trodas
04-09-2008, 10:59 AM
Once upon a time, there was a graphic card. Not fast one and on the top of that, it used all bad caps. So regardless how hard she try, she was destined to fail from the very beginning. CapXon caps made that promise:
http://i4.photobucket.com/albums/y134/trodas/RivaTNT264MB1-1.jpg
Luckily for the poor card, the user that bought her used her only very sporadically and he also connected her on quality PSU:
http://i4.photobucket.com/albums/y134/trodas/RivaTNT264MB2-1.jpg
So time went by slowly. Till something went wrong and the picture on screen get corrupted:
http://i4.photobucket.com/albums/y134/trodas/RivaTNT264MB3.jpg
Regardless that the card still show BIOS/DOS screens with text nicely, user demanded windws and installed a new GFX card. After that, his PSU ( FSP 250-60GTA ) told him "by by" and died on him. Probably sort of card-revenge that she got replaced by a ATI 9250 128MB card. PSU failure:
http://s4.photobucket.com/albums/y134/trod...SP250-60GTA.jpg (http://s4.photobucket.com/albums/y134/trodas/?action=view¤t=FSP250-60GTA.jpg)
But again luckily form the poor cardie - I pitty her, so I decided to save her and fix her. Instead of only 22uF 25V caps I used quite much better ones:
http://i4.photobucket.com/albums/y134/trodas/RivaTNT216MB4.jpg
Of course I measured firs and the whole cardie is powered from 3.3V rail.
http://i4.photobucket.com/albums/y134/trodas/RivaTNT216MB5.jpg
Rams are powered directly from the PSU 3.3V line, so if you want O/C them better, just increase the 3.3V line to say 3.5V... :D
http://i4.photobucket.com/albums/y134/trodas/RivaTNT216MB6.jpg
GPU main regulator produce 2.6V output for core and the two closest caps to the GPU support that voltage. All other show 3.3V on them. I still think that regulating from 3.3V to 2.6V is crazy as the difference is only 0.7V and that it not enought for good regulation and they should power the GPU from 5V rail, getting 2.4V to regulation = much better. But, well - too late.
http://i4.photobucket.com/albums/y134/trodas/RivaTNT216MB7.jpg
However no matter how hard I try, the board is still showing the very same picture, so recap was sucesfull, yet it did not fix the bug. I suspected from the beginning that the resistor pole used to make the 75 ohms VGA output is to blame, so I going to remove it and instead of removed CE filter caps install separate and better 75 ohm resistors and we see... ;)
http://i4.photobucket.com/albums/y134/trodas/RivaTNT264MB1-1.jpg
Luckily for the poor card, the user that bought her used her only very sporadically and he also connected her on quality PSU:
http://i4.photobucket.com/albums/y134/trodas/RivaTNT264MB2-1.jpg
So time went by slowly. Till something went wrong and the picture on screen get corrupted:
http://i4.photobucket.com/albums/y134/trodas/RivaTNT264MB3.jpg
Regardless that the card still show BIOS/DOS screens with text nicely, user demanded windws and installed a new GFX card. After that, his PSU ( FSP 250-60GTA ) told him "by by" and died on him. Probably sort of card-revenge that she got replaced by a ATI 9250 128MB card. PSU failure:
http://s4.photobucket.com/albums/y134/trod...SP250-60GTA.jpg (http://s4.photobucket.com/albums/y134/trodas/?action=view¤t=FSP250-60GTA.jpg)
But again luckily form the poor cardie - I pitty her, so I decided to save her and fix her. Instead of only 22uF 25V caps I used quite much better ones:
http://i4.photobucket.com/albums/y134/trodas/RivaTNT216MB4.jpg
Of course I measured firs and the whole cardie is powered from 3.3V rail.
http://i4.photobucket.com/albums/y134/trodas/RivaTNT216MB5.jpg
Rams are powered directly from the PSU 3.3V line, so if you want O/C them better, just increase the 3.3V line to say 3.5V... :D
http://i4.photobucket.com/albums/y134/trodas/RivaTNT216MB6.jpg
GPU main regulator produce 2.6V output for core and the two closest caps to the GPU support that voltage. All other show 3.3V on them. I still think that regulating from 3.3V to 2.6V is crazy as the difference is only 0.7V and that it not enought for good regulation and they should power the GPU from 5V rail, getting 2.4V to regulation = much better. But, well - too late.
http://i4.photobucket.com/albums/y134/trodas/RivaTNT216MB7.jpg
However no matter how hard I try, the board is still showing the very same picture, so recap was sucesfull, yet it did not fix the bug. I suspected from the beginning that the resistor pole used to make the 75 ohms VGA output is to blame, so I going to remove it and instead of removed CE filter caps install separate and better 75 ohm resistors and we see... ;)