ATX PSU +5VSB too high

[Pierre95]

Hello,

I have an LC-B300ATX. It is based on the not very usual 2003 IC PWM controller for which I have difficulties in finding datasheet. Recently I have noticed +5VSB being too high. When the PSU is cold, +5VSB starts as high as 10V then drops slowly (within minutes) to 6,6V for keeping that value, which is not acceptable.

So far as I have seen, the schematics is similar to the 250W version found here: http://danyk.cz/s_atx04a.png.

I have some basic questions about the way +5VSB is actually generated:

1) Does the +5VSB use a standalone 2 transistors oscillator for switching voltage generation or is it derived from the PWM controller output?
2) Is there some feedback to insure voltage stays as close as possible to +5V or is it unregulated?
3) What is the optocoupler labeled "COSMO" for? Is it a feedback for regulating +5VSB by adjusting the bias current of Q4?

I've noticed C18 is getting really hot and keeping a finger at its end is barely sustainable. But it may be normal temperature as this capacitor constantly gets the pulsed voltage issued at D7's cathode? D7 is also hot. Same for the COSMO optocoupler.

May be Q3, Q4 or both transistors are damaged, but I made oscilloscope recordings of the voltage at D7's anode and I can see near square alternating waveform voltage.

So which component should I check first?

Thanks a lot,

Pierre

[stj]

Re: ATX PSU +5VSB too high

the caps on the 5vstandby are bad.

[Pierre95]

Re: ATX PSU +5VSB too high

Which can explain the variation in voltage as the capacitor warms up...

You mean the 1000uF C18 capacitor connected between D7 cathode and GND I presume?

[stj]

Re: ATX PSU +5VSB too high

post some foto's

[Pierre95]

Re: ATX PSU +5VSB too high

C18 capacitor:

[stj]

Re: ATX PSU +5VSB too high

foto's of the actual psu i meant.
btw, post the whole schematic too - i want to see where that standby voltage goes as it goes up off the image.

[Pierre95]

Re: ATX PSU +5VSB too high

OK, here it is:

[eccerr0r]

Re: ATX PSU +5VSB too high

Yeah likely C18 is bad, due to ESR it's gotten worse and worse until it finally fries itself. If you have an oscilloscope you can check the voltage across it, probably spiking like mad as the ESR is so bad it can't smooth out the voltage anymore.

There is regulation - the feedback loop through the R30-R27 divider, U2 TL431 reference and the U1 optoisolator is the sole prevention to make sure 5VSB doesn't get too high, but there's no other secondary protection from the "2-transistor" oscillator to generate standby damaging voltage...

It's unfortunate - the 2003 chip seems actually fairly common, just that the datasheet isn't...

[PeteS in CA]

Re: ATX PSU +5VSB too high

The Standby section is a self-oscillating discontinuous flyback circuit. The switch MOSFET Q3, once on, can be shut off either by the voltage across R4, the 1.8 ohm resistor in the Source circuit turning on Q4 or by the optocoupler U1 turning on Q4. Once the energy stored by the primary is dumped into the secondary circuit (filtered by C18), the remaining energy in the primary leakage inductance will turn Q3 on, for the start of another cycle of charging the primary. Optocoupler U1 plus the TL431, U2, are the feedback loop. The way the TL431 is configured, it's an opamp with the internal 2.5V reference connected internally to one input, and the sample of the output voltage fed to the other input (the Ref pin).

So C18 is critical and very stressed in a discontinuous flyback design, and C8 is also important (and stressed), filtering the V(cc) for the transistor section of the optocoupler. C11 is a compensation cap, not particularly stressed, but still important. I'd replace all 3: C18 and C8 with Nichicon PW or similar; C11 with a 105C rated general purpose part from a good manufacturer.

The discontinuous flyback circuit of that basic design has been around since the late 1970s. It was used in the Apple II+ power supply and in Boschert's OL25 family, with bipolar transistor switch devices. In the early 80s Boschert refined and optimized the design and had a whole series of models ranging from 25 or 30 watts up to 165 watts.

[Per Hansson]

Re: ATX PSU +5VSB too high

Yes C8 seems to be your critical cap.
When it goes bad in these designs the 5VSB goes high.
So if you wan to fix it be sure the replace ALL small electrolytics!

[RJARRRPCGP]

Re: ATX PSU +5VSB too high

Yes C8 seems to be your critical cap.
When it goes bad in these designs the 5VSB goes high.
So if you wan to fix it be sure the replace ALL small electrolytics!

So is that just like the Bestec 250-E?

[Pierre95]

Re: ATX PSU +5VSB too high

I would like to thank all of for your answers which were very helpfull.

C8 was the faulty part. I now get 5,08V at +5VSB after C8 replacement. C18 might have been the cause of the failure but seeing a ripple-free 6,6V +5VSB, made me think C18's filtering performs well and the origin is actually before D7, so most likely in the primary of the transformer.

But without your help I would probably not replace C8 as a priority.

[Per Hansson]

Re: ATX PSU +5VSB too high

RJARRRPCGP: Do you mean Bestec ATX-250 revision 12E?

Here is some mods to make it safer:
https://www.badcaps.net/forum/showthread.php?t=6680
https://www.badcaps.net/forum/showthread.php?t=34860

And for fun
http://www.jonnyguru.com/modules.php...Story&reid=154

Yes it's failure is the same as the LC-B300ATX PSU in this thread.
In my opinion the only way to make them safe is to replace the whole 5VSB supply with for example a DM311 IC.
But to each his own

[stj]

Re: ATX PSU +5VSB too high

you should be changing all the caps, or atleast everything under 470uf

[PeteS in CA]

Re: ATX PSU +5VSB too high

I would like to thank all of for your answers which were very helpfull.

C8 was the faulty part. I now get 5,08V at +5VSB after C8 replacement. C18 might have been the cause of the failure but seeing a ripple-free 6,6V +5VSB, made me think C18's filtering performs well and the origin is actually before D7, so most likely in the primary of the transformer.

But without your help I would probably not replace C8 as a priority.

Discontinuous flyback circuits look simple, but really are not. And many components are very stressed. The big attraction is the low component count (e.g. Boschert's continuous flyback OL50 had 140-150 components; the discontinuous flyback XL50 that replaced it had about 90). The design also saves PCB space because the output transformer also serves as the "output" inductor. The core is gapped so it can store energy.

In that circuit, even though it's "only" filtering the V(cc) for the optocoupler's transistor, C8 is an output capacitor. With C8 bad, the V(cc) was unfiltered pulses and the opto's transistor probably had very little collector voltage at the time when the TL431 tried to turn off the switch MOSFET. So the switch MOSFET was being turned off by the power limit resistor in the MOSFET's Source circuit. Thus the output voltage was not being regulated. In Boschert's designs they always had an OVP SCR crowbar on the +5V, regardless of the topology. If the +5V rose to about 5.5 volts, the SCR fired, shorting the +5V and tripping current limit or power limit.

[RJARRRPCGP]

Re: ATX PSU +5VSB too high

RJARRRPCGP: Do you mean Bestec ATX-250 revision 12E?

Yes.

[momaka]

Re: ATX PSU +5VSB too high

So is that just like the Bestec 250-E?

You mean the ATX-250-12E?... Yup.

Pretty much every Deer / L&C from the early 2000's to the late 2000's used that 2-transistor 5VSB design, as did many cheap PSUs in the early 2000's as well. However, with the Bestec ATX-250-12E and the L&C PSUs, they had that critical cap, C8, which was crucial for the 5VSB regulation. And with the Bestec ATX-250-12E, it's even worse, because that cap also filters power going into the UC384x PWM controller on the primary, so that cap sees even more stress than the L&C designs.

That said, there are 2-transistor 5VSB designs that don't have an electrolytic "critical" cap in the circuit. Those designs tend to be a lot more stable, and in fact, I don't ever recall running into one that had 5VSB overshoot/over-voltage problems.

[Dan81]

Re: ATX PSU +5VSB too high

That said, there are 2-transistor 5VSB designs that don't have an electrolytic "critical" cap in the circuit. Those designs tend to be a lot more stable, and in fact, I don't ever recall running into one that had 5VSB overshoot/over-voltage problems.

I have an idea: can we bypass that critical capacitor from the traces? Or is there some more rework to do that? (without needing a DM311 circuit)

[PeteS in CA]

Re: ATX PSU +5VSB too high

If protecting the MB against the 5VSB going out of control, why not wire in an SCR crowbar circuit - a 4.7V or 5.1V zener, a 5-10A SCR, and a resistor across the gate-cathode so a little leakage current from the zener doesn't fire the SCR.

[momaka]

Re: ATX PSU +5VSB too high

I have an idea: can we bypass that critical capacitor from the traces? Or is there some more rework to do that? (without needing a DM311 circuit)

It's a lot more work than that. Basically, the whole 2-transistor 5VSB has to be designed to work without that critical cap. I've done a quick comparison on a circuit like that and one with a critical cap, and they differ in component values and layout quite a bit. Essentially, you would have to change more than half of the components in the 5VSB circuit to achieve that - at which point, you might as well go with a DM311 or other similar offline chip.

If I have more time later in the summer, I will try to copy the 5VSB circuit in my 300W Macron Power MPT-301 PSU. So far, I have posted here on BCN at least three or four 5VSB 2-transistor circuits, but they are all with a critical cap (and three are CWT designs, so they are very similar.)

But the bottom line really is: if you use good quality caps for the output and the critical cap, you won't have any issues with the PSU up to 10 years, if not longer.