Xclio 400W 5VSB slow - I mean real slow!

[Yeti49]

Hi guys,

Just discovered this great forum and hope you can help me. The above ATX supply has been working fine for years but recently the 5VSB is taking ages to come up from cold. I'm talking maybe 10 minutes! Voltage starts at about 2.8, slowly creeps up to about 3.5 and then jumps up to 4.8 after which everything is fine till next cold start. Both filter caps look fine (changed them anyway but no difference). AC into diode is what I call "squegging" (short bursts of pulses) but maybe this is normal. 5VSB comes up OK with no load but at 8 ohms I have the problem. Any help appreciated.

[tom66]

Re: Xclio 400W 5VSB slow - I mean real slow!

Is it a self-oscillating converter (two-transistor) or based on a control IC?

[Yeti49]

Re: Xclio 400W 5VSB slow - I mean real slow!

Is it a self-oscillating converter (two-transistor) or based on a control IC?

Hi Tom - thanks for the reply. It looks like a two transistor circuit, which I gather is a bit of a risk, long term (should I be worried?)!

I can only think that one of the capacitors on the primary side is leaky and becomes less so as it warms up. Certainly, aiming a hair dryer in that area brings it up real quick. Is there any history of the opto-coupler or the shunt regulator (TL431) going bad, I wonder? From further reading I gather that the circuit is supposed to work in discontinuous mode so the waveform might be a red herring.

Cheers,
Keith

[tom66]

Re: Xclio 400W 5VSB slow - I mean real slow!

Hi Tom - thanks for the reply. It looks like a two transistor circuit, which I gather is a bit of a risk, long term (should I be worried?)!

I can only think that one of the capacitors on the primary side is leaky and becomes less so as it warms up. Certainly, aiming a hair dryer in that area brings it up real quick. Is there any history of the opto-coupler or the shunt regulator (TL431) going bad, I wonder? From further reading I gather that the circuit is supposed to work in discontinuous mode so the waveform might be a red herring.

Cheers,
Keith

2-transistor circuits are strange beasts. I've played around with them a little and still don't understand them. They are usually discontinuous (most flyback converters are.)

A lot of things can stop them working properly. Optocouplers and TL431s can both fail. The TL431 can be tested: when working, it should have 2.5V between ref and cathode, I think.

Depending on the design some are OK and do not go out of control. Others will go out of control and give ~12V to the motherboard if a small capacitor on the primary fails!

[Yeti49]

Re: Xclio 400W 5VSB slow - I mean real slow!

Well, my 5VSB problem is fixed. Turned out to be the 22uF, 50V cap on the collector of the opto-coupler. Not leaky but capacitance measured about half marked value and varied dramatically with temperature. Changing the cap gives me instant 5VSB at good 5V (AC into recifier is now continuous).

However, putting the PSU back in the computer (P4 desktop, ASUS MB) I now have a different problem. It turns on instantly, does a POST beep, gives me a BIOS splash screen (most times) and then shuts down. If I hold down PS ON, PSU stays up but MB not doing anything more. All voltages look nominal and PS OK is good. Don't know whether this is related to my PSU dismantling or just a new fault. Any thoughts?

[tom66]

Re: Xclio 400W 5VSB slow - I mean real slow!

Check the supply rails for ripple using your scope. You expect under 120mVp-p for 12V. If it's too high it could be crashing the computer. 5V and 3.3V is 50mVp-p. Not sure about 5VSB, would guess the same as main 5V.

[tom66]

Re: Xclio 400W 5VSB slow - I mean real slow!

You are lucky about that capacitor as it's usually referred to as the killer cap - when that goes bad, the self-oscillating circuit can run out of control and shoot 5V up to 12V or more.

[Yeti49]

Re: Xclio 400W 5VSB slow - I mean real slow!

Hi again Tom,

Actually it wasn't anything too complicated (easy to say when you've found it - with a bit of help via Google!). When I was removing/replacing the PSU, I must have knocked the CPU fan assembly and degraded the integrity of the heat sink compound, so the CPU temperature was shutting the system down. Fixed by cleaning off the old compound, applying new and re-seating the heat sink.

I guess I was lucky with the capacitor. I think it's ESR must have gone up (it was pretty close to the heat of the primary heat sink). I presume the danger comes if it goes leaky? Maybe I should try to find a better part to replace it and move it away from the heat source.

Anyway, thanks for your help.

Keith

[tom66]

Re: Xclio 400W 5VSB slow - I mean real slow!

The danger from that cap is apparent with high ESR, low capacitance or being leaky .

Basically the voltage across that cap is used to turn off the transistor and stop the oscillation when the output reaches the set point... but when it goes bad, that voltage drops out and it can't turn off the transistor any more... so the voltage climbs, and climbs, and climbs... until the smoke leaks out of the motherboard.

It is possible to design the 2-transistor circuit such that the output sags when the cap goes bad... obviously they did that in this case, which is nice I suppose.

The cap rarely dies due to heat external to it. Internal heating is the major cause of cap failure, because the 2-transistor circuits are often poorly designed... thing is, when you spend enough on a 2-transistor to get it to behave as well as a PWM IC... you may as well just use a PWM IC! They can be found as cheap as 20 cents from some suppliers.