BSOD, bulging caps, on Gateway GT5056

I'm trying to repair an old model Gateway GT5056 Windows XP system. The symptom is that intermittently it will just freeze up in its tracks or a BSOD will appear. One of the types of BSOD says, "Machine_Check_Exception." The other says, "Kernel_Stack_Inpage_Error." I tried updating drivers and software, removing the DVD burner, and just using the bare minimum of hardware, but it still was freezing up. Usually, it will happen when watching a video. On other days, it would happen as soon as the computer booted to the desktop. Then on some days, it wouldn't freeze-up at all.

Just to make sure the hardware was ok, I ran a RAM test which passed. I cleaned out the CPU fan and heat sink. Then I visually inspected the motherboard. Hmmmm. I saw about a half dozen 1,000uF 6.3V capacitors that were bulging. In fact every capacitor of that particular value was bulging. I checked ESR and only one was out of tolerance. It measured 6 ohms while the others were less than 0.2 ohms. I replaced the one that was out of tolerance thinking I'd go back and do the rest of them if it corrected the problem but it didn't. It still freezes occasionally.

There was one particular occurrence where right before the system froze, the CPU fan switched to a very high speed so maybe it is heat related. By the way, the motherboard is a Foxconn C51GU01.

Thanks for your reply.

--
David Farber 
Los Osos, CA
Reply to
David Farber
Loading thread data ...

Get a second hand pc. I get them offered for free, the last one a HP 3GHZ 2GB dual core xp pro computer. I removed all the bloatware and personal data, and it works like a treat. People have been scared out of their wits by the "end of life" commotion. Then transfer step by step special hardware/software from the broken one(if any).

An btw, try a number of memtest programs, I had one computer failing only with one out of four test floppys. And let those run for several hours, with closed case, else the temperature will not match with usual use.

If you HAVE to use the old board, replace ALL bulging caps, they are broken, even if they seem to work.

Reply to
Sjouke Burry

Replace some of the ones that read 0.2. They are probably a bank of caps. T he common thing is to use a bunch of them in parallel. You can confirm this by meter. Of course all the negatives will read continuity, but in a bank of caps, both leads will read continuity.

If you have a bank of five caps, 0.2 ohms ESR is too high and that's one of the reasons they used so many. Another is ripple current.

When you have this, you can just replace a couple in a bank to see if it cu res the problem. In fact when it gets to that point I don't even clip the l eads off, I just let the replacement(s) stand up off the board. If the prob lem is fixed, then I sink them down and snip them, and of course replace th e rest.

The one you replaced was obviously by itself. The others are probably a ban k of five and they are probably off the output of a switching regulator. Th ey might be running at like 2 volts and anywhere from 2 to 20 amps dependin g on processor load. The math says 0.2 ohms is way too high. Say it's 3 vol ts at 10 amps. That's like a 0.33 ohm load approzimately. That could make t he ripple voltage almost half of the supply concievably. Coming off a high speed switching regulator, that is quite noisy and the data will be FUBAR. Temperature will also affct it.

If you find those five caps are just all in parallel, the ESR should be pra ctically imeasurable.

If it doesn't work, that's why you didn't cut the leads on the replacements .

Reply to
jurb6006

If you have a bank of five caps, 0.2 ohms ESR is too high and that's one of the reasons they used so many. Another is ripple current.

When you have this, you can just replace a couple in a bank to see if it cures the problem. In fact when it gets to that point I don't even clip the leads off, I just let the replacement(s) stand up off the board. If the problem is fixed, then I sink them down and snip them, and of course replace the rest.

The one you replaced was obviously by itself. The others are probably a bank of five and they are probably off the output of a switching regulator. They might be running at like 2 volts and anywhere from 2 to 20 amps depending on processor load. The math says 0.2 ohms is way too high. Say it's 3 volts at

10 amps. That's like a 0.33 ohm load approzimately. That could make the ripple voltage almost half of the supply concievably. Coming off a high speed switching regulator, that is quite noisy and the data will be FUBAR. Temperature will also affct it.

If you find those five caps are just all in parallel, the ESR should be practically imeasurable.

If it doesn't work, that's why you didn't cut the leads on the replacements.

Hi,

I'll have another look at how the caps are wired. Some are located close to each other but at first glance you wouldn't think they would be in parallel. I'll report back my findings.

Thanks for your reply.

--
David Farber 
Los Osos, CA
Reply to
David Farber

formatting link

Reply to
Jim Haynes

I repaired a roomfull of Dell computers at work that all had the capacitor plague. There are some caps in a switching supply right at the CPU chip that suffer very high AC ripple that cooks them. I replaced all the caps in that bank (identified by the bulges or leaks) and they all worked fine. I did another system where a graphics board had the same problem.

Jon

Reply to
Jon Elson

The real problem is when the ripple and noise causes data errors. It can FU BAR a harddrive or the BIOS. If it access the wrong address it can be as ba d as a power outage during a BIOS flash. I have a Dell I think board like t ha there that is destined for the trash actually.

In some motherboard manuals I remember they called it "system will hang per manently" which is about the equivalent of bricked.

Reply to
jurb6006

Motherboard caps (switch mode converters) and the power supply caps have to be as close to perfect as possible. You want the highest ripple current wi th the longest life. Price is secondary and 'good enough' never is. I don't even bother with ESR measurements. Just replace them. A problem is the new boards have polymer caps rather than 'lytics so the large variety of 'lyti cs are no longer carried my Mouser. DigiKey was not as complete as Mouser but is OK. On the good side we won't need to be replacing nearly as many ca ps in the future.

The latest motherboard I did was an IBM that had a solid plane (no thermal reliefs) on the top AND bottom of the board. My Metcal had enough heat but the little wire through the board couldn't conduct the heat fast enough. I ended up with the soldering iron in one hand and a hot air gun in the other . The holes ended up very clean and it runs like new again.

Reply to
stratus46

Yes, this is a problem. I used a Weller WMP which is a 65 W mini-pencil iron, and with the most massive tip available for it. First, dilute the no-lead solder with SnPb solder to lower the melting point of the solder. Then I heated each capacitor lead for 10 seconds or so and then levered the cap which pulled the pin back through the board a bit. Then go to the other pin and do the same. After one cycle of that, the cap is far enough off the board you can pull one lead fully out of the board, then the other. Then, I used the 65 W iron on one side and a Pace desildering tool on the other side at the same time, and it pulled the solder out of the hole. Resoldering the new caps was a breeze compared to the extraction part.

Jon

Reply to
Jon Elson

I had to use my Weller 100/140 watt gun to generate enough heat to get the capacitors out. Even with flux and SnPb solder it was a chore. The results were, one was 1.3 ohms and the remaining four measured from 4.7 to 5.0 ohms. I will order the replacements and report back.

Thanks for your reply.

--
David Farber 
Los Osos, CA
Reply to
David Farber

Yup, highly likely replacing them will fix the computer. Be sure to get high-temp, low ESR caps for the replacement. A lot of pwople are selling totally pedestrian caps as low-ESR on eBay and such, but they are actually junk, and not suitable. If the machine works with them, it won't last long. The proper units are available from Digi-Key for under $2 each. You just have to scan for the low ESR rating.

Jon

Reply to
Jon Elson

Go back and do the rest. Make sure you use top quality 105 degC replacements. Panasonic are good.

--
 (\_/) 
(='.'=) 
(")_(")
Reply to
Mike Tomlinson

Panasonic has caps that are good but also has ones that are useless for mot herboards. We bought a used VTR that had all the caps replaced. The tech us ed Panasonic NHG. These are rated for 105C but the ripple current is less t han 1/3 of the FM/FR series.

The thing that really got me was the Digikey price. The lousy caps were $80 /1000 while the excellent caps were only $85/1000. Why would anybody buy th e junk for repairs? Someone asked me about prices of the caps and is it imp ortant. I told them that on a good day I might use $20 in caps which is a s mall fraction of what I'm paid. If I'm worth it so are the caps.

Reply to
stratus46

I bought caps from the above a couple years ago. Very happy customer. Mikek

Reply to
amdx

Hi Jon,

I ordered from Digi-Key. Their part number is, 493-1467-ND.

The caps are rated at 6,000 hours, @ 105 degrees. Ripple current is 840 mA. The impedance is 87 mOhm.

There were other choices with lower impedances but they were rated at shorter lifetimes.

Thanks for your reply.

--
David Farber 
Los Osos, CA
Reply to
David Farber

It was a good thing I ordered ten of these caps because after I replaced the first half-dozen, I found another four that were slightly bulging. The ESR's of these caps varied from 1.5 to 4.5 ohms. When I put the new ones in, the in-circuit ESR's were nearly undetectable.

The results of all this were that the freezing up and blue screens never reoccurred.

Thanks for all your great replies.

--
David Farber 
Los Osos, CA
Reply to
David Farber

I've bought from here:

formatting link

--
Anyone wanting to run for any political office in the US should have to 
have a DD214, and a honorable discharge.
Reply to
Michael A. Terrell

ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.