Radeon 9800 Pro 256: Adding the LM63 Chip

[Chai]

Sorry for calling you 'nuts'. It's still a crazy attempt IMO.

I will gladly help you if you need mine. But how do I measure the capacitance?

[snipper_cr]

Hey thats a compliment

Anyways, you need a capacitance meter... and the capacitor to be measured cant be attached to the PCB...which sucks... Hopefully i can find someone who just has the data.

Supposedly using a "Bridge" you can measure a capacitor but i dunno much about that.

:: any ideas cause im lost

I may get some low level caps and work my way up... sounds like only idea to me

[snipper_cr]

Calc and Physics homework done i had an evening to myself finally!!!

Anyways i went to radio shack and picked up some .01, .1 and 1 UF capacitors.

I started with .01 and removed the resistors that wernt supposed to be on and replaced them with capacitors, .01UF. I obviously left the proper resistors on.

Well after a slight glitch...erm, ill explain later, i went back into ATI tool and things looked up for a bit.

No longer displaying the static 20/0 it was now showing 51/31 which seemd like acceptable tempatures. Strangly when ever i changed the clock speed it would sometimes read 83/63. Not sure i unplugged and called it a night, happy i made some progress. Its a lot of learning which traces go to which solder pad, which jump through the PCB to the other side, which ones are connected... which leads me to this story:

As i was removing a resistor it pulled the solder pad off... crap. I was able to catch its trace and just attach from there but it was still a scare.

However i am still at a loss for the PROPER capacitor values. Im talking with a person over@ rage3d.com and waiting a reply. Hopefully i can get the proper ones.

Im still not displaying OverDrive yet in the display adapter properties.

Also i learned that i do not -->YET<-- have fan control. I followed some traces that go to the fan port and learned there is a small micro chip i need to put for fan control. In formation on it i have no clue yet.

So thats the latest.

Anyone still think im

well ya better

[Chai]

51/31 doesn't seem right especially if you didn't mod your cooling...

[Adrian Wong]

Too bad Max_87's monitor is out. If not, he may be able to chip in..

[snipper_cr]

Working for several hours quite irritatably yielded nothing. I had several corrospondants with viperJohn over at rage3d.com, constant checking of the LM63 data sheet, solders and resolders with only the following results:

If a certain combination of resistors and capacitors are used:
20/0 in ATItool

or

51/31 in ATItool changing to 86/56 if i move the slider and set clock speed.

My next corse of action is tomorrow when im working inside my computer to pull my 9800XT out and REALLY follow traces, capacitors, resistors and such. Hopefully that could yeild some goodness.

Also the LM63 data sheet shows that pin 5/6 (the D-/D+) should be bridged wtih a 2.2uF capacitor. The highest i had was 1uF (or 33 which is WAY over). Im going to spend another 55cents to get a 2.2 uF capacitor and bridge them, although im pretty sure that that C1513 which i currently am using is that bridge.

Im actually going to remove the heatsink i have on it now cause i have a feeling i may be missing something under it all...

[snipper_cr]

I was able to do a side by side comparison between my 9800XT and the 9800 Pro 256 and took notes of a few additional parts i may need.

Other good news: I should be able to bring the fan tachometer monitored by the LM63 chip online this weekend.

After studying the information diagrams provided by national semi conductor, i was able to figure out what will be needed. A fairchild PN2222A transistor. While i was able to find an identical one at fairchild direct, i need to buy them in reels of 3000... Hey, radio shack has the TO-92 versions... might get that for 97 cents!! This will enable fan monitoring/control as voltage to the fan is regulated by the PWM section of the LM63 chip, and when absent direct voltage from the 3.3V line.

As far as getting the tempature monitoring accurate, im still working on that. It has to be something in the SMB_CLK/SMB_BUS lines that are not returning proper values. Because if the remote diode was in accurate, the local diode would still report a proper value.

Anyways, that is the latest from geek labs.
Anyone still think im

[Adrian Wong]

Woah.. Reels of 3000?

LOL! Crazy or not, this is definitely a first!

[pmg4125]

This tell me the modding is moving smoothly.

Keep it up, snipper. I may cant help you much due to lack of interest to electronics 2 years back

[snipper_cr]

New insight:

I removed the heatsink from the card in experiment and nearly shotmyself...

There were three solder pads that i had not noticed before cause it was covered by the heatsink. They are fittings for a transistor... Hope against hope i measured resistance from the head, to the D-/D+ on the LM63 chip... Resistance: 0 OHs... its a directly line not bridged.

Cant beleive i missed that!
Calc AP (BC) test tomorrow i should have some time after it but today was studying!