Need help with Shuttle SB83G5A - fans spin but that's all

[Dan81]

Re: Need help with Shuttle SB83G5A - fans spin but that's all

That said, I don't think I've ever had a successful nVidia video card reflow yet. Seems that whenever I do the full reflow on them (that is, up to proper temperatures), they just die or become worse. With ATI on the other hand, it's completely the opposite: if I didn't heat them to at least 210C, high chance they wouldn't change. Only after reaching proper reflow temps I'd sometimes get a working card... but chances so far have been about 50/50 with those.

Blasting them with 350*C or higher probably fixed both the underfill and the substrate, making them connect again. I don't know exactly, but I can tell you that it usually worked first try. One of the many 8400GS (G86 core, Gigabyte OEM btw) I had was still working, but was artefacting badly. Blasting it with the heatgun totally fixed the problem. Same goes for a 7600GT which was used as a test GPU at one time before it finally went into my C2D E6750 rig. (Abit IP35 Pro )

Speaking of which, I remember having a GT440 in here that I might have to "bake" (blast at 550*C for about 3-4 minutes actually). Maybe I'll give it a go and see how it fares after heating it that much. It's got poly caps so I don't really have to worry about them much. Cooling, on the other hand, is going to be quite a problem, as the original HSF was pretty wimpy.

[diif]

Re: Need help with Shuttle SB83G5A - fans spin but that's all

Funny you mention those. We had them in the repair shop I worked many years ago. Seemed to do more damage than good, as they were painted black and expanded quite a bit by themselves - a lot more than copper or the PCB material itself. At least in my experience, once we switched to them, I saw a decrease in successful reflows and noticed more board warping than before. So I tend to shy away from these. Leaving a board on un-tightened rail mounts so that the board is free to expand as it pleases with the heat is better IMO.


I use a microwave/oven cooking grille on which I place the board. The metal bars are thin and spaced about 0.5" (~1.5 cm) apart, so they provide decent support.


Thanks!
Nothing much still, as it's only been a month in. Just supporting the systems on a local toll road here, which mostly entails replacing coin machines and resetting/tweaking traffic controllers. Driving around that road is probably the biggest time sink - I probably clock in a good 25-100 miles per day, not counting driving to work. So yeah... lots of driving.

That's interesting to know. I saw lots of those frames being sold and just assumed they were sold for the right reason. I guess it's similar to the spade sellers of the gold rush era, they don't care if you find gold or not, they just want to sell spades.

Sounds like it's an interesting job, variety interspersed with driving. As long as you like/don't mind driving.

[Retro-Hipster]

Re: Need help with Shuttle SB83G5A - fans spin but that's all

Thanks!
Nothing much still, as it's only been a month in. Just supporting the systems on a local toll road here, which mostly entails replacing coin machines and resetting/tweaking traffic controllers. Driving around that road is probably the biggest time sink - I probably clock in a good 25-100 miles per day, not counting driving to work. So yeah... lots of driving.

That totally sounds like the kind of thing I would love to do.. It might be a bit monotonous after a while, but I'm one of those guys that does dumb projects like taking old tungsten fairy lights and converting them into LED/neon lights. lol I'm sure that sort of thing would be right up my ally.

Alas, I have already built a good career as a network admin and feel too much a pansy to try to move towards something new. ahaha (Though, I might do Java programming for the company instead eventually. <3 )

[momaka]

Shuttle XPC SB83G5A - finally fixed!

Alright ladies and gentlemen. After this project sitting on the back burner for over a year and some contemplation going on in the back of my head about how to proceed, I finally came up with a “temporary” solution :roleyes: … that may actually end up being permanent… if it ends up works permanently, anyways. LOL! I guess, we will have to see.

Basically, I did some risk evaluation, and thought the chances of messing up the board worse than it is are really high if I try to do a CPU socket reflow (let alone a socket reball.) So I went down in the garage and started tinkering with how to exert downward pressure on the CPU area to keep the board working. It took me no more than 10-15 minutes before I found some completely scrap materials that actually looked like a perfect fit for the contraption I had in mind. (Gosh! I’m such a hoarder. But at least I do eventually find a use for much of the scrap junk I keep around.) Namely, it was two pieces of wood and a threaded rod (1/4” ?) In fact, one of the wood pieces and the threaded rod were both already the length I needed. So all I needed to do was just cut the other piece of wood, drill two holes in it, and then widen an existing hole in the Shuttle case for one of the screws. The work took me maybe 20 or 30 minutes total. After that, all that was needed was a nut that fits the threaded rod and a washer - these were probably the only two “new” parts I used here (though I bought these packs of washers and nuts probably close to two decades ago.)

Anyways, here is how the “fix” ended up looking like:



Essentially, the horizontal wood piece is held by a screw on one side of the case, and the other is just wedged against the case frame. Meanwhile, the threaded rod has the nut and washer on the bottom of the horizontal wood piece. So as the threaded rod is tightened, it starts to push down on the vertical piece of wood, which then applies pressure over the CPU heatsink. This warps the board and enables whatever BGA is broken down there under the CPU socket to make contact again. Yes, it’s a complete jurry-rig setup, but it works. And if that’s not bad enough for you, notice that the threaded rod has a notch cut at the top. Why? Because I couldn’t find a matching bolt and nut with the length and thickness I needed. So when I decided to use the threaded rod, it had no head on either end. Thus, to make it “turn-able” I cut a notch at the top so it could be turned with a flat heat screwdriver. But you know what? Who cares, as long as it works… and it does.

When I first tried this contraption, I started tightening the threaded rod to see how much pressure it can exert on the board. Needless to say, the relatively fine thread on the rod allowed me to warp the board easily. At first, I put a good amount of pressure on there and tested the motherboard - it booted no problems first try. But eventually, I decided to play it the other way around: loosen the threaded rod all the way and slowly start tightening until the board starts POSTing. Did that, and found the board barely needed any pressure with the threaded rod to start POSTing. So I plugged the HDD with the original OS that the owner gave me and gave it a spin.

After over 8 years in storage, the Windows XP Pro SP2 on that HDD booted fine without a hitch. However, I found the GPU drivers were missing / never installed. Easy fix, of course. Next, I found a big library of music on the desktop and just started browsing it while also playing stuff. 2 hours later, and the PC was still running rock-solid - no crashes or anything like that.

The next day, I decided to check out some more music from that library. So I pressed the power button and the PC booted again. But then after a minute into the desktop, the screen blanked and CPU fan spun at high speed again. :\

Was my victory short-lived? Did I kill it?

I turned off the PC, took out my trusty flat head screwdriver + wrench and advanced the threaded rod about ½ to ¾ of a turn forward. Then I pressed the power button again and… hey, no more problems, amigo! The motherboard booted fine once again. After a bit of use, I tried turning off power to the PC, let it cool down, and then powered it back up later to see if the issue would re-appear with a cold boot once more. However, this time it didn’t. So I think my fix (jurry-rigged setup) works pretty well.

With that said, here are some CPU-Z screenshots of the motherboard and CPU






And the i915 onboard GPU:


While at it, I thought some of you might find it a bit funny what the two dubiously-branded memory modules (A-Tech PC3200 1GB) reported in CPU-Z:
https://www.badcaps.net/forum/attach...1&d=1608957649
https://www.badcaps.net/forum/attach...1&d=1608957649
In particular, the two memory modules above appear to have identical labels. However, one has Kingston memory chips, while the other one has… well, I forgot, but it was something different (though still name-brand chips, IIRC.)

Anyways. Among other things, I also found the motherboard has two PTC thermistors – one by the corner of the memory slots, and the other close to the CPU VRM MOSFETs heatsink. On my system, these reported temperatures under “Temp1” and “Temp3”, respectively, in SpeedFan. The CPU appeared to be reporting under “Temp2”. However, I’m not entirely sure how accurate that was, given that it varied between 29°C and 40°C while idle and under short load, with an ambient room temperature of about 18.3°C (65°F.)

On that note, this Shuttle PC isn’t quite ready to go on the “fully fixed” shelf yet. One thing I noticed is that the Southbridge heatsink runs quite hot, which also ends up heating the MOSFETs for the NB VRM I jurry-rig fixed, along with the caps in that area. The NB fan is also quite loud (and has a somewhat worn bearing.) So I’ll probably need to do some cooling mods for the Northbridge and Southbridge (maybe a big 80 mm fan?)

Another thing is that my kludged fix above with the two pieces of wood does not allow enough space for the original 3.5” HDD to be used back in the case as-is. Basically, the vertical piece of wood blocks the SATA and SATA power connectors on the HDD. So I might have to either re-locate the HDD in the 5.25” optical drive slot, or just clone a 2.5” HDD, and use that instead. (I’ll probably go for the second option, though.) As for the original 5.25” optical drive - I already have that installed and in use in another PC, so I don’t think I care about putting it back into this Shuttle PC.

At this point, fixing this Shuttle SB83G PC was more of a curiosity than anything else. Had I found the problem and fixed it 8 years ago, it surely would have been my main or secondary PC. These days, though, a 2.8 GHz Pentium 4 Prescott isn’t very relevant. Perhaps if the i915 chipset could take a Pentium D 9xx series, that would have made it a little more useful. I don’t think this motherboard officially supports anything beyond Pentium 4 Prescott 5xx series, though. So now it will just be another toy for me to play with. I’ll update the thread again as I get more done to it eventually. I might even try adding a better (but still low-power) PCI-E GPU in there. Also, the PSU is still running with the original OST RLS caps and it does run warm. So I have no idea how much life those have left (if any.) At least they are RLS series, which aren’t so terrible. Anyways, that’s all for this one so far... as usual.

[sam_sam_sam]

Re: Shuttle XPC SB83G5A - finally fixed!

Alright ladies and gentlemen. After this project sitting on the back burner for over a year and some contemplation going on in the back of my head about how to proceed, I finally came up with a “temporary” solution :roleyes: … that may actually end up being permanent… if it ends up works permanently, anyways. LOL! I guess, we will have to see.

Basically, I did some risk evaluation, and thought the chances of messing up the board worse than it is are really high if I try to do a CPU socket reflow (let alone a socket reball.) So I went down in the garage and started tinkering with how to exert downward pressure on the CPU area to keep the board working. It took me no more than 10-15 minutes before I found some completely scrap materials that actually looked like a perfect fit for the contraption I had in mind. (Gosh! I’m such a hoarder. But at least I do eventually find a use for much of the scrap junk I keep around.) Namely, it was two pieces of wood and a threaded rod (1/4” ?) In fact, one of the wood pieces and the threaded rod were both already the length I needed. So all I needed to do was just cut the other piece of wood, drill two holes in it, and then widen an existing hole in the Shuttle case for one of the screws. The work took me maybe 20 or 30 minutes total. After that, all that was needed was a nut that fits the threaded rod and a washer - these were probably the only two “new” parts I used here (though I bought these packs of washers and nuts probably close to two decades ago.)

Essentially, the horizontal wood piece is held by a screw on one side of the case, and the other is just wedged against the case frame. Meanwhile, the threaded rod has the nut and washer on the bottom of the horizontal wood piece. So as the threaded rod is tightened, it starts to push down on the vertical piece of wood, which then applies pressure over the CPU heatsink. This warps the board and enables whatever BGA is broken down there under the CPU socket to make contact again. Yes, it’s a complete jurry-rig setup, but it works. And if that’s not bad enough for you, notice that the threaded rod has a notch cut at the top. Why? Because I couldn’t find a matching bolt and nut with the length and thickness I needed. So when I decided to use the threaded rod, it had no head on either end. Thus, to make it “turn-able” I cut a notch at the top so it could be turned with a flat heat screwdriver. But you know what? Who cares, as long as it works… and it does.

When I first tried this contraption, I started tightening the threaded rod to see how much pressure it can exert on the board. Needless to say, the relatively fine thread on the rod allowed me to warp the board easily. At first, I put a good amount of pressure on there and tested the motherboard - it booted no problems first try. But eventually, I decided to play it the other way around: loosen the threaded rod all the way ........

So I think my fix (jurry-rigged setup) works pretty well.

I always find your repairs interesting but this one is over the top

I have tried things just to see if I get something to work or not

But I enjoy reading your repairs

[momaka]

Re: Shuttle XPC SB83G5A - finally fixed!

Thanks Sam!

Yeah, this one took me more than what I'm normally willing to do... but being a Shuttle PC in a cool aluminum case that no other motherboard could fit... I just kept it in storage as-is all these years and occasionally trying stuff on a rainy day... until I finally found how to get the mobo working again. If this was any other regular desktop PC, I would have re-purposed the case and scrapped the motherboard for parts a long time ago.

I feel like I learned a lot from this repair, though. IIRC, this was one of the first few motherboards that I studied in depth how the voltage rails work / get generated... and on that note, I still haven't posted the diagram (though I probably need to re-do it - it's quite messy, being one of my very early cap diagrams.) Ah well, that will come some day too. I doubt there is a high demand for info on this particular Shuttle motherboard, being so old as it is right now.