishColin
Octode
 
Join Date: May 2012
Location: Perth, Scotland
Posts: 1,762
Default Re: Non-working Commodore PET 3016

I can see clean square waves on each pin from A0-A15.

The frequency halves every pin/address line as below.

Colin.

Code:
Address line         Pin          Frequency
A0                   9            250Khz
A1                   10           125Khz
A2                   11           62.5Khz
A3                   12           31.25Khz 
A4                   13           15.63Khz 
A5                   14           7.813Khz
A6                   15           3.096Khz
A7                   16           1.953Khz
A8                   17           976.6Hz
A9                   18           488.3Hz
A10                  19           244.3Hz
A11                  20           122.1Hz
A12                  22           61.05Hz
A13                  23           30.53Hz
A14                  24           15.26Hz
A15                  25           7.632Hz
ScottishColin is online now  
Old 24th Feb 2021, 6:22 pm   #651
Timbucus
Octode
 
Join Date: Mar 2019
Location: Barry, Vale of Glamorgan, Wales, UK.
Posts: 1,362
Default Re: Non-working Commodore PET 3016

That looks excellent the CPU is counting through the whole 64K address space with no issues.
Timbucus is offline  
Old 24th Feb 2021, 6:22 pm   #652
ScottishColin
Octode
 
Join Date: May 2012
Location: Perth, Scotland
Posts: 1,762
Default Re: Non-working Commodore PET 3016

I should have said (although I don't think it affects anything) that I got the frequency numbers from the scope.

Colin.


Quote:
Originally Posted by Timbucus View Post
That looks excellent the CPU is counting through the whole 64K address space with no issues.
ScottishColin is online now  
Old 24th Feb 2021, 6:26 pm   #653
Timbucus
Octode
 
Join Date: Mar 2019
Location: Barry, Vale of Glamorgan, Wales, UK.
Posts: 1,362
Default Re: Non-working Commodore PET 3016

The scope will produce an accurate reading with a nice clean square wave so you are correct it does not make a difference.

Quote:
Originally Posted by ScottishColin View Post
I should have said (although I don't think it affects anything) that I got the frequency numbers from the scope.

Colin.


Quote:
Originally Posted by Timbucus View Post
That looks excellent the CPU is counting through the whole 64K address space with no issues.
Timbucus is offline  
Old 24th Feb 2021, 6:50 pm   #654
Mark1960
Octode
 
Join Date: Mar 2020
Location: Kitchener, Ontario, Canada
Posts: 1,264
Default Re: Non-working Commodore PET 3016

Hi Colin, that confirms the NOP test is running.

I think next is to check the prom chip selects again. First on the fast sweep time you should see the low pulses closer together than the last time you looked at them. Then slow down the sweep time until the pulses compress together and you should see a burst of 4K pulses, a larger gap, then another burst in a repeating pattern. If you look at two prom chip selects on different channels of the scope they should not both be active at the same time. Check for this on every possible pair of chip selects to make sure the address decoding is correct.

Then for each prom, while monitoring the chip select with one probe, check each data line at the 6502 socket. While chip select is low, every data line should be a good logic level before the chip select returns to a high level.

It’s a lot to test, so take it in steps and post results as you go along and we’ll add comments if we see any sign of problems.
Mark1960 is online now  
Old 24th Feb 2021, 7:01 pm   #655
SiriusHardware
Dekatron
 
Join Date: Aug 2011
Location: Newcastle, Tyne and Wear, UK.
Posts: 11,482
Default Re: Non-working Commodore PET 3016

No problem using the scope to measure frequency (I had not realised it had that feature).

In addition to Marks's suggestion re: The chip selects, repeat the same address line process but this time looking at the buffered address lines BA0 to BA15 on the right hand side of UC3 / UB3.

Scope them to make sure the size and appearance is as expected and also that the frequencies mirror those of the CPU A0-A15 lines.

Last edited by SiriusHardware; 24th Feb 2021 at 7:25 pm.
SiriusHardware is online now  
Old 24th Feb 2021, 7:02 pm   #656
ScottishColin
Octode
 
Join Date: May 2012
Location: Perth, Scotland
Posts: 1,762
Default Re: Non-working Commodore PET 3016

Mark - I'm sorry but I'm not grasping this. I know the PROM select lines are pin 20 on UD6-9, but I'm lost with the sweep time instructions. I think you're referring to the 10x/1x switch on the scope - is that right?

Could you spell it out in easier words for me please?

Sorry and thanks.

Colin.


Quote:
Originally Posted by Mark1960 View Post
Hi Colin, that confirms the NOP test is running.

I think next is to check the prom chip selects again. First on the fast sweep time you should see the low pulses closer together than the last time you looked at them. Then slow down the sweep time until the pulses compress together and you should see a burst of 4K pulses, a larger gap, then another burst in a repeating pattern. If you look at two prom chip selects on different channels of the scope they should not both be active at the same time. Check for this on every possible pair of chip selects to make sure the address decoding is correct.

Then for each prom, while monitoring the chip select with one probe, check each data line at the 6502 socket. While chip select is low, every data line should be a good logic level before the chip select returns to a high level.

It’s a lot to test, so take it in steps and post results as you go along and we’ll add comments if we see any sign of problems.
ScottishColin is online now  
Old 24th Feb 2021, 7:08 pm   #657
ScottishColin
Octode
 
Join Date: May 2012
Location: Perth, Scotland
Posts: 1,762
Default Re: Non-working Commodore PET 3016

When you say right hand side, do you mean (for example) pin 18 on UC3 and all others on the right hand side of the attached schematic?

Thanks.

Colin.


Quote:
Originally Posted by SiriusHardware View Post
No problem using the scope to measure frequency (I had not realised it had that feature).

In addiction to Marks's suggestion re: The chip selects, repeat the same address line process but this time looking at the buffered address lines BA0 to BA15 on the right hand side of UC3 / UB3.

Scope them to make sure the size and appearance is as expected and also that the frequencies mirror those of the CPU A0-A15 lines.
Attached Thumbnails
Click image for larger version

Name:	Capture.JPG
Views:	41
Size:	40.7 KB
ID:	227521  
ScottishColin is online now  
Old 24th Feb 2021, 7:25 pm   #658
SiriusHardware
Dekatron
 
Join Date: Aug 2011
Location: Newcastle, Tyne and Wear, UK.
Posts: 11,482
Default Re: Non-working Commodore PET 3016

Yes, if you follow those lines from those ICs you will see they are labelled BA0 to BA15, 'BA' standing for 'Buffered Address'.

By 'Sweep Time' Mark means the time the scope trace takes to travel from left to right across the screen, controlled by the 'Time / Div' or Horizontal control on your scope's front panel.
SiriusHardware is online now  
Old 24th Feb 2021, 10:21 pm   #659
ScottishColin
Octode
 
Join Date: May 2012
Location: Perth, Scotland
Posts: 1,762
Default Re: Non-working Commodore PET 3016

I measured AB0-AB15 at the 6502 pins and at the right hand side of their destinations of UB3 and UC3.

Each one has exactly the same frequency from the 6502 to UB3/UC3. Each one has exactly the same wave form at the 6502 pins as at UB3 and UC3.

Colin.

Quote:
Originally Posted by SiriusHardware View Post
Yes, if you follow those lines from those ICs you will see they are labelled BA0 to BA15, 'BA' standing for 'Buffered Address'.

By 'Sweep Time' Mark means the time the scope trace takes to travel from left to right across the screen, controlled by the 'Time / Div' or Horizontal control on your scope's front panel.
ScottishColin is online now  
Old 24th Feb 2021, 10:33 pm   #660
ScottishColin
Octode
 
Join Date: May 2012
Location: Perth, Scotland
Posts: 1,762
Default Re: Non-working Commodore PET 3016

I've checked combinations of pin 20 UD6/7/8/9 and see that the waves are the same, and are smack next to each other but do not overlap.

I'm struggling a bit with getting the pulses compressed together though. Sorry, but I can't work out how to do this right now. I'll carry on trying different combinations on the scope.

Colin.


Quote:
Originally Posted by Mark1960 View Post
Hi Colin, that confirms the NOP test is running.

I think next is to check the prom chip selects again. First on the fast sweep time you should see the low pulses closer together than the last time you looked at them. Then slow down the sweep time until the pulses compress together and you should see a burst of 4K pulses, a larger gap, then another burst in a repeating pattern. If you look at two prom chip selects on different channels of the scope they should not both be active at the same time. Check for this on every possible pair of chip selects to make sure the address decoding is correct.

Then for each prom, while monitoring the chip select with one probe, check each data line at the 6502 socket. While chip select is low, every data line should be a good logic level before the chip select returns to a high level.

It’s a lot to test, so take it in steps and post results as you go along and we’ll add comments if we see any sign of problems.
ScottishColin is online now  
Closed Thread

Thread Tools



All times are GMT +1. The time now is 4:36 pm.


All information and advice on this forum is subject to the WARNING AND DISCLAIMER located at https://www.vintage-radio.net/rules.html.
Failure to heed this warning may result in death or serious injury to yourself and/or others.


Powered by vBulletin®
Copyright ©2000 - 2024, vBulletin Solutions, Inc.
Copyright ©2002 - 2023, Paul Stenning.