Radeon 9200 recap and better cooling added

On graphic card Gigabyte GV-R9264DH (Radeon 9200), witch give me some SW installation problems, I tried install solid yet passive heatsink. I picked up Zalman VNF100, because it draw the temperature on upper side of gfx card, witch is good for moving the heat away. Also there is the heatsing a bit cooled down from the air movement from CPU cooler. So even the heatsink is far superior for such old card, witch did not overheat much at all, I got in the end a very cool card, been sure that the 2D artefacts it displaying ( http://www.techpowerup.com/forums/showthread.php?t=170954 ) did not come from overheating or bad heatsink to GPU contact. And finally, no fan can be heard now from it 🙂

Whole look at the installation:

It is not entierly true, that the VNF100 cooler can be used on all R9xxx cards, as Zalman claims - Gigabyte design put one small cap slightly into the way:

I decided don't bend the cap or push the cooler on it or move the cooler more to the right but rather saw off a little of the two fins:

Cluster of TEAPO and one GSC cap - they are IMHO responsible for the minimal yet existing 2D graphic quirks:

On back of the card there is one more TEAPO cap:

Main caps are, luckily, Nichicons HM 🙂

And also Nichicons HC:

Look from the top of the card is nice, I pulled off the Zalman off his logo, to make the airflow better (unrestricted):

One more look at the cluster of TEAPO caps:

Complete lookat the back side of the card - in the fins, there are there plastic dilatation support spacers, to ensure the cooler did not touch the card from back. However if you slap these small coolers on the memory chips, then it is very likely you have to remove the middle spacer:

In the overal, then nothing stand in your way - only the heatpipes not insert all the way into the base, at least not on Soltek KT600-R mainboard. Because there can be then collision with the top of the ram installed in 1st DIMM slot. That will then prevent you insert the card properly into AGP slot. After all, even the relatively big cooler can be used in computer w/o any problems - except for the two fins on the GPU base on this particular card…

Gigabyte GV-R9264DH card use these caps:
3x 1000uF 6.3V Nichicon HM d8
2x 470uF 10V Nichicon HC d10
1x 100uF 16V GSC d5
9x 100uF 10V TEAPO d5

Small caps arrived, so - time for recap! Instead of Teapo and GSC, now there are Panny FM & Rubycon LZH caps:

And it is no wonder, that the Teapos did not make it. Well, they are almost 2x smaller that Panasonic! (Teapo 100uF 10V, Panny 150uF 6.3V) And who ever seen, that some Teapo company, notoriously known to produce bad caps, can made 2x smaller cap that Panasonic can 😄

And now is the time for some stability testing 🙂 Yet - still the very same graphic quirks are there:

Replacing only small caps are obviously not enought, of the Nichicons are fake?

So, since the card still show certain (shown above) quirks in graphic, I decided to fix it by finishing the recap of the bigger caps. It also started even blacking out the image, when on website there is a flash or some moving animation… So it looks like that the card is dying (at least from this point of view).
So I choosen replacement caps to be Nichicon FP polymers, with little capacity bump when possible - they got amazing ripple currents - 5700mA and even 6100mA for the d10 ones:

Gigabyte Radeon 9200 (GV-R9264DH)

3x 1000uF 6.3V Nichicon HM d8 - 5x Nichicon FP 1000uF 6.3V d8 (Digikey.com 493-4024-1-ND)
2x 470uF 10V Nichicon HC d10 - 2x Nichicon FP 1000uF 6.3V d10 (Digikey.com 493-3710-ND)
1x 100uF 16V GSC d5 - Rubycon LZH 100uF 16V d5 (Digikey.com 1189-1148-ND)
9x 100uF 10V TEAPO d5 - Panny 150uF 6.3V d5 (Digikey.com P12917-ND)

So todays I finally soldered them into my card. I filled even the two empty spaces, however I choosed a d8 caps and there is the holes for d10 cap, so these caps are a bit higher over the board then... But it still looks pretty nicely, at least I think so?

There are the little d8 Nichicon FP polymer near ram:

Two empty spaces are no longer empty:

All these nice polymers in one shoot:

…and a detail view of the two "main" d10 1000 FP Nichicon polymers, where are before questionable and only 470uF caps, maybe real Nichicons, but who knows?

And results? Well, the card is working, the blacking of the image stop, however the quirks are still there. Just use ATItool and the monitor turn completely black, when the rotating box should appear…
So it looks like the primary cause had to be searched in the mainboard Soltec KT600-R and w/o exchange of caps on it we did not get any far with it.
This is unfortunate.
Why most of the hardware is have bad caps? 😞

I just testing the Radeon 9200 in another mainboard (rather old yet recapped suxxka - JetWay V266B - a VIA KT266A chipset, it even support SDRAMs and did not like FSB over 133MHz - it is rock stable at FSB 150, yet the IDE access become unstable, at FSB 152 no boot to Windows… gotta check if disabling 32bit access won't help a bit for faster FSB, tough) and it works just perfectly.

Not a sign of error after over 7 hours:

…and the card is only mildly warm, despite being just on the desk, with no fan around ...!!!
(But it is true, that the fps of the cube is like 33fps, witch is - sort of - minimal. Even mine AGP PNY GeForce 6800GT can go over 300fps when no other apps are running around - check the average FPS 334: http://postimg.org/image/x9m8pl4sl/ - so the performance level is a joke 🙂 )

So after that overnight torture, I just run 3DMarks - the score is laughable, because the machine is laughable too - 133x11 and Thorton core? Come on…
( http://www.hwinfo.com/forum/Thread-Slightly-wrong-voltages-on-accient-JetWay-V266B )

But the main point is, that the card works well and perfectly - and very very cold. It is still not even hot and it run some Duke3D demos (the Duke port, of course) nonstop.

Therefore any problems I encounter is the problem with the Soltek mainboard. I already noticed two bulging (and of them even leaking) Chemicon KZG caps, so recap is in order :-[
Card is more that fine now 😄