• Please review our updated Terms and Rules here

Reviving an IBM CGA Card

Carles20VT

Member
Joined
Oct 7, 2020
Messages
11
Location
Barcelona, Spain
Hello guys,

I'm trying to rebuild my IBM PC/AT 5170, and now I'm trying to revive the original video card. It's an original IBM CGA adapter.

CGA Dead card.jpg

When you plug it on the computer or any other system, it hangs it without doing anything. I put in a POST card to see what happens, but no code is shown.

I tried to see what line of the ISA bus hangs in the system. After some experimentation looking at the ISA bus pinouts, I found that the computers boot up if you cover all the data lines (D0 to D7).

I checked the CGA card diagrams fount at -0Cº but only saw a logic chip that handles these data lines to be Input or Output. I don't know how to test it or check the entire logic.

Of course, all the capacitors are checked, and none of them has been shorted. The ICs get warm after a while, and none of them is extremely hot at a touch.

¿Have anyone some idea what can I check and how? Or... if there is some list of typical issues (tantalum capacitors are one of them for sure).

I appreciate any help you can provide.
 
Have you tried cleaning the contacts and re-seating the card a few times. I use a pencil eraser on both sides
of the contacts and re-seat a few times in the slot..... have revived a few cards this way.
 
I tried to see what line of the ISA bus hangs in the system. After some experimentation looking at the ISA bus pinouts, I found that the computers boot up if you cover all the data lines (D0 to D7).
Very helpful information.

I checked the CGA card diagrams fount at -0Cº but only saw a logic chip that handles these data lines to be Input or Output.
U66, a transceiver.

So I think the problem is sure to be either:
* U66 is faulty, driving the data bus when it is being commanded not to.
* U66 is good, but is being commanded by other circuitry to drive the data bus when it shouldn't be.

On U66, pin 19 being taken low by other circuitry enables U66, and when enabled, pin 1 then controls which direction the 'driving' is occurring in.
So pin 19 is normally high, and only taken low during the brief times that the BIOS or software is interacting with the CGA card.

Is pin 19 high or low?
If high, then I suggest that you replace U66.
If low, then further detective work is required to work out why other circuitry has pin 19 sitting at low.
 
Have you tried cleaning the contacts and re-seating the card a few times. I use a pencil eraser on both sides
of the contacts and re-seat a few times in the slot..... have revived a few cards this way.

Yup! I tried it and I usually do it with the same method you describes.In this case it is not he case :(

Very helpful information.


U66, a transceiver.

So I think the problem is sure to be either:
* U66 is faulty, driving the data bus when it is being commanded not to.
* U66 is good, but is being commanded by other circuitry to drive the data bus when it shouldn't be.

On U66, pin 19 being taken low by other circuitry enables U66, and when enabled, pin 1 then controls which direction the 'driving' is occurring in.
So pin 19 is normally high, and only taken low during the brief times that the BIOS or software is interacting with the CGA card.

Is pin 19 high or low?
If high, then I suggest that you replace U66.
If low, then further detective work is required to work out why other circuitry has pin 19 sitting at low.


¡¡¡Thanks for this helpful info modem7!!!

I tested a cheap CGA clone, and I can see some small voltage decrements at some point. After trying it at IBM dead CGA, I find that it is always Hi (around 5v), but I can't see any voltage decrement. So, I can assume that the U66 IC is dead.

I just ordered a new one. In few days, you will have some news :)

The "debug" method used:
CGA Dead card Debugger.jpg

Thanks a lot for the information! I will learn a little more about logic with this small finding.
 
Only a minor update.

I have just removed the U66 and prepared the board to install the replacement.

I removed the tape and tested it. The computer boots, and nothing happens. So, it is promising! :D

5170_U66_Prepared.jpg 5170_U66_Removed.jpg
 
The IC arrived at home sooner as expected. So I just installed it when I finished the work. (I can't wait a minute more).

5170_U66_Compared.jpg

So I soldered it, put it on a test board (good known 386 board) with the POST card attached, and it started!

CGA_On386_Test.jpg CGA_On386_Test_2.jpg

Look at this!!! We have an image! And the POST card shows numbers, and the speaker counts the ram. It is promising.

CGA_On386_Image.jpg

So the IBM CGA was installed in his place, and then I was needed to set up using the Diagnostics Disk to set the "new" card. After that, I tested using the same tool to fine-tune the potentiometer on the mainboard to show the colours properly.

5170_CGA_Installed_Top.jpg 5170_CGA_Installed_Details.jpg

And after it, I tested (and played a little :p ) with these titles to test the card. Wow! CGA card indeed shows the games better-using composite! But the text is blurry. So, only games. OK, I can deal with it ;)

BigTop_CGA.jpg DigDug_CGA.jpg Pango_CGA.jpg

My dream is to have a CGA/EGA IBM monitor, but it is prohibitive to pay one, working or not.

So thanks a lot for helping with this topic. I learned a little from it. :D

​​​​​​​Now I want to solve the last mystery that I have. The dead/crazy memory expansion board!
 
I wonder how many AT owners even used the original IBM CGA card given that it won't fit in a 16-bit slot.
 
Back
Top