Troubleshooting an Artesyn power supply

[giosif]

Hello,

I've got an Artesyn NFN40-7615N1J SMPS I am trying to troubleshoot and I am in need of advice.
The problem with the supply is that it works 90% of the time but, occasionally, it starts "hiccuping" - makes a clicking noise.
The challenge is the hiccuping is very intermittent and doesn't last long (between 1 click sound up to about 5 clicks); then, it recovers and works fine for 10s of minutes.
I tried applying a load on the output in the hope to trigger the problem, but that doesn't seem to make much difference, if any at all.
I should also mention that, when the issue occurs, the output voltage drops and never goes above the intended value (i.e. regulation seems to work).

The other thing I checked was the power supply of the UCC38C42P PWM controller on pin 7 and, when the hiccuping happens, I see a bit of fluctuation on that pin of about 1V.
I'm not sure if this would be enough to consider the problem resides somewhere on the primary side.

Anyone here with similar experience that could guide me to at least reproduce the problem reliably (or, if you already know the actual source of the problem, even better)?

Thanks!

[George Edmonds]

Hi

SMPSU;s hiccuping usually means that one of the secondary supplies is overloaded by what it is powering

George G6HIG in Dover UK

[giosif]

Hi

SMPSU;s hiccuping usually means that one of the secondary supplies is overloaded by what it is powering

George G6HIG in Dover UK

Thanks for that!
Given the hiccuping happens also with no load applied, I am assuming there must be some component on the secondary side which is shorting the output.
If only the issue wasn't as sporadic...

[George Edmonds]

Hi

Normally SMPSU's need a minimum load to operate correctly or at all.

It is not surprising that the supply to the PWM varies as once the supply starts up it provides its own supply to the PWM.

Can you take some photographs of the supply to give some clues as to how it operates.

George G6HIG in Dover 

[Jay_Diddy_B]

Hi,

There will be a 10uF or so electrolytic capacitor on pin 7 of the 3842, try replacing this capacitor with a 105C low ESR electrolytic.

Jay_Diddy_B

[Messtechniker]

Hi,

There will be a 10uF or so electrolytic capacitor on pin 7 of the 3842, try replacing this capacitor with a 105C low ESR electrolytic.

Jay_Diddy_B

Also often found capacitors on the secondary side of SMPSs causing trouble (full and semi-shorts, lost capacitance, high ESR). Always worth looking in to here first.

[giosif]

Hi,

There will be a 10uF or so electrolytic capacitor on pin 7 of the 3842, try replacing this capacitor with a 105C low ESR electrolytic.

Jay_Diddy_B

Thanks!
This crossed my mind but, then, I measured its capacitance and ESR (in-circuit) and the readings were normal.
But, since the problem happens sporadically, maybe the capacitor is not 100% dead, but fails only under certain conditions.
In any case, I've now replaced the capacitor and I'm running the power supply, with a very small load of 100 mA (with no load, I see some very short dips in output voltage from time to time, although not at regular intervals).
So far so good, but let's see.

I've also measured the initial capacitor out of circuit now, and still measures fine (but, again, it could be it starts to misbehave only occasionally and when in use).

[giosif]

Also often found capacitors on the secondary side of SMPSs causing trouble (full and semi-shorts, lost capacitance, high ESR). Always worth looking in to here first.

In general, whenever investigating a power supply, I check the capacitors, particularly those around heatsinks or resistors who tend to run hot.
I checked all caps on this PSU (only in-circuit, though) and all of them checked out.
However, the problem is very intermittent, so the fault on the component might not show up all the time either.
 

[giosif]

Hi

Normally SMPSU's need a minimum load to operate correctly or at all.

It is not surprising that the supply to the PWM varies as once the supply starts up it provides its own supply to the PWM.

Can you take some photographs of the supply to give some clues as to how it operates.

George G6HIG in Dover

Sure.
Please see below.

[giosif]

Hi,

There will be a 10uF or so electrolytic capacitor on pin 7 of the 3842, try replacing this capacitor with a 105C low ESR electrolytic.

Jay_Diddy_B

Thanks!
This crossed my mind but, then, I measured its capacitance and ESR (in-circuit) and the readings were normal.
But, since the problem happens sporadically, maybe the capacitor is not 100% dead, but fails only under certain conditions.
In any case, I've now replaced the capacitor and I'm running the power supply, with a very small load of 100 mA (with no load, I see some very short dips in output voltage from time to time, although not at regular intervals).
So far so good, but let's see.

I've also measured the initial capacitor out of circuit now, and still measures fine (but, again, it could be it starts to misbehave only occasionally and when in use).

So close, but no dice, I'm afraid.
After replacing the 10uF cap, things looked rather promising: I could still see some small dips in output voltage every now and then, but they were rare and happened only when no load was applied.
If I was applying some load to the output, it looked rather stable and with no hiccups or dips.
So, I put the PSU back into the scope (ah, yes, this PSU comes from a TDS3012C scope) and let it run for a few hours - all was good.
I then turned it off from the power switch (which doesn't really turn off the PSU), let it sit for a while, then powered it on again for a few more hours - all good again.
Then, I powered it off again and let it sit once more, but then started to hear the clicking noise coming from the PSU (while the scope was off but, as I said, the PSU was still energized, as it is always the case, as long as the mains cable is plugged in the back of the scope).
Trying to power on the scope when this happens results in the LCD backlight flashing (it tries to come on, but then turns off).
So, it wasn't that cap.

[giosif]

Normally SMPSU's need a minimum load to operate correctly or at all.

A quick update on this: I found a datasheet about this power supply and this particular model (i.e. NFN...) is supposed to provide output regulation even with no load.

It is not surprising that the supply to the PWM varies as once the supply starts up it provides its own supply to the PWM.

But the supply for the PWM comes from the primary side and I would think that, unless there is something wrong there, the supply for the PWM IC should remain constant, even when the PSU is hiccuping.
Or not?

[George Edmonds]

Hi

Normally this type of power supply has a high value resistor (470K?) from the bulk storage capacitor down to a zener and smoothing capacitor  to start it up, a constant maintain supply is then generated from the main switching transformer.

The standard problem with all SMPSU's is electrolytic capacitors, both before and after the switching transformer.

George G6HIG in Dover

[xavier60]

It sounds like the UC3842's Vcc pin isn't being held up by the rectified output of the AUX winding.

[CaptDon]

I know this sounds stupid, but if you have a variac or
variable tap transformer try operating at a lower line
voltage. I have run into some 'universal' SMPS units
that act this way when the line is at around 125vac.
I didn't look at what country you are in, is your line
supply 120vac?? Also if the main filter caps on the
rectified line input have high ESR the supply will also
act this way and the effect will vary with line voltage.
With years of designing and troubleshooting SMPS
I have seen some really strange things. Units that
would only power on in the afternoon when the line
was low and units that would chirp and then finally
start in the late morning when the line was high.
I have seen units that were working fine, then
powered down and refused to start back up because
the input energy storage caps (usually a pair in series)
had high ESR.

[giosif]

Hi

Normally this type of power supply has a high value resistor (470K?) from the bulk storage capacitor down to a zener and smoothing capacitor  to start it up, a constant maintain supply is then generated from the main switching transformer.

The standard problem with all SMPSU's is electrolytic capacitors, both before and after the switching transformer.

George G6HIG in Dover

I checked and you are partially correct: it's got indeed a high value resistor (actually two, each of 470K) from the bulk storage capacitor down to a smoothing capacitor and then pin 7 of UCC38C42P (i.e. VDD), but there is no zener diode.
Also, the second winding on the primary side of the main switching transformer is rectified and then connected through a 68 Ohm resistor to the same VDD pin of UCC38C42P.

I checked all capacitors on the PSU (albeit, only in-circuit) and they all checked fine, both for capacitance and ESR.
I'm not going to exclude a bad capacitor as the possible issue here, but I think it's very unlikely.

[giosif]

I know this sounds stupid, but if you have a variac or
variable tap transformer try operating at a lower line
voltage. I have run into some 'universal' SMPS units
that act this way when the line is at around 125vac.
I didn't look at what country you are in, is your line
supply 120vac?? Also if the main filter caps on the
rectified line input have high ESR the supply will also
act this way and the effect will vary with line voltage.
With years of designing and troubleshooting SMPS
I have seen some really strange things. Units that
would only power on in the afternoon when the line
was low and units that would chirp and then finally
start in the late morning when the line was high.
I have seen units that were working fine, then
powered down and refused to start back up because
the input energy storage caps (usually a pair in series)
had high ESR.

Thanks for that!
However, I'm based in Europe, so mains are at 240V AC.
Also, as I mentioned a couple of time already, all caps check out ok for capacitance and ESR, so I don't think this is the issue here.

[giosif]

It sounds like the UC3842's Vcc pin isn't being held up by the rectified output of the AUX winding.

Hmmm... You may be onto something here: I checked the voltage on pin 7 of UCC38C42P and it is a bit over 12V with no load applied.
Only after applying a load of 400 mA or above, the VDD voltage goes above 14.5V.
At the same time, reading the data sheet for the PWM controller, I think this particular model is supposed to always operate with VDD above 14.5V - anything below should trigger UVLO.
But, if the controller would go in UVLO, VREF output (i.e. pin 8 ) should be pulled low, while I'm seeing 5V all the time.
Or am I not reading the data sheet correctly?

[xavier60]

Sometimes I get a clue from monitoring the output rails with a DSO.
Overloaded rails will come up some of the way then drop very quickly, where as lightly loaded rails will stay up longer between bursts.
If it can be seen that all of the rails come up to regulation voltage, then it's problem with the Vcc supply to the UC3842.

From my experience, the VREF pin should go low during UVLO.
The UC3842 should start at 16V and got to UVLO at 10V.

[giosif]

Just an observation the PWM controller in my power supply is UCC38C42, a slightly updated version of UC3842.
And part of the updated specs are the voltage thresholds for UVLO: 14.5V (ON) and 9V (OFF).
Now, I am not sure if VDD needs to be above 14.5V all the time or only during startup/turn-on.
If I look at the VDD pin as I apply mains power to the supply, I can see it go up to about 14.5V briefly, but then goes down to the 12V value I'm seeing otherwise (see attached picture).
I'm wondering if what I'm seeing is normal operation or not.

[xavier60]

That looks normal except maybe for that period of inactivity while the VDD is falling. This could be due to the loop recovering from overshoot.
The main thing is that it doesn't drop below 9V which will cause UVLO and needing a restart.
So this isn't the fault state at the moment?

[giosif]

Correct: this is normal operation of the power supply, specifically when applying mains power to it (i.e. power supply starting).
Reason I added this info was that I suspected something might be wrong on the PWM side (either the IC itself or its supply).
But it doesn't look like it.
So, on with the investigation.
In fact, my main challenge right now is to reliably and predictably reproduce the problem.

[giosif]

Just updating this topic, I've made no real progress, either in terms of identifying the trigger of the problem or finding the root cause.
Things I've tried since the last update:
* replaced the opto-coupler
* replaced the TL431 voltage reference
* replaced the two 330 uF electrolytic caps situated awfully close to a 220 Ohm resistor which constantly gets to about 100 degrees Celsius, under normal operation
* checked the Zener diode used to protect VDD to the PWM controller (yes, I was wrong earlier when I said there was no such diode)

The other thing I am observing is a small 330 pF cap (see picture) placed across the drain and source pins of the main power MOSFET (an IRFBC40) and which seems to get rather hot under normal operation (60-70 degrees Celsius).
Is this normal?

[xavier60]

Just updating this topic, I've made no real progress, either in terms of identifying the trigger of the problem or finding the root cause.
Things I've tried since the last update:
* replaced the opto-coupler
* replaced the TL431 voltage reference
* replaced the two 330 uF electrolytic caps situated awfully close to a 220 Ohm resistor which constantly gets to about 100 degrees Celsius, under normal operation
* checked the Zener diode used to protect VDD to the PWM controller (yes, I was wrong earlier when I said there was no such diode)

The other thing I am observing is a small 330 pF cap (see picture) placed across the drain and source pins of the main power MOSFET (an IRFBC40) and which seems to get rather hot under normal operation (60-70 degrees Celsius).
Is this normal?

I often see a 100pF capacitor across D-S. 330pF would be causing even more dissipation. These capacitors are a common failure point, breaking down and destroying a lot of other bits. Sometimes the Source resistor and control IC are damaged beyond recognition.
I think it's to reduce EMI. Another method might be by reducing the Gate drive current. Gate drive components should be checked anyway. If there is a diode, check that it's the right way around.

You either have an elusive intermittent fault or a borderline design flaw. Try increasing the size of the VDD capacitor.

[giosif]

Thank you for the explanation!
In the meantime, I believe the power supply is actually working now: with no load, the voltage still dips below 15V occasionally but, with a load applied (even at 100 mA), I cannot see the fault anymore.
I am still running the scope for a longer period of time, before calling this fixed, but all looks good so far.
What fixed it? Well, one of the things I've listed previously, I suppose.