ATX PSU turns on immediately after power connected

[ReWoP]

Hi everyone, need help to diagnose an ATX PSU!

The PSU works fine but it turns on immediately after i connect the power plug, even without having the PSU connect to the motherboard

Any ideas where should i start testing? 

[Asterix]

Hi!

This sometimes gets done when the power suply is used for testing hardware.
What happens is the green and black wire gets jumped so it starts right away.
See: https://www.silverstonetek.com/downloads/QA/PSU/PSU-Paper%20Clip-EN.pdf
If i were you i would check the 24 pins to see for a jumper wire.
If that isnt the case i would check the PCB inside.
But be careful! There are electrolytic condensators in it wich do hold charge!
Check for anything odd, like a jumper wire, a short or any residu inside, from spilling something or dirt.

If that aint the issue it is another component and i would reccomend you to measure the voltages on the outside of the psu.
Yellow is 12V
Red is 5V
Black is Ground
Colors like purple, green and more are mostly voltages like 3.3 you can google for a "Pinout".
They dont have to be exact but in range so 11.5 will be usable.

I hope you are able to fix it right now!

[ReWoP]

Hi Asterix! 

I've check the cable and there is no jumper wire on the 24 pins connector and also the voltages are all ok. Even went and test the PSU with a motherboard and it boots and works fine.

I've already removed the PCB (discharged the high voltage capacitors) and did a small inspection (mostly visual), remove some dust but didn't find any damaged components, at least nothing too obvious.

I had the idea that measuring the green wire to ground i would have something about 5v but I'm getting around half that (2.4v). Could be a damaged resistor or something?

[Manul]

I had the idea that measuring the green wire to ground i would have something about 5v but I'm getting around half that (2.4v). Could be a damaged resistor or something?

I think a lot of these supplies use tri-state logic. Voltage divider is used to set that input to the middle voltage (2.5V). In this logic middle is inactive and both low or high (0V or 5V) is active. For compatibility I guess.

[ReWoP]

I think a lot of these supplies use tri-state logic. Voltage divider is used to set that input to the middle voltage (2.5V). In this logic middle is inactive and both low or high (0V or 5V) is active. For compatibility I guess.

That makes sense, then you could pull up or pull down to signal power on...

Then could be the control IC is damaged then?

[Manul]

Many things could be wrong, not enough information to judge. I would try to trace the green wire signal, see where it goes. There could also be opto-coupler in this path.

[asis]

Hi,
You would indicate the PSU model, we can find the circuit, or a fragment.

[xavier60]

I always measure close to 5V on the green wire. Is the voltage fluctuating at all? There might be yellow glue on the PCB.

[ReWoP]

Here is some more data you asked.

The PSU model is EuroTech 580W model no. FPSPA580PSA and the control chip is WT7517
https://www.techpowerup.com/articles/160/images/WT7517.pdf

According to the datasheet the PS_ON should be 5V


Could the capacitor be damaged and acting as a resistor to ground?
I've measured the impedance between the green wire (PS_ON) to ground (with the PSU unplugged) and I've got 23.9M

[Asterix]

Hi!

I think the 5v would only be there if the psu would be enabled.
I think you should check the diodes of the pcb as well as the condensator?

Thoes this look bulky?
Also measure its capacitance if possible.

[ReWoP]

Hi Asterix!

Missed the bulky part, did you notice anything?

I think it should always be 5V because its pulled up by the +5VSB rail that is always on rather than the other voltages rails that only have voltage when the PSU is turned on.

I've already tested the diodes and other semiconductors for simple failures, like shorts or opens, and found nothing obvious...

Anyways, I will recheck the voltages for inconsistencies (following the 1st, 2nd and 3rd commandments: Thou shall always check voltages), then the capacitor to ground and afterwards other componentes around that circuit... Will post any finds.

[asis]

Hi,
First check the voltage of the CMOS battery without removing it from the socket on the PCB. In this case, the PC needs to unplug the PC from the power outlet.
If it is less than 3V, it is better to replace it immediately.
Next, go to SETUP and check the parameters for turning on the PC when power is applied. For prevention, look at all CMOS parameters and reset them to their default values using a jumper, or remove the battery from the socket and short-circuit the contacts. In this case, the PC must be disconnected from the outlet.
Has anything changed after these actions?
Vladimir.

[ReWoP]

Hi Vladimir!

Thanks for your input but it is not a "PC problem", the motherboard is working fine and its running fine with another PSU.

[MathWizard]

What happen's if you connect the motherboard, then plug in the PSU ? Does it still start   even without pressing the power button on the computer case ?


They way I've seen most these turn is
)There's a mini SMPS, that creates +12-17V Aux. power on th primary side, and powers the isolated 5V standby power and the chip on the output side.

)that 5V goes into an opto-coupler circuit, and the computer power switch either connects the 5V at the top, or hooks the botom to GND, to complete the circuit

)On the primary side, the other side of the opto-coupler, is controlling a transistor(s), that turn on or off, the highpower SMPS chip, on the hotside..............the Vcc pin of that chip is either pulled low, for off, or high, so it starts running.

[asis]

Hi,
Indicate what types of components are still on the PCB: PWM, optocouplers, type of transistors SB5V and SMPS.
You have something with an SB5V source and may have significant ripple.
Carefully test the ECAP on the primary and secondary circuit of SB5V.
Judging by the photo, you do not have PFC, and PWM is UC3842 ~ 45.

[ReWoP]

Thank you all for the input, but yesterday I wasn't able to run more tests. Ill do so today and post the results.

[Rasz]

I would just cut/isolate WT7517 PSONB pin and see what happens. At least you will know which side is messing up.

[ReWoP]

Hi all,

I've done more testing, having in consideration all the suggestions above and this is what I came up:

• Effectively all voltages where being correctly outputted as soo as the power supply was connected to mains
• The green wire (PS_ON - power supply on) to ground was about 2.5v
• The gray wire (PG - power good) to ground was 0v
• All voltages where correctly reported to the WT7517 chip and the PGI (power good input) was about 2.5V and the PGO (power good output) was 0v
• When i short the green wire (PS_ON) the PG output starts reporting the correct 5v - the WT7517 chip validades all power rails
• Traced the optocoupler controlled by the chip, to indicate the primary side to start the main power rails, and it was good and correctly flipping the power on/off states when shorting/opening the PS_ON line
• On the primary side, tracked the power state circuit and found a PNP Epitaxial Transistor (A928A) that was shorted on all terminals

Having found that, I did what we should do in this situations. Having not found a direct equivalent transistor on my transistors stash, impatiently, tried to use a "indirect" equivalent transistor that also did not worked (wasn't Epitaxial and didn't noticed that was EBC instead of ECB) and maybe broken something else on the primary because even putting back the old transistor the PSU did not start again... 

Anyways... I've ordered the correct transistor and when it arrives ill be diagnosing the primary side for any more faults since i can probably assume the secondary side is completely fine.
Lesson learned...