Flyback SMPS burn out
Hi,
we are having eventually problems with accelerated life test (ALT) in our electronic energy meter. The meter consumes about 0,3 W and uses LNK363DN in a cascode flyback configuration. The ALT is done in a climatic chamber under the following conditions:
- 240 Vac, three-phase
- 85 ºC temperature
- 87% relative humidity
- 13 A passing through each phase current bridges.
We use coating in the PCB's power supply area. The problem we have is that from time to time, some meters got burned out during the ALT. Some burn in the very beginning of the test, others burn out after many hours. We would like to understand what causes this SMPS burnout, could you help us? Please find attached our SMPS schematic and a photo of what happens after burn out. The effect we have in our boards is generally the following, as you can see in the photo attached:
- MOSFET MO1 shorted.
- DZ4 (zener between MO1 gate and source) open circuited.
- SUP1 (440V TVS diode) shorted.
- R69 (10 Ohms) burned out.
- LNK363 completely burned out, sometimes the drain pin is pulled out of the chip and lays in the track toward the source pin of the mosfet MO1.
Kind regards,
Adriano Caye
コメント
Hi Traveler,
as this device is an electronic meter, it uses current transformers (2500:1) to input currents into the meter. This current inputs are not shown in the schematic, because they don't have any relation to the power supply (analog section). We have already done such measurements in the board and found that no component is suffering overvoltage stress. The power supply is designed to support up to 300 Vac phase voltage. The MOSFET is 800V-rated. We have a serious doubt about the transformer isolation, specially at the primary winding. We found that the primary winding terminals are very close to each other and they cross themselves at the bottom of the transformer. Please find attached our transformer specs and see what you think about it.
| Attachment | サイズ |
|---|---|
| Flyback transformer specification. (110.9 KB) | 110.9 KB |
Hi Adriano,
Did you design the transformer using PIexpert? Could you please send the file to me if it is available? You might also want to check for core saturation. I could help you check the design parameters on the PIexpert file.
Also, is it possible to monitor the drain voltage waveforms during the ALT. There may be something that the drain voltage could tell us while the power supply is under ALT that will not occur during normal bench tests.
Thanks.
Hi Adriano,
Thank you for choosing Power Integrations.
Please help me understand your design. If the total output power of your design is only 0.3W, how come 13A per phase is flowing through your input?
Based on the devices that failed, it seems that the failure is caused by an overvoltage at the input. You could check the following:
1. Measure the DC level across CY1 using an oscilloscope. This is the equvalent DC input to the power supply.
2. Measure the drain-source voltage across the LNK363.
3. Measure the drain-source voltage across MO1.
Check these measurements against the rating of each device.
Hope this helps.