• Please review our updated Terms and Rules here

CBM 4032 display problem

N

nanoflite

Member
Joined
Oct 24, 2014
Messages
13
Location
Belgium
Hi,

A week ago I switched on my 4032 to find out that something is wrong with the display. All the spaces are dollar signs and the welcome text has some characters that slowly get drawn on the screen. I see some dots blinking and appearing in some characters. Probably a VRAM and character ROM problem I guess? Is there a way to figure out what is going wrong? I have access to a scope, so I was hoping that I could figure out how the video circuitry works, but I do not want to break any more things.

Here's a screenshot and a video showing the problem:
https://drive.google.com/file/d/1qgppqlp8C5aFiCOIxtX8uSHAQ4V7JCDDDg/view?usp=sharing
https://drive.google.com/file/d/1BAoGq6pr4QeCoQuYJLmIw0A9BNLsDjCRNA/view?usp=sharing

Also, when I switch on the 4032, the time to start up varies a lot. Sometimes it almost immediately comes to life, other times I see a very bright green dot on the middle of the screen for 5 or more seconds before the machine comes to life. As far as I understand, that can be a reset circuit problem?

All the best,
Johan
 
First reseat all the socketed chips, especially the character ROM.

Second, check your power supply.
 
Be careful of the bright green spot. Keep your brightness low until the problem is resolved. That can damage the phosphor of your CRT. Yes, if the delay in raster is caused by the power supply it could also affect
the reset which could cause the bizarre character display. I don't know Commodore much. Is there a reset than can be done after powered on? If ok after a manual reset then more likely it is a power supply slowly
coming up to it's regulated voltage. I would think you should have a full raster immediately even without a proper reset. Definitely leaning toward a power supply issue.

Larry G
 
There is no way to hard-reset a PET, unfortunately. I don't think any of the Commodore 8-bits had this. The CBM-II has a RESET button, but it's downright strange, and certainly not a hard-reset.

I think a 4032 has a CRTC, but I'm not certain.
 
The 'funny characters' are caused by (a) a bad contact or (b) a duff video ram or video latch or character generator ROM.

A <SPACE> character is hexadecimal value 0x20.
A <$> character is hexadecimal value 0x24.

Indicating that it is data bit 3 (counting from bit 0) that has 'changed' from a binary '0' to a binary '1'.

The fact that some of the character dots are 'changing' indicates either a bad connection or a partially faulty IC device. The data bits are 'changing' when the character is drawn on the screen - thus you perceive a different set of 'dots'. You may see two characters 'superimposed' on each other etc.

I can also see the tell-tale 'SYNTAX ERROR' meaning that the processor may be misreading the video ram - thinking there is a command there when the user didn't type anything. Again, this could either be the video RAM - or the associated address or data path. I would assume the 'SYNTAX ERROR' and the video issue are linked (initially).

Almost certainly the 'green dot' in the screen indicates either a 'power on reset' problem or that the clock oscillator is failing to start-up correctly. The 'chirp' noise and the screen display are produced once the processor is out of reset, the clock is working correctly and the processor is executing valid instructions. Until then, the monitor will be 'warming up' (although I would have thought the electron beam would have been blanked at this point)...

Dave
 
Again, I'm no Commodore expert but I'll put in my two cents since I have a schematic. The CRT tube is powered by a 12v supply. The vertical and horizontal drive is counted down by a master 16M clock
which is the clock for the rest of the cpu. That of course is all powered by the 5v supply. Reset occurs by a timer RC constant involving R15 and C67 which is 1.0 mfd 25v tantalum. That cap would be suspect for a poor reset. I would think the vert/horizontal drive would be provided to the CRT as soon as 5v is present. If the 5V was delayed that would explain the green spot initially. I think a manual reset could be done by temporarily shorting across C67 but by all means you do that at your own risk. Using like a 1K resistor would be safest. We know the processor is running but system may not have had proper reset. If reset
restores proper function, I would suspect C67. You could measure the 5V supply to see if it is absolutely 5V. If a few tenths low then that could be the problem. Just things to try if reseating chips, etc doesn't help.

I didn't see Dave's post before I posted this but kind of the same process

Larry G
 
Last edited:
Thanks for the replies! The <space> and <dollar> are indeed characters that differ in one bit (bit 3 is on). I have the impression that bit 3 is always on.

Today I installed my RAM/ROM board (http://blog.tynemouthsoftware.co.uk/2015/05/commodore-pet-romram-replacement-boards.html) and it all boots up (yeah). But the symptoms stay the same. The board does not replace the char ROM or the video RAM I guess.

I have also downloaded the PET schematics from zimmers (link in a post above) and I´m trying to understand how bit 3 can be always high. Does somebody know about a document that explains the character generating hardware of the PET? I do not think it is a CRTC problem, but probably a broken logic chip.

I also have a 8032 on storage, is it possible to swap out the char rom? Or is the char rom ok (RAM/ROM board) and is there another (logic) problem.\

The reset and green dot problem are another problem and I´m going to try to fix the graphic problems first ;-)

Cheers,
Johan
 
Chip UD8 is listed as a character generator ROM. Page 27 shows the circuit. You could measure resistance pins D0-D7 to ground. Those are the data lines. They should all read the same fairly high resistance.

Larry G
 
Last edited:
OK, some observations:
- I uninstalled the RAM/ROM board because I want a working 4032 and I want to use the board for diagnostics
- Pulling the char rom gives me a green screen where the characters are and a black square where the cursor is.
- Replacing the ROM with one from a 8032 makes it work again

So, it was a faulty char ROM. Thanks for the help.

No as I'm one char ROM short as I pulled his one from a 8032, I would like to 'burn' my own char ROMs. Does somebody know info about EEPROMs that are available now from digikey or mouser or so, or another source? I have a programmer myself.

Cheers,
Johan
 
Not to be-labor this but did you try putting the old 4032 char rom back in to see if the problem returns?
Had you tried re-seating it before? I was taught many years ago for a sure diagnosis "re-create your problem".
Especially if the part is rare or expensive you want to make sure it's actually defective.

Larry G
 
Full marks for tracking your graphics fault down.

I would agree with Larry to swap the character generator ROM back again and see if physically removing it and re-inserting it has fixed the problem. You should be able to re-introduce the fault and fix it again by substitution if it truly is a duff character generator.

I understand the character generator is a 2316 ROM - which should be identical to a 2716 EPROM (see http://ist.uwaterloo.ca/~schepers/roms.html). You should be able to read the working 2316 in an EPROM programmer - and program the contents into a standard 2716 EPROM - job done...

I have never done this myself though - so wait until we get confirmation of this from someone who has...

Dave
 
nanoflite said:
I would like to 'burn' my own char ROMs. Does somebody know info about EEPROMs that are available now from digikey or mouser or so, or another source? I have a programmer myself.
Click to expand...

Johan,
I newer programmer may not be able to handle a 2716 EPROM, but may handle a 2816 EEPROM which will also be pin compatible. They are also obsolete, but can be bought on ebay.
 
That's correct. The springs in them are weak, so they tend to not make good contact.

That may or may not be your trouble though. I would do what the others suggest and try exhaustively to duplicate and isolate the fault.
 
OK, I followed the advice and put back in the supposedly faulty Char ROM, and indeed I got back the problem. Next I put in the Char ROM from a 8032 I have in storage and the problem went away.

To be sure that the ROM is faulty and because I do not want to risk any further damage to my 4032 I read the contents of both the faulty and OK ROM and what I see is that after every 32 bytes, I get 32 bytes of zeros in the faulty ROM: https://www.dropbox.com/s/khbr95kyx82vrve/Screenshot 2015-09-16 23.25.45.png?dl=0

There are other differences as well, which I do not understand... anyhow this proves the ROM is faulty. For those who want to examine both ROM files:
OK : https://www.dropbox.com/s/1i6rn68puni8dsr/charromok.bin?dl=0
FAULTY: https://www.dropbox.com/s/7sfq6fgybd2glml/charrom.bin?dl=0

Here are two photo's of the ROMS:
FAULTY (date code 0680): https://drive.google.com/file/d/1LcGP5S3B9nubkV6Xt7OeTMw2MF0XdoN0nw/view?usp=sharing
OK (date code 4981): https://drive.google.com/file/d/1AS9dqLT8brLg5ErNhtm_hnRH3c-s_bt26g/view?usp=sharing

Cheers,
Johan
 
Johan,

It's goose is cooked!

At least you have proved beyond a shadow of a doubt that the character generator ROM is to blame - which is always nice to know that you can reproduce and fix the problem at will.

I would also like to bet that if you read the faulty ROM a few times you may read back different contents. This would account for the 'twinkling' dots on the screen when you watch it. Have a look at the pins of the ROM where they enter the IC body under a good strong light and a magnifying glass - sometimes you can see where damage has occurred. I had a fault with an Apple IIe disk controller ROM one day. It seemed to be repeating it's code every 32 bytes. I suspected an address bit stuck at '1' from the observed data bytes. When I removed the ROM - one of the IC pins got left behind in the socket! The metal of this pin had become eaten through and the pin of the ROM was effectively floating high - but you could not see the tiny air gap with the naked eye.

You could also look between the pins of the IC for 'tin whisker' formation. One of the address lines could be shorting out to its neighbour.

Dave

Dave
 
You must log in or register to reply here.
Back
Top