Dynex 350 Watt PSU self-starts

[bigbeark]

I was troubleshooting my Asus A8V motherboard (Socket 939). Finally got everything working except the board would start up as soon as the power switch on the PSU was turned on ie without pushing the motherboard power button. The Onboard sound was crackly so I tried a PCI-slot sound card. This didn't work at all and when I moved it to the lowest PCI slot the board would not boot. I swapped the PSU out and everything worked
fine with the replacement PSU, even though it had less power on the 12V rail.

I took the cover off, no obvious faults. I took time to read the PSU faqs. What I understand is that something in the 5V standby circuit is detecting ground, as if I had pushed the power button. At the same time I am losing power somewhere. With the good PSU all the PCI slots work properly.

Any hints on where to start and where on the PCB to look? I have had a rash of PSU failures of late and would like to start with this one to get some experience before tackling the others.

[momaka]

Re: Dynex 350 Watt PSU self-starts

Well, if possible, please post a picture of your PSU here. We may be able to spot something that you didn't.

Apart from that, plug PSU in the wall and check for 5V standby (5VSB). That is, check voltage between pin 9 (typically a purple wire) and any ground on the ATX connector. You should get 5V +/-5%. If that checks out and the voltage is stable, check voltage between PS-ON and ground. The PS-ON pin is typically pin #16 on a 24-pin ATX Connector and #14 on a 2-pin connector. You should get more than 3.3V here (typically between 4-5V). If not, report what voltage you get. Also, note if the fan on the PSU turns on or not when you plug it in the wall. It should stay off. Do these voltage tests with the PSU NOT connected to anything else.

[eccerr0r]

Re: Dynex 350 Watt PSU self-starts

Well if the PSU does not turn on when disconnected, it of course is something with the motherboard. Do make sure as low hanging fruit that the motherboard BIOS is not set to "restore power on power loss", that would be the easiest thing to check.

Then from there it's very hard as this circuitry should be fairly robust and shouldn't fail so not much information available about this kind of failure... Is that pin shorted on the board? You'll have to track back from the pin on the motherboard, usually an open collector transistor or open drain mosfet is used here...

[bigbeark]

Re: Dynex 350 Watt PSU self-starts

Momaka, the 5V standby (pin 9 purple) to Ground (Black) measures constant 5V.
The PS-ON ( pin 9 Green) to Ground (Black) measures 4V constant. Fan stays off. So that checks out OK.

eccerr0r, motherboard BIOS not set to restore on error. The Asus A8V motherboard works fine with another PSU.

First assumption is that something in the Dynex PSU is draining some power to ground. I'm assuming it's a MOSFET or Transistor. What's the process for testing these?

There are no bulging or leaking caps that I can see. I will post some pictures tomorrow.
Thanks for your help!

[stj]

Re: Dynex 350 Watt PSU self-starts

follow the power-on wire.
it may be a bad optocoupler

[momaka]

Re: Dynex 350 Watt PSU self-starts

follow the power-on wire.
it may be a bad optocoupler

Unlikely.
If the PSU does not exhibit the same problem when not connected to a motherboard (bigbeark mentioned PSU fan is not turning when not connected to a mobo), then PSU may not necessarily have a problem. And I think we still need to determine that with more confidence.

@ bigbeark: do you have another motherboard you can test this PSU on? Given the very limited information so far, I don't think we should conclude that the PSU is the problem yet (or if there was even one.) Some motherboards will auto-start if their CMOS was reset. My Dell Optiplex 170L does this every time the power goes out and comes back (since I didn't bother replacing the CR2032 battery when it went bad many years ago.) So getting another motherboard that has a good CMOS battery and that does not auto-start with another PSU, you can use to test this PSU. If this PSU does not exhibit the same problem anymore, then you might have just encountered a fluke or just a special case between that motherboard and PSU (maybe that PSU has a slightly weaker PS-ON signal or maybe the mobo has higher-than-normal leakage to ground on the PS-ON line that affects only some PSUs.) But if it does, then open the PSU and post some pictures. Without pictures, I don't think we can give any more advice as to what to check on the PSU - it's just too many things. Your PS-ON signal looks normal at 4V. The question is, does it stay at 4V when connected to a motherboard when the "issue" happens. If no, then there may be an open or partially open pull-up resistor on the PS-ON circuit or something similar there.

In any case, we need repeatable results. Testing some computer component X with component Y only once, then component X with component Z and drawing conclusions is NOT a good way to troubleshoot. Do more tests and introduce more components and scenarios when possible to see if that changes the problem.

[bigbeark]

Re: Dynex 350 Watt PSU self-starts

Pictures attached. Further Tests to follow

[bigbeark]

Re: Dynex 350 Watt PSU self-starts

Tested with a GigaByte GA-8IPE1000 Socket 478 motherboard with Pentium 4 HT CPU.

There was no self-starting behaviour observed. So the Asus A8V motherboard must have something that triggers the Standby Circuit. A partial leak to ground on the PS-ON Line?

I'm using a cheap analog meter powered by a 1,5v AAA battery. Is this safe to test voltages on a motherboard and PSU?

If I pull a machine out of my pile and find that the 3V battery is completely dead does that indicate a slow leak to ground on the motherboard? . Do you remove the battery when storing a machine for future use?

[R_J]

Re: Dynex 350 Watt PSU self-starts

Some motherboards can be set (in the bios) to return to last power state. If the motherboard was ON when power was removed, it will turn ON when power is restored.
On the Asus A8V it is: Restore on AC Power Loss [Last State]

[stj]

Re: BFG 1000 PSU BFGR1000WEXPSU Died Suddenly

i remove batteries before storage - they wont leak but it's a waste and you could forget it's old when you re-use the board.
they will slowly go flat if you use antistatic foam or bags btw - it shorts them out!

[bigbeark]

Re: Dynex 350 Watt PSU self-starts

Some motherboards can be set (in the bios) to return to last power state. If the motherboard was ON when power was removed, it will turn ON when power is restored.
On the Asus A8V it is: Restore on AC Power Loss [Last State]

RJ, I never have "Restore on Power Loss" set to ON.

I learned this many years ago when my wife's Dell tower rebooted after a power outage and
toasted the PSU!

[momaka]

Re: Dynex 350 Watt PSU self-starts

Tested with a GigaByte GA-8IPE1000 Socket 478 motherboard with Pentium 4 HT CPU.

There was no self-starting behaviour observed. So the Asus A8V motherboard must have something that triggers the Standby Circuit. A partial leak to ground on the PS-ON Line?

It could be that the Dynex PSU just has a marginally-designed PS-ON that can't source too much current. So that may be one reason why the A8V motherboard was playing with it. Or perhaps the A8V motherboard has a leakage to ground, like you noted.

However, it is worth mentioning that I also have some old ASUS and Intel motherboards that start by themselves if the BIOS battery has been removed (and yes, I checked on those that the BIOS did NOT have the power-on status set to "Last State" - so that's not an issue with mine and rather something done by design and only happens the first time the motherboard is connected to power.)

I'm using a cheap analog meter powered by a 1,5v AAA battery. Is this safe to test voltages on a motherboard and PSU?

Yes, it should be safe, provided it's not some extremely cheaply-made meter that is just dangerous by itself.
Also, those analog meters tend to have lower input impedance, so that is actually more helpful when testing those CR2032 coin batteries, as the meter will draw more power while measuring the battery, and is thus more likely to show if the battery is marginal or not.

If I pull a machine out of my pile and find that the 3V battery is completely dead does that indicate a slow leak to ground on the motherboard?

No.
Some motherboards tend to consume more power than others from their CR2032 battery.

Typically, most motherboards are designed to keep CMOS stored anywhere from 1 to 5 years with a fresh new CR2032 battery. Some will discharge faster, other slower. IME, older Intel-made Dell motherboards tend to suck more power and can drain a CR2032 flat in as little as 6 months. But it varies by model of mobo.

Do you remove the battery when storing a machine for future use?

Almost always.

In fact, I rarely put a battery in any motherboard, unless I know it will be going into service again.

The only exception is a test motherboard that I have for testing GPUs - that one, I keep stored and disconnected from a power source, but with the CMOS battery in there, as it has custom OC settings. As I've learned over the years, that mobo is fairly frugal with its CMOS battery usage. So between using it once or twice a week to once or twice a month, it only needs its CMOS battery replaced every 2-3 years, which I'm OK with.

But in general, it's probably better to store motherboards with the CMOS battery off. That way, even if you change hardware (CPU, RAM, and etc.) on the motherboard, it will have to re-detect everything upon power-up... vs. sometimes remembering whatever old config you had and possibly refusing to POST.

In that regard, the worst was a eVGA motherboard I got off of eBay. It ended working fine, but I couldn't get it to POST reliably at first, due to the original owner setting the bus to 400 MHz (1600 MHz FSB), which grossly trying to OC my test CPU beyond what it was stable. It did POST, though (Core 2 Quad Q6600 @ 3.6 GHz on stock voltage, lol)... just couldn't stay stable for too long.
Anyways, that just one more reason I don't like to keep motherboards with CMOS battery in storage - never know what settings I could have changed that could make something not work as desired. Much better to reset CMOS every time mobo is pulled out of storage and enter all config fresh.