Power-Up Tutorial for Dead Motherboards

[bgavin]

Does anybody have a bug shooting tutorial for dead boards?

For example, start with the Berg strip connectors that provide Gnd, +5 to turn on the PSU. When these two points are made, X happens next.

If the PSU turns on and holds on, but no POST, what does that tell me where to look next?

I want to achieve a simple tutorial that will let me determine if a board is repairable, i.e. MOSFETS, or a throw-away (dead Southbridge).

[supersabre]

Re: Power-Up Tutorial for Dead Motherboards

There are not too much information in english about repairing motherboard.

However, there are many in Chinese websites. Such as

http://www.91xiubbs.com/forum-84-1.html

There are also some chinese websites hosting repair video clip, but again it is in Chinese.

I don't know much about Russian and German, you have to check them.

there are lot of things to check if there are no POST, such as crystal, PWM, mosfet, southbridge, northbridge, I/O. It is not that easy to check if you are not familiar with the power-up sequence.

The best place is to read the forum that are specialized in repairing motherboard. If you know another foreign language, you can search that and see what comes up. However, I don't see much useful information in english website that tells the user what to look at if the board is bad.

[bgavin]

Re: Power-Up Tutorial for Dead Motherboards

Dell Dimension 4550 VRM, i845 chipset.

I'd like to start a visual tutorial 101 of the VRM using these photos. I'm trying to tie this back to AKOR's vrm troubleshooting pdf file. As comments are added, I will update the photos with text and arrows.

Q: how many phases, why?
Q: where are the meter test points for basic dianostics
Q: where is the control IC (buck controller?)
Q: differentiate the MOSFETS shown, and why they have different voltages

When I have it all done, I will make it available as a tidy PDF.

[gg1978]

Re: Power-Up Tutorial for Dead Motherboards

Two phase VRM, two top FETs and two bottom FETs per phase.. VRM phase inductors are the two large red ones, the smaller green one is part of the +12V input filter.. The VRM controller IC is the HIP6302CB right next to the lower VRM phase. The little chips that are between the FETs for each phase are what actually turn the FETs on and off.. The 4 FET's that are in a vertical line, furthest to the left are the top FETs that pull the switching nodes up to +12V, the 4 FET's that are in the next "row" to the right, flanking the two Red Phase inductors, are the Bottom FETs, that pull the switching nodes to ground. The two Red phase inductors couple the switching nodes to the Vcore plane and the row of low voltage output caps and MLCC's by the CPU socket..


I prefer to probe Vcore at the inductor outputs, but that may be difficult on this board due to the close proximity of the output caps.. You can check Vcore on the inputs of the phase inductors, but there may be more switching noise there, as that's the switching node.. There are a few MLCC's left off, but the CPU heatsink will cover those, otherwise that would be a perfect place to check Vcore..

To check the FET's in operation, you would need an oscilloscope.. But if the board has no power you can check for shorted FETs...

Each pair of top and bottom FETs are in parallel, to double current load capability..

[bgavin]

Re: Power-Up Tutorial for Dead Motherboards

Excellent, thanks!

This is a tutorial, so please spell out how to test shorted D2PAKs. I don't understand MLCC. I want to do as many checks in-circuit as possible before desoldering components and chasing a wild goose.

With the PAK oriented East-West, I find the NE leg to be Output, and the SE as Input. The absent middle leg is apparently tied to the base. I measure voltage from Base to Leg. Probing a working KG7 board with a 3+3 FET layout, I found the outputs of 1 group at Vcore, the other at Vdimm. Measuring across the associated caps show the same voltages.

AKOR's Diagnostic Procedure

Referencing AKOR's diagnostic procedure (A), where in these photos would one measure for resistance between the VRM input and ATX24 pins?

[Wizard]

Re: Power-Up Tutorial for Dead Motherboards

Pull both red coils (mark them so they are put in right way around.) And measure caps and VRM for shorts. To find which one is.

Cheers, Wizard

[bgavin]

Re: Power-Up Tutorial for Dead Motherboards

And measure caps and VRM for shorts.

Please be specific as to what you mean.

For example, D2PAK: measure base to Output should be xxx ohms. Can this be done in-circuit?

If measuring caps in the board, is the measurement reliable or affected by the cap being in-circuit?

[i4004]

Re: Power-Up Tutorial for Dead Motherboards

For example, D2PAK: measure base to Output should be xxx ohms. Can this be done in-circuit?

If measuring caps in the board, is the measurement reliable or affected by the cap being in-circuit?

i have a board here where it's rather obvious fets are shorted by measuring them in circuit. 0ohms everywhere.
(symptoms: it turns on, cpu fans spin etc., but no video...just like before changing the caps...)

as for caps...hold on...ok, i'm getting 13ohm on my ms6337le5(that vaue will probably vary with vrm design). on board with blown fets it'll probably be 0, because of fets, i think.
shorted caps will also be 0ohm, but i think shorted caps are very rare.

[bgavin]

Re: Power-Up Tutorial for Dead Motherboards

Does your cpu, vrm, nb, sb get warm? Mine stay cold.

[gg1978]

Re: Power-Up Tutorial for Dead Motherboards

Try touching each of the FETs in the VRM with power applied and a CPU in the socket.. A shorted FET will get very hot.. FET's and caps can fail either open or shorted, though caps are more likely to fail open, but if the electrolyte dries out, the damndest things can happen.. I'm sure you've seen pictures of badcaps that have more than just domed their top or bottom or spewed crap.. I"m talking about the caps that blow their cans apart.... As for a shorted FET, they can kill other parts quite promptly.. This is why i keep a junker CPU of each socket type that is designated for VRM testing..

As for testing a VRM with no CPU, the VRM will not turn on if there is no CPU in socket.. The VID bits are designated pads or pins on the CPU that are either open or shorted to ground on the package.. The VID lines are pulled up to a voltage via a set of resistors, one per VID bit, and the ones that are shorted are pulled to ground, while the others remain floated high.. This results in a binary code that tells the VRM how much voltage the CPU wants.. Of course this works differently on LGA775 and perhaps other newer type boards, as the VID bits can change during system operation..

You might find googling for the VRM spec for the board type you're working on to be helpful..

PPro was VRM 8.1
Klamath and Deschutes PII were VRM 8.2
Coppermine was VRM 8.4
Tualitins are VRM 8.5
Socket 478 P4's are VRM 9.0, i believe Prescott needed a later due to power draw

Socket 775 P4's are VRM 10 or 11, VRM 11 being the latest and greatest for 45nm C2D and C2Q goodness.

On the AMD side, they follow the intel perhaps not quite to the letter, but the general idea is the same..

[gg1978]

Re: Power-Up Tutorial for Dead Motherboards

As for your Dell board, i would get the datasheet for the Intersil VRM controller from their website, and see if the VR controller is signalling a shutdown state.. Most VRM controllers are smart enough to sense a failed FET or dropped phase, and will typically shut down..

You really can't look at the gate of the FET with a DMM.. You'll get an average DC voltage.. You need an oscilloscope to actually see what the VRM controller is telling the FETs to do.. With the system off, and all power disconnected from the board, all FETs should measure open from source to drain.. Since the FETs are connected in parallel on your board, having one bad FET in the phase will make the other FET in parallel with it appear to be shorted also.. I doubt you have a shorted top FET, as a shorted one of them will probably make the CPU burn up, and the Power supply should switch off immediately if the VRM is squeaking. As you said in your other post, you may hear some noise when the VRM is under heavy or fluctuating load, but not at idle...

I'm guessing your short is after the VRM phase inductors.


Also check for a short from Vcore to ground.. From the picture, it looks like the component designator L3 should be at Vcore.. Vcore is the massive green pad that goes from the line of VRM output caps to the CPU socket.. You can see the row of MLCC caps right next to the CPU socket also..

[gg1978]

Re: Power-Up Tutorial for Dead Motherboards

Also, a port 80 PCI test card is useful.. You can see if the board is coming out of reset, and if the BIOS is attempting to execute.. Even a board with bad caps in the VRM, but otherwise OK, can typically pop the CPU out of reset, and perhaps start executing the BIOS..

[gg1978]

Re: Power-Up Tutorial for Dead Motherboards

The VRM controller chip data is here:

http://www.intersil.com/cda/devicein...2,00.html#data

I'd take a look at pins 1 through 5, which are the VRM VID bits, pin 16, which is Vcc for the VRM controller, pin 15 which is Power_Good, the signal that indicated that Vcore is at it's correct value and stable, and pin 10, which is where the VRM controller senses Vcore from the microprocessor..

I believe in your other thread, you mentioned that you measured some shorts in the bottom VRM phase.. Have you confirmed that??

[bgavin]

Re: Power-Up Tutorial for Dead Motherboards

I don't know if I have shorts or not. The left four behave one way, the right four another. They are uniform. Now that I'm armed with more tasks, I will do them Thursday when I can do more testing. The next three days are on a contract, and I'm away from my test bench.

None of my FET are warm. Nothing is warm. The board does indeed require a processor in the socket to power up. Doh. I knew it had to read the VID bits.

This came about when I recapped the four 3300/6.3v caps. I will simply pull those out and meter them all. I will pay attention to the DVM negative lead on the stripe side... KC8 pointed out the possibility of a reversed jacket. If I have a bad cap it will be the first MCZ I've hit with this problem. It makes sense, board died after recapping. One or more must be bad. KC8 says the singing FETs are indicative of an output short.

[coolday]

Re: Power-Up Tutorial for Dead Motherboards

Admin's please continue this thread ... I've done a course on motherboard repairing but I'm a rookie ... However this thread can be helpful for me ..... I've also tried searching for books on motherboard repairing ... but couldn't find a single one

Thanks anyway

[coolday]

Re: Power-Up Tutorial for Dead Motherboards

There are not too much information in english about repairing motherboard.

However, there are many in Chinese websites. Such as

http://www.91xiubbs.com/forum-84-1.html

There are also some chinese websites hosting repair video clip, but again it is in Chinese.

I don't know much about Russian and German, you have to check them.

there are lot of things to check if there are no POST, such as crystal, PWM, mosfet, southbridge, northbridge, I/O. It is not that easy to check if you are not familiar with the power-up sequence.

The best place is to read the forum that are specialized in repairing motherboard. If you know another foreign language, you can search that and see what comes up. However, I don't see much useful information in english website that tells the user what to look at if the board is bad.

I've translated the site using google ... but how to register ... Chinese pops up nearly everywhere

[coolday]

Re: Power-Up Tutorial for Dead Motherboards

please update more tutorials on motherboard repairing

[coolday]

Re: Power-Up Tutorial for Dead Motherboards

I want some more like AKOR's Diagnostic Procedure

[precision001]

Re: Power-Up Tutorial for Dead Motherboards

I have started a thread on laptop motherboard repair. Hopefully, skilled technicians will contribute to it. From what I have learned about "dead" laptop motherboards so far, there is a chip - a maxim chip - that controls power to the board. I don't know much about this chip yet, but want to test it. Also, I want to learn some standard testing to do with a multimeter to test critical components such as power mosfets (8 pin, IC type). I have a 100 mhz scope I sniped off ebay - but so far I have only been able to use it to test crystals.

[precision001]

Re: Power-Up Tutorial for Dead Motherboards

Anyone know what might be killing mosfets on an HP DV9000 motherboard? I've replaced the Maxim 8774- thinking that was behind it, but I was wrong.