So I don't know if it's good luck or bad luck, but both of my crunchers have needed some surgery so far. Caps age and weaken, causing all sorts of weird problems, including bricks.

Started with the single-core machine. Random power-offs that were related to CPU load. So I thought maybe it was time to blow the dust bunnies out, but there were none. Upon looking around inside the case, I noticed these. All right next to the CPU socket. Went on eBay and got a 10-pack of replacement ones for $6 shipped, and replaced them. That was the second week of October 2010. That system has been rebooted once, only because some updates installed that required a reboot.

Then the main cruncher started having weird issues three weeks ago. Whilst crunching and playing a game, it would freeze..but not really. The cursor could still move around, the keyboard lights worked, I could alt+tab and bring up the ctrl+alt+del menu, but there was zero disk activity and it wouldn't recover. Figured it was time for the dust bunnies on that rig, too. Again, nothing of significance (thanks to the 11 5-1/4" filters). Decided it was time to deploy the Ghost image for a clean install and start over, but that meant swapping my optical drive from one onboard controller to the other (one of the controllers is not bootable, but the one that is causes the system to lock up in Windows when any disc is inserted and read), and I noticed a bulged capacitor. Read the numbers off of it, and eBay'd some new ones.

Picked a date for surgery, and got to it. Found three of them that needed to be replaced. Fortunately all the same values. Upon looking around at all the others, I noticed one more. That is charred ooze coming out of it, and right next to RAM banks. Ah-ha! That could be causing weird freeze issues (I remember long..long ago back in the SD-133 days I accidentally poked a DIMM and got a nearly identical freeze). Dug around in my parts drawers and couldn't find a 1500uF cap. All I had that was over 1000 was the left-over 3300's for the single-core machine. I remember during my research that the voltage rating can be higher, but the opinion is mixed regarding a higher capacitance being fine. Decided to gamble and did it anyway.

Guess it worked, because it has been four days of crunching and playing games and not a single hiccup. That also means that the soldering for the 1500/3300 one worked out after all. Upon removing the old one, I ended up destroying one of the contact pads on the bottom of the board for that. I looked really closely and there were no tracings leading up to it on the back of the board, just the top and possibly one of the inner layers (if there are even any inner layers). Ended up putting a glob of solder on the top side and heating it from the top until the iron could poke through enough to heat the capacitor lead from the bottom, pulling solder down with it.

Here's all four of them lined-up. And the toll 24/7 crunching for just over four years takes on the lacquer (note the discolored areas). Chipset and some voltage regulators at the top-left of the picture, and bottom-right is all the voltage regs for the CPUs.

So that's my story. I would have just RMA'ed the board, but for one.. I don't want to be down for three weeks, and two, the warranty expired over a year ago anyway. Here's to another four years of crunching!

---

Linux laptop:
record uptime: 1511d 20h 19m (ended due to the power brick giving-up)

You did a nice job noticing those caps and replacing them. You are right that you can get caps with greater voltages. I always do that when I can get them to fit. Changing the capacitance can cause other problems, but it is up to the individual circuit design on how tolerant it will be. Many times you can get an electrolytic capacitor, which the ones yopu pictured are, drying out. When you have several of them it becomes difficult to troubleshoot, but you can take a capacitor of equal value and the same of higher voltage, making sure you have the polarity right, and touch the leads to the solder pads of the suspect cap. If the suspect has a voltage leak, the one you hust touched to the back side should clear up the problem. This is really tough on a mother board, and may require building a gig to hold it while you mess around with it.

Anyway, very nice job on identifying and replacing those caps!

Steve

---

Warning, addicted to SETI crunching!
Crunching as a member of GPU Users Group.
GPUUG Website

I remember during my research that the voltage rating can be higher, but the opinion is mixed regarding a higher capacitance being fine. Decided to gamble and did it anyway.

A value that high will be mainly intended as a reservoir (ripple smoothing) capacitor, so a higher value is generally fine. Small values around chokes (inductors) tend to be for filtration & altering the value would alter the frequency.

A 3300uF capacitor will generally have a higher ESR ( Equivalent Series Resistance ), which will make it long term more susceptible to what caused the 'drying out' or 'boiling' in the first place. That factor is 'ripple current' which the reservoir is there to smooth.

From the looks of your replacements, they are probably Low-ESR high temperature variants, so despite the higher values will probably be better than the originals anyway IMO. If you were concerned, you could add some small value ceramic bypass capacitors to aid with the high frequency components, such as 10nF and 100nF monolithics in parallel with those... But I suppose that depends on how long you really want it to last, versus your time & effort etc.

Next time, I suggest avoiding desoldering on multilayer circuit boards, because you can rip out the through plating. Instead, you can cut/pull off the top of the old capacitor leaving only the existing legs, and solder to those.

Jason

---

"Living by the wisdom of computer science doesn't sound so bad after all. And unlike most advice, it's backed up by proofs." -- Algorithms to live by: The computer science of human decisions.

Next time, I suggest avoiding desoldering on multilayer circuit boards, because you can rip out the through plating. Instead, you can cut/pull off the top of the old capacitor leaving only the existing legs, and solder to those.

Jason

Excellent idea!

Steve

---

Warning, addicted to SETI crunching!
Crunching as a member of GPU Users Group.
GPUUG Website

Thanks for the feedback and input.

The replacement caps (both the 820's and the 3300) are low ESR, and increased from 105C to 125C, and from 6.3v to 10v.

Removing components from multi-layer PCBs do make me nervous, since I know it can be easy to botch and then end up with an expensive paperweight. I prefer things to look neat though, rather than using the old component's leads as stand-offs to solder to, plus two of the 3 820s have to be as low as possible due to the GPU.

I don't like that my chipset runs at 75-83C, but that's all I can do with it. Passive heatsink (as seen in one of the pictures), but I did remove that generic thermal compound and put on some Arctic Silver 5 instead. I used to have a 740-gram solid-copper cooler that stood about 3 inches high and had a 40x25 fan on it, but that was with the old single-slot video card.. couldn't use it anymore when I upgraded to an 8800GT. That old cooler kept the chipset under 40C though. I did purchase a really cheap low-profile active cooler, but it is still about 4mm too tall to fit under the GPU. I thought about sanding the fins down just enough, but decided it wasn't worth it. Just another fan that will end up making noise in a year or two and need to have some motor oil put inside to make it quiet again.

---

Linux laptop:
record uptime: 1511d 20h 19m (ended due to the power brick giving-up)

Excellent idea!

I can't take credit for that one. That's Class 3 military/aerospace rework training.

---

"Living by the wisdom of computer science doesn't sound so bad after all. And unlike most advice, it's backed up by proofs." -- Algorithms to live by: The computer science of human decisions.

                       ---[Not quoted]---

A 3300uF capacitor will generally have a higher ESR ( Equivalent Series Resistance ), which will make it long term more susceptible to what caused the 'drying out' or 'boiling' in the first place. That factor is 'ripple current' which the reservoir is there to smooth.

From the looks of your replacements, they are probably Low-ESR high temperature variants, so despite the higher values will probably be better than the originals anyway IMO. If you were concerned, you could add some small value ceramic bypass capacitors to aid with the high frequency components, such as 10nF and 100nF monolithics in parallel with those... But I suppose that depends on how long you really want it to last, versus your time & effort etc.

Next time, I suggest avoiding desoldering on multilayer circuit boards, because you can rip out the through plating. Instead, you can cut/pull off the top of the old capacitor leaving only the existing legs, and solder to those.

Jason

I can only agree with you on this, mobo's are multilayered and soldering has to be done carefully, better to avoid it. And sometimes adding a 100nF condensor
enhances stabillity.
And there are bad-manufactored capacitors (electrolytes), they also used in video cards, PSU's . And they've gotten smaller, too
The ripple causes high amps, load and unload, one good reason to use/make
high frequency, 22KHz or higher, PSUs. Capacitance can be 1000 smaller if 50KHz
is used.
(Audio Amplifiers and sometimes 10mF is used.(10,000uF))

---

Removing components from multi-layer PCBs do make me nervous, since I know it can be easy to botch and then end up with an expensive paperweight. I prefer things to look neat though, rather than using the old component's leads as stand-offs to solder to, plus two of the 3 820s have to be as low as possible due to the GPU.

Only you can decide if the risk is worth it in that kind of situation :) You don't get the same choice when you're working on life support equipment.

hehe, I should post a picture of the switchmode powered car amplifier that I built when I was seventeen. Looked like a dog's breakfast, wires everywhere, but outlived 3 commercial amplifiers in my Ford Falcon. neatness is cool too, but cleanliness & robustness rule for reliability.

[Edit:] BTW ,the caps can also lie down, or be at strange angles.

Jason

---

"Living by the wisdom of computer science doesn't sound so bad after all. And unlike most advice, it's backed up by proofs." -- Algorithms to live by: The computer science of human decisions.

On an old slot1 board I was overclocking to the Nth degree. I had one of the electrolytics explode & shoot acidic paper all over the place. That was a fun mess to clean up... I ended up soldering wires to the remains of the cap leads and locating the new cap elsewhere.

---

SETI@home classic workunits: 93,865 CPU time: 863,447 hours
Join the [url=http://tinyurl.com/8y46zvu]BP6/VP6 User Group[

On an old slot1 board I was overclocking to the Nth degree. I had one of the electrolytics explode & shoot acidic paper all over the place. That was a fun mess to clean up... I ended up soldering wires to the remains of the cap leads and locating the new cap elsewhere.

ROFL, I love the eerie silence as you stare around at the feathery floaties. And the smell :P

---

"Living by the wisdom of computer science doesn't sound so bad after all. And unlike most advice, it's backed up by proofs." -- Algorithms to live by: The computer science of human decisions.

On an old slot1 board I was overclocking to the Nth degree. I had one of the electrolytics explode & shoot acidic paper all over the place. That was a fun mess to clean up... I ended up soldering wires to the remains of the cap leads and locating the new cap elsewhere.

ROFL, I love the eerie silence as you stare around at the feathery floaties. And the smell :P

It will be fine as long as you get all of the smoke back in! Wait... some just went out the window :(

---

SETI@home classic workunits: 93,865 CPU time: 863,447 hours
Join the [url=http://tinyurl.com/8y46zvu]BP6/VP6 User Group[

On an old slot1 board I was overclocking to the Nth degree. I had one of the electrolytics explode & shoot acidic paper all over the place. That was a fun mess to clean up... I ended up soldering wires to the remains of the cap leads and locating the new cap elsewhere.

Many years ago I was running sound for a local band. I was in charge of pyrotechnics also, so I set up several non-polar capacitors, and hit them with 120 VAC when it was time. The bangs went off perfectly, and nobody got hurt!

Steve

---

Warning, addicted to SETI crunching!
Crunching as a member of GPU Users Group.
GPUUG Website

I tried changing caps of my Epox NF2 board. Soldering process went OK, but the voltage regulators had got some damage. The board itself was a good buy, though it was a bit expensive at the time of purchase. It was sad to abandon that board, it served many years in my main rig and it was my only board which could OC my mobile Athlon.

---

Many years ago I was running sound for a local band. I was in charge of pyrotechnics also, so I set up several non-polar capacitors, and hit them with 120 VAC when it was time. The bangs went off perfectly, and nobody got hurt!

Steve

That sounds similar to a prank I heard about in automotive repair shops 30+ years ago. Back when there was a large capacitor on distributors. I heard stories of charging it up with 50,000 volts, folding the leads down on opposite sides, get the new guy's attention and toss it to him across the shop. Pow!

Capacitors have so many uses. Some are completely safe and legitimate.. others, well not so much. :D

---

Linux laptop:
record uptime: 1511d 20h 19m (ended due to the power brick giving-up)

Usually by the time capacitors become a problem, it is near time to retire
the motherboard anyway.

If you are comfortable soldering on a circuit board.. have fun! You can save a few bucks. But not something I would recommend for most computer users.

---

Janice

Normally I would agree.. but four years seems a little short for a motherboard's lifespan. I mean, I have a Socket A board that I bought in 2002 and used for crunching for five years, and it is still going strong today (not crunching).

This rig is scheduled to be succeeded once a decent selection of Bulldozer hardware is out. It will continue on with crunching and being a part of my network, but as a NAS (probably 15 2tb disks in a raid6 + hot spare). I just need it to be a reliable main rig until then. Built it to keep up with everything I ask it to do for five years, and it is still doing it beautifully..only 7 months to go.

---

Linux laptop:
record uptime: 1511d 20h 19m (ended due to the power brick giving-up)

Normally I would agree.. but four years seems a little short for a motherboard's lifespan. I mean, I have a Socket A board that I bought in 2002 and used for crunching for five years, and it is still going strong today (not crunching).

This rig is scheduled to be succeeded once a decent selection of Bulldozer hardware is out. It will continue on with crunching and being a part of my network, but as a NAS (probably 15 2tb disks in a raid6 + hot spare). I just need it to be a reliable main rig until then. Built it to keep up with everything I ask it to do for five years, and it is still doing it beautifully..only 7 months to go.

With 4 machines at home I sorta do about a 4-5 year upgrade cycle. So about 1 a year that way. :D Right now I'm trying to decide if I should retire my mobile C2D machine as I made it way back in 2007. By upgrading it with an Atom into a NAS with 6HDs, or trying to find some sort of PCIe x1 RAID controller that has more than 4 ports for it. Then next year upgrade my main system from a C2D to whatever ivy bridge turns out to be, or maybe Haswell/rockwell stuff.

---

SETI@home classic workunits: 93,865 CPU time: 863,447 hours
Join the [url=http://tinyurl.com/8y46zvu]BP6/VP6 User Group[

I try to stick to 1 per year myself. 2 desktops, 1 notebook.
I put it off last year and I am feeling it.

They still work past that, but I start to ask "why?".

---

Janice

To soft^spirit's last post ...

I am a tight wad at heart, and my well stuffed pocket book (very rare now-a-days) determines my replacement cycle.

Have 4 rigs including 2 server based PCs, and the oldest (a dual P3@933MHz) on W2K and with RAID-1 mirror is still running hot after 10.5 years.

---

It's good to be back amongst friends and colleagues

As to the thread title question, Yes. In 2006 I noticed brownish ooze coming from several capacitors in the set near the CPU that are associated with the on-board voltage coversion. I took pictures of location 1 and location 2.

I'm an electrical engineer with a moderate amount of practical lab experience, but replacing components on a motherboard is way past my comfort level, so I sent the board off to a guy who does this replacement professionally. He replaced all the suspect capacitors, not just the ones with current visible ooze or bulge. I suggest that, by the way. Most likely some more capacitors on the original poster's board will show symptoms quite soon.

Electrolytic capacitors are a life-limited component. Even if you don't have the infamous "stolen recipe imperfectly copied" issue, the flat fact is they deteriorate with age and use. The higher the temperature and the higher the ripple current, the more rapid the deterioration. So, yes, overclocking is relevant.

My particular board was operating with an overclocked Gallatin, which burned a lot more power at my operating condition than the typical processor for which the board was designed. I doubt my case cooling was of the very best, either.

My capacitors were Taicon, and were not on any of the worst "bad cap" lists.

---