SMPS Troubleshooting Design based around Viper22A

[oddbudman]

smps troubleshooting

Hello,

I have been developing an emergency light which uses a Viper22A SMPS for its power supply. I have been having some problems with the circuit blowing up after a few minutes of operation.

The current design is very close to AN1735.
https://www.st.com/stonline/products/literature/an/9682.pdf

From what I can see the SMPS IC is failing at the Source Pins (burn marks can be seen on these pins after a failure). When it fails it is quite spectacular with a flash, then the circuit fuse blows.

I was thinking that perhaps this problem is caused by the RC snubber circuit in the design (there is no diode used in the snubber circuit).

The circuit load on the secondary is quite small (around 7W) and the circuit does not seem to be running excessively hot.

When the circuit is functioning regulation is very good with around 1% Voltage ripple on the DC output.

Attached is our circuit diagram for the power supply. (Please note that Test Switch Port goes to ground, SMPS port is not used)

Any assistance you could give on what could be causing this problem or what things I should be checking would be greatly appreciated.

thanks in advance,

oddbudman

 

[E-design]

viper22a

Are you using your own transformer construction? You may have too much leakage inductance (you can measure it) causing a destructive spike when turning off, on the Vdrain of the chip. I will start by measuring the Vdrain voltage with a good scope. Sounds like voltage punch-through to me. Layout with any SMPS is critical. You don't want currents circulating in a large loop area. If you followed the suggested layout then you should be OK here.

 

[oddbudman]

viper 22a

Can you please explain in a little more detail what you mean? Or point me towards a reference which discusses this topic.

I have also added quite a lot of solder to the drain pins and small drain plane to provide heatsinking.

Anyway, I have attached the relavant Mains side of the SMPS PCB layout. This may clear up any queries you had on the PCB design.

Thanks,
oddbudman

 

[E-design]

troubleshooting smps

I would do the following:

1) Run the unit off a variac with a isolation transformer while monitoring the voltage between Drain and Source of the chip with a good scope. Start at a low safe voltage with maybe a lighter load and check for the magnitude of the voltage spike each time the device switch off. Increase the load and voltage in small increments to see what happens to this voltage spike. (Vspike=Ldi/dt) where L in this case is the leakage inductance of the winding + any track inductance connected to the device.
2) If this turns out to be too close for comfort, remove your transformer and measure the leakage inductance on this winding with a LCR meter (short out all the other windings of the transformer). I think the spec says it shouild be less than 19uH. Take corrective action if out of spec.
3) Look at your layout area for your critical current path. (you have a gnd track in the middle, while on the sample layout they kept it outside) Compare the current-loop area with their layout that I marked in blue. You want this area as small as possible to keep the added inductance from the tracks minimal.

I also see they allow a lot more copper area for cooling the device. I am sure you will find your problem in one of these steps.

Good luck

 

[oddbudman]

circuit troubleshooting

Thanks for your tips E-design.

I rigged up an Isolation transformer for the unit so I could CRO it. The RC snubber wasn't sufficient for the transformer we were running. I have now changed to a RCD based snubber and peak Vds voltages are down to 536V easily clearing the 730V limit on the SMPS chip.

Anyway I'm going to do a 16 hour burn in test tonight just to make sure things are working well.

 

[E-design]

viper22a transformer

This indicates that the leakage inductance is higher than it should be. Better snubber will work, but you pay with efficiency!