PSU question

[VJ]

Hello,

I have a strange psu question. On the Enermax website, there is a calculator to determine how big a powersupply you need.

http://www.enermax.outervision.com/

The footnote on the page mentions capacitor aging as a cause for decreased wattage output, and allow one to compensate for it in the selector. It is unclear to me if this keeps degrading, or reaches a certain minimum after some degradation. Any thoughts?

And more practically: if my psu is 6 years old, how much wattage can I still expect to get from it?

Thanks!

Jörg

[TransformX]

if my psu is 6 years old, how much wattage can I still expect to get from it?

Enough

[VJ]

Hehe... good one...
It is just that I have been experiencing some weird issues every now and then, and I wondered if this could be the cause.

[Chucky Cheese]

BIOS Hardware Monitor

Hehe... good one...
It is just that I have been experiencing some weird issues every now and then, and I wondered if this could be the cause.

Do the voltages look OK in the system bios? Are they within IIRC 10% If you have a DVOM try and measure the 5 volt rail at a legacy power connector(Molex 4 pin). You don't want to have your 5 volt rail lower than 4.95 volts.

I just helped a friend of mine that had her power supplys 5 volt rail was at 4.4 volts and she was having all kinds of wierd problems, especially with magic jack and transferring music to her ipod.

cc

[VJ]

All the voltage are spot on, all of the time... (using the mainboards test software)

The problems I see are intermittent, and I have found some sources hinting at the mainboard/videocard combination running a OS that is not fully compatible. It mainy is a problem of the system not warm booting (so the reset button blanks everything, but the system doesn't boot; cold boot is not an issue), allthough I sometimes feel it may not be stable under full load (a bit unsure).
I used the Enermax calculator, and with the capacitor aging at 10% (lowest value) it states that I need 507 W. Increasing it to 20 % (which they suggest for 1 year+, it increases to 553 W. The PSU I have is this one, 510 W continuous (650 W peak).

Jörg

[Chucky Cheese]

Memory ???

All the voltage are spot on, all of the time... (using the mainboards test software)

The problems I see are intermittent, and I have found some sources hinting at the mainboard/videocard combination running a OS that is not fully compatible. It mainy is a problem of the system not warm booting (so the reset button blanks everything, but the system doesn't boot; cold boot is not an issue), allthough I sometimes feel it may not be stable under full load (a bit unsure).
I used the Enermax calculator, and with the capacitor aging at 10% (lowest value) it states that I need 507 W. Increasing it to 20 % (which they suggest for 1 year+, it increases to 553 W. The PSU I have is this one, 510 W continuous (650 W peak).

Jörg

I guess I am unsure then...could it be the memory stick in your first slot ?? IIRC the first meg of memory is not flushed on a warm boot...memtest pass?

cc

[VJ]

I'm guessing Enermax will overspecify the power needed; also, I couldn't select my videocard, so I choose another one (but it requires more watt), so I should have some headroom.
Memtest passes without problems. It is also ECC memory (mainboard is an old server mainboard). Some time ago, I found some description of the warmboot issue on the website, in combination with a Matrox Parhelia. And I have the impression it only occurs when rebooting during a system install, or when using maybe OSes that are not perfectly supported (Knoppix, ...). It really showed these problems with Vista, but hasn't showed it since I put a trial version of Server 2009 on it. So perhaps it is very critical on mainboard/chipset drivers?

I'd like to turn it into a file server (probably Suse) and adding some harddrives later on.

[Chucky Cheese]

MB Caps ??

Have you taken a close look at the caps on the MB, specifically close to the processor ? Those are the VRMs' capacitor bank for regulating the core logic.

I am out of ideas, I guess it could be software though. It seems as though you could get the combination(..driver/os level) "right" no matter what OS was installed.

cc

[VJ]

Also checked those...
I'm also guessing software, esp. since it hasn't really occured anymore since installing Server 2008 (which after all is a server OS, on a server mainboard), but it is a very strange problem... It occured most when using Vista and Knoppix 5.x.

Regardless, I suspect it may be best not to add too much hardware (e.g. a couple of harddrives without upgrading the PSU)...


Jörg

[Chucky Cheese]

Server OS

I guess that maybe a server OS has timing algorithyms that "sync" with that ECC memory versus a non server OS might have trouble with it! I'm just guessing though!

...anyways, good luck!

cc

[VJ]

Just for reference, this is the entry on the FAQ website of the manufacturer:

http://www.supermicro.com/support/faqs/faq.cfm?faq=3431

But my Parhelia is at the latest bios version...

[Chucky Cheese]

MB Bios

not knowing what MB bios works with what parhelia bios that "fix" seems kind of ambiguos because it seems as though a hard reset with the reset switch would let the system boot normally.

by chance does that board have the server works chipset? the reason i ask is we use steelcloud pcs' with super-micro MBs' for high speed real-time image capturing for address resolution(OCR) and binary transfer and they have matrox video cards and a proprietary frame-grabber. They are very reliable and do not get shutdown very often, but when they need it they have to be powered off and the ac plug pulled for a brief time(10-15 sec)...not reset or soft rebooted. Technical documentation we get to support this equipment states that this is due to the power_on signal that is monitored by the chipset, the vendor was not specific as to the nature of the failure.

I know it is totally different circumstances but I can't help but wonder. myself and another tech think the power_on signal is dropped during reset or soft booting and doesn't get "seen" unless it comes from the power supply, almost always it will freeze before it tries to start NT4 if we don't power it off...does that make any sense?

well, anyways its working so we can't argue with success

[VJ]

It has an E7505 chipset... but it is indeed a striking coincidence (?) !

[Chucky Cheese]

Hmmmmmmm

well then...i think that it is an inherant flaw of the chipset. I rebuild these pcs with parts from our supply chain and i have experimented quite extensively with different OSs' that aren't part of the machines configuration but always with the hardware that it needs to be part of the machines network(...each machine has 38 of these pcs') and the behaviour is fairly consistant.

these pcs' are "single task" other than tcp/ip functions...over all they perform quite well and are rugged and reliable and for what we use them for i don't think we could get any better.

[VJ]

FWIW, Vista and the other OSes on which it occured, are not in the compatibility list either. Ok, neither is Server2008, but that didn't exist at the time. Vista was just around the corner, so the support website would have been updated for it.