12th Jan 2021, 4:44 pm | #41 |
Octode
Join Date: Mar 2020
Location: Kitchener, Ontario, Canada
Posts: 1,294
|
Re: Commodore PET 2001
Is A12 staying high or low, or is it floating at some in between level?
If its staying low it might just be executing code from an even 4K block of memory. Staying high would be an odd 4K block. I’d expect A12 low to be a better sign as this might indicate access to zero page, stack in ram and an even block of rom, possibly something in the initialisation waiting for something to be ready. It might be better at this stage to check the outputs of the ‘154, to see which devices are being addressed. |
12th Jan 2021, 5:14 pm | #42 | |
Octode
Join Date: Apr 2018
Location: Newbury, Berkshire, UK.
Posts: 1,287
|
Re: Commodore PET 2001
Quote:
|
|
12th Jan 2021, 5:51 pm | #43 | |
Nonode
Join Date: May 2004
Location: Halifax, West Yorkshire, UK.
Posts: 2,587
|
Re: Commodore PET 2001
Quote:
Thanks also to Mark for his input. A12 is low but all the other address pins are showing healthy looking Hi-Lo activity. I'll certainly have a look at the decoder's outputs. Alan |
|
12th Jan 2021, 9:47 pm | #44 |
Nonode
Join Date: May 2004
Location: Halifax, West Yorkshire, UK.
Posts: 2,587
|
Re: Commodore PET 2001
Have been doing a bit more probing including on the A12 line and embarassingly discovered that when I turned up the 'scope's screen intensity a little, faint signs of pulsing emerged. I've been using a fairly basic Hameg 203-5 (20MHz) so thought I'd have a closer look using my Tek 100MHz machine. This revealed that there is a fast positve pulse (about 4V) on A12 repeated roughly every 10µs. Each pulse only lasts about 1µs. On the other address lines the pulse length is variable but longer. I'm hoping this means that the CPU is likely to be ok. Any thoughts?
Alan PS I guess this tale of woe demonstrates that not all analogue oscilloscopes are equal. |
12th Jan 2021, 10:13 pm | #45 |
Octode
Join Date: Apr 2018
Location: Newbury, Berkshire, UK.
Posts: 1,287
|
Re: Commodore PET 2001
Read / write pulses will be about 1uS. Check the 16 outputs of the 74154 to see what pages are being accessed...
|
25th Jan 2021, 6:49 pm | #46 |
Nonode
Join Date: May 2004
Location: Halifax, West Yorkshire, UK.
Posts: 2,587
|
Re: Commodore PET 2001
Took a break from the PET but have now looked at the activity on the 74154 decoder's output pins although I started by looking again at the outputs of the 6502 which go solely to the 74154. These are inputs as far as the 74154 is concerned and are designated A, B, C & D. The attached photos show the signals on these pins where the first image is A and the second is B, C & D (they're all much the same). The vertical axis was set at 0.1V whilst using a x10 probe. The timebase was set at 2µs. I should mention that adjacent pulses were just off the screen in the first shot so the interval between them was about 10µs. The other thing I noticed was that the pulse length of A roughly doubled as the ICs warmed up. I confess that I don't understand the significance, if any, of these things.
Turning to the decoder's outputs, those designated 2 through to 12 plus 14 and 15 showed a simple continuous ≈5V output with just the usual hash superimposed. Ouputs 0, 1 & 13 exhibited waveforms similar to those at inputs B, C & D (see second photo) but with a greater amplitude. I'm hoping that someone has the patience to study these findings and is able to offer a fog clearing interpretation. Alan |
25th Jan 2021, 7:46 pm | #47 |
Octode
Join Date: Mar 2020
Location: Kitchener, Ontario, Canada
Posts: 1,294
|
Re: Commodore PET 2001
I think it’s odd that the pulse length on A is changing. Try measuring the clock on the 6502 at pin 37, 39 and 3, just to see if they are also changing.
It looks to me that the 6502 is running, but maybe not running the code it should. Take a look at the data lines on the 6502, pins 26 through 33, to see if they are being driven to healthy high and low levels. Also worth doing the same for the address lines, 6502 pins 9 through 20, and 21 through 25, using the better scope. Do you have an eprom programmer? Unless anyone else has any better ideas, it may be worth trying to verify the ROMs, though its probably going to mean getting or making an adapter. |
25th Jan 2021, 8:42 pm | #48 |
Nonode
Join Date: May 2004
Location: Halifax, West Yorkshire, UK.
Posts: 2,587
|
Re: Commodore PET 2001
Thank you again for your input Mark. I can confirm that all the data and address lines (except A12) showed good solid signals with both 'scopes. Have also checked the clock a good few times and it seems to be running at a consistent 1MHz.
No eprom programmer I'm afraid and I don't currently have enough knowledge to use one. Can always learn of course. Still considering the possibility of a bitfixer ROMulator. Total cost to import would be around £50 from your homeland I believe. Still haven't fully investigated the possibility of poor pin contacts somewhere or failed RAM chips. Alan |
25th Jan 2021, 10:54 pm | #49 |
Nonode
Join Date: May 2004
Location: Halifax, West Yorkshire, UK.
Posts: 2,587
|
Re: Commodore PET 2001
This is probably going to sound very simplistic but I'm trying to understand why the 6502's A12 pulse could be so short lived when compared with the signals generated by the other address lines. Working on the assumption (true or otherwise) that this implies a fault condition I'm trying to figure where the fault lies. Since A12 is only connected to the A input of the 74145 decoder the obvious conclusion is that either the 6502 or the 74145 is causing the problem. However, is there something else (RAM or ROM perhaps) connected to the 6502 that could cause such a short pulse? Seems unlikely to me as all the other pin activity appears to make sense but then again I'm fishing in the dark!
Alan |
25th Jan 2021, 11:50 pm | #50 | |
Octode
Join Date: Mar 2020
Location: Kitchener, Ontario, Canada
Posts: 1,294
|
Re: Commodore PET 2001
Quote:
Looking at Tindie the Omni romulator is available from some guy in Thames valley. I have no idea if these are a potential solution. That short pulse is probably from a memory access. At 1mhz the 6502 clock will be approx 500ns high and 500ns low. The 6502 only accesses memory during one phase of the clock, so a 500ns pulse is not unexpected. If it was fetching code from an address with A12 high then I’d expect it to be high for more than 500ns, but if its just reading or writing to memory or an input or output port, then you might see those 500ns pulses. I wonder if its polling an input port, waiting for a ready signal of some kind. Can you see any 500ns low pulses on any of the odd outputs of the 74154? Pins 2, 4, 6, 8, 10, 13, 15, 17. You might need to change the trigger settings on your scope to see the low pulses, or increase the sweep time. |
|
26th Jan 2021, 11:36 am | #51 |
Octode
Join Date: Apr 2018
Location: Newbury, Berkshire, UK.
Posts: 1,287
|
Re: Commodore PET 2001
If its a I/O port they are all in $E8xx, (1110 1000 xxxx xxxx) so A12 would be low. So its either accesses to RAM >4k or ROM at $Bxxx (Expansion rom, not usually installed) $Dxxx (Basic) or $Fxxx (Kernal).
I would guess the last 2 - perhaps you could check for activity on 74154 pins 15 & 17 at the same time as A12? This would verify the 74154. |
26th Jan 2021, 11:54 am | #52 |
Nonode
Join Date: May 2004
Location: Halifax, West Yorkshire, UK.
Posts: 2,587
|
Re: Commodore PET 2001
Working from my notes of yesterday pin 15 (output 13D) has square wave type activity but pin 17 (output 15F) is simply a continuous high. A12 consistently displays the trace shown in my previous post. Does that help?
Alan |
26th Jan 2021, 12:30 pm | #53 | |
Octode
Join Date: Apr 2018
Location: Newbury, Berkshire, UK.
Posts: 1,287
|
Re: Commodore PET 2001
Quote:
If you monitor pin 17 while you reset the computer does it show activity? the first thing the processor is supposed to do is access the reset vector at $FFFC & $FFCD which sends it to $FD38 ("Old" v1 Roms, somewhere else for "New" ROMS) for the reset & initialisation. I suppose its possible the Kernal ROM is corrupt and its vector at $FFFC is sending the processor to some random place in the BASIC ROM where it loops. Given the screen isn't clearing (which I think is fairly early in the reset sequence) then this scenario would seem more likely, in addition to the fact that the interrupts do not seem to have been enabled. |
|
26th Jan 2021, 12:37 pm | #54 |
Octode
Join Date: Apr 2018
Location: Newbury, Berkshire, UK.
Posts: 1,287
|
Re: Commodore PET 2001
Another thought - given the machine has never worked, have you checked the ROMS are in the correct sockets? If the previous owner had been dabbling with it they may have refitted them in the wrong place...... The ROMS all have a serial number on them, and you could cross-reference them with the archive at http://www.zimmers.net/anonftp/pub/c...pet/index.html - the ROMS should have a serial number of the form 9014xx and the archive shows which address they are supposed to be in.
|
26th Jan 2021, 5:58 pm | #55 | |||
Nonode
Join Date: May 2004
Location: Halifax, West Yorkshire, UK.
Posts: 2,587
|
Re: Commodore PET 2001
Quote:
Quote:
My impression from the chap I bought the PET from all those years ago was that he'd bought it new and that it had failed after a couple of years so he just put it to one side until his wife insisted that he got rid of it. She certainly had a big smile on her face when I carried it away. My view, then and now, is that I'm the first person to have opened the case since it left the factory. Quote:
Although I have so far removed the CPU and the four essential RAMs for contact cleaning I'm wary of completely removing more ICs because of the known unreliability of Commodore's sockets. As a next step I'm inclined to think in terms of partially lifting the ROMS and soaking the contacts with a little IPA before re-inserting fully. Any further suggestions/analyses most welcome as always. The battle continues! Alan |
|||
26th Jan 2021, 6:33 pm | #56 |
Octode
Join Date: Apr 2018
Location: Newbury, Berkshire, UK.
Posts: 1,287
|
Re: Commodore PET 2001
The Kernal roms are UH4 (F000-F7FF 901447-06) and UH7 (F800-FFFF 901447-07)
|
26th Jan 2021, 6:39 pm | #57 |
Octode
Join Date: Apr 2018
Location: Newbury, Berkshire, UK.
Posts: 1,287
|
Re: Commodore PET 2001
Many of the retro-computing people on Youtube recommend "De-oxit" for contact cleaning with its seemily magical properties for healing sockets and switches. Its a tad pricey (about £14 a tin I think).
|
26th Jan 2021, 6:40 pm | #58 |
Nonode
Join Date: May 2004
Location: Halifax, West Yorkshire, UK.
Posts: 2,587
|
Re: Commodore PET 2001
Got them located, thank you. Took me a while to latch on to Commodore's grid referencing system when I first set out on this adventure.
Alan * Post #56 refers Last edited by ajgriff; 26th Jan 2021 at 6:59 pm. Reason: Post reference added |
27th Jan 2021, 1:01 pm | #59 | |
Nonode
Join Date: May 2004
Location: Halifax, West Yorkshire, UK.
Posts: 2,587
|
Re: Commodore PET 2001
Quote:
Alan |
|
27th Jan 2021, 1:57 pm | #60 |
Dekatron
Join Date: Aug 2011
Location: Newcastle, Tyne and Wear, UK.
Posts: 11,556
|
Re: Commodore PET 2001
a.j., do you have kind of EPROM reader? Even if you have and it doesn't support / feature the devices used in the PET, you can make an adaptor to make them 'look' like a more generic EPROM device in order to read them out.
The state of your PROMs will always be a possible area for doubt unless you can verify them somehow. It may also be an idea to undertake the (admittedly laborious) job of checking that every pin of every socketed IC has continuity to the points it is supposed to be connected to, given the notoriety of the 'bad' sockets used in these. |