VGA Repair report collection \ VOGONS

VGA Repair report collection

Topic actions

First post, by tehsiggi

Posted on 2025-04-25, 11:58

tehsiggi Offline

Rank Newbie

Rank: Newbie
Posts: 17
Joined: 2025-04-14, 13:37

Dear all, as a follow-up to Hercules Radeon 9700 Pro repair report I'd like to start a thread where others and I can share their VGA repair reports.

I hope that this will help and encourage others to repair graphics cards, mainly in the "retro" perspective from my end. (Whatever that means for everyone).

I have a couple of repairs in the pipeline and would append the reports to this thread, so that they're all in one place.

This first post will take care on a Radeon 9800 Pro 128MB with reference design:

### First Impressions ###

The card itself was sold on ebay with the status of "working, but showing artifacts". Physically the card appeared in a very good condition, which is not a surprise, as the seller had cleaned it already.

The artifacts themselves appeared right from boot, they were flickering patterns, which could even completely disappear from time to time.
Just for completeness I checked if all signals appear to be free of shorts, which they were:

### Diagnosis ###

The patterns directly appeared to me as memory related, so I booted up FreeDOS and started R3MEMID to check the memory. The visual appearance of the menus and the running first test can be seen here:

The test result:

1
2R3MEMID version 1.07, (c) Copyright ATI Technologies Inc, 2003
3Log file generation enabled to .\R3MEMID.LOG ...
4Reference data file (RDF) loading disabled ...
5[1  ] Fill                                                : FAIL
6    Error ID 0VB001
7  1024 x  768 - 32 bpp ( 60 Hz): TEST FAILURE
8    failing bit : MDD0 32 33 34 35 36 37 38 39 ... 
9

According to the result, the memory chip on memory channel D starting with bit 32 is throwing errors. This is the second memory IC on that channel and same as on the Radeon 9700 Pro is designated as U59.

The memory IC is again a K4D26323 type, which I will replace with a K4D26323RA-GC2B

### The repair ###

Using pre-heating and hot-air at 390°C. I isolated the IC surroundings using kapton-tape, to make sure to reduce thermal stress.

Here is the PCB with the chip removed:

Cleaned pads:

I made sure all excess solder is removed using solderwick and used isoprop to clean off any residue.

Then I soldered on the new memory IC and cleaned the PCB around it.

### Testing ###

I ran R3MEMID again, with success:

3DMark2001SE works well too:

I ran a multi hour burn-in test using S.T.A.L.K.E.R. which was rock-solid.
The memory ICs run around 55°C on an open desk, without any heatsinks.

So again, BGA memory failure.

Reply 1 of 1, by tehsiggi

Posted on 2025-04-29, 18:48

tehsiggi Offline

Rank Newbie

Rank: Newbie
Posts: 17
Joined: 2025-04-14, 13:37

Today I am taking a look at a Radeon 9000 Pro in reference design.

This card was bought as working, though I had the impression from the offering that the card it pretty beat up.

### First Impressions ###

The card arrived in a pityful condition. It was very dirty on pins and pads. Basically every electrolytic capacitor had scratches on it, some of them had significant dents.

It was in dire need of a cleaning and some love.

### Diagnosis ###

Though the card was offered as working, it was behaving weird. In windows, even though the driver installed fine, the UI was sluggish and there appeared to be no hardware acceleration. Furthermore running 3DMark2001SE showed that a) the card appeared to be unstable and b) the result screen had graphical errors.

So after the first test, I took a look at the card in detail.

There is one capacitor missing at the VDDC regulator. According to pictures it is 470µF/16V. The 3.3µH inductor of the VDDC regulator is also broken.

There is also a blocking capacitor missing at one memory IC.

On the back, there are 0 Ohm jumper resistors missing and broken, which are parts of the AGP signal lines.

### Repair ###

Since all electrolytic caps looked bad and some had dents, I decided to basically re-cap the card. I am by far no fan of "recapping for the sake of recapping", but this card deserved better.

The missing capacitor was replaced by scraping off some silkscreen from the PCB above the missing pad. Then I was able to solder the new capacitor to it.

I also removed the old 3.3µH inductor and replaced it with a new one.

The broken AGP signal line jumpers have been removed, pads cleaned and new 0 Ohm jumpers have been installed.

### Testing ###

After the repair the card works just fine again. No more instability or graphical glitches. And the 2D mode was properly accellerated again and windows ran smooth.

### Parts used ###

12 x 22µF / 16V - EMZA160ADA220MD61G - Europe Chemi-Con
2 x 47µF / 6.3V - EMZA6R3ADA470MD61G - Europe Chemi-Con
4 x 1000µF / 10V - EEUFM1A102 - Panasonic
3 x 470µF / 16V - EMZA160ADA471MHA0G - Europe Chemi-Con
2 x 100µF / 16V - EEEFK1C101P - Panasonic
1 x 100nF / 50V - C0603C104K5RACAUTO - KEMET
1 x 3.3µH - PISM-3R3M-04 - Fastron

Go to top of page Go to top of page

Back to Video