Another PET with issues

[MerrickD]

I noticed that there seem to be a lot of threads about broken PETs. Allow me to add another...

It's an 8k PET, early model with the built in tape drive and chiclet keyboard. It has the 6540ROM/2114RAM motherboard and up until recently it's been working great.

I attached an image of what comes up on the screen when the machine is powered on. Typing any characters usually works fine although sometimes the typed character will be repeated a line down. Pressing the return key at any time locks up the computer. Pressing shift + CLR doesn't clear the screen completely.

All the socketed chips were reseated with no improvement and the voltages from the power regulators are all fine (within about 0.1 volts). I'm assuming this problem is caused by one or more bad ROMs, but I'm wondering if there is something else to check for first. I have access to an oscilloscope if that will help.

I appreciate any help you guys can provide.

 

[tezza]

Eww, looks nasty.

PETS seem to be fixable although diagnosis can be a mission. I'm not the expert here, but others on this forum certainly are. Is the RAM socketed? A complete swap out of RAM would eliminate at least that part of it.

Tez

 

[dave_m]

I attached an image of what comes up on the screen when the machine is powered on. Typing any characters usually works fine although sometimes the typed character will be repeated a line down. Pressing the return key at any time locks up the computer. Pressing shift + CLR doesn't clear the screen completely.

As Tez mentioned, one or both of your video RAMs may be bad (C3 & C4). It is good they are 2114s as it will be easy to find replacements.

The return key thing sounds scary. I hope it is not a bad ROM. It is odd for two things to go wrong with a working PET.

 

[tezza]

The return key thing sounds scary. I hope it is not a bad ROM. It is odd for two things to go wrong with a working PET.

Interestingly Dave you might recall my own PET 3032. The original board on that one had not one but TWO faulty ROMS! Before this was diagnosed I had received a replacement board from Anders so it didn't stop me having a working PET.

I managed a long winded method of checking the ROMS using the now-working PET, and I've recently confirmed that diagnosis with my Willem programmer. Imagine if you couldn't check them though, how difficult that diagnosis would have been.

When more than one issue is present it certainly complicates things.

Tez

 

[MerrickD]

Swapped out the video ram with two known good 2114s. Didn't fix the problem, but I noticed that the image is slightly different now, with different characters displayed than last time, So no worse no better.

 

[dave_m]

Swapped out the video ram with two known good 2114s. Didn't fix the problem, but I noticed that the image is slightly different now, with different characters displayed than last time, So no worse no better.

Well, so much for an easy fix.

I'm surprised there were sockets for the video RAM. Do your ROMs and main RAM have sockets?

 

[MerrickD]

Yes, luckily all the ram and roms are socketed including the video ram and the 6520s, 6522, and cpu also have sockets. Everything appears original too, except for one ram chip which has an "s" stamped on it instead of "mos" like all the others.

 

[MikeS]

Well, so much for an easy fix.

I'm surprised there were sockets for the video RAM. Do your ROMs and main RAM have sockets?

Yeah, I would have put a (very small) wager on the video RAM; not a stuck bit though, since the same character is mis-displayed differently in different places.

But while we're there, what do you get when both video 2114s are removed, and when only one and then the other is put back?

If that xx67 *YTES FREE is really 7167 then maybe the main RAM has at least passed the POST.

 

[MerrickD]

When both of the video rams are removed, the display is a checkerboard across the entire screen. for the other two cases of either C4 or C3 removed, I attached some images. (sorry about the blur)

 

[dave_m]

If that xx67 *YTES FREE is really 7167 then maybe the main RAM has at least passed the POST.

Good catch 'eagle eyes'. Then maybe Merrick's machine is a candidate for E's tester ROM? They are both in the SF Bay area.

 

[Eudimorphodon]

Good catch 'eagle eyes'. Then maybe Merrick's machine is a candidate for E's tester ROM? They are both in the SF Bay area.

If anyone thinks it might be useful I can't think why it wouldn't work on an original PET except... those use 2k ROMs, right? I could assemble another version of it that orgs at F800 instead of F000 so it'd fit into a 2716 (and from there into one of those 6540 ROM adapters.) The code itself is well under 1k.

 

[dave_m]

If anyone thinks it might be useful I can't think why it wouldn't work on an original PET except... those use 2k ROMs, right? I could assemble another version of it that orgs at F800 instead of F000 so it'd fit into a 2716 (and from there into one of those 6540 ROM adapters.) The code itself is well under 1k.

Oh, I forgot about the 6540 ROMs. Yes, an adapter would have to be made or purchased.

 

[MerrickD]

I would love to try out the test ROM but the 6540 to 2716 adapters are a bit pricey. I was able to get another set of original ROMs but it didn't fix the problem. Also swapping the ram chips for new ones didn't fix the issue either. I'm going to start looking for bad traces around the RAM and video RAM (it looks like there were some repairs done previously).

 

[MikeS]

I would love to try out the test ROM but the 6540 to 2716 adapters are a bit pricey.

Jim Brain sells his bare boards for $1.50 and last I heard would sell an assembled one for about $5.00 (you supply the 2716); an easy mod to use 2716s only doesn't use any parts other than socket & pins:
http://store.go4retro.com/products/ROM-el-6540.html

I was able to get another set of original ROMs but it didn't fix the problem. Also swapping the ram chips for new ones didn't fix the issue either. I'm going to start looking for bad traces around the RAM and video RAM (it looks like there were some repairs done previously).

Sounds like you and Eudimorphodon can swap tales about previous repairs... ;-)

 

[dabone]

Jims adapters are for the c64 roms to a 27128,27256,or 27512 not a 2716 adapter.

The 6540 adapters are 30.00 each with a preprogrammed 2716 rom.

That's why I was so glad to be able to order the pet ram/rom board. All the rom versions and a ram upgrade at alot less cost.

Later,
dabone

 

[MikeS]

Jims adapters are for the c64 roms to a 27128,27256,or 27512 not a 2716 adapter.

Did you take the trouble to actually look at the link I posted? The "...6540.html" should give you a clue what they're for even if you didn't bother. ;-)

You're probably thinking of this one:
http://store.go4retro.com/products/2364-Adapter.html

The 6540 adapters are 30.00 each with a preprogrammed 2716 rom.

I assume you're talking about Matt d'Asaro's version; a little expensive but certainly an option if you can't supply your own 2716.

That's why I was so glad to be able to order the pet ram/rom board. All the rom versions and a ram upgrade at a lot less cost.

Definitely a nice board, although it looks like there may not be any more available. There was also a simpler similar board that replaced the ROM set but I can't find the link at the moment (see below).

FWIW, one of my 2001-8 PETs uses one of Jim's $5 6540 adapters to replace all the ROMs with a 27256 containing two switch-selectable versions of BASIC (although it did require connecting 5 or six address jumpers to the board; no cut traces though, so easily reversible).

With a DIP switch to select the image you can also use it to replace any of the 14 ROMs one at a time for testing.

I haven't gotten around to trying it, but I think replacing the RAM with the same adapter will be even easier.

Found the link to one of the other ROM replacement boards:
http://www.zimmers.net/anonftp/pub/cbm/documents/projects/memory/pet/PET2001_ROM-Adapter.zip

But there was at least one other one...

 

[dabone]

I didn't know about that converter. You should really document how you used that to upgrade the roms. It would be a nice alternative. But doesn't Jim's adapter require surface mount flash?

Later,
dabone

 

[Eudimorphodon]

Discussions about ROM adapters aside...

Just spitballing here, but based on the troubles my PET was having I'm wondering if the OP might be having problems similar to mine with writing to VRAM. The hand-drawn schematics on Zimmers for the static-board PETs are sort of giving me a headache but it looks to me like it uses a similar arrangement of 74LS244s. Depending on what tools MerrickD has available I'd look hard at tracing the read and write enable lines of all of them (pins 1 and 19, whether it's read or write depends on where it is, at least on the dynamic board PET) and make sure they're running where they're supposed to. If the ones in front of VRAM are floating they could only intermittently be allowing writes to succeed.

This is probably a dumb question and I'm not sure how to phrase it, but... with the Commodore's screen editor is there an input buffer outside of the screen RAM? IE, if I use the cursor keys to go back to a previous line, edit it, and hit "enter", does the editor read VRAM to construct the command it processes, or is there a floating line buffer elsewhere in memory? Hrm... I can check this in xpet...

Okay, it *does* read the VRAM. (I did this:

PRINT "HELLO"

followed by a:

POKE 32977,2

which changes the first "L" in "HELLO" in the print command to a "B". If I cursor back up and hit "enter" on that line it prints "HEBLO".) I wonder if the reason the computer is hanging when "return" is hit *could* be that the editor reads garbage from VRAM and as a result gets lost and confused.

(EDIT: It worries me a little that the OP mentions "Typing any characters usually works fine although sometimes the typed character will be repeated a line down.". I'm getting that phantom cursor in the wrong place on my PET, and of course the cursor updating is the result of the CPU writing to VRAM. I might still have bugs in *my* VRAM write circuitry.)

 

[carlsson]

I believe the PET as well as later VIC, C64 etc have a list of logical lines, which means relative pointers to the screen memory. At least on the later machines, if you enter a program line that causes a line wrap across the boundary, the list is reconstructed so e.g. logical line 3 spans two physical screen lines and logical line 4 begins where originally line 5 was to be found.

The Basic interpreter indeed reads the screen memory, which might be confusing to refer to as VRAM as most other computers with RAM + VRAM have different uses for those and not always the CPU can access VRAM directly. On the Commodores, to most part all RAM is the same and the location of the screen memory is a matter of software configuration. Of course it is practical to the RAM chips which normally - by default - holds the screen memory as video RAM.

 

[Eudimorphodon]

The Basic interpreter indeed reads the screen memory, which might be confusing to refer to as VRAM as most other computers with RAM + VRAM have different uses for those and not always the CPU can access VRAM directly. On the Commodores, to most part all RAM is the same and the location of the screen memory is a matter of software configuration. Of course it is practical to the RAM chips which normally - by default - holds the screen memory as video RAM.

The PET has fully dedicated RAM for the screen buffer (which is different from most Commodores, I grant), so I figured it's fair to call it VRAM. If there's a better shorthand for "the separate memory-mapped chips which are dedicated for screen refresh" I can try to use it.

(I know once upon a time VRAM often referred specifically to dual-ported RAM that allowed simultaneous reading and writing, so I suppose calling plain-old 2114s or their equivalent "VRAM" is technically wrong from that standpoint. Maybe use a small v, for v(ideo)RAM?)