PDA

View Full Version : PET got seriously ill...



track18
March 15th, 2009, 06:59 AM
Hi there,

recently my CBM 3032, which used to work fine previously, all over sudden ceased to function and started exhibiting the dreaded "garbled screen" symptom... I was quite sad when this happened since I have so many happy memories about this kind of machine -- after all, it was the second computer type (and the first one featuring a CRT and an alphanumeric keyboard) I learned working with back during my school days.

However, reading about several PET ressurections described in this forum raised my hope that also this baby might get revived to its fullest. Alas, though knowing some *very* basics about electronics and digital circuitry, I definitely will depend on the advice of people being more educated in this kind of matter.

So, here's what I've found out so far:


- Powering on: Screen filled with garbled characters.


- Switching off and back on very quickly (approx. < 0.5 s): garbled characters, then (after 1 s or so) blank screen. It seems that in this scenario the boot sequence at least passes the "clear screen" routine. So the $F000 ROM seems principally ok, since this is where the CPU fetches the RESET vector from ($FFFC/$FFFD), and it seems to point to something reasonable. The "clear screen" routine itself seems to be located in the $E000 ROM -- at least this is where I found something that looked pretty much like some code for filling the video RAM ($8000 - $83E7) with space character screen code ($20) when I scanned the corresponding VICE emulator ROM image for the LDA #32 byte sequence (A9 20):



e246 A9 20 LDA #$20
e248 9D 00 80 STA $8000,X
e24b 9D 00 81 STA $8100,X
e24e 9D 00 82 STA $8200,X
e251 9D 00 83 STA $8300,X
e254 CA DEX
e255 D0 F1 BNE $E248


So one might assume that the $E000 ROM is also in good shape. No assumptions can yet be made about the other two ROMs, though.


- Re-seating the ROMs didn't improve the PET's condition; so far I haven't dared to re-seat the CPU and the I/O chips since I am not too dexterous when it comes to pulling DIPs out of sockets without bending the pins... Unfortunately neither the video, nor the "normal" RAMs are socketed.


- I purchased a logic probe and had a look at the voltage levels at all the CPU's pins. During a normal powering on, I get the following



Pin Pin
VSS 21 L ?PH 20 A11
A12 22 ?PH ?PH 19 A10
A13 23 ?PH ?PH 18 A9
A14 24 ?PH ?PH 17 A8
A15 25 ?PH ?PH 16 A7
D7 26 ?PH ?PH 15 A6
D6 27 ?PH ?PH 14 A5
D5 28 ?PH ?PH 13 A4
D4 29 ?PL ?PH 12 A3
D3 30 ?PL ?PH 11 A2
D2 31 ?PH ?PH 10 A1
D1 32 ?PH ?PH 09 A0
D0 33 ?PL H 08 VCC
R/W' 34 HPH LPL 07 SYN
N.C. 35 - H 06 NMI'
N.C. 36 - - 05 N.C
Ph0 37 P H 04 IRQ'
SOV 38 H P 03 Ph1
Ph2 39 P H 02 RDY
RES' 40 LH L 01 VSS


The letter codes next to the pin numbers describe the sequence of observed signals during start-up. The first letter stands for the first second after power-on, the second letter for the second second (<= no typo ;-), and the last letter tells the signal level which then lasts on. L and H stand for 'low' and 'high', respectively, and 'P' means 'pulsating', i.e. the probe continuously detects L/H transitions. '?' just means that different address and data lines exhibit very different behaviour during the first second; some of them give a H, some a L, and some others something inbetween.

Remark: According to the clock pins (PhX), there is a continuous heartbeat. However, the SYN pin stops emitting pulses after the second second, indicating that the CPU ceased to execute commands. This is in accorance with the observation that both the address and the data bus stop talking to the system at the same time point.

Remark: The last thing the CPU did before halting seems to read (R/W' is H) %11100110 = $E6 from address $FFFF. That's in so far very nice since this is exactly the value that should reside at this address according to the VICE ROM image (and which is the hi-byte of the IRQ vector $E61B). However, what puzzles me is that the IRQ line remains high all the time -- shouldn't there be like 60 incomming interrupt requests per second?


- Since switching the power on and off too often too quickly is supposed to be unhealthy for the CRT circuitry, I decided to install a reset button which would connect GND (from pin 1 of the user port) to capacitor C68 of the power-up reset circuit (well, that's what I was told). The thingy works fine, I see CPU's RES line going L for about one second when pushing the button, and I get the blank "powering off and on quickly" screen.

Here's the what I normally find at the CPU after doing such a reset:



Pin Pin
VSS 21 L P 20 A11
A12 22 P P 19 A10
A13 23 P P 18 A9
A14 24 P P 17 A8
A15 25 P P 16 A7
D7 26 P P 15 A6
D6 27 P P 14 A5
D5 28 P P 13 A4
D4 29 P P 12 A3
D3 30 P P 11 A2
D2 31 P P 10 A1
D1 32 P P 09 A0
D0 33 P H 08 VCC
R/W' 34 HP P 07 SYNC
N.C. 35 - H 06 NMI'
N.C. 36 - - 05 N.C
Ph0 37 P ? 04 IRQ'
SOV 38 H P 03 Ph1
Ph2 39 P H 02 RDY
RES' 40 LH L 01 VSS


Remark: According to the activity found on the R/W', address and data lines, the CPU *does* do something. The question remains: what? And again there seem to be no interrupt requests -- the IRQ line remains constantly either H or L.

Remark: Occasionally, after hitting reset the CPU seems to get stuck in a state in which it doesn't talk on to the bus anymore. I missed the opportunity to check all data and address lines, but I suspect it might be the same jam state the CPU's gets stuck in after a cold power-on.


- Months ago I connected a small audio amplifyer to the user port M pin (CB2) for audio output. When pulling up the volume, I hear some interference from the PET's motherboard as a humming or hissing sound of different tones. For instance, after powering on, I first hear the typical 50 Hz of our national AC grid, during the second second there's a kind of beep, and afterwards it's the 50 Hz humming again. When hitting reset, after a second I frequently also hear the beeping "power-on" sound, followed by a lasting hissing which I suppose to be generated by the CPU busily accessing the bus. (Sometimes, however, there's barely any sound other than the 50 Hz, even though the bus lines are busy.) Funny enough, I do not always get the "power-on beep" during reset; sometimes I get a more noisy sound, sometimes just a clicking noise.


- *Very* occasionally (I thing it has happened only twice or thrice) I get another funny screen after quickly re-powering the PET or after hitting reset. It is mainly blank and shows some intermittent lines of garbage, including the words "IGNORED" and "REDO FROM START". Don't remember to ever have them encountered during my school days, though... ;-) [Don't know yet how to show images in-line, but I attached a "screen shot".]



So - for any hints where to go next I'd be really grateful...

per
March 15th, 2009, 07:33 AM
- *Very* occasionally (I thing it has happened only twice or thrice) I get another funny screen after quickly re-powering the PET or after hitting reset. It is mainly blank and shows some intermittent lines of garbage, including the words "IGNORED" and "REDO FROM START". Don't remember to ever have them encountered during my school days, though... ;-) [Don't know yet how to show images in-line, but I attached a "screen shot".]



So - for any hints where to go next I'd be really grateful...

I don't know too mutch about the PET, but this sounds like some stuff in ROM somehow get transfered to video-memory. I don't know how this may happen, but if you dump the ROMs, I'm sure there is a similar string in one of them.

track18
March 15th, 2009, 08:42 AM
Per,


I don't know too mutch about the PET, but this sounds like some stuff in ROM somehow get transfered to video-memory. I don't know how this may happen, but if you dump the ROMs, I'm sure there is a similar string in one of them.

not quite... At least when hunting the VICE ROMs for, say, 09 07 0d (which would be the screen codes for "IGN") there is no hit. On the other hand, the corresponding ASCII (or rather: PETSCII ;-) sequence 49 47 4e *is* present in the ROM at $CC03. In fact, memory around $CC03 looks like



cbc0 a2 2a c8 b1 77 f0 6d c8 b1 77 85 3c c8 b1 77 c8 .*..w.m..w.<..w.
cbd0 85 3d b1 77 aa 20 03 c8 e0 83 d0 dd 4c 52 cb a5 .=.w. ......LR..
cbe0 40 a4 41 a6 0b 10 03 4c 3a c7 a0 00 b1 40 f0 0b @.A....L:....@..
cbf0 a5 0e d0 07 a9 fc a0 cb 4c 1c ca 60 3f 45 58 54 ........L..`?EXT
cc00 52 41 20 49 47 4e 4f 52 45 44 0d 0a 00 3f 52 45 RA IGNORED...?RE
cc10 44 4f 20 46 52 4f 4d 20 53 54 41 52 54 0d 0a 00 DO FROM START...
cc20 d0 04 a0 00 f0 03 20 6d cf 85 46 84 47 20 aa c2 ...... m..F.G ..
cc30 f0 04 a2 00 f0 66 9a 8a 18 69 04 48 69 06 85 21 .....f...i.Hi..!
cc40 68 a0 01 20 ae da ba bd 09 01 85 63 a5 46 a4 47 h.. .......c.F.G


All in all, this seems to indicate that it's not just ROM somehow miraculously mirrored into video RAM, but some ROM routine for printing out a text string (thereby converting PETSCII to screen code) does get executed by the CPU.

BTW, I just googled for the meaning of those messages -- the ?EXTRA IGNORED error occured when entering too many (comma-separated, IIRC) items when a BASIC program was waiting for INPUT, and a ?REDO FROM START message was issued when a string value was entered where a numeric value was expected.

Regards --

track18

MikeS
March 15th, 2009, 09:07 AM
That sure is a detailed diagnosis!

Can't help much, but it's always a good idea to check and clean the power connector pins (and the voltages of course); also, you don't have to remove the chips (and risk bending the pins) to re-seat them - just lift them a bit and push them back down, with a little sideways pressure lengthwise.

Unfortunately problems with 30xx's aren't usually that simple...

Good luck!

cosam
March 15th, 2009, 09:07 AM
Wow, quite a story! Before we start analysing the symptoms, though: Do you have the means to check the power supply? Any chips (especially RAM, ROM and the "big four" at the top) running hot? Or not warm at all?

I'd definitely try reseating the larger chips. You can even leave the PIAs out and replace them one at a time to see if either of those is playing up.

track18
March 15th, 2009, 10:02 AM
Hey,

that's amazing -- so many kind replies so shortly after my initial posting!


Wow, quite a story! Before we start analysing the symptoms, though: Do you have the means to check the power supply?

Hm, I don't have an oscilloscope, but my cheap multimeter should do for checking the power supply, shouldn't it? (I haven't searched the web too thoroughly for PET power supply specifications yet, and so far haven't found any which I were able to interpret in a way to give me any clue regaring which voltage should appear at which pin of J8... Any concrete values I should check for?)


Any chips (especially RAM, ROM and the "big four" at the top) running hot? Or not warm at all?

I checked that the other day -- the CPU does get a little warm, the I/O chips stay almost cool, and so do the RAM chips. In fact, the only things that do get significantly hot are the four ROMs -- which is barely surprising if my hypothesis is correct that the CPU does continuously execute some ROM code, is it?


I'd definitely try reseating the larger chips. You can even leave the PIAs out and replace them one at a time to see if either of those is playing up.

<sigh/>I'll do it... hoping my fingers don't get too sweaty... ;-) I guess I'll first try to...


...just lift them a bit and push them back down, with a little sideways pressure lengthwise.

It might take me a day or two, though, to follow all your generous tips since apart from the PET I have a couple of other babies to attend to -- two of them even humanoid and just 8 weeks old! :-)

Regards

track18

MikeS
March 15th, 2009, 10:44 AM
Just put the meter on the 3 test points (12V + 2x5V) or one of the related chips, and wiggle the power connector a bit to make sure there's no noise or intermittent. This was a common problem with the original PETs (as were bad RAM & ROM chips), but not so much with the later dynamic memory boards. Still, never hurts to check.

cosam
March 15th, 2009, 11:08 AM
Hm, I don't have an oscilloscope, but my cheap multimeter should do for checking the power supply, shouldn't it? (I haven't searched the web too thoroughly for PET power supply specifications yet, and so far haven't found any which I were able to interpret in a way to give me any clue regaring which voltage should appear at which pin of J8... Any concrete values I should check for?)
Don't worry about measuring at J8 - that is raw power which hasn't been through the rectifiers/regulars yet. As MikeS pointed out, you want to measure at the test points. I take it you have the schematics (http://www.zimmers.net/anonftp/pub/cbm/schematics/computers/pet/2001N/index.html)? If your meter has an AC setting, try measuring with that too. The reading will give you an idea of any ripple riding on the DC voltage.

As for popping the big chips: if you have a long, thin screwdriver, you can get wiggle the back end of them out by coming in from the holes for rear ports. As long as you take your time and alternate between ends, you should be fine.

tezza
March 15th, 2009, 11:38 AM
Sorry to hear about your PET. It's always a blow to turn on a beloved machine only to find it suddenly doesn't work (I've just had this experience with my IBM 5051).

The 5051, with it's built in POST messages, seems a little easier to diagnose though.

One other thing to try would be piggybacking an equivalent RAM chip over the existing RAM one at a time and see if that changes anything. Sometimes (not always) this can show up a faulty RAM chip. I diagnosed a faulty Osborne 1 by doing this

If you do this make sure you have the IC in the correct orientation.

Tez

Bungo Pony
March 15th, 2009, 03:42 PM
The only thing I could suggest is making backups of ANY programmable chips inside these old computers. When I was working in the electronics industry, we had Eproms and EEproms go goofy after being in use for a decade or more. Re-programming the chips usually fixed the problem.

MikeS
March 15th, 2009, 10:06 PM
The only thing I could suggest is making backups of ANY programmable chips inside these old computers. When I was working in thek18 electronics industry, we had Eproms and EEproms go goofy after being in use for a decade or more. Re-programming the chips usually fixed the problem.
Commodore stuff is pretty well covered in that department, at least the standard ROMs; they're all archived and on the WWW. But there were quite a few third-party ROMS, and I don't think they are all archived; anything in yours, track18, that didn't come from the factory?

track18
March 16th, 2009, 10:50 AM
Hi everyone,

so... I did it -- re-seated the "big four", but no difference in behaviour. (Well, perhaps I should be grateful things didn't get worse...)


Just put the meter on the 3 test points (12V + 2x5V) or one of the related chips, and wiggle the power connector a bit to make sure there's no noise or intermittent. This was a common problem with the original PETs (as were bad RAM & ROM chips), but not so much with the later dynamic memory boards. Still, never hurts to check.

I suppose mine is a dynamic mem board -- all the CBM 30xx had one, didn't they? But anyway...


Don't worry about measuring at J8 - that is raw power which hasn't been through the rectifiers/regulars yet. As MikeS pointed out, you want to measure at the test points. I take it you have the schematics (http://www.zimmers.net/anonftp/pub/cbm/schematics/computers/pet/2001N/index.html)? If your meter has an AC setting, try measuring with that too. The reading will give you an idea of any ripple riding on the DC voltage.

...next challange will be to check the power supply. (Yes, I've seen those schematics before -- but could anybody please give me a clue where to look for those test points?) Just wondering if my meter's 200 V AC setting will be capable of detecting tiny voltage ripples...


Sorry to hear about your PET. It's always a blow to turn on a beloved machine only to find it suddenly doesn't work (I've just had this experience with my IBM 5051).

Thanks for the sympathy -- and all the best for your little big blue precious...


The 5051, with it's built in POST messages, seems a little easier to diagnose though.

*g* Somewhere I read the garbled screen actually *was* the PET's POST message -- saying something went wrong...


One other thing to try would be piggybacking an equivalent RAM chip over the existing RAM one at a time and see if that changes anything. Sometimes (not always) this can show up a faulty RAM chip. I diagnosed a faulty Osborne 1 by doing this

Osborne 1? The mother of all laptops? ;-) Cool...

As for the piggybacked RAM, yes, I've read about this method before -- the only problem is: where to get such an IC from? (Wait a minute -- I still must have some dysfunct VC1541 board -- perhaps I'll find some suitable RAM there? And maybe even one of the VIAs and/or the 6502 is still ok...)


The only thing I could suggest is making backups of ANY programmable chips inside these old computers. When I was working in the electronics industry, we had Eproms and EEproms go goofy after being in use for a decade or more. Re-programming the chips usually fixed the problem.

Well, as MikeS says in...


Commodore stuff is pretty well covered in that department, at least the standard ROMs; they're all archived and on the WWW.

...it shouldn't be any problem to find dumps on the Web. For instance, as mentioned in my previous posts, I've already had a look at the kernal and basic ROMs that come with the VICE emulator. Let's just hope whoever may currently hold the copyright for the old Commodore stuff doesn't start sueing harmless vintage enthusiasts just because.


But there were quite a few third-party ROMS, and I don't think they are all archived; anything in yours, track18, that didn't come from the factory?

Nope, there are no extra ROMs, and when the machine was still working I hadn't realised anything that looked more advanced than BASIC 2.0 ;-)

Now have a good night, and thanks for all the encouragement!

track18

cosam
March 16th, 2009, 11:16 AM
so... I did it -- re-seated the "big four", but no difference in behaviour. (Well, perhaps I should be grateful things didn't get worse...)
Did you try removing the PIAs completely? It's a long shot, but it solved a blank screen for me.


...next challange will be to check the power supply. (Yes, I've seen those schematics before -- but could anybody please give me a clue where to look for those test points?) Just wondering if my meter's 200 V AC setting will be capable of detecting tiny voltage ripples...
Voltage regulators are on page nine. All of the test points (which I also never found written on the board) correspond to one end or the other of various diodes, which you should be able to locate using the parts layout picture. Let's just hope the ripple doesn't register on a 200 VAC scale! ;-)


As for the piggybacked RAM, yes, I've read about this method before -- the only problem is: where to get such an IC from?
If you really can't find any I have a few here which I thankfully never needed.

tezza
March 16th, 2009, 01:26 PM
As for the piggybacked RAM, yes, I've read about this method before -- the only problem is: where to get such an IC from? (Wait a minute -- I still must have some dysfunct VC1541 board -- perhaps I'll find some suitable RAM there? And maybe even one of the VIAs and/or the 6502 is still ok...)


Yes, these 4116 ICs (I'm assuming that's what they are) are relatively common in early 80s equipment and also on some older PC cards.

Tez

track18
March 17th, 2009, 10:05 AM
Hi Cosam,


Did you try removing the PIAs completely? It's a long shot, but it solved a blank screen for me.

I reinserted the chips in the following order: UC5, UC6, UC7 (I understand that is VIA, CIA2, CIA1) -- but no BASIC welcome message appeared. Do you think it might be worth trying to remove CIA2 while leaving CIA1 plugged in?


Voltage regulators are on page nine. All of the test points (which I also never found written on the board) correspond to one end or the other of various diodes, which you should be able to locate using the parts layout picture.

Yepp, I think I got them. TP10, by the way, *is* explicitly labelled on my PET's board. For TP5 and TP6, I just probed the "anode leg" of the corresponding diode. Ground was taken from user port pin 1 or the internal cassette interface. And here's what I found (meter set to 20 V DC):

TP5: 5.something V, raising to 8.7 V over a periode of 2 hours (initially raising with a rate of a few mV / s)
TP6: 12.something V, raising to 20.3 V
TP10: -5.something V, falling to -8.9 V



Let's just hope the ripple doesn't register on a 200 VAC scale! ;-)

You shouldn't have made fun on this issue... :-) Here's what my meter found:

TP5: 17.9 V
TP6: 43.4 V
TP10: no AC

But this just can't be true, can it? I guess my meter's AC setting is actually meant for mains voltage and would give cr*p readings for small AC riding on a DC baseline.

BTW, I just purchased an oscilloscope through some big, fameous internet auction house -- this should help clarifying these readings.


If you really can't find any I have a few here which I thankfully never needed.

Hey, thanks in advance -- I'll fall back on this kind offer if my PET's disease turns out not to be power-supply related and my broken 1541 floppy disk controller's RAM doesn't turn out helpful.

All the best --

track18

cosam
March 17th, 2009, 02:38 PM
I reinserted the chips in the following order: UC5, UC6, UC7 (I understand that is VIA, CIA2, CIA1) -- but no BASIC welcome message appeared. Do you think it might be worth trying to remove CIA2 while leaving CIA1 plugged in?
I don't think it would matter, as long as at some stage you had both PIAs out at the same time. I don't think it'll run at all without the VIA though. You should still get BASIC without either PIA inserted and if one was the cause of your problem, removing it should have some effect.


TP5: 5.something V, raising to 8.7 V over a periode of 2 hours (initially raising with a rate of a few mV / s)
TP6: 12.something V, raising to 20.3 V
TP10: -5.something V, falling to -8.9 V
Huh? Are you sure? If the voltage was that high, I'd expect a lot of parts to be getting rather warm ;-) How about at the other +5V regulator (there are two)?


But this just can't be true, can it? I guess my meter's AC setting is actually meant for mains voltage and would give cr*p readings for small AC riding on a DC baseline.
I think it depends on your meter and how it measures AC. I have a digital one that gives that sort of wacky readings when set to AC too.


BTW, I just purchased an oscilloscope through some big, fameous internet auction house -- this should help clarifying these readings.
Well, you certainly sound serious about getting this thing fixed ;-) With a scope and a logic analyser you are better equipped to repair this thing that I am!

track18
March 18th, 2009, 12:25 PM
Hi!


Huh? Are you sure? If the voltage was that high, I'd expect a lot of parts to be getting rather warm ;-)

Well, as sure as my 10 € multimeter allows me to be... Just checked again: right after switching on the voltages are approx. what they ought to be, but the values are raising quickly... I bet in another two hours I'd end up with the same high voltages I reported yesterday.


How about at the other +5V regulator (there are two)?

Have had no time to check -- had to work all day. Tomorrow there'll be a lot of family business to be dealt with, and Friday will be another work day for me. I guess I'll have to wait until the weekend before proceeding with further diagnostic measures...


I think it depends on your meter and how it measures AC. I have a digital one that gives that sort of wacky readings when set to AC too.

Maybe the osci will help clarifying things -- when it arrives. (The guy I purchased it from hasn't given me any payment instructions so far...)


Well, you certainly sound serious about getting this thing fixed ;-)

You bet ya! ;-) I'd love to write some sort of music sequencer for it!


With a scope and a logic analyser you are better equipped to repair this thing that I am!

Oh, I certainly don't have a logic analyzer, just a logic probe! This is a little device that tells me whether a line is high, low or changing its level -- nothing more!

Regards --

track18

MikeS
March 18th, 2009, 04:51 PM
Time to replace the battery in your meter I think...

track18
March 19th, 2009, 01:36 AM
MikeS,


Time to replace the battery in your meter I think...

good point -- it's a recharcheable, and that's exactly what I'll do: recharge it...

Regards --

track18

track18
March 22nd, 2009, 10:56 AM
Hi,

hope you had a weekend as enjoyable as I did... Lots of family and friends for visit -- but little time to look further into the PET business...


Time to replace the battery in your meter I think...

You were absolutely right -- after recharging its battery, the meter now shows reasonable DC values which also don't change over time:

TP5: 5.0 V
TP6: 11.6 V
TP10: -5.1 V

Regarding the AC values, they appear to be equal to approx. twice the DC values:

TP5: 10.2 V
TP6: 24.8 V
TP10: 10.5 V

As for TP10, I only get a reading other than 0 V AC if I reverse the meter's electrodes so that the DC setting yields a positiv value. Dunno, maybe for AC this cheapjack thing only determines the maximum positive value it encounters within each measuring interval and just displays the double value... Well, the osci should resolve this issue once it arrives.

Meanwhile I wonder if it might be worth digging further into the state of the PIA and VIA chips. On the one hand, I might try to figure out if their pins exhibit sensible readings using the logic probe. (The difficulty, however, will be to figure out which readings would be sensible -- information about the PET's inner workings is scattered so much more sparsely over the Web as about, say, the C64's!) On the other hand I might "sacrifice" one of my working 1541 drives and just see if swapping a known-to-be-ok VIA into the PET might help.

Regards --

track18

cosam
March 22nd, 2009, 12:56 PM
You were absolutely right -- after recharging its battery, the meter now shows reasonable DC values which also don't change over time:

TP5: 5.0 V
TP6: 11.6 V
TP10: -5.1 V

TP6 is a bit low. Probably not bad enough to cause trouble, but I'm not sure what the tolerances are on these machines. How about the other +5V? There's no test point, but its the output of VR3.


Meanwhile I wonder if it might be worth digging further into the state of the PIA and VIA chips. On the one hand, I might try to figure out if their pins exhibit sensible readings using the logic probe. (The difficulty, however, will be to figure out which readings would be sensible -- information about the PET's inner workings is scattered so much more sparsely over the Web as about, say, the C64's!) On the other hand I might "sacrifice" one of my working 1541 drives and just see if swapping a known-to-be-ok VIA into the PET might help.
Is the VIA socketed in the 1541? If so it's worth a try - you might just strike it lucky. If not, I'd personally not start dismantling stuff yet. I'd start with the CPU itself - try to figure out what's sensible there, then work outwards. For example, you reported a lack of interrupts before, which could definitely cause this kind of problem. Using the schematics and logic probe, you may be able to work out which part(s) is/are causing that. You won't be able to verify everything without the scope, but a logic probe can still get you a long way.

track18
March 24th, 2009, 04:19 AM
Cosam,


TP6 is a bit low. Probably not bad enough to cause trouble, but I'm not sure what the tolerances are on these machines. How about the other +5V? There's no test point, but its the output of VR3.

still haven't had time to check this out -- it is amazing how time flies if there are two little creatures in your household which are in constant need for food, fresh dipers, and entertainment...



Is the VIA socketed in the 1541? If so it's worth a try - you might just strike it lucky. If not, I'd personally not start dismantling stuff yet. I'd start with the CPU itself - try to figure out what's sensible there, then work outwards. For example, you reported a lack of interrupts before, which could definitely cause this kind of problem. Using the schematics and logic probe, you may be able to work out which part(s) is/are causing that. You won't be able to verify everything without the scope, but a logic probe can still get you a long way.

Yes, at least in one of the drives there is a socketed VIA -- otherwise I would also be reluctant to heat up the soldering iron... As for the missing IRQ signal, this *might* be the result of a broken IO chip. However, if I'm not mistaken, those chips (including the IRQ timer) need to be programmed by the CPU in order to function -- and if the corresponding block of boot code doesn't get executed due to some other defect, there'll hardly be any IRQ. So I guess I'll really try to learn more about the interplay between the CIAs, the VIA and the CPU in order to figure out what to expect when probing their pins.

Fortunately there's no deadline associated with my PET's (hopeful) revival... ;-)

All the best --

track18

Terry Yager
March 24th, 2009, 10:04 AM
Cosam,

still haven't had time to check this out -- it is amazing how time flies if there are two little creatures in your household which are in constant need for food, fresh dipers, and entertainment...

track18

Woah! You gots pet monkeys? I always wanted one...

--T

track18
March 24th, 2009, 12:50 PM
Terry,


Woah! You gots pet monkeys? I always wanted one...

you better think it over twice before you get one (or two) for yourself...! ;-)

track18
March 30th, 2009, 10:20 AM
Hi everyone,

here's the result of the research I managed to do regarding my PET over the last week:


Power supply
============


TP6 is a bit low. Probably not bad enough to cause trouble, but I'm not sure what the tolerances are on these machines. How about the other +5V? There's no test point, but its the output of VR3.

VR3 yields some nice 5.1 V -- so everything ok here, at least for the DC part. Though I got my scope a couple of days ago, I wasn't able yet to check for any AC rippling on the DC since the scope came without a probe head. I ordered one and it should arrive anytime soon.


IRQ
===

Seems in the PET the IRQ signal doesn't get generated by an IO chip timer (like the CIAs' in the C64), but by the video circuitry. Normally, the VIDEO ON signal gets fed into PIA1 via CB1 and triggers its IRQA and/or IRQB output to emmit an L pulse. For my PET, my logic probe told me that the VIDEO ON line does do something, but there's no pulses on the IRQA/IRQB pins -- in correspondence with the observation that there are no IRQs detectable at the processor's IRQ input.

Another thing that seems odd to me is that all the PIA1's port B pins remain H no matter which keys I press -- despite the 74145 BCD-to-decimal decoder UC9 dragging its pin 1 (= 0) line down to L (in accordance to the PIA1's PA0 - PA3 output after reset). If the PET's keyboard scanning routine works somehow similar to the C64's, I'd expect this L signal to be passed on to one of the PB input lines when a key is pressed. However, since all PB pins remain H, I suspect that this port is set to output, rather than input mode.

All in all I conclude that PIA1's initialisation routine gets only partiall executed, and only after hitting the reset button (after a cold start, PA0 - PA3 are all H).


Other IO chips
==============

Have only briefly had a look at their pinouts and what to expect there, but everything looked reasonable at a first glance.


ROMs
====

It seems that after a reset only code located in the $E000 and $F000 ROMs gets executed -- the CS1 lines of $C000 and $D000 never get L. Can be confirmed when looking at the 74154 4-line to 16-line decoder/demultiplexer.

This finding concides well with the observed screen blanking after reset ($E246), but the missing start message. Scanning the VICE PET ROM image suggests that the ### COMMODORE BASIC ### string is located at $E1C4 and its printout initiated at $E196. The actual STROUT routine does, however, live at $CA1C -- in one of those ROMs which apparently never get accessed. (There is no commented disassembly available for BASIC V2.0 PETs, is there? I only found original PET and 8000 series ROMs disassemblies.)

I keep wondering whether it is normal for those ROMs to get so hot. Could anyone who owns a working PET perhaps have a look at their machine and verify/falsify this? Would be great, thanks in advance...


Things getting worse
====================

- More and more often, and in particular after machine has been switched on for a while, the reset button doesn't have any effect -- the PET seems to get stuck in the same state as immediately after power-on. In particular, the screen doesn't get blanked and the CPU stopped doing something (according to SYN, address and data bus pin).

- The (still garbled) screen starts to flicker, mainly horizontally.


What to do next?
================

- Check power supply for AC ripples once the oscilloscope probe head arrives.

- So far I haven't swapped CPU and VIA with working ones from a 1541. I would test the ICs' functions by checking whether the 1541 still works after the swap -- this procedure should protect the 1541's chips from any potentially harmful conditions that might exist in the broken PET.

- Together with the probe head I ordered a breadboard. It should be possible to readout arbitrary addresses from the ROM chips by plugging it into the board and manually applying the corresponding L/H values to the address pins, shouldn't it? Anything else to take into consideration (apart from protection against static electricity)?

- Check more detailed what's going on around PIA2 and VIA

- Learn more about what actually happens (or at least: should happen) after powering on a PET.

- Any further ideas welcome...


Best regards --

MikeS
March 30th, 2009, 05:40 PM
I've found one of these indispensable when working on PETs, Apples, AIM65s and other 6502 boards. If you've got a scope all you need is a 40 pin IC socket; I'd avoid a wirewrap socket because they tend to stretch and loosen the socket. It goes between the CPU and the CPU socket and is wired to generate NOPs, which exercises all address locations; you'll quickly see any problems with address decoding, stuck lines and it can even help find data line problems.

http://www.ctalkobt.net/files/6502-NOP-test_0.pdf

Good luck!

tezza
March 30th, 2009, 06:29 PM
Hi everyone,

here's the result of the research I managed to do regarding my PET over the last week:

(<snip> ..extremely detailed findings....)



I find it disturbing that I actually understand some of this now! Only some mind you, and not enough to be any help :)




I keep wondering whether it is normal for those ROMs to get so hot. Could anyone who owns a working PET perhaps have a look at their machine and verify/falsify this? Would be great, thanks in advance...



Mine get quite hot. Not hot enough to take your fingerprints off though, like a shorted RAM chip.

Keep at it. You obviously know what you're doing and I would say it's probably just a matter of time before you find the fault. Hopefully there is only one.

Tez

track18
March 30th, 2009, 10:56 PM
MikeS,



http://www.ctalkobt.net/files/6502-NOP-test_0.pdf
Good luck!

that looks like a really handy device! I certainly will try to build one as soon as I can spare some time to drop by at the local electronics store!

Thanks --

track18
March 30th, 2009, 11:14 PM
Tez,


I find it disturbing that I actually understand some of this now! Only some mind you, and not enough to be any help :)

I thought I knew what 'disturbing' means, but maybe as a non-native speaker I am missing a subtle undertone? I usually find it the opposite of disturbing if I understand something (except, of course, when understanding something disturbing ;-) )


dis·turb (d-stūrb)
tr.v. dis·turbed, dis·turb·ing, dis·turbs
1. To break up or destroy the tranquillity or settled state of
2. To trouble emotionally or mentally; upset.
3.
a. To interfere with; interrupt.
b. To intrude on; inconvenience.
4. To put out of order; disarrange.




Mine get quite hot. Not hot enough to take your fingerprints off though, like a shorted RAM chip.

Thanks a lot for almost burning your fingertips for me! Seems my ROMs don't get hotter than yours, so I'll re-insert the ROM business further towards the queue's tail.


Keep at it. You obviously know what you're doing and I would say it's probably just a matter of time before you find the fault. Hopefully there is only one.

Oh, thanks for your trust in my expertise... In fact, I find digital electronics not so hard to understand (it basically boils down to truth tables), but as soon as something as simple as a capacitor comes into play, I'm quite lost...

All the best --

Terry Yager
March 30th, 2009, 11:51 PM
Tez,



I thought I knew what 'disturbing' means, but maybe as a non-native speaker I am missing a subtle undertone? I usually find it the opposite of disturbing if I understand something (except, of course, when understanding something disturbing ;-) )


dis·turb (d-stūrb)
tr.v. dis·turbed, dis·turb·ing, dis·turbs
1. To break up or destroy the tranquillity or settled state of
2. To trouble emotionally or mentally; upset.
3.
a. To interfere with; interrupt.
b. To intrude on; inconvenience.
4. To put out of order; disarrange.



Definition #2 works for tez, with a shading of #1. He's so smart he scares himself...

--T

tezza
March 31st, 2009, 02:54 AM
Tez,
I thought I knew what 'disturbing' means, but maybe as a non-native speaker I am missing a subtle undertone? I usually find it the opposite of disturbing if I understand something (except, of course, when understanding something disturbing ;-) )


Yea, sorry. It was a subtle undertone, and a mild attempt at humour. It's unsettling to suddenly understand something which in the past you felt you had no chance of understanding.



Oh, thanks for your trust in my expertise... In fact, I find digital electronics not so hard to understand (it basically boils down to truth tables), but as soon as something as simple as a capacitor comes into play, I'm quite lost...


Me too! My last two repairs have both involved faulty capacitors that on the outside appeared completely undamaged!

Tez

track18
April 6th, 2009, 05:49 AM
Hi there,

I finally got the probe for my scope and even found some time to do some measurements... So here's what I found:

Power Supply
============

With the scope, the voltage regulators' output look like perfect flat lines -- "smooth as an android's butt", one might say ;-)

Clock Signal
============

The signal entering the CPU at Phi0 looks like shown on the attached 'phi-in.jpg' image:

http://vintage-computer.com/vcforum/attachment.php?attachmentid=1589&stc=1&d=1239025296

Not exactly a perfect square wave, but I guess those transient oscillations after each level change should well be within tolerance, shouldn't they?

The signals emitted by Phi1 and Phi2, however, looks a bit funny (sensu "strange") to my eyes. Phi1 is shown in the 'phi-out.jpg' file:

http://vintage-computer.com/vcforum/attachment.php?attachmentid=1590&stc=1&d=1239025439

Phi2 looks virtually identical (should be phase-shiftet, though; however, lacking a second probe, I cannot confirm this on the scope). Does anyone know if the signal is supposed to feature this "kitchen-knive's blade" shape?

Boot Sequence
=============

Image 'rw.jpg' shows, at a larger time scale, the output of the CPU's R/W line:

http://vintage-computer.com/vcforum/attachment.php?attachmentid=1591&stc=1&d=1239025439

Apparently, the CPU goes through the same cycle of operations over and over again; this is confirmed by the signals detectable at the address and data lines, which also exhibit a similar periodicity at the same time scale.

I can well imagine the CPU getting stuck in such an infinite loop while in vain trying to initialise some hardware component -- writing to some of its registers (when R/W goes L) and waiting for some readout to yield a certain value. I guess knowing what exactly the CPU is attempting to achieve here might help identifying the faulty component. To this end, I have the following idea, comments about which I'd greatly appreciate:

I understand the 6502 CPU only executes commands as long as its READY input is H. So dragging this pin to L should halt the CPU, right? IIRC, I read somewhere that MOS Technologies boasted that their 6502 would *always* display something meaningful on its address/data bus. So while halted, it should be possible to read out the CPU's address and data bus using the scope or the logic probe. When repeating this procedure several times, at least some of the readouts should point to the I/O address of the hardware component in question and/or the locations within the ROM where the repeatedly executed commands reside.

In order to be able to explicitly apply an L or H level to READY, I plan to just use a 40-pin socket with its pin no 2 removed. Placing this socket between the CPU and its original socket, I should be able to apply L/H to READY at will, using GND and +5 V taken from somewhere on the motherboard.

So what do you think? Does this sound reasonable, or did I miss something critical? (With a little luck, I'll be able to drop in at our local electronics store tomorrow to buy two 40 pin sockets -- one for the described procedure, and one for the NOP testing device recently suggested by MikeS.)


My last two repairs have both involved faulty capacitors that on the outside appeared completely undamaged

Uh, sounds nasty... Is there some sort of standard procedure recommended for identifying those little beasts?

All the best --

cosam
April 6th, 2009, 06:47 AM
Apparently, the CPU goes through the same cycle of operations over and over again; this is confirmed by the signals detectable at the address and data lines, which also exhibit a similar periodicity at the same time scale.
I'd have thought this would be true even on a fully functional machine if it's sitting idle. Would indeed be good to know where it's looping.


In order to be able to explicitly apply an L or H level to READY, I plan to just use a 40-pin socket with its pin no 2 removed. Placing this socket between the CPU and its original socket, I should be able to apply L/H to READY at will, using GND and +5 V taken from somewhere on the motherboard.
You should be able to get away with a simple wire attached to pin 2. If READY is usually high, just touching the wire to ground will pull it low. Remove the connection to ground and whatever was pulling it up before should do the same again.

Unless you get a definitive answer in the mean time, or someone beats me to it (I'm not home right now) I'll take some measurements tonight and compare them to your photos.

chuckcmagee
April 6th, 2009, 07:41 AM
Just for starters, that PHY-IN photo sure doesn't look good to me. If the scale on the scope is 0 - +5 then the overshoot going negative (to zero) looks very bad to me. I remember something about +0.8 as being the TOP of a zero value. Those oscillations start out greater than a 0,8 range. So, pin is going from a good zero into "undefined" land and back to zero, etc. I have seen "normal" square waves. You get a little spike of overshoot on the leading edge, nothing like your picture, and the start of the negative going leg tends to be rounded off, so voltage starts dropping more slowly then drops like a rock. However, I have never seen one where the middle of the square wave has cute little slowly decresing amplitude waves on it.

track18
April 6th, 2009, 08:34 AM
Cosam,


I'd have thought this would be true even on a fully functional machine if it's sitting idle. Would indeed be good to know where it's looping.

you're, of course, right -- however, I would expect the idle loop to be entered *after* the BASIC promt has been output (which obiviously doesn't happen on my machine). Furthermore, in VICE the CBM3032 seems to loop through



.C:e29d A5 9E LDA $9E
.C:e29f 85 A7 STA $A7
.C:e2a1 F0 FA BEQ $E29D


when idle, which seems significantly shorter than the loop I observed with the scope. For instance, in the above code, there seems only one write operation involved. (As far as I understand, the loop only fetches the number of chars currently stored in the keyboard buffer from $9E and stores it in the 'disable cursor blinking' flag at $A7. This is repeated until there are characters waiting for being processed.)


You should be able to get away with a simple wire attached to pin 2. If READY is usually high, just touching the wire to ground will pull it low. Remove the connection to ground and whatever was pulling it up before should do the same again.


Usually I am a little bit reluctant directly connecting something H to ground without a resistor limiting the current. However, I just had a look at the schematics, and -- tataa! -- READY is connected to +5 V via R14! :-)


Unless you get a definitive answer in the mean time, or someone beats me to it (I'm not home right now) I'll take some measurements tonight and compare them to your photos.

I'll see if I do find the time to check out the READY thing tonight. Thanks for offering to do some measurements on your machine if my loop analysis doesn't come up with something useful!

Regards --

cosam
April 6th, 2009, 11:37 AM
Well, my Phi-in and out both look just like yours. R/W looks different (I only get the first longer dip in the series, not the following three shorter ones) but that would be logical if you're in a different part of the ROM.

Let us know if you manage to read an address off the halted CPU.

track18
April 7th, 2009, 03:20 AM
Hi there,


Well, my Phi-in and out both look just like yours.R/W looks different (I only get the first longer dip in the series, not the following three shorter ones) but that would be logical if you're in a different part of the ROM.

Let us know if you manage to read an address off the halted CPU.

sorry, didn't manage to check out *anything* yesterday... Many thanks for comparing your PET signals with my findings -- although I already had begun hoping that those oscillations superimposed on the L phases (which, as chuckcmagee pointed out in


Just for starters, that PHY-IN photo sure doesn't look good to me. If the scale on the scope is 0 - +5 then the overshoot going negative (to zero) looks very bad to me. I remember something about +0.8 as being the TOP of a zero value. Those oscillations start out greater than a 0,8 range. So, pin is going from a good zero into "undefined" land and back to zero, etc. I have seen "normal" square waves. You get a little spike of overshoot on the leading edge, nothing like your picture, and the start of the negative going leg tends to be rounded off, so voltage starts dropping more slowly then drops like a rock. However, I have never seen one where the middle of the square wave has cute little slowly decresing amplitude waves on it.

and depicted in

http://vintage-computer.com/vcforum/attachment.php?attachmentid=1624&stc=1&d=1239382657

occasionally really enter nowhere land) were the cause of the trouble... That would, at least, have been a hint where to search next.

As for your single dip in the R/W signal, this fits nicely to the STA operation in a healthy PET's idle loop.

I'll try to read out some addresses ASAP -- no idea when that might be...

Regards --

MikeS
April 7th, 2009, 02:39 PM
Can't help ya with your problem, but those traces look suspiciously like a mis-adjusted probe (or low-bandwidth scope); are you using a high-impedance (x10) probe and is it properly compensated?

track18
April 10th, 2009, 11:40 AM
Hi everyone,

finally some pieces fell into place... Here's what I found over the last few days:


Clock Signal
============


Can't help ya with your problem, but those traces look suspiciously like a mis-adjusted probe (or low-bandwidth scope); are you using a high-impedance (x10) probe and is it properly compensated?

The photos I posted do indeed show measurements done with the probe set to 1x, but repeating the tests with the x10 setting (after proper compensation, of course ;-) ) gives very similar results, i.e. again with damped oscillations riding on the pulses.


CPU and VIA Replacement
=======================

I dared swapping the PET's CPU and VIA with their counterparts from a working 1541 disk drive (one at a time). The drive was still working with either of the PET chips, and the PET was still misbehaving with the drive's ICs. I take this as an evidence that the PET's CPU and VIA are, in principle, ok.


Halting the CPU
===============

I used GND (from the internal cassette port) to drag the CPU's READY line to L (accessed at pin 23 of J4), as discussed in previous postings. Most of the time I found $FFFF on the address bus, sometimes something weird like $90FF or $AAFF, and frequently something in the $E6xx range, in particular $E61E, $E620, $E621, $E624... and sometimes $E6FF. Apart from the latter, these addresses lay within the IRQ handler, which starts at $E61B and looks like



.C:e61b 48 PHA ; save A, X and Y on stack
.C:e61c 8A TXA
.C:e61d 48 PHA
.C:e61e 98 TYA
.C:e61f 48 PHA
.C:e620 BA TSX ; get stack pointer
.C:e621 BD 04 01 LDA $0104,X ; get status register from stack
.C:e624 29 10 AND #$10 ; check for break flag
.C:e626 F0 03 BEQ $E62B ; no break: skip next instruction
.C:e628 6C 92 00 JMP ($0092) ; handle BRK
.C:e62b 6C 90 00 JMP ($0090) ; handle hardware interrupt


$90/$91 contains the hardware IRQ vector ($E62E), while $92/$93 points to the BRK handler, which resides at $FD17. I'll come back to this later.

One thing, however, that disturbs me (sensu definition #2 ;-) ) is that the CPU didn't resume its work after releasing READY from GND -- only a reset could persuade it do something again.


ROM Test
========

As announced earlier, I plugged the ROM chips into my breadboard, wired them to provide power and appropriate chip select signals (as shown below)

http://vintage-computer.com/vcforum/attachment.php?attachmentid=1621&stc=1&d=1239379225


and manually applied various values to their address lines in order to read out the contents of the corresponding memory locations. Here are images of me reading out bits 5 (L) and 6 (H) of address $C000 (which contains $40):

http://vintage-computer.com/vcforum/attachment.php?attachmentid=1622&stc=1&d=1239379238
http://vintage-computer.com/vcforum/attachment.php?attachmentid=1623&stc=1&d=1239379245

For each ROM, I sampled at least the addresses $x000, $x001, $x002, $x003, $x004, $x008, $x010, $x020, $x040, $x080, $x100, $x200, $x400, and $x800 and compared the readouts with the VICE ROM images.

The $C000, $D000 and $E000 ROMs turned out fine -- but when reading out the first few bytes of the $F000 ROM I only got zeros -- where I should actually find some PETSCII-coded error messages! I continued taking some more samples than for the other ROMs, and a pattern seemed to emerge (values in brackets indicate the contents of the corresponding VICE ROM image address):



$F000: 00 (54)
$F001: 00 (4f)
$F002: 00 (4f)
$F003: 00 (20)
$F004: 00 (4d)
$F005: 00 (41)
$F006: 00 (4e)
$F007: 00 (59)
$F008: 20 (20)
$F009: 46
$F00A: 49
$F00B: 4C
$F00C: 45
$F00D: D3
$F00E: 46
$F00F: 49
$F010: 00 (4c)
$F011: 00 (45)
$F018: 49
$F019: 4C
$F020: 00 (4f)
$F021: 00 (50)
$F028: 20
$F029: 4E
$F040: 00 (a0)
$F048: 50
$F080: 56 (52)
$F081: 4F
$F088: 54
$F100: 00 (8d)
$F108: A9
$F180: DF (5F)
$F181: F2 (D0)
$F200: 00
$F208: D0
$F400: 00
$F408: 90
$F800: 00
$F808: F6


I got a $00 in all cases where A3 was L, except for $F080 and $F180 -- here I got non-zero, though equally wrong values. So this ROM seems heavily corrupted!

I also tried to figure out how exactly my ROM's false values should theoretically affect the boot process and if this was compatible with my PET's symptoms. So first for the relevant vectors:



$FFFC: D1
$FFFD: FC
$FFFE: 1B
$FFFF: E6


Apparently, the RESET ($FFFC/$FFFD => $FCD1) and IRQ ($FFFE/$FFFF => $E61B) vector bytes still were what they ought to be. However, wenn following the RESET vector to its target address, I realized that I did not only get loads of false values, but these values occasionally also changed over time! More precisely, immediately after turning on power, I frequently got values with many bytes set (like $FF, $FB etc.), and during the next 30 or so seconds more and more bits turned L, sometimes, but not always, until the values actually exptected were reached. For instance, for the first seven addresses I found values like



$FCD1: FE/FB/B6 (A2)
$FCD2: FF (FF)
$FCD3: FA/9A (9A)
$FCD4: F9 (D8)
$FCD5: 78/20 (20)
$FCD6: FE/DE (DE)
$FCD7: FD/E1 (E1)


Now sooner or later one of the false values almost certainly must lead to a CPU jam (=> no activity on the data/address bus and R/W, which is what I encounter immediately after turning on the computer) or a BRK instruction ($00), which will cause a software interrupt.

Now let's again have a closer look at the interrupt routine, which lives in one of the "good" ROMs (at $E61B). Since we came from a BRK, it will try to branch to the BRK handler, which is at $FD17 in the bad ROM. Reading out this addres yields -- tataa! -- another false value, namely $00. Consequently, any initial BRK will cause an infinite number of follow-up BRKs -- and this is most likely what I observe as recurring pattern in the R/W signal:

http://vintage-computer.com/vcforum/attachment.php?attachmentid=1591&stc=1&d=1239025439

The first, somewhat longer dip, probably corresponds to the BRK instruction itself, which writes the 2 byte program counter and the processor status to the stack, and the subsequent three smaller dips are caused by the three PHA instructions at the start of the interrupt handler.


What next?
==========

Now having most likely identified a major problem with my PET, the question ist: where to get a new $F000 ROM from? Well, the supermarked next door doesn't have them on stock -- but I could perfectly live with a surrogate like a non-original, modern-type EPROM containing the original code. From the http://vintage-computer.com/vcforum/showthread.php?t=14984 thread I've learned that some of you guys are quite competent regarding EPROMs -- do you know some nice introductory web resources where one could learn something about EPROM types and programming in general and with regard to substituting damaged vintage ROMs?

Happy easter (if you happen to celebrate this holiday)! --

tezza
April 10th, 2009, 12:56 PM
Well done on the diagnosis...and all from first principles!

A corrupted F000 ROM was one of the main problems with my Pet too.

Tez

cosam
April 10th, 2009, 01:37 PM
Nice work! I suspect someone here will be able to fix you up with a replacement ROM. Otherwise they do pop up now and again on eBay.

If you're really stuck I can program one for you but (a) you'd need to supply the EPROM and (b) it'd have to be adaptable to my el-cheapo programmer (which doesn't support the old PET chips directly). So you'd have to be really stuck, I think ;-)

MikeS
April 10th, 2009, 04:49 PM
With all those Commodore fans and experts in Holland, Germany, etc. I'd have thought there'd be all sorts of offers...

It's pretty trivial to make an adapter to use a 2732 instead of a 2532, in case that opens some doors, and a programmer's not that hard to build either, for that matter; you'll need one sooner or later...

Here's a simple reader and programmer:
http://www.larwe.com/zws/products/index.html

If all else fails, PM me and I'll see what I can do from over here in far-off Canada...

Where's Anders? He's closer and he has *everything* ;-)

mike

carlsson
April 11th, 2009, 02:59 AM
I am here, but previously I didn't know what the fault was.

After doing an EPROM inventory, I came up with the following:

5x 2516
12x 2716
2x 2532
7x 2732
12x 2764
5x 27128
2x 27256

It appears one of my two programmers can read/write 2532 chips, so in worst case I could spend one of the last two for this cause.

Does anyone know what to do with 2516 and 2716 EPROMs by the way? They're 2 kB each and while pin compatible with 2532 and/or 2732, rarely you have much use of these small chips.

cosam
April 11th, 2009, 09:47 AM
Does anyone know what to do with 2516 and 2716 EPROMs by the way? They're 2 kB each and while pin compatible with 2532 and/or 2732, rarely you have much use of these small chips.
You mean apart from using them as PET editor or character ROMs? ;-)

track18
April 14th, 2009, 12:39 AM
Hi everyone,

thanks for all your kind offers to help me out with my faulty ROM! However, we had some friends visiting us over easter, one of the being an electric engineer, and he reckoned he might be able to burn me a replacement EPROM. We'll give this a try, and I'll let you know how it worked out.

But of course MikeS is right when he says


you'll need one sooner or later...

regarding EPROM programmers! I guess I shall get into this sooner or later. (One of the first things I'd like to change is the C64's startup colors -- light blue on dark blue doesn't exactly cause eye cancer, but they couldn't have come up with a less readable colour combination, could they? ;-) )

Thanks again for all your support (and mind, I have some more vintage problems up my sleeve)!

track18
April 22nd, 2009, 10:45 AM
Hi there,

still trying to figure out which kind of modern and easily available EPROM could be used as a replacement for my faulty F000 ROM... In spite of the plethora of documents available about the PET, I find it difficult to find out something useful about the original ROMs (they are 6332/6316s, aren't they?), not to mention possible replacements EPROMs -- I'm even unable to find again the web page I got the ROM's pinout from...

I understand the appropriate replacement would be a 2532 (pin-compatible) or 2732 (would need an adapter)? Well, the closest match I could find in some electronics stores' catalogues was called 27C64. I guess it should be possible to fit one of these into the F000 ROM socket by some amount of wiring (and perhaps an additional 74xx for getting the CS levels right) -- or did I miss something?

As for building an EPROM programmer/reader, I am still searching for something nice... Something that could be connected via the serial port would be great, and Linux support is kinda must. Does anyone know such a device?

Regards --

MikeS
April 22nd, 2009, 09:11 PM
Well, a 2764 will need a few more jumpers, but not too bad; you'll just use half of it. If you look up the specs for a 2532 and a 2764 you should find that except for the end pins one side will correspond exactly, while a few pins on the other side will need to be rerouted; no need for any extra parts other than a couple of sockets and some wire.

What I do is take a 28 pin socket and bend out the pins that will be rerouted; then I take a 24 pin socket and solder thin wirewrap wire to the pins that will need to be rerouted. Poke the 28 pin through a piece of paper to insulate the rerouted pins and plug it into the 24 pin.
Make the necessary connections, plug the programmed 28 pin EPROM in the top and plug the whole thing in.

I'd send you a picture but my cheap camera doesn't take close-ups very well. If you really need them I suppose I could look up the connection detail, but it should be pretty obvious; just tie the unused address pin to ground or Vcc, depending on which half you program (or program both halves the same).

Hope that's not to confusing. Did you have a look at the link I sent to a cheap & simple programmer?

carlsson
April 22nd, 2009, 10:30 PM
If you like a pre-programmed 2532, I'll be able to help. However as noted in the other thread, if Bacon2002 actually ended up with a surplus F000 ROM chip, perhaps the two of you could deal unless he wants to keep it as a spare in case more ROMs in his PET go bad.

MikeS
April 23rd, 2009, 10:04 PM
Yeah, that's what I thought, that the one you'd sent wasn't needed and was going to be passed on to our friend here; what happened to that one?

track18
April 28th, 2009, 12:49 AM
Hi everyone,

the other day, the mentioned friend unexpectedly dropped by and brought me his EPROM burner. Together with this...


Well, a 2764 will need a few more jumpers, but not too bad; you'll just use half of it. If you look up the specs for a 2532 and a 2764 you should find that except for the
[...]
the unused address pin to ground or Vcc, depending on which half you program (or program both halves the same).


excellent and detailed explanation I should be able to get going as soon as I find the time to order some 2764 chips.


Hope that's not to confusing. Did you have a look at the link I sent to a cheap & simple programmer?

Yes, I did; interesting stuff, which might serve as an introduction to this topic. However, with all these "development discontinued" notes datet from the last millenium, I'm somewhat unsure in how far the presented information is still up-to-date. Do you have any experience with those 'EPROMr's?


If you like a pre-programmed 2532, I'll be able to help. However as noted in the other thread, if Bacon2002 actually ended up with a surplus F000 ROM chip, perhaps the two of you could deal unless he wants to keep it as a spare in case more ROMs in his PET go bad.


Yeah, that's what I thought, that the one you'd sent wasn't needed and was going to be passed on to our friend here; what happened to that one?

Hey, I missed that one... that would, of course, be great -- on the other hand, I really don't want to predate on other people's valuable spare parts collection. So... Bacon2002, do you read this? If so, what do you rekcon? Do you think you could live without a spare F000 ROM? And where do you actually live -- can I afford the shipping costs...? ;-)

Best regards --

carlsson
April 28th, 2009, 02:50 AM
Bacon's in the UK.

As for 2764 chips, you should be able to find them almost anywhere vintage. Just make sure those are erased, or otherwise you need an UV lamp or EPROM eraser before programming them. I suppose the easiest way would be to duplicate the 4K of code twice on the chip.

track18
April 28th, 2009, 04:36 AM
Carlsson,


Bacon's in the UK.


ah, cool, that's not too far away.


As for 2764 chips, you should be able to find them almost anywhere vintage. Just make sure those are erased, or otherwise you need an UV lamp or EPROM eraser before programming them. I suppose the easiest way would be to duplicate the 4K of code twice on the chip.

Yepp, my friend also borrowed me his EPROM erasor and gave me some used EPROMs -- haven't had a closer look yet, though, hence I don't know if they're suitable.

As for the 2764, I wouldn't mind spending a couple of Euros for a new one. I found offers for new 27C64 ICs which, I guess, should be fine since -- to my understanding -- CMOS chips usually go well with TTL levels, don't they?

Regards --

dave_m
April 29th, 2009, 11:41 AM
Hopefully, Bacon_2002 will send you the actual ROM which would be best for your restoration. If he does not, send me your email and shipping address by private message on this forum, and I will send you a 2532 EPROM programmed with the kernal code. Make sure to let me know which version you want.
-Dave

track18
May 1st, 2009, 01:15 PM
Dave,


Hopefully, Bacon_2002 will send you the actual ROM which would be best for your restoration. If he does not, send me your email and shipping address by private message on this forum, and I will send you a 2532 EPROM programmed with the kernal code. Make sure to let me know which version you want.

thanks for your generous offer, which I'll gladly accept in case everything else fails! :-) I'll try to contact Bacon_2002 over the next days and beg him for his spare ROM. Perhaps I can persuade him to leave it to me if I promised him a can of original German sauerkraut in return... ;-)

All the best --

tezza
May 2nd, 2009, 12:54 PM
If you can't find a ROM, I saw this item on ebay today.

Commodore PET ROM Chip Adapter (MPS6540 MOS6540 6540). Replaces any MOS ROM chip in the Commodore PET 2001! Item number: 260391689552 . $30US or near offer.

It's not just the adaptor but the seller will program it with the ROM image you want.

Tez

cosam
May 2nd, 2009, 01:00 PM
Perhaps I can persuade him to leave it to me if I promised him a can of original German sauerkraut in return... ;-)
Or maybe threaten to send a can if he doesn't deliver the ROM? ;-)


Commodore PET ROM Chip Adapter (MPS6540 MOS6540 6540). Replaces any MOS ROM chip in the Commodore PET 2001! Item number: 260391689552 . $30US or near offer.
My word - I bought a whole PET for less than that!

MikeS
May 2nd, 2009, 03:16 PM
...Yepp, my friend also ... gave me some used EPROMs -- haven't had a closer look yet, though, hence I don't know if they're suitable.
...

So, have you looked yet? Maybe you have what you need?

track18
May 3rd, 2009, 01:11 AM
Hi everyone,


If you can't find a ROM, I saw this item on ebay today.

Commodore PET ROM Chip Adapter (MPS6540 MOS6540 6540). Replaces any MOS ROM chip in the Commodore PET 2001! Item number: 260391689552 . $30US or near offer.

It's not just the adaptor but the seller will program it with the ROM image you want.

Tez

well, as Cosam pointed out in


My word - I bought a whole PET for less than that!

that seems a bit overpriced to me, too. I'm not *that* desperate yet -- and I'm actually quite confident that I will be able to build an EPROM adapter myself (with all the good advice I already got from you guys).


So, have you looked yet? Maybe you have what you need?

I got a collection of approx. 20 EPROMs of all shapes and sizes, tow of them being labelled 'NMC27C64Q' -- haven't figured out yet what 'NMC' and 'Q' stand for, but I guess those would be the most promising candidates.


Or maybe threaten to send a can if he doesn't deliver the ROM? ;-)

Appropriately prepared (with bay leaves, juniper berries, a dash of white wine and diced apples [I'm talking about fruits here ;-)]), it's a delicacy that goes particularly well with bratwurst. :-)

Regards --

dave_m
May 3rd, 2009, 08:55 AM
labelled 'NMC27C64Q' -- haven't figured out yet what 'NMC' and 'Q' stand for

NMC is a Fairchild Semiconductor part
Q is the 28 pin DIP with a quartz window

-Dave

MikeS
May 3rd, 2009, 10:18 AM
Well, you've got the eraser, you've got reasonably close EPROMS and I assume you've got (use of) a programmer; all you need is a couple of sockets and/or some thin wire and you're in business. What else did ya get; maybe there's something even closer?

dave_m
May 3rd, 2009, 12:59 PM
Commodore PET ROM Chip Adapter (MPS6540 MOS6540 6540). Replaces any MOS ROM chip in the Commodore PET 2001

I looked at the website and it appears to be a good solution for 28 pin ROMs
(6540) but doesn't the 3032 PETs have the 6332 ROM with the 24 pin package?
-Dave

track18
May 3rd, 2009, 11:07 PM
Dave,


I looked at the website and it appears to be a good solution for 28 pin ROMs
(6540) but doesn't the 3032 PETs have the 6332 ROM with the 24 pin package?
-Dave

mine definitely has 24-pin package ROMs!

Regards --

alker
May 5th, 2009, 12:15 AM
Hi all,

I did not follow the complete thread now but if you want to repair your PET 2001 Motherboard the best solution is this PET RAM/ROM Expansion from Nicolas Welte / Germany.

http://freenet-homepage.de/x1541/hardware/petram.html

http://freenet-homepage.de/x1541/hardware/petram-p.html

it is a board that completely substitutes the RAM and ROM and you won't have any troubles again.. I use one of this expansions and it is perfect also for diagnosting problems. Nicolas is a nice guy.. I wanted to buy another expansion from him but he collects orders to build new ones ;-)

I paid around EUR35/$50 for the board 2 years ago .. and it is worth it

track18
May 11th, 2009, 11:58 AM
Sometimes an image says more than a thousand words...

http://www.vintage-computer.com/vcforum/attachment.php?attachmentid=1726&stc=1&d=1242071653

(N.B. I made the typical lend vs. borrow mistake -- so thanks to Christian for *lending* me his programmer... ;-) )

Details yet to come!

Regards --

tezza
May 11th, 2009, 08:56 PM
Excellent! Well done.

Tez

MikeS
May 12th, 2009, 11:19 AM
Ditto: congratulations!

What did you finally do? Find a ROM or a 2532 or adapt a 2732/64/?? ?

m

track18
May 14th, 2009, 12:03 AM
Hi all,

here are some photos of the 27C64 adapter I built using a 24 and a 28 pin socket, wire-wrap connectors, some wire and a piece of veroboard:

http://www.vintage-computer.com/vcforum/attachment.php?attachmentid=1737&stc=1&d=1242283127
http://www.vintage-computer.com/vcforum/attachment.php?attachmentid=1738&stc=1&d=1242283135

Initially I tried the procedure described by Mike_S in


What I do is take a 28 pin socket and bend out the pins that will be rerouted; then I take a 24 pin socket and solder thin wirewrap wire to the pins that will need to be rerouted. Poke the 28 pin through a piece of paper to insulate the rerouted pins and plug it into the 24 pin.

but when stacking all these sockets directly on top of each other so that they seemed safely connected, there was hardly any gap left a wire could be routet through. (Maybe I partly misunderstood the directions?) The wire-wrap connectors allowed me to circumvent this problem, and bending out the pins which needed rerouting still did the trick.

Here's the adaptor plugged into the PET's motherboard:

http://www.vintage-computer.com/vcforum/attachment.php?attachmentid=1739&stc=1&d=1242283145

It does indeed not look too pretty... To this end, though being very happy the PET is, in principle, working again, I would indeed prefer a less bulgy solution. Since I haven't managed to get in touch with Bacon, I'd love to fall back on Dave's kind offer to send me 2532 over from sunny California. (Dave, do you read?)

Alternatively I might once more dive into the "etch-your-own-PCB" business -- although I remember it to be a really messy business when I last tried it some 25 years agio... (I was etching a board that would allow you to plug a second SID into the C64, according to instructions published by the renowned German 64'er magazine :-) )

As for Nicolas' RAM/ROM expansion...


I did not follow the complete thread now but if you want to repair your PET 2001 Motherboard the best solution is this PET RAM/ROM Expansion from Nicolas Welte / Germany.

...I've encountered his hardware pages during my research regarding the PET's repair, and though the expansion board looks like a really smart device, it is somewhat beyond what I'd (currently) consider an appropriate modification for restoring a vintage computer's functionality. But that may change in future... ;-)

(BTW, the name Nicolas Welte somehow rang a bell -- is it possible that this guy has already been publishing Commodore-related stuff back in the eighteis? Books, magazines?)

So what's left to do?

- The keyboard definitely needs some cleaning. Some keys ceased to work at all -- for instance "CRSR Left/RIgh", which is pretty annoying.

- Currently the audio amplifier gets its supply voltage directly from the raw power supply connector (J8, it is?) -- which is probably the reason for the huge amount of continuously audible hissing and humming. I guess there must be some better way to drive the amplifier?

- After some minutes after powering on, the PET's screen starts to flicker vertically (not horizontally, as I initially stated by mistake). Don't know what to do about this yet, and I am admittedly not particularly keen on messing around with such a high voltage device as a video ciruit...

- Find a convenient way for transfering files between a PC and the PET. Reviewing http://www.vintage-computer.com/vcforum/showthread.php?t=14878 should reveal something.

So... I guess this is it. Is there a way to mark this thread as SOLVED? :-)

Regards --

cosam
May 14th, 2009, 01:52 AM
Good to see you're up and running again! Sure, a drop in replacement would be nice but working is always better than not working in my book.

- After some minutes after powering on, the PET's screen starts to flicker vertically (not horizontally, as I initially stated by mistake). Don't know what to do about this yet, and I am admittedly not particularly keen on messing around with such a high voltage device as a video ciruit...
Do you know which version of the video board you have? There are schematics on the Zimmers site which include lots of test points you can check with a scope. I got some great help from the guys here fixing a similar problem. Oh, and check the power for ripple in the monitor - I overlooked that and went on a wild goose chase...

I understand your concern about the high voltage stuff but as long as you're careful you should be OK. You shouldn't need to be near any of the dangerous bits and you can always power off and discharge things before hooking on probes or otherwise poking around.

dave_m
May 14th, 2009, 03:32 PM
, I'd love to fall back on Dave's kind offer to send me 2532 over from sunny California.

No problem. Use the "private message" (top right on this page) to send your shipping address to "dave_m". Do you need the 901465-03 or the -22 part?

As far as repairing the keyboard, there are several good solutions mentioned on the forum. One that worked well for me was the "Rubber Keypad Repair Kit" by MG Chemicals P/N 8339. Here is a link:

Rubber Keypad Repair Kit (http://www.mgchemicals.com/products/8339.html)

From their website, they show a distributor in France:
SDEP
550, AV. De la mauldre
ZA de la couronne des prčs
78681 Epone, France
Phone: 33 (0)1 30 90 00 00, Fax: 33 (0)1 30 90 09 09, contact@sdep.fr <email>

MikeS
May 14th, 2009, 11:25 PM
Well, I use much thinner wire, so there's no problem soldering between the sockets. But the only thing that matters with a kludge is that it works until you can do it properly, and yours does that just as well.

Sometimes the pads and contacts only need a good cleaning; if not, then the repair kit would probably do the job nicely. Alternatively, I've had pretty good success punching little disks out of aluminum foil and gluing them on with a touch of contact cement; you can also use a PCB repair "pencil" or even the metallic paint used to repair rear window defroster tapes.

But it looks like you're well on the way; good luck with the "final touches".

track18
May 17th, 2009, 12:53 PM
Hi there,


There are schematics on the Zimmers site which include lots of test points you can check with a scope. I got some great help from the guys here fixing a similar

[...]

should be OK. You shouldn't need to be near any of the dangerous bits and you can always power off and discharge things before hooking on probes or otherwise poking around.

yes, I've seen the schematics, as well as the test oscillograms -- this information will come handy once I dig into the video circuitry issue. Considering my usual degree of absent-mindedness, I'm still a bit worried about work-safety, though...

As for cleaning the keyboard...


As far as repairing the keyboard, there are several good solutions mentioned on the forum. One that worked well for me was the "Rubber Keypad Repair Kit" by MG Chemicals P/N 8339.


Sometimes the pads and contacts only need a good cleaning; if not, then the repair kit would probably do the job nicely. Alternatively, I've had pretty good success punching little disks out of aluminum foil and gluing them on with a touch of contact cement; you can also use a PCB repair "pencil" or even the metallic paint used to repair rear window defroster tapes.




...loads of interesting stuff -- however, I guess I'll first give the method described at http://www.geocities.com/davegostelow/PETkeyboard/PETkeyb.html a try, hoping the key's conductive film is still in good shape. Well, we'll see.

All the best --

dave_m
May 17th, 2009, 04:01 PM
I guess I'll first give the method described at http://www.geocities.com/davegostelow/PETkeyboard/PETkeyb.html a try, hoping the key's conductive film is still in good shape. Well, we'll see.


I looked at the method by Mr. Gostelow and it is very comprehensive but if you break one of the many plastic parts you will be in deep yogurt. I would just clean the top of the keyboard with a brush and maybe a dab of isopropyl alcohol if the keycaps are dirty rather than take it all apart. The main job will be on the back of the keyboard. The printed circuit with the matrix of connections is gold-plated and probably OK as is. If not, a small amount of contact cleaner or alcohol will make it good. The real problem will be the reconditioning of the carbon tips on the rubber key bottoms. Age has given it a high resistance. Rubbing pencil lead on the tip will probably work at least for a while, but for a permenent fix, you need to bond a conductive material to the tip in such a manner that it will not flake off with usage and cause a short across the matrix later. Mike's trick of gluing a small disk of aluminium foil should work fine. If you decide on conductive paint, make sure you put a thin layer of super glue on each tip first, then wait a few minutes such that it is still slightly tacky, then put the paint over the glue. The paint should not tend to flake off due to good bonding. Also, be careful not to get the conductive paint on any part of the rubber key that does not have glue as it might flake off with usage.
-Dave

track18
June 12th, 2009, 07:30 AM
Dear all,

today I received a parcel from California... :-) It contained a 2532 EPROM programmed with the Version 2 $F000 kernal code, and it works like a charm! It also looks much better than the clumsy adapter I initially built for the 2764 EPROM:

http://vintage-computer.com/vcforum/attachment.php?attachmentid=1788&stc=1&d=1244820118

Manymanymanymanymanymanythanks to Dave! (Also for the additional EPROM -- cool, now I have a spare part in case one of the other ROM chips says good bye, too!) I'll get your sauerkraut on its way during the next days... ;-)


I looked at the method by Mr. Gostelow and it is very that does not
[...]
have glue as it might flake off with usage.

I haven't had time to dig into this business any further so far, but I guess I'll go for the super glue / conductive paint method once my little boys let me (and all the other to-do items which have meanwhile piled up are dealt with...).

All the best --

track18

tezza
June 12th, 2009, 10:52 AM
Ahh.. that's nice.

Nothing like the elegance and clean lines of a brand new replacement ROM.

Tez

dave_m
June 12th, 2009, 09:03 PM
I'm glad the EPROMs arrived in good condition. It looks like it took about 25 days to get a small parcel from California to Germany by regular mail. It must have travelled by a slow boat. I sent the extra EPROM for your future use.
Best regards, Dave

pinkponger
June 26th, 2009, 10:24 PM
Is there a way for me to jump to the last page of a thread? I pretty much browse the Glock Talk most recent post page, and I have to click on the thread and then click to the last page. Does my question even make sense?

dave_m
August 1st, 2009, 02:30 PM
Dear all,

today I received a parcel from California... :-) It contained a 2532 EPROM programmed with the Version 2 $F000 kernal code, and it works like a charm! It also looks much better than the clumsy adapter I initially built for the 2764 EPROM:

http://vintage-computer.com/vcforum/attachment.php?attachmentid=1788&stc=1&d=1244820118

Manymanymanymanymanymanythanks to Dave! (Also for the additional EPROM -- cool, now I have a spare part in case one of the other ROM chips says good bye, too!) I'll get your sauerkraut on its way during the next days... ;-)

All the best --

track18

Today I received a fairly heavy parcel from Germany. It contained a vacuum sealed package of sauerkraut along with many packages of local sweets, chocolate and biscuits. Also enclosed was a nice letter giving the history of the various items as well as a wonderful recipe for preparing the sauerkraut with bratwurst.

I will enjoy these goodies from a friend.
-Dave

MikeS
August 1st, 2009, 05:56 PM
Today I received a fairly heavy parcel from Germany. It contained a vacuum sealed package of sauerkraut along with many packages of local sweets, chocolate and biscuits. Also enclosed was a nice letter giving the history of the various items as well as a wonderful recipe for preparing the sauerkraut with bratwurst.

I will enjoy these goodies from a friend.
-Dave
Well, I didn't really contribute anything but I'd love a copy of that recipe... waddya say?

dave_m
August 1st, 2009, 07:38 PM
Mike,
Will do. I'll scan the letter right away.
-Dave

dave_m
August 1st, 2009, 08:28 PM
Here is the letter
1952

dave_m
August 1st, 2009, 08:29 PM
Here is the letter:
1912

MikeS
August 2nd, 2009, 01:37 AM
Thank you!

But you're treading on thin ice there; I see that even a post in the "completely off-topic rants" section was criticized for not being about vintage computers (among other things) so a sauerkraut recipe in the Commodore section is really asking for it... ;-)

But my mouth is watering just reading it, despite the "farting Santa Claus" reference ;-)

track18
August 21st, 2009, 04:49 AM
Hi there,



But you're treading on thin ice there; I see that even a post in the "completely off-topic rants" section was criticized for not being about vintage computers (among other things) so a sauerkraut recipe in the Commodore section is really asking for it... ;-)

being back from holiday, recovered from swine flu (or something similar) and with kids not too demanding at the very moment, I thought I'd bring this thread back on track by mentioning that meanwhile I've bought some conductive paint, but don't know when I'll find the time to apply it to the PET's keys.

As for building an EPROMmer, I recently read an article about an inexpensive and easy-to-programm microcontroller board called Arduino (http://arduino.cc/) in the German c't magazine. I wonder if this was a suitable device for controlling the burning process, or if there are even Arduino-based EPROMmer designs already out there.

Finally I am glad to hear the sauerkraut parcel made its way to the US -- wasn't sure whether the FDA would allow importing all the kinds of food it contained... BTW, I hereby grant permission to publish my private letter to Dave on the internet ;-) I must, hoever, add an erratum: Haribo's founder's name was *Hans*, not Harald Riegel.

All the best --

dave_m
August 21st, 2009, 11:41 AM
As for building an EPROMmer, I recently read an article about an inexpensive and easy-to-programm microcontroller board called Arduino (http://arduino.cc/) in the German c't magazine. I wonder if this was a suitable device for controlling the burning process, or if there are even Arduino-based EPROMmer designs already out there.



Hi track18,

I don't see a lot of programmable I/O pins on the Arduino. At least for the older EPROMs you will need to control all the address and data lines as well as chip select to the device to be programmed. You will also have to have a programming voltage source of 25V for very old parts like the 2532 or 21V for parts like the 2764A, or 12.5V for CMOS types.

Here is a link to a board with a lot of I/O:
Teensy (http://www.pjrc.com/teensy/index.html)

Here is a link to a data sheet for the old 2532 to see how to program it. It needs a precision 50 milliSecond pulse on the chip select for each address to be programmed.

TMS2532.pdf (http://www.jrok.com/datasheet/TMS2532.pdf)

dave_m
August 21st, 2009, 02:06 PM
BTW, I hereby grant permission to publish my private letter to Dave on the internet ;-) I must, hoever, add an erratum: Haribo's founder's name was *Hans*, not Harald Riegel.



Sorry, you are quite right. I did not request permission to post the letter. I have edited the location. I hope this is not a secret family recipe. :)

x3r13x1z
September 23rd, 2009, 09:39 AM
When I bought my Commodore PET 3032 on an auction, I had the same problem with the screen output (garbage characters filled the screen). I didn't know what to do, but I tried to pull out every single IC out of their sockets, and then put them back again. I swiched on the computer, and voila! It's now working! Try, it may solve your problem.

dave_m
September 23rd, 2009, 10:55 AM
Track's problems involved a bad ROM and other things, but yes, re-seating the socketed chips is a very good thing to start with. It wipes the oxide layer off the pins and allows for good connections.

MikeS
September 23rd, 2009, 09:30 PM
I've got an 8K 2001 that consistently powers up saying 'HOUT GOSUB' (presumably the tail end of the error message). Replaced all the socketed chips with known good ones, no change, so I put it aside for a rainy day.

But this thread has inspired me to have another look one of these fine days; meanwhile, anybody got any good ideas how to proceed? It looks like someone's already replaced a number of chips, but they look pretty good.

Probably time to build a logic analyzer...

dave_m
September 23rd, 2009, 09:56 PM
This is a tricky one. You might try to power up with ground on pin 5 of the user port. This should get to the monitor. If so, check for the contents of RAM using the .M command (.M 1000 1020), and see it you can edit memory with FFFF and 0000 patterns and see if there are stuck bits.

MikeS
September 23rd, 2009, 10:06 PM
This is a tricky one. You might try to power up with ground on pin 5 of the user port. This should get to the monitor. If so, check for the contents of RAM using the .M command (.M 1000 1020), and see it you can edit memory with FFFF and 0000 patterns and see if there are stuck bits.
Ah, forgot to mention that it's unresponsive. But even if it weren't, it's got BASIC 1 ROMs, so no monitor. The chips are OK, so if it's that kind of problem it'd have to be the buffers (or the sockets of course, but it's 100% consistent and repeatable while bad sockets are usually intermittent).

dave_m
September 25th, 2009, 09:52 AM
The chips are OK, so if it's that kind of problem it'd have to be the buffers (or the sockets of course, but it's 100% consistent and repeatable while bad sockets are usually intermittent).

OK, I was thinking of my 8032 where no RAM is socketed.

If all RAM has been replaced, you should make sure the -5V bias for the RAM is OK assuming they are the 4116 dynamic chips. Is the kernal ROM on sockets also?

MikeS
September 25th, 2009, 08:10 PM
OK, I was thinking of my 8032 where no RAM is socketed.

If all RAM has been replaced, you should make sure the -5V bias for the RAM is OK assuming they are the 4116 dynamic chips. Is the kernal ROM on sockets also?
--
It's a 2001-8, the original with notoriously unreliable single-supply 6550 RAMs and almost-as-bad 6540 ROMs, but I do have two of them and swapping chips confirms that they're all OK.

I've been beta testing a little adapter that Jim Brain is going to sell that replaces a 6540 ROM with a standard 2716 and with a few jumpers and a gate I managed to replace the whole set with a single 27128; interestingly it now boots but there are some anomalies in the display that lead me to suspect that the problem is with the memory addressing. I'll check it out tomorrow and maybe replace the RAM with a 6264 while I'm at it.

Surprising considering their bad rep, but with all the testing of Jim's board and the sick PET I've probably removed and replaced every RAM and ROM chip literally dozens of times and AFAIK not had a single problem with the chip sockets (even with mildew growing on the caps from sitting in a damp basement)

Getting there...

dave_m
October 11th, 2009, 11:07 AM
I've been beta testing a little adapter that Jim Brain is going to sell that replaces a 6540 ROM with a standard 2716 and with a few jumpers and a gate I managed to replace the whole set with a single 27128; interestingly it now boots but there are some anomalies in the display that lead me to suspect that the problem is with the memory addressing. I'll check it out tomorrow and maybe replace the RAM with a 6264 while I'm at it.



Mike,
Any update? Or did you let yourself get distracted by other projects? :)
-Dave

MikeS
October 12th, 2009, 02:28 PM
Mike,
Any update? Or did you let yourself get distracted by other projects? :)
-Dave
Yeah, distracted by other projects and 'real life'. I only dug out the PETs to test Jim's ROM adapter and thought I'd have a quick look at the dead one while I was at it; then some CP/M and Tandy M100 stuff distracted me and the coming winter is making some outstanding 'projects' around the house more urgent.

But maybe your reminder is what I needed to get me back on track... Thanks!

mike