Galax 3090 bad duty cycle on mem phase

[pedrolopesbe]

I got this 3090 wich I own, same board as the common 388_50 boardview.
I'm new to SMD work, and in a moment of stress trying to solder a 0402 filter cap that I've knocked out I've made a big solder mess in the "U42" drmos.
As past is past, after getting a pre heater now I succesfully unsoldered the chip but I'm having problems soldering it back again.
At the last time I've worked on it, I had to reflow it as it was not giving any out signal, measuring using the Hz function in my multimeter.
Now I get a reading of something like 1.7kz while other phases 3.09khz if I'm not mistaken.
Voltages and PWM seems to be getting to the chip fine, I could measure again to be sure, but it's the same chip that controls the other ones, right?

One little thing: I can't measure the 27MHz crystal, don't know why... (I'm using a needle tip attachment, but the phases measure fine)
The card boots and gives video, but artifacts when driver is loaded.

Looking for some guidance, where to look at...

Thanks

[piernov]

Phases are switching at several hundred kHz. Look at the spec sheet of your multimeter. Need an oscilloscope.

[pedrolopesbe]

Phases are switching at several hundred kHz. Look at the spec sheet of your multimeter. Need an oscilloscope.

My multimeter reads up to 30mhz, its a hikari 2090
I'm not buying an oscilloscope right now, if I really need one I could rent it.
But I don't think one would make or break this repair, right?

[pedrolopesbe]

I've tried soldering a new drmos in place.
Phase measure fine but still getting artifacts, didn't measure it again to see if the phase died again (the past time I've had a bad hz reading, the card booted but then the phase died, so its an option.)

[ktmmotocross]

I've tried soldering a new drmos in place.
Phase measure fine but still getting artifacts, didn't measure it again to see if the phase died again (the past time I've had a bad hz reading, the card booted but then the phase died, so its an option.)

dont bother u have bad core. if u have bad memory 3090 will not give picture

[pedrolopesbe]

dont bother u have bad core. if u have bad memory 3090 will not give picture

bruh, I hope not. The card didn't show any problems before the shitshow I've made.

I'll be sending this board to a professional when the guy responds me.

[pedrolopesbe]

Oh, something I didn't mention before: I had errors on all mem bank other than C and D if I'm not wrong while the phase was out. With the phase I get no errors but any test that stress the card fails. Not sure if it would help tho.

[ktmmotocross]

what u write dont make sense. better measure it again with scope

[pedrolopesbe]

what u write dont make sense. better measure it again with scope

The gpu phases reads 3.09khz, mem reads 3.89khz, what doesn't sound right about it?

For renting a scope I'd need a valid enterprise registration and having problems with mine right now. If there's the need of one to be sure, I'll just send the card to someone that knows what he's doing 😫

[pedrolopesbe]

Something came through my mind: Maybe, some other mem phase got shorted while running the card with one less phase.
I've tried the short finder method dicky96 teached in his youtube channel. (https://www.youtube.com/watch?v=P_2GGNr4q1s)

There's these test points for the DRMOS:



Measuring the U17 DRMOS gate test point to the inductor showed 1.13V drop while the three others 1.33v drop. I'll be replacing this DRMOS in hope this may solve my problem!

I supose the other mem phase that received 12V from the psu got overwhelmed and got shorted or damaged. (there's 4 phases, 2 get power from pci and two from psu)

I'll be back for updates.

[pedrolopesbe]

I'm back for updates!

Well, for a change I did fucked it up again lol. The new DRMOS I've installed in the other phase now doesn't work as well, nice

I've tested the frequency again and found out that the phase that I've said was working had a unstable frequency and wrong duty cycle compared to the other 2 mem phases on the pci power.

Seems like I gotta get my soldering together, this card ain't going to repair it self!

I wonder why the one drmos I've "successfully" soldered is giving out the wrong signal.

I think you guys can agree with me that the problem won't be in the controller, correct?

Short finder still shows a lower reading in the U17 drmos I've mentioned above.

[ktmmotocross]

check input signal. u can have bad controller