x800 Pro Vivo needs rescue!

[Marlowe]

Okey I had a pretty bad soldering session today.. Was going to replace the resistor I earlier put on for the VGPU with a variable one.. Ended up unsoldering the frickin' resistor on the pcb. Thought I could manage with soldering the VR with correct value straight on the IC.. So I did that. At the same time I connected some wires for VDD and VDDQ measurement at those capacitors.. something that somehow proved extremely difficult, 'cause the solder wouldn't melt when first applied.. so I heated up those capacitors PRETTY bad..

I don't know what might cause it, but inserting the gfx in the mb causes the computer to turn off after 0.5 seconds, and then not be willing to boot without resetting the psu/motherboard.

I would really like some advices! Thanks

[CaTalyst.X]

Set the 10K trimmer to about 5K and get back to me

-CaT

[Marlowe]

Hmm Ok.. I'm actually having it at about 420 ohms at the moment, since according to techPowerUp that is the original value of the resistor.. I've soldered at pin 3 and 4 from the bottom left on the IC by the way. But trying 5k now.. :) Thanks

[Marlowe]

Well, the computer powers on at least, but I'm afraid the card is dead.. There is no picture and the Asus bios lady repeats "system failed cpu check - system failed cpu check". I can't make out if it's maby GPU she says.

I measured the VGPU by the way.. and guess what.. 0,00 volts.. :(

[Holst]

Sounds like you fried the VGPU circuit.

It might be repairable with some messing about.

Can you take a pic (or is your soldering too embarasing)

[Marlowe]

Hehe a picture is no problem! I would have done it already, but I don't have a camera at the moment. I will fetch one from a friend today if I'm lucky.

I'll report back soon!

[Marlowe]

OK, the picture lies! It's not really that ugly.. :D

http://img.photobucket.com/albums/v205/graastein/DSCN5001mod.jpg

See also
http://img.photobucket.com/albums/v205/graastein/DSCN4999mod.jpg
http://img.photobucket.com/albums/v205/graastein/DSCN5000mod.jpg

[persivore]

Can you check the resistance between pin 17 and your Vgpu read point, and pin 17 and ground, and check if the resistance between pin 17 and ground relates to your varaible resistor value.

[Marlowe]

Looking at that picture.. The R1597 is gone, and so is one of its pads. When the resistor came loose, the pad was stuck on it, and so did a small copper wire that you see sticking out of the R there. What I am thinking is maby there should be connnection between that wire and one of the two points up to the right, a connection wich I maby broke when braking loose that wire..? I could try to reestablish the connection there.. What do you guys think?

[Marlowe]

Can you check the resistance between pin 17 and your Vgpu read point, and pin 17 and ground, and check if the resistance between pin 17 and ground relates to your varaible resistor value.
Pin 17 would be the one I soldered to, right? I will do that in a moment! Just have to go home again from visiting my folks ;)

[persivore]

Looking at that picture.. The R1597 is gone, and so is one of its pads. When the resistor came loose, the pad was stuck on it, and so did a small copper wire that you see sticking out of the R there. What I am thinking is maby there should be connnection between that wire and one of the two points up to the right, a connection wich I maby broke when braking loose that wire..? I could try to reestablish the connection there.. What do you guys think?
I needed you to check the resistance between the points I said to confirm what you just posted. From that, we can work out which resistors have been damaged/disconnected, and work out the right place to solder new ones to :)

Edit: Sorry, I didn't see the second post...pin 17 is the pin that you soldered to.

[Marlowe]

Hehe sorry for the confusion, I didn't see your post when posting my first! :D But as you asked, I just measured and some things are much clearer now:

As CaTalyst.X suggested earlier, the resistance between pin 17 and 18 is set to 5,00 kΩ.
The resistance between pin 17 and the vgpu read point is 5,57 kΩ.
The resistance between pin 17 and ground is also 5,57 kΩ..
and the resistance between the vgpu read point and ground is approximately zero! :eek:

There I have one very obvious error. This explains the "short circuit feeling" I had earlier when the psu shutted off after a few seconds. I guess the next step will be to find where that short circuit has appeared!

That must be later tho, since the lady is coming over any second now.

Thank you very much for the help so far!

[CaTalyst.X]

Hrmm i think that is fixable. If you cant get her done, sent it to me and ill do it for cost of return shipping.

-CaT

[Marlowe]

Hehe that is very kind of you, and I will consider that if I can't do it myself :)

However to send it away would be defeat and a loss to me if I know I can do it myself! I'm sure you all understand that :)

Do you guys have any suggestions or is this secrets of the trade ? ;)

[CaTalyst.X]

Ohk try soldering a 470 Ohm Fixed Resistor from pin 17 to the left side of R1592. Then solder a 10K VR set at max in parallel with that resistor.

-CaT

[Marlowe]

That will result in 449 Ω. The 10 k VR can be set to that.. Why the fixed resistor?

[CaTalyst.X]

Using the fixed resistor because in the normal mod, you solder a 10K vr in parralell w/ R1597, and now you just have the VR. Adding the Fixed Resistor might fix it :P Just give it a try.

-CaT

[Holst]

I think the circuit is probably more complex that you realise.

If you had two cards this would be a relatively "simple" fix as you could just continuity check everything.

Can you link to the vmod guide you used please.

[Marlowe]

I agree with you, Holst. If I had a similar card, I could check on that and on mine where there should be a connection and where there should be some resistance value. Or did you think of any other method?

I could also measure the capacitanse of those capacitors in the VDD and VDDQ circuits too, just to check. That would eliminate that area as the problem totally and I could concentrate on the VGPU circuit. (that is if I didn't heat up any other nearby components too bad, but I don't think so)

The friend I borrowed the camera from earlier happens to own a Powercolor x800 Pro Vivo just like mine! So I might be very lucky there! I will ask him very kind today.. and maby later today I will be able to do the measurements! :D

I used the vmod guide over at techPowerUp! (http://www.techpowerup.com/articles/overclocking/voltmods/101), and the part that concerns us is this following picture. The only difference is that the R1597 is gone, and the 10k VR is connected directly to pins 17 and 18 on the IC, where both sides of R1597 had a direct connection to. But obviously something is not as it should be..!

http://www.techpowerup.com/articles/101/images/vgpu.jpg

[Marlowe]

Okey I got hold of the card and did some measuring.. things looks clearer now! Here are the measurements: Working card first, then mine:

R1597 measured to 415 Ω .. my vr set to same
vgpu read out to ground : 4,0 Ω .. 3,5 Ω
Pin 17 to vgpu: 4,0 Ω .. 988 Ω (!!) :)
17 to ground: 0,4 Ω .. 991 Ω

C213 : 1670 Ω .. 1620 Ω
C215 : 17,1 Ω .. 17,1 Ω

The 'C' parts were not capacitors as I first thought, but regular resistors. They both looked very fine, so the problem is not there. The resistance between Pin 17 and the vgpu readout is where we think the error is. According to the data sheet (http://www.fairchildsemi.com/ds/FA/FAN5240.pdf) of the 5240mtc IC, pin 17 senses the output voltage and regulates according to that.. I might have broken the connection there and this can be causing the problem I think.

What I suggest is to connect Pin 17 and VPU readout. We think this should solve the problem. But I want to hear you guys' opinion before I do it :)

Thanks a lot!

[CaTalyst.X]

Lol if they are C parts, they are capacitors, They might read out as having that resistance because they are still soldered in circuit. None of those resistance values really mean anything except to compare where there should and shouldnt be some type of resistance. You would need to unsolder R1597 to get an accurate readout.

-CaT

[Marlowe]

Those C parts are in the VDD and VDDQ power circuits and don't have anything to do with R1597 :) But yes, they might be capacitors, but it doesn't really matter.

What do you think about my suggestion? I'm sitting here extremely impatient, ready to heat up the iron :D

[Holst]

I dont understand why people dont do a VID mod on this rather than this ugly vsense one...

Anyhow thats not all that important.

Here is the PDF for the fairchild 5240
http://www.fairchildsemi.com/ds/FA/FAN5240.pdf

Pins 17 and 18 are what you have been messing with.

If we look at fig 1 in the PDF you can see that 17 and 18 are not indipendant, they both connect to vcore (17) and ground (18)
Resistor R sets the feedback.

As I dont have a card, and reading the traces is awquard you cant say for sure if your card is set up exactly the same as in the PDF, its probably not. But the basic operation will be the same.

So what am I getting at...

Basically you must have broken pin17 away from ground AND possibly from the caps as well...

Soldering 17 to ground will "probably" make it work. But I would want to make sure that the caps between 17 and 18 are still intact as well. if they are not grounded you may get instability, or it might not work at all.

Difficult to say exactly what to solder to where without having a look at the card mysefl.

But you should aim to get everyhting exactly the same as your friends card.

[Holst]

Can you please check continuity between thost caps and Vcore and ground at both ends please.

If they are the same on both cards then you are ok (i suspect they will be)

If they are not then they may be routed through the damaged trace, and will also need to be repaired.

[Marlowe]

I cincerely thank you for the time and at the same time I say sorry, for it has all been waste.

I have just ruined the card totally.

I am very sad and very sorry.

Thank you all for your time.

[persivore]

I think that the feedback path between the feedback pin and the regulator output has become...detatched. Try connecting a 1k fixed resistor between pin 17 and the Vgpu read point, and leave the varaible resistor connected as it is atm (between pins 17 and 18). This will probably give you a very low Vgpu, but it should tell you if the Vgpu regulator is still alive and you should be able to adjust the votlge by trimming the varaible resistor.

Edit: Sorry, I didn't see your last post. Sorry to hear that dude :(

[Holst]

You have a PM, dont give up just yet :D

[Malves]

Marlowe,

Listen to Holst, bro. Don't give up, yet.:)

Holst,

Good to see you back in action. Heh, I still remember first time I sent you a PM asking how to work vmods out.:)

[Marlowe]

Both the pads under R1597 is gone. Tried to fasten the variable resistor directly to pin 17 and 18, but without success.. it's too tight. Got some solder over some ic pins.. while trying to take it away, my "solder vakuum" device hits downwards when pushing its button and BRAKES pin 19 and 20 on the ic.. Tried to bridge them with solder whitout success.. just leaving solder between the pins wich I cannot take away.. This was too small for my equipment and my skills.. I can only fix it if I have a microscope and can build bridges with my thoughts!

I'm very sorry for dissapointing you great guys with my terrible skills.

[Holst]

Hehe Malves, now you have probably done more mods than me :)

You have another PM marlowe, it can be fixed. (probably :P )