View Single Post
Old 05-02-2019, 07:31 PM   #11
Badcaps Veteran
momaka's Avatar
Join Date: May 2008
City & State: VA (NoVA)
My Country: U.S.A.
Line Voltage: 120 VAC, 60 Hz
I'm a: Hobbyist Tech
Posts: 9,388
Talking Shuttle SB83G5A - she's alive!

Before anyone else says it... YES, HOLY ANCIENT THREAD REVIVAL, BATMAN!

It's not without a reason, though! - the Shuttle SB83G PC in this thread is alive!!!
Well, sort of. I'll elaborate.
She's not exactly fixed yet. But at least I found the problem, and you're not going to believe it - bad BGA in the CPU socket, just as I suspected right from the beginning. (that is, after I recapped the motherboard and fixed the NB VRM)

So how is it that I found about it now, you ask? Simple: pure luck.
After doing a full recap on this motherboard... more than 6 years ago, mind you ... I needed some short 6.3V 1500/1800 uF caps recently. Since I didn't have any in stock, I remembered the recap I did in this PC and pulled it out from storage (yes, probably shocking to some of you that I even kept it this long), then pulled most of the good caps from its motherboard. I figured it's dead forever and I would probably never be able to revive it, so I started pulling various cables and parts from this system all these years. Who knew I would prove myself wrong!

After pulling some of the good caps from this motherboard for use on other projects, I still felt bad for it. Thus, I shoved back in a few of the original capacitors that were still good (namely, the OST RLS on the CPU VRM), along with a few excess pulled Teapo SC. I powered it on, and nothing had changed - mobo still dead and POST card stuck at "FF". (Previously, I mentioned in my above posts that it was stuck on "00". But over the years, I found that this POST card will usually give "00" in dusty PCI slots and/or ones with tarnished contacts, until the card is inserted a few times to clean them. After doing this, I got a stable "FF" and no BUS activity on the PCI slot, other than when PWR or RESET buttons were pressed.)

Also, the ghetto Northbridge VRM repair shown above in post #8 kind of started to bother me. Even though the motherboard was dead, I decided to re-do it a bit neater and maybe give the PC another go at testing.

I still didn't have any SOIC-8 MOSFETs for a proper replacement, so I just re-arranged/re-soldered the TO-252 MOSFET, like so:
Yes, that handiwork is not beautiful by any means. But hey, the MOSFET still fits and the circuit still works! (Note: capacitor CE83 was re-installed after taking these pictures - a Rubycon MFZ 6.3V 820 uF from an Xbox 360.)

I checked the voltage after re-doing the above MOSFET, and the NB VRM still had proper voltage (about 1.52V this time.)

But apart from that, the motherboard was still "dead", with the POST card showing "FF" and CPU voltage still incorrect at almost 1.6V. In my last post here, I mentioned that I should try another CPU... which I did (another Pentium 4 CPU) many years ago. I just didn't bother updating the thread because nothing had changed in terms of symptoms. This left only three things I suspected could be wrong: bad Northbridge (due to the VRM burning out the first time), bad BIOS, or bad BGA in the CPU socket.

While bad BGA in the CPU socket seemed unlikely, I will tell you why I had a suspicion this was the issue: when I was taking this PC apart for the first time, there was a rubber foot between the motherboard and the case, right under the CPU socket. I imagine due to heat, it had its glue badly melted. Because of this, the motherboard was still stuck in the case, even after I had removed all screws. I remember I rotated and flexed the motherboard quite a bit before the glue on that rubber foot finally gave up. And this is when I suspect I had cracked the socket BGA. But due to the bad caps on the motherboard and later the shorted NB VRM MOSFET, I just didn't want to believe this was the issue, even though this incident with flexing the motherboard was always on the back of my mind all along.

So when I saw the wrong CPU voltage this time in addition with the POST card stuck in "FF" and no change in behavior regardless of whether or not I had RAM in the system... something in the back of my mind put two and two together that perhaps it is a BGA issue with the CPU socket. Thus, I decided to insert some pressure on the CPU cooler and try powering the motherboard that way. I don't know why I didn't think of doing this before, as I've suggested it numerous times to other people here when doing motherboard troubleshooting.

So with everything still connected the same way, I pressed hard on the CPU socket and then pressed the power button. Miraculously, this time I saw the POST codes on the POST card change. After two seconds, the motherboard turned off, then auto-powered again, ran through different POST codes, and then... BEEP! - my LCD screen turned On and I got a picture on it, saying default CMOS settings running!

Just to make sure I wasn't day dreaming or imagining things, I decided to re-run all these tests. First, I turned Off the PC. Then, I tried powering the system On without inserting pressure over the CPU socket/cooler. Unsurprisingly, the system didn't boot and POST card was stuck again on "FF" code. I powered it down, inserted pressure over the CPU, then powered On again. But I wasn't imagining things - the PC POSTed successfully again!

With all that being said and done, I think this confirms it that the BGA for the CPU socket was (hopefully) my last issue with this PC. Now the question is whether I should try re-flowing the CPU socket BGA or just insert something underneath it to put pressure on the board. The reason I am hesitant to do a reflow is because I would need to remove all of the electrolytic caps in the area (again.) Leaving them in would surely pop them, as they are connected with thick Coper traces to the CPU socket. Also, there's always a chance something could go wrong with the reflow and thus mess up the motherboard even more. In particular, I've never done any large CPU socket reflows like this, and I know the process requires quite a bit of heat. This means there's a considerable chance I could warp the board when doing so and actually make it irreparable. On the other hand, propping the motherboard with something underneath it for applying pressure under the CPU socket could also warp the board over time and make it impossible to do a successful reflow/reball on the CPU socket. So that's why I'm not keen on doing that either.

Yes, I know this is only an old (and now quite worthless) Pentium 4 motherboard that's arguably worth spending any time on. But please take a moment to understand that I am doing this for fun. I've come so far (and so close), that I really just have to finish it now. Hence, all the pondering above about what would be the better way to proceed. With that said, does anyone have any suggestions regarding the above matter?

Also, I see that I never posted a "cap map" for this motherboard, which is something I always do with my other recaps. Funny, because I did draw one many many years back (and scanned it too.) I don't imagine anyone would be recapping such an old PC anymore... but why not, right?

Anyways, cheers!

I'll update this thread again when I think of a way to fix the CPU BGA socket issue. Might not be very soon, but we'll see.
Attached Images
File Type: jpg Shuttle XPC FB83 - NB Vcc VRM repair (1).jpg (141.1 KB, 22 views)
File Type: jpg Shuttle XPC FB83 - NB Vcc VRM repair (2).jpg (171.9 KB, 24 views)

Last edited by momaka; 05-02-2019 at 07:39 PM..
momaka is offline   Reply With Quote