B21 is ferrite bead in 1806 SMD case.
Samething like BLM41PG600SN1L
Package/Case: 1806 (4516 metric)
Impedance: 60 Ohms
Maximum DC Current: 6 A
Tolerance: 25 %
Maximum DC Resistance: 10 mOhms

Just short these pads and that's it

Appreciate the info! I will get around to doing that shortly and report back to see if I get it working or not.

Hint.
If there is no SMD 0 ohm jumper with "three zeros" on the marking, then simply flatten the tinned wire with a diameter of 0.5 mm with a hammer and cut the tape by knife.

Wow thanks for info shevalier i have same missing component aswell as C1129 cap . However my 9800 working fine without those components so op might have another issue.
Btw if anyone could tell c1129 nominal id be much apreciated .

ciornyi wrote on 2023-03-11, 12:56:

Wow thanks for info shevalier i have same missing component aswell as C1129 cap . However my 9800 working fine without those components so op might have another issue.
Btw if anyone could tell c1129 nominal id be much apreciated .

Its a 9800Pro ..Id be worried the VRM caps are getting flaky, this card was notorious for just up and dying and the VRM caps are usually the main reason. Thing is you can try replacing them but that doesn't always fix the issue, I have three dead 9800Pros that cant be resurrected..even recapping didn't work. I figure if I had gotten the cards before the caps failed fully I may have been able to switch them out before they died and perhaps stopped the death of the entire card.

ATI really did end up with some bad components on both the 9700Pro and 9800Pro cards, both die out of the blue. Though one might make the argument that the issues with the 9700Pro were totally ATI fucking up the GPU die Shim.

Some will say I'm over stating the issues with these cards but eBay is littered with the corpses of these cards or untested models that even the seller refuses to test because of their notoriety at dying.

So if you have a working one, look after it !

ciornyi wrote on 2023-03-11, 12:56:

Btw if anyone could tell c1129 nominal id be much apreciated .

I don`t see this capacitor on PCB, foto needed.

TrashPanda wrote on 2023-03-11, 13:00:

Its a 9800Pro ..

fast and very compatible for XP and older OS.
Just 4 lack, for the old version of the board (new - with turbine cooler)
- no temp monitoring. Because of this, the card is very hot. Change cooler needed.
- no fan speed control
- weird powering GPU core- combining 3.3V and 5V for current balancing.
- open ferrite core coil - very big magnetic noise. At microATX MB its problematic for some soundcard. My MX300 diamond monster hate this.

Love this card

Here it is found another missing cap. Thanks in advance.

ciornyi wrote on 2023-03-11, 16:47:

Here it is found another missing cap. Thanks in advance.

0.1 to 1u @ 16V 0805 case, X5R or X7R dielectric.
Or don't solder. It`s just EMI filter

TrashPanda wrote on 2023-03-11, 13:00:

Its a 9800Pro ..Id be worried the VRM caps are getting flaky, this card was notorious for just up and dying and the VRM caps are […]
Show full quote

ciornyi wrote on 2023-03-11, 12:56:

Wow thanks for info shevalier i have same missing component aswell as C1129 cap . However my 9800 working fine without those components so op might have another issue.
Btw if anyone could tell c1129 nominal id be much apreciated .

Its a 9800Pro ..Id be worried the VRM caps are getting flaky, this card was notorious for just up and dying and the VRM caps are usually the main reason. Thing is you can try replacing them but that doesn't always fix the issue, I have three dead 9800Pros that cant be resurrected..even recapping didn't work. I figure if I had gotten the cards before the caps failed fully I may have been able to switch them out before they died and perhaps stopped the death of the entire card.

ATI really did end up with some bad components on both the 9700Pro and 9800Pro cards, both die out of the blue. Though one might make the argument that the issues with the 9700Pro were totally ATI fucking up the GPU die Shim.

Some will say I'm over stating the issues with these cards but eBay is littered with the corpses of these cards or untested models that even the seller refuses to test because of their notoriety at dying.

So if you have a working one, look after it !

Which caps are the ones affected on it? After putting a bridge between the two pads its still not being picked up by atiflash (it was a mac flashed card at one point so trying to reflash it back to pc) and I pretty much have nothing to lose and am willing to try fixing it however I can.

If you're comfortable with it, a CH341A and a 9800 Pro BIOS image should work as well.

theamtrakvirus wrote on 2023-03-11, 22:49:

it was a mac flashed card at one point so trying to reflash it back to pc

As far as I remember, there are still configuration resistors to move.
https://dokumen.tips/documents/radeon-9800-pc … ion.html?page=5

shevalier wrote on 2023-03-12, 06:03:

theamtrakvirus wrote on 2023-03-11, 22:49:

it was a mac flashed card at one point so trying to reflash it back to pc

As far as I remember, there are still configuration resistors to move.
https://dokumen.tips/documents/radeon-9800-pc … ion.html?page=5

Yep, this came in a bundle of G4 parts like 10 years ago so it does indeed have those missing. Can I just bridge those with solder or do I need actual resistors to put in?

theamtrakvirus wrote on 2023-03-12, 08:46:

Can I just bridge those with solder

Nope.
Often bootstrap inputs pin become outputs after startup.
For example, Horizontal Sync Out is a strap for HDMI audio. If it is shorted to ground or power, then it will either not work or burn out.
Something like 10kΩ would be adequate enough

Well, I figured it probably wouldn't hurt to test it out before I messed with it further and grabbed my MDD G4 and threw it in. It does actually output which is a nice relief knowing that this thing already had a whole different capacitor knocked off in shipping that I had already replaced however it isn't perfect. Its artifacting a bit in OS 9.2. Going off the pic I took is it worth continuing after I maybe reflow it? It was running a bit warm since I hadn't changed the thermal paste yet since I was assuming it was dead. I also plan on reballing in the future as well.

theamtrakvirus wrote on 2023-03-13, 00:54:

I also plan on reballing in the future as well.

This, if its not ball between DIE and PCB GPU

shevalier wrote on 2023-03-13, 05:18:

theamtrakvirus wrote on 2023-03-13, 00:54:

I also plan on reballing in the future as well.

This, if its not ball between DIE and PCB GPU

Reballing a 9700Pro would be an interesting task, its got that stupid metal shim around the die which would complicate the reballing as it would and does act as a huge heat sink. The 9800 might be a bit easier but it too has the metal shim around the entire die though its thinner than the 9700, I wonder if its at all possible to remove the shim reball and then put the shim or an alternative back.

TrashPanda wrote on 2023-03-13, 05:30:

Reballing a 9700Pro would be an interesting task, its got that stupid metal shim around the die which would complicate the reballing as it would and does act as a huge heat sink.

Not really, the actual surface area of the shim is way too small to make a meaningful impact. The actual PCB and its ground plane remains the only significant heatsink when it comes to repairs like these, which is why preheating is essential.
Anyway, that shim can be easily removed, although I would definitely leave it on, to keep the card looking as original as possible. There are cooling solutions that work fine with the shim.

Having said that, reballing should only be done if there's actual evidence that points to bad solder joints underneath the GPU. More often than not, the issue is somewhere else.

bloodem wrote on 2023-03-13, 07:34:

TrashPanda wrote on 2023-03-13, 05:30:

Reballing a 9700Pro would be an interesting task, its got that stupid metal shim around the die which would complicate the reballing as it would and does act as a huge heat sink.

Not really, the actual surface area of the shim is way too small to make a meaningful impact. The actual PCB and its ground plane remains the only significant heatsink when it comes to repairs like these, which is why preheating is essential.
Anyway, that shim can be easily removed, although I would definitely leave it on, to keep the card looking as original as possible. There are cooling solutions that work fine with the shim.

Thats good to know, from my own reading it seemed like the shim on the 9700pro was a design mistake on ATIs part.

TrashPanda wrote on 2023-03-13, 07:37:

Thats good to know, from my own reading it seemed like the shim on the 9700pro was a design mistake on ATIs part.

Oh, it certainly was a mistake! The shim on the Radeon 9700 PRO actually prevents any flat heatsink from making good contact with the core, which is why a thicker TIM was used to fill the gap.
I'm not sure if the issue was addressed with the Radeon 9800 PRO, though (there are conflicting reports floating around). I have three of these myself (the Hercules 3D Prophet 9800 PRO - arguably the best looking version), but two are still sealed in their original packaging, and the third one (which was also sealed, but I couldn't resist and had to open it 😁 ) has only seen a few days of actual use, so there's no point in removing the heatsink (the thermal transfer is still perfect... for now).

I do know that the problem was fixed with the X8xx series, since I've personally restored quite a few of these.