ATX PSU troubleshooting - mysterious flyback shutdown?

[JeanF]

Hi,
I have an old Corsair CX400 (also called 400CX) that I would like to repair. It was already cheap when I bought it new 10 years ago and it is obviously worth nothing at all now, but I'm still interested in repairing it as a learning process. I'm a bit stuck, could anyone please suggest what I should try next?

The main switching part for the main 12/5/3.3 V etc rails is fine when I provide the main control chip with power from an external bench PSU. On its own the standby PSU is unable to provide enough current (on the +15V rail on the primary side) for the main control chip, so the PSU never starts.

If I remove the main PSU control chip from the board, thus removing the ~15 mA load on the +15 V rail, then the standby PSU is able to give 5 V and 15 V. With the slightest load on the +15V rail, even one or two mA, the voltage on the +15V rail drops to only a few volts.

Even in this "no load" scenario, there is ripple on the +15 V rail and I can hear a light clicking noise at a low frequency (around 16 to 17 Hz). I don't know if this is normal or if this is the standby controller going into one of its protection modes.



I haven't been able to find a schematic, if anyone has it that could be great. Apparently this model is based on the Seasonic S12ii series (but I couldn't find any schematic either). Also apparently Antec EA380 or EA430 or EA500 are very similar (still no luck)

there are pictures of (some parts of) the board, here and these pictures are much prettier than what I could do myself (there are a few others here). However I will happily take new ones of specific parts of the circuit if anyone asks.

I've been poking around for a long time but I didn't find anything useful... so what should I do next? (I could just trash it... but I don't like to give up!)
Thanks !

The relevant datasheets are attached to the post.

[JeanF]

Here is a bit more context, including what I've already checked.

When I removed it from the PC it was in, it was not reliably starting and whenever it started, it would shut down soon after. Then it wouldn't start at all.
Here is what I did:

- visual inspection: nothing obvious
- checked all the electrolytic capacitors by desoldering one leg each time, with my multimeter in capacitance mode and with a basic capacitor impedance meter (ELV ESR1) : could not find any values that are dead giveaways, all capacitance values seem to be OK and all impedances are very small, often below the resolution of my meter.
- checked the two 431 references : they seem to regulate OK

here I should have checked the power supply pins of the two main ICs (FSQ0165R for the standby 5 V and CM6800G for everything else) because "thou shall check your power rails" but I didn't... Anyway I took the necessary precautions that were discussed many times to probe safely on the primary side, and found out that the CM6800G PFC and PWM output were dead, so because I'm a newb I assumed the chip was dead. Long story short, the chip was OK, replacing it changed nothing...

Then I found that the CM6800G was in fact not powered, because the 15 V rail provided on the primary side by the standby PSU was down. Again I (too quickly and wrongly) assumed the control chip (FSQ0165R) was dead. Which was peculiar because I could swear I had a working +5 Vsb on the secondary side at some point. Strange... Long story short, the chip was OK, replacing it changed nothing...

I found this thread where people are trying to help someone troubleshoot another CX400, I thought the issue could be similar. Someone suggests the rectifier or the filter cap on the +5Vsb rail could be faulty. On my unit they look OK (diode tested with diode mode and cap tested just like the other ones, capacitance is within range and impedance is below the resolution of my meter). By despair, I replaced the 3300uF 10V that is on the +5Vsb rail by a 2200u (I know, too small) I had laying around. Then, the +15 V rail is now a solid 16 V, no more clicking ! How is that? But, still not enough current available to power the CM6800. When I put it back on board, the +15V rail collapses again...

When I put another 2200u in parallel (because I hear that it is generally a bad idea to replace a cap by a smaller one), the clicking and ripple are back... I don't understand what is happening.In the badcaps thread I mentioned, user fzabkar says the rectifier could be "leaky". Can I check that with an oscilloscope ?

I've been going round in circles and I'm a bit lost... Thanks for any help.

[JeanF]

Hello again,

as I mentioned earlier, there is an additional winding on the small SB transformer, which provides around 15 V to power the two switching controllers.

I was previously saying that I couldn't draw any current from the 15 V side, thus the main controller couldn't start. In fact, there is probably nothing wrong with the circuit on the 15 V side, it's just a matter of power-up sequence. I was trying to draw current from the 15 V rail too soon.

Normally, when mains is switched on, the standby switcher comes to life first; that creates the 5 Vsb rail. Then there is a little bit of circuitry on small daughterboard (contains a HY510N voltage supervisor and a few discretes) that powers up, and only then, if the PS_ON green wire is grounded, the "protection" optocoupler (is there a proper name for it?) turns on, which in turn switches on the 15 V rail to the CM6800G via a NPN transistor, allowing the main PSU to start.

So if I wait until the 5 Vsb turns on before I load the 15 V rail, then no problem.

But, still, there are some issues :

- there is quite large (I think) ripple on the 5 Vsb rail, about 0.2 to 0.4 V pk-pk, at a frequency of around 300 to 600 Hz, with the frequency and amplitude increasing with the load on that rail: (200 mV/div, 0.5 ms/div) :


I don't have the exact noise specs for this particular model on the 5VSB rail but I think that amount of noise seems excessive when compared to other models.

- Also, the +5 Vsb rail is rated at 2.5 A max by the manufacturer, but at around 1 A, it shuts down. That can't be right!

Any hints about where to look ? I checked/swapped all capacitors I could find on the 5Vsb rail, as well as the Schottky rectifier, but still no luck. Could this oscillation come from the TL431 ? Thanks for any advice!

- Finally, there appears to be a minimum load required on the 5 Vsb rail, for the 15 V rail to stay at 15 V. There is a 1k load resistor already on board, but that's barely enough. With that resistor alone, the 15 V rail is about 10-11 V. The CM6800G is specified for a Vcc of 10 to 18 V, 15 V typical, but that gives only a small margin. And that could explain why, back in the day when I took the PSU out of the PC, it wouldn't start at all when outside of the PC (with the famous paperclip trick), but would start intermittently when plugged to the motherboard. At the time, I tried to load the 3.3, 5 and 12 V rails, but I didn't try to load the 5Vsb rail. I don't know if this is a fault or if it's a design "feature", I would need another (working)  CX400 to compare...

[JeanF]

Hi,

here is the schematic that I drew for the part of the PSU i'm trying to troubleshoot: (click to open larger image)


so (not unexpectedly) this is very similar to the reference schematic provided by Fairchild in the FSQ0165R datasheet that I attached above.

I have a very large ripple at the output, so large that it's not even ripple anymore, it's more like a sawtooth waveform, at a low frequency (around 300 to 600 Hz) compared to the switching frequency (around 57 kHz). I didn't take new scope screenshots, but it's the same shape as previously attached, and the amplitude of the sawtooth can be even more ridiculous, more than 3 V at full load. I'm pretty sure that the IC shuts down several hundred times a second, which makes audible noise, because one of its protections triggers. But I don't know which one... According to the datasheet (see above), it has Overload Protection (OLP), Over-Voltage Protection (OVP), Abnormal Over-Current Protection (AOCP), Internal Thermal Shutdown (TSD).

I've tested a vast majority of the components on the board *** and couldn't find a culprit, so either I'm not looking hard enough, or my measurements are not trustworthy, or both...

I should mention two more things:

- this is voltage dependent. I lifted pin 2 of the IC and brought power directly to it with a variable bench power supply. When the supply voltage of the IC is lower than normal (around 9 to 10 volts instead of 15), the switching starts correctly and the output is regulated correctly at 5 volts. Then I tested the current capacity with a variable load, and the supply could deliver 2.5 A @5 V as specified. So could this be the OVP triggering ? The IC is monitoring the voltage on the "sync" line and triggers OVP at 6 volts. I carefully looked at the waveform on this pin and couldn't see it going above 6 V, but I'm not sure, my scope doesn't have storage and my eyes aren't fast enough to record transients...
When I crank up the supply voltage, the ripple is rapidly increasing and the current capability of the supply decreases. at around 11.5 V I can still draw 2.5 A, but there is a strong audible noise (which is not present when Vcc is lower). At 15 V, the PSU shuts down at 1 A.

- this is also load dependent. When the PSU is started with a low voltage supplied to the control IC as described above, I've noted that:
      - at no load, the low frequency ripple disappears, I have a smooth 5 V (with a small amount of 57 kHz ripple which is fine)
      - when I crank up the current, the low frequency ripple (and sound) comes back, but its amplitude and frequency are increasing with load.
      - at some point when the load is "enough" (that "enough" depends on the Vcc of the control IC), then the low frequency ripple disappears and the output is clean again.

I'd be very glad if someone could point me in the right direction, or give any advice that could help me to find what's going on! Maybe tell what component you would suspect first, or maybe give me tips to find out whether it's really one of the protection modes of the IC that triggers or not.

Thank you again !

*** : tested OK and/or swapped with new ones to see if it changed anything, with no luck (but will happily test again if you smell something dodgy!)
U101
U102
C122, C123, C104
D121, D103, D102
D101, ZD102
+ all resistors

[madires]

From what you've found out already the problem seems to be the 15V auxiliary supply. So I'd check all components in that part of the circuit. Another thing to check is the primary's high voltage DC after the active PFC to make sure that the high voltage rail is fine. A low voltage there would cause a lower voltage for the 15V rail.

[JeanF]

Thank you for your reply,

as the problem was also present when I was powering the IC directly from my bench PSU, with its pin 2 disconnected from the 15 V circuitry on board, I wasn't suspecting the 15 V supply too much. But I will certainly leave no stone unturned so I just checked them all:

(with my DMM respectively in resistance, capacitance and diode mode, and with a variable bench PSU + a resistor for ZD103)
C103 : 1 µF
ZD103 : 16 V
R105, R106, R107: values are correct
C104: 100 µF
D102, D103 : OK

for C108 I think it's too small for the resolution of my DMM in cap mode.  In the IC datasheet they suggest a 12 k / 33 pF filter at that spot in the circuit, here Corsair fitted a 33 k resistor so if the cutoff frequency is about the same, the cap should be about 12 pF which is indeed too small for my DMM. Just in case it had gone open circuit, I swapped it with a 10 pF and that changed nothing.

at C103 I have a steady 16 V.

BTW at this point in the debugging the active PFC is not active because I've disconnected the main IC which does active PFC + PWM for the other ATX voltages. So the DC bus is 325 V. That shouldn't be a problem because that's the same situation as when the PC is off.

Thank you again, I look forward to hearing your comments and other replies!

[madires]

Maybe I got it wrong, but you said that the voltage of the 15V auxiliary supply breaks down with a tiny load of a few mA?

[JeanF]

Indeed I said that at first, but I was wrong... I didn't edit my posts because I had the feeling it would be difficult to follow. In my third post (labelled "Reply #2") I tried to explain the situation. The Q0165R can start with its startup resistor, and put enough charge into C103 for Vcc to rise above 12 V, then it runs normally and the 5 V is regulated. From this point on, a few more mA can be drawn from the 15 V rail (which should be enough to allow the CM6800G to finally start)

At the beginning I was trying to draw current from the 15 V rail too soon. With a bit of time to stabilize its OK.

For now I've discarded the CM6800G and the main switching PSU, I'm focusing on the standby PSU driven by the FSQ0165R,

The remaining problems are:
-distorted (nearly sawtooth) waveform on the 5 V output, with associated strong audible buzzing
-the 5 V output can't sustain the rated 2.5 A, at even less than 1 A the voltage goes down to maybe 1 or 2 V in the troughs. the average voltage is way out of spec and the IC finally shuts down. (... unless the IC is undervolted as stated above)

[madires]

I see. So the next step would be to check the feedback loop.

[JeanF]

Thank you for your help.

It turns out that C107 had gone open circuit... Although I was "sure" that I had checked it before and it was OK. I obviously must have goofed up when reading the silkscreen markings which are quite small and close from one another...

I've replaced it with a 47n cap as per the Q0165R datasheet reference circuit, problem solved ! There may be room for improvement with the value but it works well enough for now

The main switching power supply which is used to generate the other ATX voltages still has problems, that I caused myself by making stupid mistakes, but at least there is progress. And I have some more homework to do before asking here

[madires]

  An open low value cap in the feedback loop is rare. Can't remember that I ever encountered such a fault.