How to troubleshoot semi-dead system?

[rtsai]

Last night my machine died, somehow. It's not completely dead, because the Ethernet port LED is on, and if I turn off the PSU, the Ethernet LED goes off. However, I can't boot the computer. When I hit the power button, nothing happens.

How do I begin troubleshooting? It doesn't seem to be either the PSU or the MB, because of the Ethernet LED still working. Does that leave the CPU? Or maybe a faulty connection between the power button connector and the MB? I tried unplugging everything and plugging everything back together and it didn't seem to help.

The system:

A64 3500+ Winchester.
XP-120 + Yate Loon D12SL-12.
Soltek SL-K8TPro-939.
Antec SLK3000B.
Seasonic Super Silencer 400 Rev. A3
2 x SP1614C (non-RAID).
Yate Loon D12SL-12's for intake and exhaust.
Generic CD-RW drive.

All fans (CPU, intake, exhaust) were undervolted to 5V.

I can't boot the system, so I can't get into the logs to see what was going on before the machine died.

[Chris.Day]

Remove everything except for minimal memory (1 stick), CPU, graphics card. Clear CMOS and see if it boots.

[andyb]

I dont like to be the bearer of bad news..........

It's not unusual for LED's to light up when the machine is NOT on, just plugged into the mains, mine powers the ethermet for example, and various other machines power the "Num-Lock" key.

Do as Chris suggests.


Andy

[rtsai]

Remove everything except for minimal memory (1 stick), CPU, graphics card. Clear CMOS and see if it boots.

I've done this (system only contains PSU, CPU, RAM, VGA, cleared CMOS, reseated RAM and PSU connectors), and only LED2 illuminates (this LED serves to say that the DIMMs are plugged in and "working"). The main LED (the one that shows POST codes during boot) briefly flickers, then remains dark.

So the PSU is definitely supplying at least minimal power. The power button is working (because LED2 reacts to me pushing it).

This says to me that at least the motherboard is having problems; it ought to be able to POST even without a CPU, right? I'm not certain; this is the first system I have built.

So is the problem more likely to be with the motherboard or with the CPU?

To ask the question another way, which one should I try to RMA for replacement first? Both were purchased in Feb 2005, so I am well beyond the standard 30-day refund policy, but I am well within the standard 1-year manufacturer's replacement policy. The motherboard purchased at NewEgg; the CPU (retail, not OEM) was purchased at ZipZoomFly. Even better, the retail CPU has a 3-year warranty instead of the 1-year warranty.

What are the symptoms of an overheated CPU (can the Winchester 3500+ A64 CPUs burn themselves to death, or do they have built-in protection)? I seriously doubt this is the case, but I'll look for symptoms anyway. I was regularly getting reported diode temperatures of 60C under load. Adjusting for the more typically reported thermistor temperatures, this should be about 40-50C, so that should be well within normal operating temperatures (spec'd max 65C, presumably thermistor temperature).

Thanks for your help!

[andyb]

It could be the PSU or mobo, or both.

Most mobo's wont post without CPU or RAM.

Have you got the extra 4 pin mobo power connector plugged in.???

The PSU could be dead, but still be supplying the power for the LED's, remember their is, 3.3V 5V and 12V.

take the heatsink off of the CPU, if the thermal paste covers all of the CPU, it was making contact, now you just have to put it back.

It could be a short, they happen from time to time, to test, take the mobo out of the case, and test it on the anti-static bag it came in on the desk.

Andy

[darthan]

Just because the LEDs on the board work doesn't mean the PSU is working perfectly, it only means the +5VSB (one of those other rails that you never really think about) is working. Try running a paper clip from the green wire on the ATX mobo connector to any black wire and see if it powers up and plug in a fan to make sure it actually provides power.

try andyb's suggestions too and then I'd say you should RMA the mobo first.

Also, no system will POST without a CPU or RAM. The Power On Self Test involves checking to make sure that both of those are working and, more importantly, a motherboard can't run its BIOS code without a CPU to execute it and RAM to put it in.

[m0002a]

If you system is dead quiet, I don't know what you are complaining about.

[rtsai]

Try running a paper clip from the green wire on the ATX mobo connector to any black wire and see if it powers up and plug in a fan to make sure it actually provides power.

I removed the 24-to-20-pin adapter and used a bent paper clip to make contact between the green wire (there's only one; it's the fourth one away from the edge) and a couple different black wires. In all cases, I got the same behavior where the PSU fan "tried" to spin up, but then died. The PSU makes a tiny electrical clicking sound, the fan maybe turns 30 degrees and then dies. None of the fans I plugged in to the PSU made any movement.

Am I doing this wrong? Or does this mean my PSU is dead? The silver lining is that Seasonic also has a 3-year RMA period (trying to remain optimistic in the face of missing however many weeks of TV that my now-dead Myth box will be missing).

Ironically, earlier I had asked about testing out a PSU without other components just to "play" with it, and had I actually followed through with it at the time (other components unexpectedly arrived the following day), I would now know what it's like to "play" with the PSU.

Thanks again!

If you system is dead quiet, I don't know what you are complaining about.

[StarfishChris]

RE: the PSU. Plug a (spare!) hard drive into it, most need loading on the 5v rail and that might be why it cuts out almost immediately.

If you system is dead quiet, I don't know what you are complaining about.

I thought the emphasis here was more on 'quiet' than 'dead'... still, his CPU is running at 0v, I calculate that as 0W dissipation, which is excellent performance for a 3500+.

[rtsai]

RE: the PSU. Plug a (spare!) hard drive into it, most need loading on the 5v rail and that might be why it cuts out almost immediately.

I actually happen to have an old Quantum 8GB 5400RPM IDE drive lying around, so I plugged that in and again used the paper clip to connect green to black. This time the PSU fan didn't do anything at all. I turned off the PSU, unplugged the hard drive, and turned the PSU on again, and the fan did its 30-degree-turn-then-die thing again.

Is this quite conclusive evidence that the PSU is dead? Now I wish I had played with it before when it was working so I'd know what to expect now. You folks have been great!

At this point I am going to try disassembling my Dell 8300 and swapping in its PSU (I'll first try the paper clip thing, then the spare hard drive, then a full system power-up). Does anyone happen to know for sure if these newer Dell machines' PSUs are standard ATX?

[rtsai]

OK. I took out my Dell PSU, plugged the old Quantum hard drive into it, used the paper clip to connect green to black, and the hard drive spun up. I think my PSU is definitely the problem here. I'm not going to bother trying to retrofit it into my Myth box (the Dell serves the more serious purpose of tracking household finances).

Thanks for everyone's help!

Next question: how ethically-bad of me would it be to run to the nearest electronics store (where I purchased this PSU), buy an interim PSU while waiting for the Seasonic RMA replacement, and then return it when the RMA comes in? I can't really bring myself to do it, but I suppose if enough other people tell me that everyone else does it ...

[m0002a]

Next question: how ethically-bad of me would it be to run to the nearest electronics store (where I purchased this PSU), buy an interim PSU while waiting for the Seasonic RMA replacement, and then return it when the RMA comes in? :( I can't really bring myself to do it, but I suppose if enough other people tell me that everyone else does it ...

No, not everyone does it. I don't know under what basis you could return the interim PSU, unless it is defective, and even then a lot of stores would only replace it.

In another thread, everyone is complaining about Newwegg.com prices, but they have a liberal return policy. If other people are abusing the return policy, it drives up the cost for eveyone else.

[rtsai]

I don't know under what basis you could return the interim PSU, unless it is defective, and even then a lot of stores would only replace it.

Actually, a different scheme doesn't involve RMA at all. I just buy an identical item from the same store, then return the old defective one as defective, using the new receipt. All I've effectively done is extend the store's nominal 30-day return policy. The store is presumably also doing returns to its suppliers for manufacturer's warranties ...

[m0002a]

Actually, a different scheme doesn't involve RMA at all. I just buy an identical item from the same store, then return the old defective one as defective, using the new receipt. All I've effectively done is extend the store's nominal 30-day return policy. The store is presumably also doing returns to its suppliers for manufacturer's warranties ... :(

That sounds more reasonable, assuming there is no problem with store tags or prices stickers. Still, not all stores will take returns on electronic gear, so check their store policy.

Personally, I would just wait, which is what I had to do recently, and suffered through a problem with the Newegg.com RMA computer system which lost any record of the item being returned. Or I would ask for a refund RMA instead of a repair RMA, and then buy the PSU locally.

[alglove]

I would also just wait. Ssome stores actually keep track of item serial numbers (barcode on box), so the "buying a new one and returning the old one with the new one's receipt" can actually backfire on you.

Have you contacted Seasonic for the RMA already? I was wondering if they could work out some cross-shipment arrangement, where they mail you the replacement immediately, asking you for a credit card number in case they do not get the old unit back. You will still be without a power supply for a few days, but it will cut down on the wait time considerably, and it is perfectly ethical.

[andyb]

What I do with most of the resellers I use in the UK with a dead product is just buy a new one, and get a fefund on the old one, as it's dead, they dont usually mind (some companies actually suggest that you do this to speed up the returns rate).

All it means to you is that you have paied for 2 PSU's and wont get the refund on the first one for a week or 2.


Andy

[rtsai]

I essentially wanted to buy a new one and get a refund on the old one. The problem is that Microcenter (the B&M store I purchased from) will only provide refunds/exchanges for products (for any reason) for up to 30 days, and it's been 4 months.

I've just gone the route of Seasonic RMA. Exercising their cross-shipment policy is not conveniently available on-line; by the time I got around to finding it, getting a phone number, etc., my return shipment could have already made it there (California mailing address). I suspect that for the hassle of making it any more conveniently available online (take a credit card number, ship to you, etc.) they may as well run their own on-line store and sell directly to consumers (in essence allowing me to buy a new one and get a refund on the old one).

I'll follow up with my thoughts on the RMA process; I'm hoping it goes as smoothly as it did for MikeC with his defective Seasonic samples way back when.

[rtsai]

I'll follow up with my thoughts on the RMA process; I'm hoping it goes as smoothly as it did for MikeC with his defective Seasonic samples way back when.

I finally received my RMA replacement today. There was no indication of their findings on my returned unit; there was only a note that said "ATE MACHINE TEST OK"; I'm not sure if that refers to the returned unit (e.g., nothing was actually wrong) or the replacement unit (e.g., they confirmed replacement was functional before shipping out).

But my replacement has the exactly same failure symptoms (won't boot computer; completing circuit fails to spin up PSU fan).

I saw in this thread that a weird combination of video card + PSU yielded a failure, so I unplugged all devices except for fans, connected PSU to motherboard, and tried to boot, again to no avail.

This is very disappointing. I am going to try swapping in the PSU from my main system to see if that works. But I assume that there are many other happy Super Silencer 400 owners?

[Ralf Hutter]

But I assume that there are many other happy Super Silencer 400 owners?

Hell yeah! I've got four of them and they all run great.

[rtsai]

I'm still waiting for a replacement motherboard from NewEgg, but I feel I must give mad props to Seasonic customer service. I had initially thought that my replacement PSU was dead, and obtained another RMA number. Then I confirmed that the PSU was in fact fine.

I just received a phone call from them asking if I had ever shipped out my dead unit for replacement; I had let them know there was nothing wrong and that the RMA could be canceled.

Go Seasonic!

[wisefool]

>It could be a short, they happen from time to time, to test, take the >mobo out of the case, and test it on the anti-static bag it came in on the >desk.

Do not test components on anti-static bags.

The ones with little lines running on them are somewhat conductive. That said, I've powered up computers with no problems on them before, just been lucky. I also fried a couple of boards.

Another tip I learned the hard way. Modern computers (that is anything more advanced than say, a 486?) supply power to the mainboard even if it's powered off. Unplug PSU before messing with components.

Best way to test would be with a multimeter. I'm sure the CPU and memory sockets have the standby power. Not that sure about the PCI/AGP slots.