Shuttle SD36G5M - Mobo died, recapped, still issues

[racer2086]

Hi everyone,

New to the forum. Great wealth of knowledge on here, wow! I'm hoping some of you can help me out. This is gonna be long as I throw out my back story here.

I have an older Shuttle XPC SD36G5M that I bought off ebay for cheap a couple years ago. I use it as a home theater computer and has been running fine until 2 weeks ago. It went to sleep and never woke up again. Pushing the power button would get the fans to jerk and that was it.

I bought a new power supply figuring that was the problem, but had the same results. That's when my internet searching brought me to this website. There was one OTC cap that was visibly bulged on top. So I decided to recap the whole board.

The first thing I did was replace the bulged cap and see if that would fix things. The fans spun for about 1 second this time before the board would turn off. Then I recapped everything else and sure enough it fired up and worked!

Then, again, the computer went to sleep after some online streaming, I woke it up and shut it down and it hung at shut down. Ever since, it has been freezing whenever doing something somewhat "intensive," such as opening Google Chrome, installing Windows, or running a memory test (freezes at 10% every time, no matter what stick or how many are installed). It would also hang at one point just copy pasting files. Something is unstable, maybe RAM? It will stay in the BIOS no problem. It also seemed to have no issue sitting on the Windows desktop without anything running.

What I did notice is that the whole side of the board where all the caps are and where the PSU plugs in gets very hot. Abnormally hot. So hot that all my new caps get very hot to the touch. Today, I turned the computer on from being off for a day and felt around. I found one SMD Mosfet(?) that was so scorching hot it burned my finger (right next to the first cap that was bulged and replaced). Not long after having the system up, all the adjacent caps got pretty hot. Heat radiated from there and seemed to slowly heat everything on that side up.


Now for all my questions...So from reading it's possible the mosfet is toast. Would that happen from the cap going bad? Is the hot mosfet the actual problem now or something else causing it to draw a ton of current? And if I replace it, what are the chances something else is damaged and the system will still be unstable?

Lastly, I replaced all caps with Nichicon. After reading some more, my concern is with the ESR. Most are HV, some are HM, and some are HE Nichicon caps. Are those acceptable or part of my problem?

Thanks so much for the help!

[RJARRRPCGP]

Re: Shuttle SD36G5M - Mobo died, recapped, still issues

I don't even consider installing Windows nor running Chrome to be "intensive", unlike say, Prime95, Linpack or x.264. If you can't even complete the Windows installation, then it's xtremely unstable!
No more stable than Charles Manson!

[piernov]

Re: Shuttle SD36G5M - Mobo died, recapped, still issues

What is the motherboard model number?
Send a picture of the area that gets hot.
The MOSFET could be shorted or there may be a short further down the line, it depends on what the circuit does.

[racer2086]

Re: Shuttle SD36G5M - Mobo died, recapped, still issues

I think its the VRM section. I'll upload a pic shortly. Motherboard is FD36 v1.1

Lol yeah loading windows isn't "intensive" but it's more than idle, which it seemed fine doing. For the most part he computer boots and will load Windows, so I'm hoping that means most things are fine and there is just some small part still burned out.

[racer2086]

Re: Shuttle SD36G5M - Mobo died, recapped, still issues

Ok here are 3 pictures progressively more zoomed in to give you an idea where this is on the board. In the first, I circled the area. The second picture has the arrow to the cap that was originally bulged and the chip that is very hot. Third is circling something coming from under the chip that just doesn't look right. It looks almost metallic with a flashlight like solder, but not sure how that would be.

BTW, the cap next to the chip there is intentionally angled, since the Nichicon caps were slightly wider than the originals, so it was hard to fit everything right down on the board with so little room to spare.

I may replace the caps right around there again because of the heat they experienced. I don't know if they got damaged. I have the burn on my finger to prove how hot that damn thing got in less than a minute of being turned on.

[Agent24]

Re: Shuttle SD36G5M - Mobo died, recapped, still issues

That's a MOSFET. The metallic bit underneath it looks like where the solder has started melting and coming out! I'm not surprised you burned your finger on it. It should not be getting that hot.

Inspect the board where you worked in it and be sure there are no shorts, for example some solder could be bridging pads.

Otherwise, that MOSFET or another connected to it could be damaged or the controller driving it could be bad, or both.

[racer2086]

Re: Shuttle SD36G5M - Mobo died, recapped, still issues

So, I tested the MOSFETs according to this post on this forum: https://www.badcaps.net/forum/showpo...4&postcount=19

Q53, the one with solder under it, measured the following:

Pin 1-2 = 16.2 ohms
Pin 1-3 = 14.6 ohms
Pin 2-3 = 1.8 ohms

Q54 next to it:

Pin 1-2 = 27.5 ohms
Pin 1-3 = 3.6 ohms
Pin 2-3 = 24.9 ohms

According to the MOSFET testing post, these are shorted. I tested identical MOSFETs on other parts of the board and they measured in the Mega Ohms with one pin combo measuring around 30 ohms. So am I safe to assume that these two things absolutely need to be replaced? And am I safe to assume that they are the root cause of the system instability and heat?

[Agent24]

Re: Shuttle SD36G5M - Mobo died, recapped, still issues

Something does not sound right with those readings, yes, you should have similar readings to the other MOSFETs on the board, but it's harder to analyse the results without knowing which pins are which.

Can you list which numbers you used for which pins? (Gate, Drain, Source)? Did they match those in the attached datasheet?

One of those MOSFETs should have the Source pin connected directly to ground. Is it Q54 or Q53? Also, one should have its drain connected directly to the 12v rail. Which one is that?

[racer2086]

Re: Shuttle SD36G5M - Mobo died, recapped, still issues

Yes I looked up the datasheet for it. My pin labels match the datasheet. Pins 1, 2, 3 left to right. Pin 1 = gate, pin 2 = drain (which is just a stub going to no where on these, but is continuous with the large pin 4 drain on the datasheet), pin 3 = source.

So it seems for Q53, pin 2-3 (source to drain) is at 1.8 ohms or practically continuous. For Q54 pins 1-3 (gate to source) is at 3.6 ohms, which seems to indicate a short between the gate and source, no?

I'm not exactly sure how to tell which is hooked to 12v. But I figured that pin 11 of the 24 pin connector being 12v would be a start. I used the common leads of the 4 pin power connector to test ground.

Drain of Q53 (the large part soldered to board) is 0.2 ohms to 12v. The Source (pin 3) of Q54 is 0.2 ohms to the common lead of the 4 pin ATX connector - ground.

Worth noting, the drain of Q54 and pin 3 of Q53 are only 1.8 ohms from 12V. That seems very low?

I wish I understood this better to provide you some better info.

[Agent24]

Re: Shuttle SD36G5M - Mobo died, recapped, still issues

I think you understand it quite well actually. Your info is good anyway. OK. So Q53 will be the high side MOSFET, and Q54 the low side.

At the very least, It seems Q53 likely is bad. The low side MOSFET is harder to check in-circuit because the CPU or chipset presents a low resistance to ground in parallel, but a dead short will of course be obvious. The 24.9 ohms you get there would likely be normal, but the readings show what looks like a short to the gate also, so Q54 is probably bad too.

I would remove both MOSFETs at this point and re-test them out-of-circuit, to see if any readings change.

[momaka]

Re: Shuttle SD36G5M - Mobo died, recapped, still issues

And if any of these MOSFETs are bad, I would also suggest replacing the PWM controller that drives them.

[racer2086]

Re: Shuttle SD36G5M - Mobo died, recapped, still issues

Which part is that?

[Agent24]

Re: Shuttle SD36G5M - Mobo died, recapped, still issues

Which part is that?

I've circled a controller in green on this image. You would have to trace out the connections to see if it's the right one for those particular MOSFETs.

[racer2086]

Re: Shuttle SD36G5M - Mobo died, recapped, still issues

I have the Mosfets on order from eBay from China :/ can't get them in the US. Hopefully they are real.

Can I test that controller somehow? Or should I just order one of those also and replace it?

[momaka]

Re: Shuttle SD36G5M - Mobo died, recapped, still issues

I have the Mosfets on order from eBay from China :/ can't get them in the US. Hopefully they are real.

Can I test that controller somehow? Or should I just order one of those also and replace it?

If you purchased only one set of replacement MOSFETs, you might want to replace the controller then, just to be sure it doesn't blow your new MOSFETs. Otherwise, you'll be waiting on a new order of FETs to arrive (unless you feel like kludging something together with any other MOSFETs you may have on hand, which may not turn pretty).

That said, unfortunately, there is no test that you can do on the controller to tell if it's good. I've had seemingly good controllers blow several MOSFETs before I wizened up and replace them too. Only then did I manage to take care of the issue.

However, looking at the picture of the controller that Agent24 highlighted above, replacing that controller is going to be a pain in the butt due to being SMD. You will definitely need hot air station for something like that. If you don't have one, then just install the new MOSFETs and pray that everything works.

And who knows - it just might. I've had PWM controller blow roughly only about 30% of the time when a MOSFET blew up on a motherboard or a video card.

[racer2086]

Re: Shuttle SD36G5M - Mobo died, recapped, still issues

The FETs I ordered came in a 10 pack, so I can definitely just give it a shot. I don't want to damage anything else though, but it seems like just the FET will blow.

I do have a hot air station, so I could replace that piece. I'll look up the part number to see how easy it is to get.

Thanks for the help so far, everyone!

[racer2086]

Re: Shuttle SD36G5M - Mobo died, recapped, still issues

So, I replaced the two mosfets that were in question. There was some separation of the board under the fet that got really hot. I re-epoxied the top layer back down and it seemed like there was no affect on the traces. Everything seemed ok.

So when I first powered it up, I got the XPC splash screen, then the no keyboard error before the computer turned off. Now it just power pulses it seems. The motherboard LED blinks on and off and you hear the fans speed up and slow down.

So, I have no idea what is wrong at this point, and it's beyond my knowledge. Maybe the components I got are bad. Maybe something else is blown out. I just want to get a good replacement board, recap it, and then be able to use my computer again.

Anyone know where I can find one? Nothing on ebay seems to match.

[momaka]

Re: Shuttle SD36G5M - Mobo died, recapped, still issues

So when I first powered it up, I got the XPC splash screen, then the no keyboard error before the computer turned off. Now it just power pulses it seems. The motherboard LED blinks on and off and you hear the fans speed up and slow down.

Does the power supply fan pulse as well? If so, that would indicate your MOSFETs may have shorted again or something is drawing excessive current.

So, I have no idea what is wrong at this point, and it's beyond my knowledge. Maybe the components I got are bad. Maybe something else is blown out.

Like mentioned before in the thread, it's quite possible the PWM controller could have gone bad with the MOSFETs (or caused the MOSFETs to go bad if it went bad first), which is why we recommended you replace that too.

Of course, it could also be something else that is blown too that is causing those MOSFETs to go bad / overheat (I'm dealing with a similar problem right now on a Radeon HD3870 video card that had a shorted MOSFET).

I just want to get a good replacement board, recap it, and then be able to use my computer again.

Anyone know where I can find one? Nothing on ebay seems to match.

If I find one, I'll let you know. But so far, it looks like Shuttle motherboards are quite hard to find. I've been looking for one for my Shuttle XPC FB83, which also suffered a blown FET (Northbridge VRM) due to badcaps. I fixed the blown FET (and VRM controller), but the PC was dead. The only few Shuttle motherboards I saw on eBay were priced beyond the moon and not even guaranteed to work or had a ton of bad caps (seriously, ???). On top of that, they weren't compatible with my Shuttle either. So I shoved that whole computer in storage and used various bits and pieces out of it for other builds.

That's not to discourage you from looking, but I do want to give you a fair warning of what to expect. Besides, if your Shuttle mobo appears to boot at all, you may have a better chance at fixing that than finding a replacement board.

[racer2086]

Re: Shuttle SD36G5M - Mobo died, recapped, still issues

Just to finish off this thread, I ended up just buying a whole Shuttle SG33G5 computer off eBay for $50. Components and all. It's the same "newer" model as my "daily driver" Shuttle XPC I built years ago, which has a motherboard they marketed as having an "all solid capacitors motherboard design". Not sure what that means, but all the caps on the board are those all-metal-can caps. Knock on wood...mine has been running flawlessly for 10 years.

So I swapped that newer board, Core2 CPU, and RAM over to the HTPC case with bad board, which essentially is now a massive upgrade over the older board and Pentium D chip. I put the dvd-rw drive into my current daily computer as an upgrade and added the card reader that came in it to the HTPC for kicks.

Now I have an old WD hard drive, an extra video card, and an extra case with power supply left. If anyone wants the case let me know. It's just the metal frame with no face.

[momaka]

Re: Shuttle SD36G5M - Mobo died, recapped, still issues

Do you still have the old dead motherboard by any chance? I am a bit curious to try to fix/play with it and hopefully have something to put into my Shuttle case, as I think my motherboard is worse.

As for your empty case... flip it on eBay for a few $$??