Diagnostics advice please.

[JonJ678]

Hey. Having some trouble with my computer. I've been updating this as I go along, to keep things clear in my mind as much as to ask for help. It's become quite long, hopefully it's sensibly formatted now. Testing method is now listed at the bottom, under subheading 'attempts'.

I have a working amd system, 7750be, ecs board, 1gb ddr2. I have a broken intel system, q9550, new p5q premium, 2x2gb ocz reaper. Each are using a pc p&c 860W. Excellent psu.

I believe one psu to be partially damaged. The other I assume is fine as it has suffered through no trauma.
I believe both sets of ram to be fine, as the suspect one passes memtest. I will test at 1066 when I can.
I assume both motherboards are fine, as one has given no problems and the other is new.
The graphics card is known to be working, tested on the amd system. Similarly the monitor works.
I think the q9550 is dead, and am running out of ideas for how to test this.

Background
This is after an rma of the motherboard, the previous one died with a burning electrical smell. I wanted to send the processor and ram along with it for rma, as burning electrics suggested to me that other components may have died. The supplier declined this request. I hope that the P5Q is not DOA, I'm reasonably sure it won't be.

Conclusions
I'm getting there, but very slowly. Mostly ruled out ram, as it runs fine at 800 at least. Gfx checked. Can't test cpu or motherboard. One psu known good, the other has a broken 8 pin line.
I think the processor is dead, but would like to test further before rma'ing it. Anyone sent one back direct to intel? Bought october last year, retail, so warranty should be good. Not lapped thankfully.

Thus far I have not managed to get the machine to post, the closest I've come is fans spinning and a pump working off a molex connector. No beeps at any point either, and the speaker is connected.

Attempt 1
Leds on motherboard are lit up as they should be
When pressing power, nothing occurs. No beeps, neither psu nor cpu fan spin.
This is with processor, ram, heatsink, psu, motherboard.
If ram is removed, no beeps.
If psu is started using a wire, psu fan spins.
Next, I've tried it with 1gb of generic ram, and it does better. Fans spin, no post. This points at the ram being faulty. However, I am currently running said 'faulty' ram in the machine I took the 1gb stick from. It passes half an hour of memtest at 800mhz (rated at 1066) and I can't test it at the higher speed in this box.

Attempt 2
This differs from the previous in that I'm using known-good ram, have connected a graphics card and a monitor.
All fans spin and lights as appropriate. No image on the screen. After 5 seconds approx the fans turn off briefly then spin up again, I think this is an oddity of my graphics card though as I've seen it before. No beeps.
Removed the ram. Still no beeps.

Attempt 3
Changed monitor cable. Fans no longer spin at all, but I'm certain I didn't change anything else. Removed card, and removed ram, no change. Computer is now back to doing nothing whatsoever when plugged in.

Attempt 4
Attached graphics card to secondary computer. Bit of a mission as it's water cooled and the amd computer isn't. Typing on the result now, so I'm going to rule out damaged graphics card. Again this is using the better/possibly damaged set of ram.

Attempt 5
Swapped psus over. My amd machine is now running quite happily, apparently it will run on either psu with no ill effects. Neither will run the intel system, but in different ways.This is with the known good ram.
PSU 1/ No life when power is pushed whatsoever. PSU 2/Lights on, fans spin, pump works, no post/beeps/life on screen. Propose that part of my power supply burnt out when the system died, and the amd system does not use this part. This is supported by the 4+4 pin cable working fine with the intel system, and either half of it working fine with the amd, but the 8 pin cable not working with the intel system. Credit to pc p&c, part of their psu is dead and the rest has kept right on going.

appendix
Q9550 [no means of testing]
Asus P5Q [just back from rma]
4gb ocz reaper [passed 30 mins memtest and normal use in secondary machine]
PC P&C 860W [one working fine, second has a broken 8 pin cable but otherwise looks good]
8800gt [tested and working in secondary machine]

What can I test further/what conclusions may I draw?

It's been a long night playing with this. I've sent ocuk a webnote to see what they suggest wrt rma.

 

[JonJ678]

Unfortunately students with desktops instead of laptops are rare, and students who haven't vanished home for the summer are rarer still. I'm trying to track down an available 775 processor but not having enormous luck so far. It's probably possible to pick up a 775 pentium for very little money if borrowing fails.

I'm fairly confident that it's narrowed down to either motherboard or cpu, the ram may also be faulty but as I'm currently using a different set I'm ignoring this. Reassuring to hear you've reached the same conclusion as to where to place blame.

Difficult. Cpus very rarely die, but the cpu power cable on the psu is the dead one and the motherboard is a new one. What should a motherboard be expected to do with a dead cpu?

 

[JonJ678]

sounds like the motherboard it should still post will no cpu or at least beep

Are you sure on this one? I'm struggling to find anything on it. If a motherboard with no cpu in should beep and mine doesn't, that's a strong hint that its the board that's down which would be the first piece of evidence either way. My motherboards manual says, for error codes,

one beep, all good
continuous then two short, repeated, no memory
continuous then three short, repeated, no gfx card
continuous then four short, repeated, 'hardware failure'

Which suggests it has no way of telling if the cpu is absent. I'll try this later, worth a shot. If it beeps without the cpu but not with it, we have our first real piece of evidence.

I still wouldn’t rule out memory incompatibly as a long shot - as memory is a b*stard and if it can rock the boat it usually will find a way.

It is OCZ with Asus, which has a bad rep. The memory was fully compatible with its predecessor, but may be now faulty and can't be trusted. The 1gb stick of Kingston is unlikely to be incompatible though, would be very bad luck if one set is damaged in a way such as to become incompatible and the other just doesn't work out of the box.
I liked your turn of phrase here, it made a stressed man smile.

If you can’t get your hands on a cpu via any of your student colleagues, I would try getting hold of a 4-Pin Power Supply to 8-Pin ATX motherboard adapter just so you can stop switching PSU’s over to run a tests.

I'm moved to using the known good psu with the Intel machine largely for this reason, the maimed one is running the amd system as, to be honest, I have no idea how damaged it is and I'd rather it killed the £120 amd system. I'll open it up and look for signs of damage in a bit. Hopefully members market will loan me a cpu for a fee, Ive put a note up there in case.

What does your gut tell you that the problem is? You probably suspect one component more than another, even if you can’t explain why…

Processor. The 8 pin cable burning out is the only hint I've got either way; I consider it a fairly firm one. The board shorted near the top end of the ram slots, towards the cpu, and the 8 pin cable largely powers the cpu. That suggests it briefly had to cope with enough voltage to burn out a fuse/cable in the psu which could easily knock the processor flat.

I wish I was good enough at electronics to trace the region of failure back to which components it connected, ram to cpu is based only on physical location and the one patch of damage I could see. I'm quite surprised the ram is working at all really, let alone passing memtest at 800 5-5-5-15. All 4gb recognised too.


Cheers for the responses guys. I'll see if something comes up on MM, otherwise I'm leaning towards rma with a note saying "I'm sorry if this processor isn't dead, however I'll be livid if you replaced my dead motherboard with another dead motherboard", though more sensibly worded. It's likely to take a similar length of time to tracking down a processor or motherboard anyway, and I think ocuk will waive the tenner fee if I rma a working component under the circumstances.



p.s. The processor was clocked at 3.6ish a while ago, but was actually down to stock when this occurred. Was just planning to have another attempt at this, I think the limit was my refusal to go over 1.35V or use load line calibration rather than anything else but I had hoped not.

 

[JonJ678]

Ah well. Taking the cpu out didn't change anything, but had to be worth a try. At least it's being very consistent, if given a working psu it will light up. If power is pressed, fans will spin. It will do nothing else, whatever is attached or missing

I asked ocuk, and received:

1) Clear the CMOS on the motherboard (This can be found in the motherboard manual)

2) Strip the machine back to the minimal components needed to boot the system.
The Motherboard, CPU, PSU, GFX card & 1 stick of memory (in slot 1)

3) Rotate the memory stick through slots 1 - 4 until you have a picture. If still nothing then change to a different memory module.

In fairness I hadn't reset the cmos. Didn't help mind, but still. I love the optimism of 'until you have a picture'. So ocuk are also inclined to blame the memory. Hopefully they'll have further ideas.

The worst case scenario is that you have a DOA MB and you have a corrupt CPU that you have yet to identify due to lack of known working components to test it in. All coupled with the added doubt of memory incompatibility. Christ that is a depressing paragraph… :/

Thanks, that's a much nicer way of looking at it. It is indeed possible that all of it is dead, still. Don't know whether to laugh or cry.

I think we all have our pet hates. I mistrust hard drives, but then memory hasn't been too cruel to me yet. I just don't understand how a file can sit on a hard drive for months then be corrupt when I try to open it. At least with a cd you can blame the scratches in the surface when it inexplicably no longer works.

I'm certainly learning things, and that's always good. An ek bolt down kit works well with a true, but you can no longer fit a fan. A motherboard, my one at least, doesn't beep to complain of a missing cpu but will object to a missing gpu. Power supplies can burn out in part when the rest keeps on working. All good general knowledge, we shall see where this leads. Perhaps I'll become one of the elite few to have a cpu die at stock settings.

 

[JonJ678]

Happy to hear another vote for cpu. Not heard anything back from ocuk and its the end of the working day, so I'll see what tomorrow brings. Asking rjkoneill is a good suggestion. I'll drop him an email tomorrow. Too late for me to write coherently now.

Cheers guys

 

[JonJ678]

You're encouraging me towards windows 7, I'm encouraging you towards ubuntu. Fair exchange, we'll both learn from it

Fairly stuck. There's a processor on offer in the mm for 10 inc, if ocuk are prepared to offer to test & replace the cpu if faulty that remains the best option. It would also mean I'd have some form of computer working while intel take however long they see fit for the rma.

So far the retailer has offered me an rma on the new motherboard and refused me one on the cpu, I suspect I'm going to end up rma'ing it to intel and sulking. Both retailer and intel are pointing at the other guy and saying 'it's his problem' which is a bit rubbish, I may be going down the bleak road of legal advice.

Unemployed student about to spend a summer paying rent without a student loan. Poor does not start to cover it, tomorrow I'm going to spend a day in town offering to work as absolutely anything. A tenner on the credit card is a hit I can just about take but am not pleased about.


I suspected other components to be dead from the start (burning electrical smell and bare copper newly visible on the board...), and asked to send mb/cpu/ram all off to be tested. The retailer declined, and so all this started. What do people think about this? I understand opposing multiple item rmas, but equally it wouldn't take more than a couple of minutes longer to test the cpu as well once the components were on the bench, and it would have saved me many hours.
I feel they should have offered to test the cpu in the first place. Instead they're still declining to, and I'm confused and annoyed by this.

 

[JonJ678]

It doesn't power up almighty15. Indeed it behaves exactly the same with the cpu removed as it does with it in place.

I've bought a pentium 4 for a tenner so shall see if the computer works with this. If it does, I'm going to find out how long a rma to intel takes, and learn to overclock pentium 4s in the mean time. Should be here in a couple of days.

I've decided to be rational about this. The entire rma process has been handled by a single guy, and it's unfair to assume the rest of the company behaves the same way. Considering future course of action, suspect I'm going to register an official complaint against said guy.

 

[JonJ678]

I wish to state clearly that this problem is not with OcUK.

Oh my. I have indeed got an update, new processor arrived today (cheers optitech, ill leave trust for you shortly). In short, the computer still does not work. I'll recap, for my own clarity as much as anything else.

I have a tested and known working graphics card, psu, ram, hard drive. Indeed I'm using these now with a fairly terrible amd system. I have a diagnosed faulty q9550 and p5q motherboard, and a known working pentium 4.

Motherboard + either processor = fans spin up, no post.
Mother board + q9550 - ram = fans spin up, no post
Motherboard + P4 -ram = fans spin up, motherboard beeps that it has no ram
Motherboard + either processor -graphics card = fans spin up, no post

Lack of graphics card should lead to beeps according to the manual. From this I conclude that the q9550 is more dead than the pentium 4. As the processor is known working, and as the board does not beep as it is supposed to without gfx card, I conclude that the board is faulty.

I therefore have a dead q9550 which has been repeatedly refused for rma. I also have a dead motherboard, despite being freshly back from rma. I am postponing my complaint as at present I think all I'd do is scream abuse down the phone. I'll register a formal complaint via the website complaints system, give it several hours then phone and try not to scream at them.

As an interesting example, here is the last exchange. Names removed, but to the letter how the exchange went.

8pm Thursday last week, I sent

--------------------------------------------------------

I do not believe the motherboard to be at fault.

I remind you that I first asked for the cpu to be tested alongside the motherboard and you declined. You therefore knew that I have reason to believe the cpu to be faulty. I cannot believe that *blanked out* would ship an untested motherboard to a customer under these circumstances, especially when this action required formal complaint to achieve.

I formally request to return an almost certainly dead Q9550 to you for you to test and replace if faulty. I am not interested in waiting for 2 weeks for you to test it.

Intel are prepared to accept a direct rma if the retailer sees fit to shirk this duty.

Regards.

--------------------------------------------------------

This was not friendly, but given this is now over a month on I consider it reasonable. Here, astonishingly, is the response I got at 5pm today. Monday.

--------------------------------------------------------

Thank you for your webnote, I apologise for any inconvenience caused by this. and im sorry for not been able to accept both items back at the same time for testing together sir. this is something we are looking into benn able to do in the very near futher.

--------------------------------------------------------

I kid you not, that is word for word the response I received today. Spelling mistakes and all.

 

[JonJ678]

Ah curses. I was sure diagnositcs were finished, but you're quite right. The cpu may still be working, though the evidence that it isn't is starting to stack up somewhat. I think I draw the line at buying a new motherboard for diagnostics however. I'm distracting myself by watercooling the 'temporary' amd mb/ram/cpu in the meantime, it's slow but it's definitely quiet. I remain astonished that they didn't test the board before shipping it.

I've dealt with two people so far, the first was well spoken but didn't resolve anything, the second is quoted above. There was also a formal complaint dealt with by management which lead to sending me a new board the next day, fair play for that. Shame they sent me a dead one though.

Previously I've had no problems with the company at all, indeed a question I asked last week about which heatsinks were on their current batch of reaper was quickly and professionally answered. The fellow actually checked the stock in the warehouse for me. Bizarre.

I've submitted a complaint which should be dealt with by management, as keeping everything written spares me recording phone calls. No word yet, but they've only had it for half a working day. On the bright side my girlfriend is very excited by all this as she's training to be a lawyer, and I haven't seen her show interest in my computer since I built it. I'm pretty confident small claims will back me on this, but I like the company and still don't really want to cease trading with them.

A final complication is that the P5Q premium which I'm still very fond of is just starting to be difficult to find, the retailer don't stock it anymore. Quite exciting though, it might finally all be resolved this afternoon.

Cheers for all your help Plec, very grateful for it.


edit: The potentially suspect ram causes my amd machine to shut down under prime 95 blend, stock settings except for ram running slower than stock. So that's not looking good either.
Close of business today, no response.

Amd machine now refuses to post with the ocz ram, and then refused to post with the 1gb stick. Scared the hell out of me, gone back to using the 1gb stick it came with after a cmos reset and trying to post with no ram. I'm not putting the ocz ram near another computer, it just isn't worth it.

 

[JonJ678]

I don't quite follow you here. When disaster struck the ram, processor, motherboard, psu and mb speaker were connected. I'm very grateful my graphics card wasn't. Now that the ram is being flakey in a different system, why would that count as further evidence against the processor?

I'd have thought the motherboard blowing could hurt the ram without involving the cpu at all, so I'm interested as to your reasoning here.


8 pin cable burning out
same behvaiour with cpu present as absent
working processor getting more life out of the board
and now perhaps the faulty ram

not sure I have any more evidence against it yet, might post it to a friend for testing

edit: actually it's possible that I was sent a working motherboard that was then savaged by the processor, so if retailer had tested it perhaps I have an answer to whether a faulty processor can break a motherboard. This would be interesting, and probably worth knowing.

 

[JonJ678]

Epic post, thank you.

Flaky seems a fair description. It passes memtest but the system crashes and burns during prime blend, whereas it does just fine with the 1gb stick. I think if the ram was briefly hit by a large voltage it's reasonable for it to continue to function, just not very reliably. Prior to this it was just fine, running quite happily at 1100 or so undervolted. So yep, I think the motherboard hurt it.

Motherboard giving up on voltage regulation does sound about right now I think about it. It must have drawn a hell of a lot of current to burn out the 8 pin cable though, the psu is single rail so it's the cable itself which died. Credit to the psu, if it can output enough power to burn through its own cables Im quite impressed. It might be much less exciting and just an internal fuse of course, hard to check.

I'm glad you believe the cpu is not a motherboard killer. I'm worried enough that it might be that I won't put it in a friends computer, I hope when it eventually gets rma'd it doesn't kill any test equipment.

I'm not sure I'm after a break, I'd like the retailer to pull their finger out though. They've had a formal complaint for three working days now and a post in their support forums since the 24th. Taking the **** really. I shall have to find a phone that isn't on pay as you go and call them.

Cheers again man

Now four working days.

Phoned today (saturday). A suitable manager isn't in, I've left my number and with some luck they'll call on monday.

 

[JonJ678]

Yeah, its a bump. Significant update though, and beats making a new thread.

A week ago I was asked to send the board and processor back for rma. I replied to noreply@etailer, and then today started writing an angry email because id been ignored. Thankfully I caught my mistake and sent a slightly more sheepish one along with the one Id failed to send last week.

Ram has not been rma'd, processor and board were shipped back yesterday and probably got there today. Postage is hopefully being refunded, we'll see how that goes.

I still want to change to gigabyte, as id rather rma to manufacturer and I believe theyre in milton keynes. On the other hand, I really liked that board before it killed everything, and quite want the system I was so fond of two months ago back. So probably happy with either. If I'm sent another dead motherboard I'm going to go ******* mental.

Happy days

 

[JonJ678]

Hey man. No news back from them yet, sadly but unsurprisingly. I shall be patient for a little while further, then try phoning again. Last time I was told the manager was out, and he'd call me back... I'm still waiting on that.

I hope to be moving to i7, but we shall see. Strange that my rma doesn't have any sort of priority associated with it after this long. I think the minimum list of parts Id need would be memory, backplate, mounting plate. The worry isn't the pump (I'd hope the 18W ddc can cope) but the radiators, as apparently i7 runs very hot. It's a feser 240 and two feser 120s stacked, so pushing my luck a bit. Even on the stock cooler it would do better with cad than the current box, and that's getting a bit pressing now.

Sad to hear you're going to be deprived of internet access, I hope this means you're going somewhere exciting rather than impending electrical failure. I'll let you know how this turns out, enjoy your trip

 

[JonJ678]

Well, an update at last. It looks like they're going to make it to the 2 month anniversary sadly, but I have a gigabyte UD5 and an i7 D0 stepping heading my way. So that's good. I need to find the bracket for my waterblock and some ddr3 ram. Unfortunately the corsair I've set my heart on is about £100 more than I can spare at present, trying to work this one out. I might try selling absolutely everything I have remaining that isn't actually used in my computer. Needs some thought.

On a related note, the psu that experienced this hit in the first place gave up the ghost today. Computer does nothing whatsoever with this one connected, but works fine with a different one so that's definite. That puts it at motherboard, processor and psu which died when the board went. Fairly certain the ram has had it but not found a way of testing it. Still, a definite step forwards.

 

[JonJ678]

Happy owner of a gigabyte UD5 and i7, D0 stepping. Process took just over two months. Google finds nothing of use from the batch number, so we'll see how it goes.

Just need to get some ddr3 and I'll have an actual working computer again. Happy days

 

[JonJ678]

Glad to see you've found broadband again, enjoy your trip?

It's nice to be proved right, and definitely nice to have a swifter computer. All worked out well. I found myself part time employment and am using the money to buy pieces of computer. Happy days.

6gb of corsair sitting in a board being inefficiently and noisily cooled by the stock hsf after I foolishly overtorqued an ek block. Still, it's running at last and I've got a couple of weeks to learn the bios before putting the board under water. plus a consequence of the breakage is that I can justify importing this:

Which is just beautiful. I'm a happy man

edit: shame on me if I can't take it to 4ghz, after some reading around I'm pretty sure a 240 and two 120 radiators are going to cope just fine. If the £50 corsair effort can do it I'm sure my loop can. Worst case is I'll need a second pump to overcome the ludicrously restrictive loop, and end up moving one of the 120's to a more sensible place. I'm glad I'm using an 8800gt though.

 

[JonJ678]

Thanks for the kind words.

I'm still withholding judgement on the corsair. It has a couple of issues that I haven't seen answered. No means to replace the liquid which evaporates out through the tubes is one, and how what is essentially a 120 radiator with a pump is great but a normal 240 loop is 'adequate' for the second. Surface area is the king here, and the corsair has half the recommended. So I will wait and see.

Cheers, I'm looking forward to getting back to water. Going to order the remaining pieces today I think. The intel stock fan is loud, and I've caught a cable in it once already leading to the computer shutting down. No harm done, but water feels safer. Mainly moved to it to for myself to learn something about thermofluids

I think I'll post a build log of it, not sure it'll attract any interest since I probably don't have any questions to ask, but I'll enjoy writing it.

Shame the die cast top is going to take another couple of weeks to get to me, I think I'm going to order the chipset blocks today.

How did the 12gb i7 monster you were building for your friend work out? If I missed the build log I'd love a link to it

 

[JonJ678]

I'm pleased that it's delayed rather than cancelled, looking forward to the build. SSDs really are amazing, my one is in the step mothers new computer at present because I failed to get XP onto their hard drive quickly enough. She's currently the happy owner of a ridiculously overspecced office pc (antec P180, e8400, 30gb vertex), and I think will be sad when the ssd is gone.

I dont think I count as needing 12gb of ram as such either, I'll find uses for it certainly but I'm currently on 6gb with no real issues. I think its the empty slots which get to me, a sense that it's still not finished. I have a suspicion that at no point in the last 18 months have I had a computer which is not partially disassembled, really pushing to have this done before uni starts up again. Looks possible.

I fear the reservoir idea is actually a lot more boring than it sounds. Screwing a funnel into a fillport would achieve basically the same thing. However I don't trust the xspc reservoir, and am unlikely to use it in the final spec. There's a very good thread on here about water pressure, but the conclusion from it was pretty much that standard practice is so for good reason. The debate is whether a second ddc would improve a very restrictive loop. I'm going to solve this experimentally since I can't work it out. Loop temps are either flow rate or radiator limited. So I can undervolt the pump and the fans independently and determine from this whether another pump is worth it. Hard to say at present.

Not sure I'm going to end up doing anything particularly strange this time around. I'm putting daft numbers of things onto a single loop, and seriously considering watercooling part of the psu, but thats about it. I've made a very crude start on a build log, but its nothing remotely exciting yet. Still in the planning what tests to do phase. My ek block is going to be lapped, and I'm going to try liquid metal tim for alloying fun.

Actually I didn't rma to intel (thankfully it turns out), I had a long chat with one of their support people when considering using 'liquid metal'. Its a gallium alloy which bonds rather well to heatsinks and processors, offers significant temp decrease over as5. It's very electrically conductive, very fluid, and attacks most things. Eats aluminium at a scary rate. Also tends to cold weld the processor to the heatsink. I was trying to find out if this would void my warranty (a tim which has alloyed into the ihs ), and subsequently discovered that absolutely anything voids the warranty so I may as well lap it.

With regards the beautiful copper, I'm struggling to import it. No sign of it yet, and no eta. So hopefully a cheerful Polish chap is going to machine one for me and ship it. Roughly the same cost as a genuine ek one, but equally the same cost as the copper were I to attempt it. If it's good, which I strongly suspect it will be, I may be sending him models of chipset blocks to remake in copper form. Anyhow, the cpu is likely to be lapped, with gallium on it, then a milled and lapped ek supreme with a copper top. So the processor is damned well going to be cold

Off to write up a list of tests to run and order chipset blocks. Catch you later man