9" CRT vc 12" CRT

[giobbi]

Dave,

thank you for you interest (well, thanks all you people for the great support!)

Unfortunately I'm totally profane about low-level programming and hardware, so what you told it isn't totally obvious for me (what a shame!)

Well, I need you tell me what to do next. I can't wait for your diagnostic!

thank you again!

-- Giovi

-----------------------EDIT :::: IMPORTANT CLUE (at least, I believe it's important!).

I looked at the part lists from Zimmers and I see there are two alternative ROMs for the edit ROM in UD8: 901447-24 (E000-E7EE) and 901474-01 (E000-E7FF ? <- marked by hand on the schematics). So I programmed a 2716 with the 901474-01 and I tested it.

WITH THIS .BIN THE COMPUTER IS WORKING! BUT:

- the keyboard has an incorrect layout (maybe it's the .bin image for the business keyboard...?)
- Sometimes it shows just 15359 bytes free and sometimes 31743 bytes free.

I've tried to program a 2716 with the other image (901447-24 but this way it stucks: garbage -> black).

Getting closer!

 

[dave_m]

-----------------------EDIT :::: IMPORTANT CLUE (at least, I believe it's important!).

I looked at the part lists from Zimmers and I see there are two alternative ROMs for the edit ROM in UD8: 901447-24 (E000-E7EE) and 901474-01 (E000-E7FF ? <- marked by hand on the schematics). So I programmed a 2716 with the 901474-01 and I tested it.

WITH THIS .BIN THE COMPUTER IS WORKING! BUT:

- the keyboard has an incorrect layout (maybe it's the .bin image for the business keyboard...?)
- Sometimes it shows just 15359 bytes free and sometimes 31743 bytes free.

I've tried to program a 2716 with the other image (901447-24 but this way it stucks: garbage -> black).

Getting closer!

Yes, the -01 is for the business keyboard. The -24 is for the normal graphics keyboard. I assume you have a graphics keyboard? You should re-examine the E socket and then reflow the E Socket solder joints if necessary.

During power up, the PET checks its memory by writing and and reading a $55 (01010101) and then a $AA (10101010) pattern in each location. If there is a read mis-compare, the PET will stop there and only declare that much memory free. You may have soft errors in the upper 16K memory chips. You can do some poke and peeks in the address space above 15359 and see what bits are bad to learn what 4116 chips to replace.

 

[giobbi]

Yes, the -01 is for the business keyboard. The -24 is for the normal graphics keyboard. I assume you have a graphics keyboard? You should re-examine the E socket and then reflow the E Socket solder joints if necessary.

Yes, I have the graphics keyboard (with graphics symbols like the VIC-20 one).

I checked the E socket (assuming you mean the edit ROM socket) and, just to be sure, all the other ROM sockets. I found no problem there: every pin of every ROM is connected with the same PIN of every other ROM, except for pins #20 that lead to the DM74154 chip. I checked its socket too and it seems to be ok.
I also checked the four 74244 (B3, C3, E9, E10) and I didn't find anything wrong.

A question: if the ROM sockets have a problem, shouldn't get stuck with the -01 edit ROM too?
Assuming the ROM sockets are ok, what could be else?

During power up, the PET checks its memory by writing and and reading a $55 (01010101) and then a $AA (10101010) pattern in each location. If there is a read mis-compare, the PET will stop there and only declare that much memory free. You may have soft errors in the upper 16K memory chips. You can do some poke and peeks in the address space above 15359 and see what bits are bad to learn what 4116 chips to replace.

I can't do any poke/peek, I can't do any keyboard command, really: with the -01 edit ROM (the only one that actually makes it to work) I haven't a full keyboard: not all the symbols are available, even assuming I can type a command with a wrong keyboard.

Two Questions:
- shouldn't the petester say "b" if some RAM chip is wrong?
- can this RAM problem affect the boot, preventing it to correctly boot with -24 edit ROM?

 

[dave_m]

A question: if the ROM sockets have a problem, shouldn't get stuck with the -01 edit ROM too?
Assuming the ROM sockets are ok, what could be else?

Sometimes installing a new chip will flex the board enough to get a good connection so it was worth a look.

I can't do any poke/peek, I can't do any keyboard command, really: with the -01 edit ROM

So even with the -01 EPROM you do not have a blinking cursor? That may mean your 60 Hz interrupt is not working. That is where you get the keyboard scan.

Two Questions:
- shouldn't the petester say "b" if some RAM chip is wrong?
- can this RAM problem affect the boot, preventing it to correctly boot with -24 edit ROM?

Both the PET RAM test and pet tester are just quick tests and do not always find soft errors. No, the problem found by the PET was in upper memory which should not effect boot. Let us think about the next test. Very tricky problem, but we are getting closer.

 

[giobbi]

Sometimes installing a new chip will flex the board enough to get a good connection so it was worth a look.
So even with the -01 EPROM you do not have a blinking cursor? That may mean your 60 Hz interrupt is not working. That is where you get the keyboard scan.

Sorry, I didn't explain well: with -01 eprom I do have a cursor and the keyboard is working; but key map is incorrect, so when I type Return I get "?" on screen, when I type "!" I get "2", "sdf" returns "qwe" and some keys simply doesn't do anything. In other words, I haven't the full char map available on the keyboard.

I can send you a screenshot taken by my cellphone if you need.

Both the PET RAM test and pet tester are just quick tests and do not always find soft errors. No, the problem found by the PET was in upper memory which should not effect boot. Let us think about the next test. Very tricky problem, but we are getting closer.

ok. Btw actually it booted with 31743 bytes free, so the ram problem is intermittent.

Tricky problem, yes, but if it was too easy it wouldn't be funny ;-)


btw the boot is "### commodore basic ###" and not "*** COMMODORE BASIC ***"; if I'm right this difference should tell something about the basic revision... Don't know if it can be useful or not for diagnostic purposes...

 

[dave_m]

I suggest you get another 2732 and erase it for a full 30 minutes. Then reprogram it from the -24 file on Zimmers. I think the only difference between the -01 and -24 is the keyboard mapping. I would think the return key would be the same code for both, but maybe not. Steve Gray or Mike_s would know.

Yes the ### indicates the upgrade ROM set BASIC 2 which is fine for the 3032.

When we get the machine booting correctly, we should run a long memory test and verify that the RAM is perfect.

 

[giobbi]

did you mean 2716?

btw I found the return key, is it in place of the "P", if I remember correctly. But not where it's expected to be.

 

[dave_m]

Yes, 2716 although a 2732 can safely be used if the 2K data contents are programmed into the upper half of the EPROM. This is due to a quirk in the pinouts. A 2532 can not be used in the E000 socket in 3032 machines or it will 'step on' the I/O address space (but it can be used in the E000 socket in 8032 machines due to extra logic in the chip select circuit). See how much trivia you are learning about PETs

 

[giobbi]

Yes, 2716 although a 2732 can safely be used if the 2K data contents are programmed into the upper half of the EPROM. This is due to a quirk in the pinouts. A 2532 can not be used in the E000 socket in 3032 machines or it will 'step on' the I/O address space (but it can be used in the E000 socket in 8032 machines due to extra logic in the chip select circuit). See how much trivia you are learning about PETs

Really strange. I tried both 2716 and 2732+adapter without any results (garbage -> black).

I'm wondering if I'm doing some error programming the eproms?
This is the dialog map where I select what kind of image I'm opening (here with the parameter I use):


I'm also downloaded again from Zimmers the bin image I'm using, just in case; no changes. still black...

Yes, I'm learning a lot! Well, it's quite easy to find something I don't know about PETs

 

[dave_m]

Really strange. I tried both 2716 and 2732+adapter without any results (garbage -> black).

No, in the E000 socket, an adapter for the 2732 or 2716 is not wanted, it will not work with an adapter. Just plug it in. The 2K data must be programmed at $0800 to $0FFF in the 2732 EPROM.

I will convert the binary file to an Intel format and post for you. All EPROM programmers work well with the Intel format.

 

[dave_m]

The EPROM file 901447-24.hex in Intel hex format can be downloaded HERE

 

[giobbi]

The EPROM file 901447-24.hex in Intel hex format can be downloaded HERE

Hi Dave,
thank you for the hex file. Unfortunately the problem wasn't the bin vs. hex format OR I did something wrong programming the eprom.

I tested both 2716 and 2732A *without* adapter (one more thing I didn't know...)

I programmed the 2716 starting at 0000$ and the 2732 starting at 0800. In this case I tried to fill the empty part with both FF or 00 hex.
I also tried to use the .bin image from Zimmers in the 2732 starting at 0800 (who knows?).

Nothing changed. There's a poltergeist inside this old boy, I'm sure...

Why, oh why, it works with the wrong edit ROM and it doesn't with the right ROM ? Where's the difference???

 

[dave_m]

Why, oh why, it works with the wrong edit ROM and it doesn't with the right ROM ? Where's the difference???

Giovi,
It is hard to read the 2732/2532 adapter you used, but is it possible pin 21 on the board socket (+5V) inadvertingly came in contact with either the chip select signal (Select E) or perhaps the address line BA11? It may be possible one of those signals may be damaged. A shot in the dark would say to replace the chip select decoder D2 (74154). We are graspimg at straws right now. I wish you had a scope. Hang in there. You are still very close to fixing this.
-Dave

 

[giobbi]

Giovi,
It is hard to read the 2732/2532 adapter you used, but is it possible pin 21 on the board socket (+5V) inadvertingly came in contact with either the chip select signal (Select E) or perhaps the address line BA11? It may be possible one of those signals may be damaged. A shot in the dark would say to replace the chip select decoder D2 (74154). We are graspimg at straws right now. I wish you had a scope. Hang in there. You are still very close to fixing this.
-Dave

Dave, I'm not using any adapter in the E socket (UD8) and however I tested the .hex file you passed me either with 2716 and 2732 directly into the socket.
When I did this test, I left the others original ROMs (901465-01/-02/-03) in their sockets, so no adapter was used.

However the adapter 2532 -> 2732 I'm using when I'm checking the other ROMs (or when I use the petester.bin) was made this way:


(no diodes, just wires)

I believe I got it from the Ray Carlssen's site. It works fine with the petester.bin

I checked many times every ROM socket pin to be sure there is no short between them. I didn't check only adiacent pins but I also the adiacent traces; i.e. I checked pin 21 with pin 22 and 23 but also with pin 2, 3 and 4; and I did it with all the pins.

Even knowing it was redundant (almost all pins are connected) I did this check with every ROM socket. Then I checked that every pin of every rom is connected with other roms, except for the pin #20. I'm having multimeter nightmares at night ;-)

I can read about 760 ohm between ROMs pin 21 and 20 (one direction only) and 0 ohm between 21 and 24 (but it should be ok, shouldn't it?)

Yes I've already replaced the DM74154 with a good one.

The only way to have a scope is to buy it, unfortunately. And it will cost more than the 3032 :-( --> there are some cheaps but one at 20 MHz or more is quite expensive...

There a question that's driving me nut: what is the difference between edit ROM -24 (standard keyboard) and edit ROM -01 (business keyboard) ? This last one seems to work fine, so doesn't it mean the circuitry is ok? If I had some bad/wrong contact or some other problem concerning sockets, eprom programming, etc., shouldn't it appear with the business keyboard eprom too?

 

[MikeS]

No, unless you plan to do a fair bit of testing/repairing/building/experimenting I think an oscilloscope would be a waste of money and might not help here anyway.

When you originally calculated the checksums of the ROMs were you using the original ROMs or images from Zimmer's? I wonder if the 901447-24 image on Zimmer's could possibly have an error. One would think that it would have been found by now, but you never know; I ran across a checksum discrepancy in another ROM the other day, but haven't looked into it to see where the difference was.

But here's another thought: why not burn a set of BASIC 4 EPROMs and see if they work by any chance? We're actually back to your original idea of using "8032" ROMs, but with the appropriate E ROM; I think these are the correct ones:
901456-20, 21, 22 and 23 (4k, 2532/2732) and 901447-29 (2k, 2716)
Check the locations; they are not in sequence.

 

[giobbi]

No, unless you plan to do a fair bit of testing/repairing/building/experimenting I think an oscilloscope would be a waste of money and might not help here anyway.

When you originally calculated the checksums of the ROMs were you using the original ROMs or images from Zimmer's? I wonder if the 901447-24 image on Zimmer's could possibly have an error. One would think that it would have been found by now, but you never know; I ran across a checksum discrepancy in another ROM the other day, but haven't looked into it to see where the difference was.

But here's another thought: why not burn a set of BASIC 4 EPROMs and see if they work by any chance? We're actually back to your original idea of using "8032" ROMs, but with the appropriate E ROM; I think these are the correct ones:
901456-20, 21, 22 and 23 (4k, 2532/2732) and 901447-29 (2k, 2716)
Check the locations; they are not in sequence.

Mike, while you're typing this post, I was doing exactly that test! I extracted all the ROMs from my spare 8032 board, except for the edit ROM.

And now I have some more clues:

Also the Basic 4.0 works only with the edit ROM for the business keyboard (edit-4-b.901474-02.bin) and not with the one for normal keyboard (edit-4-n.901447-29.bin).

So there is something concerning the normal keyboard (whatever it is...) that is preventing it to complete the boot.

Actually I tend to exclude a ROM image on Zimmers: it would mean both images are wrong *and* my original ROM is broken. Quite improbable...


>>> is there any specific IC for a given keyboard? Maybe my board has a wrong component and I didn't see it....


-------------------------------------------EDIT-----------------------------------------------

there is some info about business/graphic keyboard at:
http://www.zimmers.net/anonftp/pub/cbm/schematics/computers/pet/2001N/320351-1.gif

and something about jumpers at:
http://www.zimmers.net/anonftp/pub/cbm/schematics/computers/pet/2001N/320351-6.gif

but I can't understand how set the right assy and how to check it on my board...

 

[MikeS]

No, I didn't think it was likely that the image was bad, but when nothing else makes sense we grasp at straws...

I can't think of anything that's different between the two respective ROM versions other than the table (and the keyboard of course) and AFAIK they don't access different memory or I/O areas.

Grasping at another straw: does disconnecting the keyboard make any difference?

A real puzzle; hope we don't have to wait another 12 years for the answer...

 

[MikeS]

there is some info about business/graphic keyboard at:
http://www.zimmers.net/anonftp/pub/cbm/schematics/computers/pet/2001N/320351-1.gif

and something about jumpers at:
http://www.zimmers.net/anonftp/pub/cbm/schematics/computers/pet/2001N/320351-6.gif

but I can't understand how set the right assy and how to check it on my board...

I think that the various jumpers are all just related to memory size.

As to the keyboard wiring, AFAIK everything is the same except for the matrix connections on the keyboard and the corresponding different decoding in the 'E' ROM.

I asked Steve (who's the resident keyboard guru) if he has any ideas; just to recap, is this the correct situation:

901465-01,02,03 and 901447-24: does not work; screen clears garbage and stays blank.
901465-01,02,03 and 901474-01: works, with intermittent upper memory errors

901465-20,21,22,23 and 901447-29: does not work.
901465-20,21,22,23 and 901474-02: works; also intermittent memory errors?

 

[giobbi]

yes, I tested with and without keyboard (normally I don't connect the keyboard).

Please take a look to the links in my previous post; there's a parts list that's discriminating assy between graphic keyboard and business keyboard, and some jumpers on the parts layout image (see the three pair of jumpers marked as "B" where they're showing you need to cut the horizontal tracks and put the jumper vertically).

I'm believing there is a difference on the assembly between graphic and business keyb. But I'm not skilled enough to understand the schematics :-(

 

[MikeS]

...I'm believing there is a difference on the assembly between graphic and business keyb. But I'm not skilled enough to understand the schematics :-(

I *think* those are the "Cassette Jumpers" on sheet 3 (note that detail B refers to the Internal Cassette versions only). When the cassette drive is internal it becomes Cassette #1 and the external cassette port on the rear which is normally Cassette #1 becomes # 2 instead; that's what those jumpers select.

Have you considered changing keyboards? ;-) I guess the 9" monitor would look funny on top of the 8032 base? ;-)

How about this: connect the keyboard from your 8032, plug in either set of ROMs with the appropriate (working) business KB ROM, put the corresponding graphic E ROM into the spare 'A' socket and calculate the checksum.