I guess you could if you have the BGA reballing equipment for it, if you dont then .. its cheaper to just buy another FX5500, I wouldn't attempt replacing the core without some reballing gear and a hot bed as its exceptionally difficult.

These cards are dirt cheap so yeah without the right gear its just not worth it in either time or resources and even with the right gear Im still not convinced it would be worth the time.

Trashbytes wrote on 2024-05-17, 13:17:

I guess you could if you have the BGA reballing equipment for it, if you dont then .. its cheaper to just buy another FX5500, I wouldn't attempt replacing the core without some reballing gear and a hot bed as its exceptionally difficult.

These cards are dirt cheap so yeah without the right gear its just not worth it in either time or resources and even with the right gear Im still not convinced it would be worth the time.

Problem is where I live, retro hardware is overpriced. Forgot to say the fried card (before i killed it), for the couple of seconds, was displaying but the screen was moving to the right and then it went blank again. Would flashing a new BIOS, on the card with artifacts, make an difference in behavior?

If the card has resistors on memory data lines, I would measure all of them to make sure they are not open.

you can use old MATS/MODS for DOS to test vmem on NVIDIA cards

https://drive.google.com/drive/folders/1pFuLL … kJDv4TiJKE_GWNw

it will show you which exact memory bank is bad. btw also check the capacitors/resistors around the memory ic.

dm- wrote on 2024-05-21, 10:07:

you can use old MATS/MODS for DOS to test vmem on NVIDIA cards

https://drive.google.com/drive/folders/1pFuLL … kJDv4TiJKE_GWNw

it will show you which exact memory bank is bad. btw also check the capacitors/resistors around the memory ic.

Hmm handy, thanks for this 😀 Do you know which generations this covers?

dm- wrote on 2024-05-21, 10:07:

you can use old MATS/MODS for DOS to test vmem on NVIDIA cards

https://drive.google.com/drive/folders/1pFuLL … kJDv4TiJKE_GWNw

it will show you which exact memory bank is bad. btw also check the capacitors/resistors around the memory ic.

Found 2 missing resistors. What i believe one is a 15ohm and the other one is 100ohms under U3. Ill replace them and see if that changes anything. Also, thanks for the link. I though only MONTEST is the memory tester for retro hardware.

I've been trying to use VMTCE to test memory faults but the Geforce FX memory layout is somewhat more complicated than the 1st gen AGP cards like the voodoo 1
Accurately troubleshooting video memory faults with VMTCE
It hasn't really worked well for diagnosing faults on a Geforce FX 🙁

For anything where the component value isn't clear, you could look at the P162 schematic which is what most FX5200 / FX5500 regular-height 128-bit cards are based off of. And there's the MS-8919 schematic which is similar but with a tv-encoder chip.

Replaced the missing components but there's another issue. My install of XP decided it was a good time to get corrupted. Ill try to resolve this issue.

Stopping working on this card as it seems like I worsened the condition of it to the point that it damages main boards. Thanks for help everyone.

Decided to do some further diagnostics instead of making it for parts. The 100 ohm resistor drops the resistance on all 100 ohm resistors near memory to 51 ohms. Any ideas? The replaced resistor measured correct resistance before soldering on.

kotel wrote on 2024-06-20, 11:17:

Decided to do some further diagnostics instead of making it for parts. The 100 ohm resistor drops the resistance on all 100 ohm resistors near memory to 51 ohms. Any ideas? The replaced resistor measured correct resistance before soldering on.

That's usually normal. 50-ish Ohms is the parallel resistance of two 100-Ohm resistors. I bet if you probe some of the other 100-Ohm array resistors near the other memory chips, you'll find the same thing (but if not, then inspect your soldered array resistor carefully, as you may have bridged the pads on two or more resistors.)
In general, though, it's OK for a resistor to read lower resistance in-circuit. Resistors only fail open-circuit or high-resistance. So if you find one that measures higher resistance (higher than its 5% tolerance) in-circuit, then it likely is bad.
Since you had chipped/broken/missing SMD components on your card, I would suggest to inspect every component as carefully as possible. For example, ceramic caps can become shorted if struck hard. Most ceramic caps filter power rails, though... so if any was shorted, they would either smoke itself out or short-circuit the power rail... but either way, the card would not work at all if that happened. However, there are a very very few ceramic caps near the RAM chips that are sometimes used for signal coupling (same function as the small ceramic caps near the edge connector on the back of PCI-E graphics cards). If any of these are shorted, you could get artifacts or other similar anomaly. So worth checking these with your meter while at it. After a few rounds at it, you should be able to tell which ones filter power rails on your card and which don't. The ones that do filter power rails will have the same resistance to ground as the MOSFETs / voltage regulators that connect to the GPU V_core, MEM Vdd, and etc - i.e. typically a few Ohms to a few hundred Ohms.
Testing the SMD components on a video card is all about symmetry / pattern recognition. If you take a close look around all of the SMD components around memory chip, you'll usually find that the other memory chips have the same components connected to them. The layout may be a bit different (or not) around each memory chip, but that's about it. So for example, if you find that there are five resistors near one memory chip, and two measure 50 Ohms while the others measure 100 Ohms, you should see that pattern for the same 5 resistors around another memory chip.

momaka wrote on 2024-06-20, 23:55:

That's usually normal. 50-ish Ohms is the parallel resistance of two 100-Ohm resistors. I bet if you probe some of the other 100 […]
Show full quote

kotel wrote on 2024-06-20, 11:17:

Decided to do some further diagnostics instead of making it for parts. The 100 ohm resistor drops the resistance on all 100 ohm resistors near memory to 51 ohms. Any ideas? The replaced resistor measured correct resistance before soldering on.

That's usually normal. 50-ish Ohms is the parallel resistance of two 100-Ohm resistors. I bet if you probe some of the other 100-Ohm array resistors near the other memory chips, you'll find the same thing (but if not, then inspect your soldered array resistor carefully, as you may have bridged the pads on two or more resistors.)
In general, though, it's OK for a resistor to read lower resistance in-circuit. Resistors only fail open-circuit or high-resistance. So if you find one that measures higher resistance (higher than its 5% tolerance) in-circuit, then it likely is bad.
Since you had chipped/broken/missing SMD components on your card, I would suggest to inspect every component as carefully as possible. For example, ceramic caps can become shorted if struck hard. Most ceramic caps filter power rails, though... so if any was shorted, they would either smoke itself out or short-circuit the power rail... but either way, the card would not work at all if that happened. However, there are a very very few ceramic caps near the RAM chips that are sometimes used for signal coupling (same function as the small ceramic caps near the edge connector on the back of PCI-E graphics cards). If any of these are shorted, you could get artifacts or other similar anomaly. So worth checking these with your meter while at it. After a few rounds at it, you should be able to tell which ones filter power rails on your card and which don't. The ones that do filter power rails will have the same resistance to ground as the MOSFETs / voltage regulators that connect to the GPU V_core, MEM Vdd, and etc - i.e. typically a few Ohms to a few hundred Ohms.
Testing the SMD components on a video card is all about symmetry / pattern recognition. If you take a close look around all of the SMD components around memory chip, you'll usually find that the other memory chips have the same components connected to them. The layout may be a bit different (or not) around each memory chip, but that's about it. So for example, if you find that there are five resistors near one memory chip, and two measure 50 Ohms while the others measure 100 Ohms, you should see that pattern for the same 5 resistors around another memory chip.

Thanks for the reply. For now, I really don't care about the artifacts. All I need to know (for now) is why and how does the card damage the SB and/or NB that the board doesn't detect GPU's and other components like HDD's. All i can think of now is something on the card has to pull the data line (or some other line) to the SB and/or NB that it disrupts proper functioning of the mainboard.