"Power tends to corrupt, and absolute power corrupts absolutely."


Techgage Staff
Staff member
I'll be honest. My quest for PC power has corrupted my PC, perhaps irreparably. If you're a serious computer enthusiast then you have certainly heard at least one horror story from someone you know that was using a cheap power supply, that often end up not being so "cheap" in the end.

This story is a bit different. Instead of two cheap power supplies, we had two expensive, highly recommended and highly regarded units that both failed, and for wildly different reasons. Within the time span of two months, no less. As these were review samples neither unit's 5-year warranty is applicable, but both of these units were supposedly some of the best.

The first one is the Cooler Master 1,100W UCP (Ultimate Circuit Protection). This unit was the first 80 Plus Silver power supply to reach market, and was highly recommended based on build quality and safety protections. It had seen two years of use. I can at least vouch for the safety protections as working properly.

The problems began with the system overclock showing instability that was hard to point down. Gradually as the days went by the problems go worse, I'd lower the overclock or adjust voltages to stabilize it again. Probably too quickly I assumed that the CPU I had been using for over a year at 4.2GHz was degrading, because no other logical explanation seemed to fit.

Individually one-by-one I'd replace or switch every single PC component including the PSU to try and resolve the problem. Thanks to hindsight it is clear that when I switched out the PSU, I mistook the GPU driver crashes I'd been experiencing as a sign that replacing the PSU had not resolved the error... so I reinstalled the CM 1,1000W power supply. Duh! Eventually the system got so unstable that at fully stock settings the system would generate Linx errors, Prime errors, and any other stability checking program I'd care to try.

Eventually I noticed a new sound that I had not been able to hear before over the sound of the window AC unit in the PC room, and pinpointed it to the PSU itself... a cycling "bzzzt-tic, bzzzt-tic, bzzzt-tic" sound greeted my ears, only faintly audible even with the AC turned off. I opened it up to find out the cause... and this is why you should never open a PSU yourself unless you have some electrical knowledge and training, including the proper tools like a high voltage detector. :eek:

Visually, the unit looks perfectly fine. All caps are intact and pristine, all voltage rails test within specifications, all Vregs are clean. (I had tested inside the system under load as well previously). Then the high voltage detector warned me about the cause... even unplugged, the primary heatsink was generating a voltage drop that shouldn't have been there.

So I plugged it in and fired it up, so far so good, only a 0.1 millivolt reading on the secondary heatsink, which is well within norms from my understanding... (sorry for the photo quality, quick and dirty shots on the green countertop for safety reasons)

Hmm, perhaps not. Even unplugged it would be dangerous to have touched the primary heatsink due to the residual voltage/charge, but this unit is actually powered on here:

The voltage readings spike as high as -097.7 Volts every other second or two. This combined with the sound effects was enough to prove to me this PSU had failed. All of that said, I can say that the safety protections were working properly because the system was never damaged. It was unstable, but no components had failed or been damaged, unlike later...

Next up, was a PC Power & Cooling 750W Quad Silencer. I had this unit for about three years.Technically I still have it, but I'll never be using it in a PC ever again... As of last week, in the middle of a game of Supreme Commander no less, amidst all the smoke and explosions I noticed the special effects were getting a bit too real... I could detect ever so faintly the sickly sweet twinge of singed components, or perhaps dust. I alt-tabbed and shut off Folding@home and tried to find the source of the smell but couldn't, so I resumed the game figuring the central heating system had been mistakingly turned on... Around thirty minutes later nothing was amiss, the game ends and I resume the Folding@Home application (which was generating only a 67% processor load at the time, and is far less demanding than say, Linpack stabiltiy testing with Linx). Hmm, there's that smell again, and it's definitely electronic and not dust... so with resignation I power down Windows, shut off the machine and unplug it and proceed to remove the PSU.

Upon unplugging the EPS12V connector I immediately find the cause, and it isn't pretty. After removing the motherboard I could see the full extent of the damage.

What is interesting is that this was a gradual overload, not a surge. Secondly I never actually once noticed smoke, and the smell was never very strong. Technically both the board and PSU still work (remember I shut the PC down normally), but suffice to say I am using a different board and PSU both now because of this. Having just had to fight to unlatch the connector inside the case I knew it was fully seated and latched, so that wasn't the cause.

From the looks of it I am guessing that there was a load imbalance on the middle 12V wire, as it is noticeably more melted than the others, which are mostly scorched/singed than anything else. This wire started to melt, and when it failed the current load jumped to the remaining three 12V wires instead, which promptly were overloaded and began to fail in turn.

As for how/why this occurred, I am not sure and really would like to ascertain the cause. Could the motherboard have imbalanced the CPU power load onto one pin, or the PSU itself? Furthermore this event strongly reminds me of an ad campaign Antec was running about not all power supplies using OCP on the 12V rails... after seeing this I am wondering if this unit lacked any 12V OCP protection for the EPS12V connection. Given the reputation of PC Power & Cooling, I find it highly surprising if this was the case.

Hm, high voltage warning. Yeah I can believe that. I wonder if the name means 750 watts delivered straight to the EPS12V connector? :p

So, end of the story... I'm now considering Corsair's new line of AX750-AX1200 professional grade power supplies. The Core i7 920 & a borrowed Core i7 930 were used during the CM UCP escapades, but it appears the recent addition of a 980X was a bit more than the PC Power & Cooling unit could handle.... thankfully no 980X processors were harmed in the making of this thread. :eek:
Last edited:


Soup Nazi
How many times have you used that Seasonic 750W? I say Seasonic as they're the OEM of the Silencer series of PC P&C. After several rounds of plugging and unplugging a PSU from components the pins will open up and cause the current to go beyond tolerable levels which results in what we've seen here.

Aside from the smoked EPS connector that PSU is fine. You could likely return it and have them replace the connector if you wanted to salvage it.


Techgage Staff
Staff member
How many? For every month I don't mess with it, there probably was another I performed multiple installs. The connector fit tightly, it's always the most difficult one to remove partially due to the positioning in the case/board., and partly because it had such tight tolerances.

So your believe one or two of the pins shifted and wasn't making proper contact? This is interesting as I have observed this problem with basic 4-pin molex connectors all the time. I frequently have to reseat those because a pin will get pushed out of the plug itself, which isn't just annoying but extremely dangerous. The thing is the pins on these power connectors have much tighter tolerances than any molex, and after checking again to be sure, none of the prongs have been pushed back from the plug even slightly as visible in the photos.

You're right, the unit can probably be repaired given it doesn't appear to have overloaded the PSU itself, just the motherboard. But I don't trust the PSU anymore, and I'm not sure it's worth the repair cost + shipping ontop of that. I'd like to know with dead certainty what occurred here before I'd even consider using it again on non-junk hardware.

That said, I am using a spare CM 700W UCP I had as my last 24pin capable PSU right now, and with my old X58 Gigabyte motherboard. Exact same overclock settings on the 980X, but even using 12 threads of LinX the EPS12V connector is not anywhere close to overheating. Something was definitely wrong with the RIIE, the EPS12V plug/connector, or the PSU itself that caused it to nearly fry.


Soup Nazi
No, I don't think the pin got shoved out, I think it got opened up some causing it to arc and get hot and causing the other three to pass more current than they're rated for which caused the damage to them.

I've seen the same thing happen on ATX +12V 4 pin connectors as well as PCI-e 6 pin connectors. They're all passing a good deal of current on high end parts and when one pin fails the rest follow suit especially when the connector is pulling max draw or close to it.

Running that PSU on a lower end CPU would most likely have never been an issue. When you stressed it with more draw, well the results speak for themselves.


Techgage Staff
Staff member
Ah, alright. Thanks for your input Madmat, it's certainly possible and it does look like one pin failed first causing the rest to follow.

From the results I've seen the 980X draws the same power as a Core i7 975, at least at stock. I've been running the 980X at the exact same overclock as my 920... I guess I should actually grab an extension cord so I can test draw from the wall with a P3 power meter.


Tech Monkey
I just got around to reading this thread. I am a connector guy, meaning that I analyze via software, all kinds of connectors for a living. Typically for high frequency & wide bandwidth, but analysis of current carrying capacity versus temperature rise does occur (DC in other words).

That said ... the connector appears to have experience considerable i2r (current squared times resistance) power loss. Translating that, the connector got hot (obviously) and per your description it has been subject to more wear than typical. And, that means the original plating has probably worn off and/or any corrosion (we call them contaminants) accumulated to cause excessive contact to contact (the "point" of contact) resistance.

You guys that make the multiple installs for reviews, aka plugging/unplugging the PSUs, I would advise to plan to be very familiar with connector pins & the appropriate tool(s) to properly crimp them ... solder is better. Although if it actually the contact to contact resistance being the issue, then soldering the wire to the pin probably won't help ... but it could be either until you really know. It seems to me that an expensive tool could be shared amongst you via UPS, Fedex, or what ever. Just my $.02