Troubleshooting a cycling SMPS

Thread Starter

Yami

Joined Jan 18, 2016
311
Hi guys,
I am trying to troubleshoot a switch mode power supply which keeps cycling, the output voltage jumps around. The model of the power supply is ASTEC LPT42 - https://datasheet.octopart.com/LPT42-Artesyn-datasheet-8318749.pdf
Its a power supply based around the IC AS3842. I looked for any faulty components on the secondary side bulged caps etc. Couldn’t find any obvious faults. Following is the result of voltage measurement I took at the control IC
COMP = 0V
VFB = 0V
ISENSE = 0.260V Varies
RT/CT = 0.02V
GND = 0V
OUT = 0V
VCC = 11V to 18V keeps varying a lot
VREG = 0.08V
The power supply keeps trying to start up, voltage measured across the main filter cap also varies slightly during the failed start ups.
I was hoping that you could give me some more ideas on what to look for.
Thanks in advance for the help.
 

GetDeviceInfo

Joined Jun 7, 2009
1,732
Do you have Scope?
After a capacitor check, I typically pull the main switching element, then scope controller waveforms with injected feedback. Had a couple with a bad cap on the controller IC power. Use an isolation transformer while testing.
 

dendad

Joined Feb 20, 2016
3,598
Don't just go on a visual for the caps. Remove them and measure the capacity. Dried out caps is the main fault. Often that is the cause of other things popping as large voltage transients can be there without the caps working.
 

Irving

Joined Jan 30, 2016
957
It probably (hopefully) doesn't need to be said, but don't go probing the hot (mains) side of a SMPS with a normal scope probe as its typically 400+ volt above the incoming neutral rail. You will fry the probe, are quite likely to fry the input ASIC of the scope and potentially do yourself some serious injury. You need a special high voltage, differential, probe to see anything there.
 

dendad

Joined Feb 20, 2016
3,598
I use an isolation transformer and have a series incandescent mains 40W lamp in series to limit the smoke while testing (under no load).
Please keep safe. If in doubt, don't!
 

Irving

Joined Jan 30, 2016
957
Isolation transformer deals with common mode element - as long as you remember to use it - but most scope probes only good for 250 or 300v so still need a HV probe.

Joe Smith YouTube video on DIY probe:

Dave Jones (eevblog) video on how not to blow up your oscilloscope:
 

Thread Starter

Yami

Joined Jan 18, 2016
311
Thanks everyone for the response. I have got an isolation transformer connected and a series light bulb connected as well. Unfortunately do not have access to an oscilloscope at the moment.
The board is covered with a conformal coating thats makes my life that much harder. I have taken the ESR measurements from most of the capacitors on the secondary side all of it checked ok. Had to scrape all the leads before testing :/
Regarding having connected a load - I have got an almost identical power supply send by the manufacture and it seems to work with no load connected. So I am assuming the faulty one should also work with out a load.
Another question that I wanted to ask was - how can I be sure that its an issue with the secondary (cold) side, or an issue with the controller IC. I noticed that the VCC for the chip kept varying a lot, varied so much that I was not able to read the values. Here is what my thought process was - if there was an issue on the secondary side that is a short or an issue with the feedback etc... at least the VCC for the chip would be stable. Am I on the right track?
Again thanks for the help.
 

Irving

Joined Jan 30, 2016
957
If you have a working one, then you can do comparative measurements between the two.

On the primary side - carefully - check the DC volts across the big smoothing caps. It should be relatively easy to trace the AC input, bridge rectifier and DC smoothing. The voltage on the caps should be around 1.4 * AC volts.

Where the problem might lie is in R9/C4 in the schematic below. This is the trickle circuit that powers up the chip until sufficient volts is generated by the secondary supply through R8 from the switching transformer. If that doesn't happen the chip will shut down and restart. Some comparative measurements around there might shed some light on things.

screengrab-20200709-073403.png
 

Thread Starter

Yami

Joined Jan 18, 2016
311
If you have a working one, then you can do comparative measurements between the two.

On the primary side - carefully - check the DC volts across the big smoothing caps. It should be relatively easy to trace the AC input, bridge rectifier and DC smoothing. The voltage on the caps should be around 1.4 * AC volts.

Where the problem might lie is in R9/C4 in the schematic below. This is the trickle circuit that powers up the chip until sufficient volts is generated by the secondary supply through R8 from the switching transformer. If that doesn't happen the chip will shut down and restart. Some comparative measurements around there might shed some light on things.

View attachment 211667
Here are the measurements at the IC from the good power supply
COMP = 2.1V
VFB = 2.4V
ISENSE = 0.1V
RT/CT = 2.1V
GND = 0V
OUT = 0.094V
VCC = 11.02
VREG = 4.9V

and from the faulty supply
COMP = 0V
VFB = 0V
ISENSE = 0.260V Varies
RT/CT = 0.02V
GND = 0V
OUT = 0V
VCC = 11V to 18V keeps varying a lot
VREG = 0.08V

Voltage at the main filter cap is 319V, but it does vary slightly when the power supply tries to start.

With regards to C4 - The IC and the capacitors are on a small breakaway board, cannot access it with the ESR meter clips. However I did measure the capacitance using DMM and it showed the correct value. I am going to de-solder the board to closer inspection. I have had issues with C4 on other faulty P/S but those times when I checked capacitance value using a DMM it would always show a lower reading (that is the cap in circuit).
 
Last edited:

Dodgydave

Joined Jun 22, 2012
9,322
Pin 2 is the Voltage feedback, pin 3 is the Current sense, as a temporary measure you can short out pin 3 to ground to see if it kicks up, the voltage feedback is from the Tl431 on the output through an Opto coupler.

From your readings I would say the Voltage Feedback is wrong, looks like it's not getting a signal from the Opto, or the chip is blown.
 
Last edited:

Thread Starter

Yami

Joined Jan 18, 2016
311
Pin 2 is the Voltage feedback, pin 3 is the Current sense, as a temporary measure you can short out pin 3 to ground to see if it kicks up, the voltage feedback is from the Tl431 on the output through an Opto coupler.
I have never done this before. What should I expect?
Thanks
 

Thread Starter

Yami

Joined Jan 18, 2016
311
The current sense pin is to protect the MOSFET and diodes from excess over-current draw on the output load.
What would happen if there actually is a short on the secondary side?
By start up did you mean get the proper voltage at the outputs?
Thanks
 

Irving

Joined Jan 30, 2016
957
That Vsense value suggests the drain current in the main MOSFET is on the high side.. VREG feeds the remaining circuits in the chip, if thats zero then COMP will be zero as the comparator is not powered up, so there's no feedback loop. The PWM is free running, output volts go up, hence VCC goes to 18v and then current sense shuts it all down. If VREG isnt being grounded by a faulty C5 then I'm thinking the chip is toast.
 

Thread Starter

Yami

Joined Jan 18, 2016
311
Ok umm heres an update, I replaced the IC 3842 turned on the power there was a pop sound and the light bulb turned on to full brightness.
I quickly disconnected the power supply - the mosfet has shorted out and the new IC is also shorted out (all the pins to ground)
 

AlbertHall

Joined Jun 4, 2014
10,258
Another question that I wanted to ask was - how can I be sure that its an issue with the secondary (cold) side, or an issue with the controller IC. I noticed that the VCC for the chip kept varying a lot, varied so much that I was not able to read the values. Here is what my thought process was - if there was an issue on the secondary side that is a short or an issue with the feedback etc... at least the VCC for the chip would be stable. Am I on the right track?
You can't be sure where the fault is. The supply powers up but then detects an error condition. That might be overcurrent, overvoltage, undervolatge on primary or secondary side. The protection circuit then shuts down the supply and it attempts to restart. Repeat ad infinitum.

The principle source for the chip VCC will be (probably) from a winding on the main transformer. For the brief period the supply is running this voltage will be present but will disappear when the supply shuts down.
 

Irving

Joined Jan 30, 2016
957
Another question that I wanted to ask was - how can I be sure that its an issue with the secondary (cold) side, or an issue with the controller IC. I noticed that the VCC for the chip kept varying a lot, varied so much that I was not able to read the values. Here is what my thought process was - if there was an issue on the secondary side that is a short or an issue with the feedback etc... at least the VCC for the chip would be stable. Am I on the right track?
As I said above, Vcc comes to the controller via R9 until the MOSFET is switching and then its powered off a transformer secondary via diode and R8/C4. That's meant to be nominally 12v maintained by the zener on the VCC pin. The fact you were seeing VCC climb to 18v was concerning as that suggested there was a problem with the zener. Also the REG pin is a bypass capacitor for the internal 5v regulator for the Ct/Rt timing loop. The fact that it was at zero volts suggested the regulator or the capacitor C5 were having, or had had, a bad time. On the face of it the chip was toast, but there were indications that other parts were toasted too. That seems to have been borne out. What is strange that the oscillator was running enough to get the switching started yet VREG was seemingly zero, which should have stopped the oscillator running.

Just a quick thought. The circuit I posted above is from an ASTEC application note for a secondary side error amp/feedback loop. It disables the chip's internal comparator by grounding VBF and provides opto-isolated feedback to the output of the internal error amp at the COMP pin. However, you measured 2.5v on VBF on the good PSU. It may be that its been left floating in this design and/or feedback is being handled differently..

The chip has an internal zener of 30v, but the external one limits the gate voltage from the OUT pin. What I'm surmising is that the external zener was suspect or has become disconnected when you replaced the chip; This meant Vgs of the MOSFET was no longer restricted and might have been to high for the MOSFET which has now failed.

What MOSFET was it?

The following would be my repair strategy... make of it what you will, its not the only approach.

After removing MOSFET & Chip but before replacement, check there is a high but no o/c resistance from + side of bridge to VCC, for no short to ground on VREG pin and after applying power, voltage on VCC pin (warning, should be ~11v, but could be 300-ish if zener dead).

If all OK then replace chip. but not MOSFET.

Power on, check VCC =~ 11v, VREG = 5v, VFB=2.5v?, drain pin of MOSFET at ~+320v, gate should be toggling between 0 and 10v but without a scope you won't see that, it may appear to be around 5v as the duty cycle should be at maximum.

If all good replace MOSFET.
 

Thread Starter

Yami

Joined Jan 18, 2016
311
Thanks for the help everyone - here is what I have done after the shorting of the mosfet and the control IC.

1. I replaced the IC
2. Another transistor (picture attached) SA673AB was shorted. So I replaced it with a A1015 thats what I had available - its a pnp transistor in the path of the Mosfet's gate. I couldn't find a similar schematic with a pnp transistor to share here. Also could someone please confirm that I have the part number correct, the markings were a bit odd I did a lot of searching online to come to the pnp conclusion.
3.Replaced the capacitor at the VCC pin.
4. I powered the circuit with out the mosfet first, to see whether anything blows etc. Powered up nothing blew. I did measure the voltage at VCC pin. It didn't stabilise never went above 8V. Should it have stabilised without the Mosfet?
5. Soldered the Mosfet back
6. Same thing happened again. IC,MOSFET,pnp transistor shorted.

I measured the resistance of the transformer windings which connects to the mosfet, read almost 0 ohms. I know it'll be a low value but does 0 ohms sound right?

So finally can I conclude that the transformers winding has shorted out and if that is the case why didn't the mosfet and everything shorted out before. I should remark that the original IC was AS3842 when I replaced it, I replaced it with UC3842A.

@Irving - sorry I was typing this message while you posted. Do you think I should do the steps you mentioned provided what has happened thus far? Thanks
p.s the mosfet was an IRF840

Edit: by 0 ohms at the transformer windings I actually mean 0.3ohms. My probes are 0.3ohms.
 

Attachments

Last edited:

AlbertHall

Joined Jun 4, 2014
10,258
Another transistor (picture attached) SA673AB was shorted. So I replaced it with a A1015 thats what I had available
Especially with TO92 or similar transistors you need to check that the pinout is the same. Those two do seem to have the same connections so it should be OK as a replacement.
 
Top