I'm not sure about the resistor, but it appears that the values of SMD ceramic capacitors is determined by their size, so one that same size should have the same values. https://resources.pcb.cadence.com/blog/2023-s … tor-size-charts

Here is a good tutorial on soldering SMDs: https://www.youtube.com/watch?v=NycbZMKaxRc

This image is from an Asus 9200 Se for sale on Ebay - both of those spots are populated, but the image doesn't have sufficient resolution to know what the value of that missing resistor is.

R400=6.8K
C53=100nf

Karbist wrote on 2024-04-20, 18:53:

R400=6.8K
C53=100nf

Is there a method for determining these values that you could share with those of us who aren't very experienced at this sort of thing?

Repo Man11 wrote on 2024-04-20, 19:36:

Is there a method for determining these values that you could share with those of us who aren't very experienced at this sort of thing?

I have a bunch of ati 9250, so that was easy to find the value but even without a similar card you can easily guess the correct value,
the Q3 and Q4 are horizontal and vertical sync outputs therefore the value of R399 and R400 should be the same.

Thank you for your replies @Repo Man11 and @Karbist

Karbist wrote on 2024-04-20, 18:53:

R400=6.8K
C53=100nf

Thank you for the values. I am assuming the smd cap is a ceramic one with no specific polarity marked. Can you please confirm? Thanks.

stealthjoe wrote on 2024-04-21, 04:41:

Thank you for the values. I am assuming the smd cap is a ceramic one with no specific polarity marked. Can you please confirm? Thanks.

Yeah mlcc cap size 0603.

Update:
Had soldered the SMD resistor and capacitor today in the respective places (C53 & R400). However the issue is still present.

Any thoughts as to what might be the issue?

That looks like the kind of checkerboard pattern that is created by a failing RAM module.

Try applying pressure to the RAM chips whilst the card is running and see if you can see any difference.

Measure every zero ohm and 56 ohms resistor around the ram chips on both side of pcb, they are in form of single resistors and resistor arrays.
it's annoying task but you might get lucky and find an open resistor.

Considering the card it is, from an economical standpoint just buy a new one. Now from a challenge standpoint, it’s probably a bad RAM chip from what I have seen in youtube videos.

I wouldn't say that type of RAM is particularly known for going bad... so check solder on all pins... if good and shiny and nothing cracked off the board, consider that it might be a symptom of the power supply to the RAM being not as good as it should be somehow, either noise/filtering or lower volts than spec or something.

Karbist wrote on 2024-04-25, 17:37:

Measure every zero ohm and 56 ohms resistor around the ram chips on both side of pcb, they are in form of single resistors and resistor arrays.
it's annoying task but you might get lucky and find an open resistor.

I measured all the zero and 56 ohm resistors around the memory chips and they seem to be fine. One thing I noted is that there are some resistors marked 512 below the ram chips. As per my understanding this is supposed to be 5.1 k ohm. However on measuring these it showed 2.6 k ohm. It is the same case with all these 512 resistors and with 2 different multimeters. Is this the expected value?

paradigital wrote on 2024-04-25, 16:59:

That looks like the kind of checkerboard pattern that is created by a failing RAM module.

In the case of a failing ram module, any idea if another alternate easily available module could be used here? I checked for the same part number on ebay and it seems to be unavailable atleast to my country.

stealthjoe wrote on 2024-04-26, 09:52:

One thing I noted is that there are some resistors marked 512 below the ram chips. As per my understanding this is supposed to be 5.1 k ohm. However on measuring these it showed 2.6 k ohm. It is the same case with all these 512 resistors and with 2 different multimeters. Is this the expected value?

Yeah they are in parallel with each other, that's normal.

stealthjoe wrote on 2024-04-26, 09:55:

In the case of a failing ram module, any idea if another alternate easily available module could be used here? I checked for the same part number on ebay and it seems to be unavailable atleast to my country.

Try to find the datasheet of the original RAM chip and then look for one that has similar specs. Might help to search Mouser or any other distributor, they have parts search where you can input parameters.

stealthjoe wrote on 2024-04-26, 09:52:

As per my understanding this is supposed to be 5.1 k ohm. However on measuring these it showed 2.6 k ohm. It is the same case with all these 512 resistors and with 2 different multimeters. Is this the expected value?

Yes, it's normal for resistors to sometimes measure lower resistance in circuit. This is because other paths in the circuit could be putting other resistors or components in parallel with them, making the overall reading lower. Only if resistance reads higher than what is printed, you should be concerned.

Since you card had damaged SMD parts on it, there's a good chance there may be more. I can see nothing else appear to be missing. However, hard shock can sometimes crack ceramic caps and cause them to short out. Now, if any of them shorted out on a power rail, they would either burn up or take the power rail down, causing the card to not work at all. However, there should be a few ceramic caps around the RAM modules that are not for filtering a power rail and are rather used as signal coupling. If these short, the signal going to the memory chip(s) can become corrupt, leading exactly to the issues you see. I managed to save a Radeon HD6850 a few years back with this exact problem (shorted ceramic cap near one of the RAM chips.) To check for this issue, basically perform a resistance test across all of the small ceramic caps around the memory chips and the back of the GPU chip. Note that some will show a very low resistance reading - these are likely to be power rail filters for either the GPU core voltage or the RAM V_dd voltage. Once you test a few, you'll know which ones are which, because the one on the GPU V_core, for example, will have the lowest resistance reading, and it would be consistently the same for all ceramic caps that are connected to it. Same with the RAM. Outside of these two, if you find any ceramic caps reading a "hard" short-circuit (less than a few Ohms), you should note them down (and maybe post them here too.)
I have the same exact Asus Radeon 9200 video card, so I could double-check these values for you to see if they are a fluke or not. Just have to find where it is (since I moved somewhat recently, and there's still lots of non-essential stuff sitting in boxes.)

I'm all for fixing things, and I know it has been a few weeks since your original post, but I would recommend returning it if you still can, then getting something else. Problems like these can take many hours to diagnose and in the end may not be fixable (bad GPU for example), so think about how much your time is worth.

A Radeon 9200 SE with a 64bit memory bus is practically the least desirable AGP card you could find, so unless you have a very specific need for this card (there shouldn't technically be any since these have no special features that other cards don't have) you could probably get something better that already works. Just for a point of reference, a 9200 SE with 166Mhz 64bit memory only has about 15% more memory bandwidth than an ATi Rage 128 Pro and it is the slowest (common) DirectX 8+ card you can buy.

If you can find even a 64bit FX 5200 it would be a significant improvement in most cases... a 128bit DX9 level card of some kind would be even better and may not even cost any more depending on the listing. A Radeon 9550, 9600 or 9600 Pro (stay away from all 9xxx SE models or cards with only 2 memory chips) are fairly common and affordable and will run circles around 64bit cards. Or, for a tiny bit better backward compatibility with 90s games at the cost of some performance, an FX 5500 (make sure it is actually 128bit) is a solid option.

Of course there are lots of other options if you have the budget for them (Geforce4 Ti series for example), but there's no need to break your budget just to get a decently working late-90s to early-2000s gaming PC up and running.

Ozzuneoj wrote on 2024-05-05, 06:08:

I'm all for fixing things, and I know it has been a few weeks since your original post, but I would recommend returning it if you still can, then getting something else. Problems like these can take many hours to diagnose and in the end may not be fixable (bad GPU for example), so think about how much your time is worth.

+1

Those are indeed valid considerations.
Personally, I often try to fix just about everything I get, even stuff that isn't worth it. But not everyone likes doing those kinds of repairs (or has the time for it)... so if the seller has a return policy and the card wasn't listed for parts / as is, then let the seller know. It could also help the next person on Ebay, if for example, the seller takes a little more time to test the hardware next time before putting it for sale... or listing it as "untested" and willing to sell at a reasonable / reduced price.

Ozzuneoj wrote on 2024-05-05, 06:08:

A Radeon 9200 SE with a 64bit memory bus is practically the least desirable AGP card you could find, so unless you have a very specific need for this card (there shouldn't technically be any since these have no special features that other cards don't have) you could probably get something better that already works.
...
If you can find even a 64bit FX 5200 it would be a significant improvement in most cases...

I don't know about that.
In my own tests with my 9200 SE (64-bit) and a 64-bit FX5200, both are pretty terrible cards for any games past 2001 time frame, especially at resolutions above 800x600. The 128-bit FX5200 is only slightly better, but not by much, since nVidia didn't include the memory bandwidth features that are present in the FX5600 (and yes, I also have a 128-b FX5200 and a FX5600. 😉 )
For late 90's games, though, I find the 9200 SE and FX5200 to be just fine... well, at least if you don't mind gaming at 1024x768 max. Seems like both of these cards choke really badly with any higher resolutions. Perfect for a "high-end" Pentium II / low-end Pentium 3 build, IMO. 😀

Ozzuneoj wrote on 2024-05-05, 06:08:

A Radeon 9550, 9600 or 9600 Pro (stay away from all 9xxx SE models or cards with only 2 memory chips) are fairly common and affordable and will run circles around 64bit cards.

+1
I think the Radeon 9550/9600 cards are really the best from the 9k series, particularly because they don't need much power and are easy to cool down, thus allowing them to be much more reliable compared to the 9700/9800 series. Just point a slow-turning fan towards your 9550/9600 card (at least the ones with a passive cooler) and it will run nice and cool. Moreover, Radeon 9550/9600 cards are still relatively easy to find and don't usually cost too much ($15-20 shipped, or cheaper if you're patient / find one at a time when people aren't looking for them.)

Ozzuneoj wrote on 2024-05-05, 06:08:

Or, for a tiny bit better backward compatibility with 90s games at the cost of some performance, an FX 5500 (make sure it is actually 128bit) is a solid option.

FX5500 is essentially the same thing as the FX5200.