Hi all,

I just purchased the following:
3.3 GHz Intel 2500K CPU, LGA1156 socket
Asus P8P67 motherboard
2 x 4GB Kingston HyperX DDR3 1600 RAM
Saphire 6950 2GB GPU

I am putting that together with the following existing components:
OCZ "GameXStream" OCZGXS700 700W PSU
SATA DVD Burner
various SATA HDDs
Antec AMG3700 case

Last night I was feeling greedy so I put everything together in the case, connected all of the headers, plugged in all of the drives, etc. (I know, that's just asking for trouble). Trouble is what I got; a first press of the power button got a quick jerk out of some of the fans, and then nothing. Anything after the first press had no response whatsoever; I had to cycle the PSU's power switch to repeat this cycle. Suspecting power trouble, I removed the power-hungry 6950 GPU and replaced it with a passively cooled nVidia 8400GT. Attempting to power on again, I got the same symptoms.

It was late, so I went to bed, and I tried a more scientific approach tonight. I now have the motherboard out of the case, with only the CPU, CPU heatsink/fan, and one stick of memory installed. The PSU's 24-pin ATX cable and the dual ATX 12V connectors are plugged into the motherboard. With or without the 8400GT installed, I get the same symptoms - only the briefest turn of the fans before everything turns off again, and this only happens once between PSU power switch cycles.

I have removed the CPU heatsink, reseated the CPU, and reaseated all of the power cables. The same symptoms persist. I have a strong feeling that this high-quality 700W PSU is certainly up to the task of starting up a lightly loaded 2500K, P8P67 motherboard, and 1 stick of RAM, without any hard drives or any GPU.

RMA motherboard?
RMA CPU? (needs to be done quickly because of Newegg's 30-day CPU RMA policy)
RMA both?

I'm not an Intel expert. Maybe you've tried this:

Reset BIOS using on board jumper or push button. Maybe the BIOS jumper is already "mis-located". Maybe the BIOS battery is not secure or is dead (does not seem likely -- but, hey, we're looking for fixes that are less trouble than an RMA -- I fixed a friends machine with a battery problem -- just a few minutes to steal a battery from another MB).

Try a different PSU if you have one -- just to verify that component is not the issue. A bare system should boot on less than 250W -- heck my entire HTPC with Phenom II quad runs on a 300W PSU.

Try a different GPU or the built in one (doesn't Sandy Bridge have some sort of GPU built in?? -- I thought one model did). Again to eliminate GPU as an issue.

Try a different memory stick just in case you picked the "bad" one.

Some other set up switch on the MB not clicked in the right direction?


I think the BIOS itself would partly POST and put something on the screen -- if the CPU were bad. So, if none of the above works, and the system does not ever even begin to boot, maybe its the MB? I've had that symptom many years ago and it was the MB in that case (again, I don't know Intel MB peculiarities).

OK, update...


I put the old system back together so I can at least have a computer for basic tasks.
Asus P5P motherboard
Intel Wolfdale 2.53 GHz CPU, LGA 775
2 x 2GB DDR2 RAM

With the same OCZGXS700 PSU, when I power on the CPU fan kicks and then stops, and the PSU fan turns, but the LEDs inside the PSU don't stay lit; they flicker on and off. Something is definitely up with the power supply...

I have a 350W Raidmax PSU that's running my server. It has one ATX12V plug. I don't believe its 350W rating, but I will yank it and ty it on both motherboards.

OK... the el crapo Raidmax "350W" PSU will POST both systems (old and new)... even though it only has a 20-pin ATX connector and a single 12V ATX connector (both boards have 24-pin ATX receptacles and the new board has an 8-pin 12V ATX connection).

So, it's the OCZ PSU causing my problems. Any idea what I did to it and how I undo it?

Seriously, it worked fine in the old system, I took apart the old system, built the new system, and now it won't work in either system. How does that happen?

Allen, thanks for the prodding to try a different PSU.

EDIT: The 700W PSU does not respond to the "paperclip trick" (shorting the green cable to ground to get the PSU to turn on while it's not connected to a motherboard) so unfortunately I think it kicked the bucket. I am all ears for any possible revival techniques.

Hello Joe - Yes, your OCZ can be brought back to life but you must have a fair knowledge of electricity and electronics first. You will also need a decent multimeter and quality small gauge soldering iron. For most, the time input is simply not worth the money they spent to purchase the PSU in the first place. About 4 to 20 people also die in the USA every year when they exceed their own knowledge limits and make contact with the larger capacitors and leave their souls moving heavenward while their willies are left behind blackened and charred. Simply put, if you dont know what you are doing, I dont recommend it. And from what you describe, there is no secret 'jump-start' trick to easily fix your inexpensive Chinese made OCZ. It sounds to me like the AC side is fine but you have a DC distribution problem that needs circuit checking. It could be a simple bad solder or a much more complex board problem. If after reading this, you want to possibly invest 3 to 30 hours to fix your 100 dollar unit, PM me and I will give you a rundown of all the steps.

S! - WD
(.... yes, I received a BSEE)

Interesting that the PSU should go bad while "moving it". Still, stuff happens -- i.e. it had to be something.

If you decide to buy a new one, some suggestions (which you may already know):

650W rating at 40C+ is enough (it's what runs my system). My full system with the OC hex core CPU at 100 percent and the Crossfired GPUs at 100 percent draws only around 500W (by actual measurement). And, AMD says 650W runs a system with two Crossfired HD5870s. Some PSUs still are rated for room temperature and can fail when supplying rated voltage at system temperatures -- or voltage can drop.

Get a single-rail PSU. Older high wattage PSU did it by splitting the 12V power into rails. Rails can be overloaded. That reduced the power available to the GPU and some GPUs need a lot of power these days. Or if too many lower power components were put on a single rail the power limit of the rail could be exceeded.

My XFX is made by Seasonic and it runs my system below, so I can vouch for it. Seasonic is one of the highly regarded manufacturers of PSUs. Rated at 650W at 50C. Usually on sale. Today its $73 delivered.

Anyhow, there are other good PSUs. I would get one from Newegg for $73 or less (now that I know I could get the XFX for that). Brand name, good customer comments, rated at least 650W at 40C+. PSUs are not a place to skimp with an "off brand". BTW not all XFX are made by Seasonic -- so, I cannot vouch for other models of XFX (or anything else I haven't personally owned).

XFX PSU

Once it powers up, you'll have a great system

Just a random thought, could there be a dip Switch on the back (that one that selects US/ EU Power)? and could it have been moved to the wrong position or halfway?

other than that it definitely looks like it failed.

SkateZilla,

I thought of that. This PSU is a "full range" PSU, however, with no voltage switch. It accepts AC input from 110V to 240V.

New PSU on the way soon. Thanks for the recommendation, Allen; that looks like a good one and the price is right. Unfortunately I won't have it for this weekend's SimHQ race, so I need to come up with an alternate method of participating... I will see if my old motherboard can drive the 8400GT in a gaming environment with only a 20-pin ATX connection and graphics options turned down.

WhistlinggDeath,

Thanks for the offer. While I do think I could try to fix it with guided assistance, it's not worth it to me.

When a guy named "WhistlinggDeath" warns you about capacitors killing you - I think it is just best to step away from the computer. Sounds too much like a movie setup to me..

Wow, some bad luck.

Yeah, kinda. It was pointed out to me today, though, that components like PSUs tend to die when "given an opportunity". This particular system hadn't been rebooted in about 2 months, and even before then reboots were almost always warm or near-warm.

Once my SSD died, I had my system shut down for 6 or 7 days, with the PSU unplugged. So for all I know it may have died anytime in the last few months; it was just waiting to get cold.

Joe - No worries, for most the effort involved is not worth the money they spent on the PSU in the first place; dont blame you. Even experienced techs and EE students can have trouble rebuilding a PSU because they dont have the board logic available to em usually. Also, no disrespect, but things dont die "when given an opportunity"; they are (with a few small exceptions) digital devices and current either flows or it does not with no in-between. Of course heat, over-volting, high loads, etc ... exacerbate flaws, but under normal conditions the opportunity is usually to perform normally.

S! - WD

BTW - You, Dart and Ken King were never on my SimHQ shoot down on site in HL list. I actually found many of your articles to be thorough and good reads, ... and I am pretty da*n picky. Keep up the good work.

Thanks for the kind words regarding articles.

My understanding from computer professionals (I ma not one) is that components are more likely to die when they change state. For example, the server tech for the small company at which I work recommends that we reboot our server, if necessary, on Friday afternoon, not Monday morning. Components are most likely to malfunction when being power-cycled, and thus if our server goes down on Friday afternoon we have two days to fix it before it gets in the way of office production on Monday morning. This was the concept I applied to the failed PSU.

It is what I mentioned above with specific regards to load. And yes, I agree with your server admin. For example, you come into work on Monday morning, boot up a system and suddenly dozens to hundreds of capacitors have a differential placed across them very quickly. If for some reason that load is slightly above tolerances then of course, .... failure. That being said, decent room with average temps, no super overclocking or over volting, and some fans, the expectation is for business as usual.

Thanks to this thread I was able to get my new rig up and running.

I too have the same power supply as the OP and save for the memory, my system is exactly the same.
Had the same problem....was it the boards? the CPU?

I ended up sending two of the Asus boards back and then switched to Gigabyte P67X model. Still ran into the same problem..except no posting to bios.

I thought for sure it was the CPU, until I saw this thread.

So I ended up buying a Thermaltake TR2RX, 750Watt PS. Computer booted right up and I was able to finish up my build.

Interesting...

For me the trouble continues. I got an RMAed SSD last week (the failed component that started this whole upgrade process) and the SSD that OCZ shipped me didn't work! It wasn't recognized as detected on my new system and it would prevent my old system from completing a POST.

A second RMAed SSD showed up at my office today, so hopefully the end is near. If this one doesn't work I'm gonna go looney.