FLYBACK SMPS PROBLEM

[ravilite]

hello guys ,
Here is the situation, I am working on a 90 watts , 1050 mA , 72 V flyback based constant current smps which is used for outdoor LED lights,
Mosfet used 12A 800V
O/P diode used - Average forward current 8 A , Peak revers voltage 400V
In this design diode and Mosfet are getting damaged in pairs with in a month of operation. I din't find any burnt marks or any cracks on the diode or on the Mosfet they just look like a fresh components.
To protect the the diode and Mosfet I have used RCD snubber circuit. but still they are getting damaged.

I am unable to understand the reason of Mosfet and diode failure.
can anyone please tell me the possible reasons of output diode failure in flyback SMPS.

 

[kubeek]

Welcome to the forum, please show your schematic. Without that it is very hard to say anything.

 

[ravilite]

This is the schematic , used diode and Mosfet I have mentioned above

 

[kubeek]

What is the ratio of the transformer?

 

[ci139]

i'm not quite shure but perhaps
https://www.google.com/search?q=flyback+reverse+avalanche+avoiding+-patent
https://www.google.com/search?q=flyback+reverse+avalanche+avoiding+-patent&tbm=vid
------
adding up my reading list
https://www.onsemi.com/pub/Collateral/SMPSRM-D.PDF
https://www.microsemi.com/document-...iting-techniques-for-switching-power-supplies
https://www.infineon.com/dgdl/mosfet.pdf?fileId=5546d462533600a4015357444e913f4f
https://www.infineon.com/dgdl/an-937.pdf?fileId=5546d462533600a40153559ea1481181
? https://industrial.panasonic.com/content/data/SC/ds/ds4/AN8026_E_discon.pdf -- i just made a fly over the d/s at my computer ...
??? https://hackaday.io/project/16384-enhancement-to-depletion

 

[ronsimpson]

Mosfet and diode failure

It is hard to know if the diodes or the MOSFET died first. Probably the diodes. Diodes=short will cause great stress on the MOSFET.
What is the temperature of the diodes and MOSFET?
Do you know if the core (transformer) is near saturation?
----------------
diode current. I think one diode will work. Average current is 1A and you have 3x5A of diodes.
I can not see how much ring there is on the primary and secondary.

Are you in continuous or discontinuous mode?

 

[ravilite]

Transformer turns are
Primary turns : 45
Secondary turns : 28
Transformer core is PQ2620

The diode I am using U16C40C diode having T-220 package which have two internal diodes out of which only one is getting damaged .

Constant current power supply is working in critical conduction mode.

How can I check if the transformer is getting saturated. And how the transformer saturation can damage the Mosfet and output diode.

 

[ravilite]

The power supply I am Using have short circuit protection, whenever output gets shorted input watts drops t0 0.5 - 0.2 watts .
I have measured voltage across the mosfet drain and source during normal operation voltage is around 180V and when output is shorted voltage climbs to 310V .
The mosfet I am using is of 800 voltage .

 

[ronsimpson]

How can I check if the transformer is getting saturated.

Here is a picture of primary transformer current. (or the voltage across the Source resistors) I added a red line. If there is any hint of the current going up at the end then you have saturation. (gets worse at high load and after the transformer gets hot)

And how the transformer saturation can damage the Mosfet and output diode.

If the core saturates the transformer inductance drops, current gets high, and out of control current. .....
This is hard on the MOSFETs.

 

[schmitt trigger]

Waveforms, we need waveforms!
At least a pair: the Vdrain to ground, and I drain (measured as a voltage across the sense resistors).
If these can be made simultaneously with the scope in two channel mode, even better.

Goes without saying: an isolation transformer is required to make these readings safely.

Also make sure that:
1) your probes have been compensated and are in 10X mode
2) use short, short, short grounding leads to avoid picking up stray noise.
3 don't crowd the screen with readings. What we want to see is the waveform's shape, with as much detail as possible.

 

[schmitt trigger]

The second set of measurements will be more difficult: Measuring how the output current shares across the three paralleled output diodes.

I suspect that the problem could lie there. Most likely they are not sharing current equally, across all the possible operating conditions.
Could you replace these diodes with a single, higher capacity one?
Or; could you add a very low value, 0.1Ω resistor in series with each individual diode?

 

[ronsimpson]

Measuring how the output current shares across the three paralleled output diodes.

There are three 5A diodes. It should work with only one. I do not see "sharing" is a problem.

There are snubbers on the primary and secondary but with out scope pictures we don't know if they are good.
Maybe the transformer is hand wound.
Probably this is a bread board with long wires. (we just don't know)

 

[ravilite]

The second set of measurements will be more difficult: Measuring how the output current shares across the three paralleled output diodes.

I suspect that the problem could lie there. Most likely they are not sharing current equally, across all the possible operating conditions.
Could you replace these diodes with a single, higher capacity one?
Or; could you add a very low value, 0.1Ω resistor in series with each individual diode?

I have not used Three separate diodes instead I have used U16c40c diode which has 2 internal diodes and out of which one is getting damaged

 

[ravilite]

There are three 5A diodes. It should work with only one. I do not see "sharing" is a problem.

There are snubbers on the primary and secondary but with out scope pictures we don't know if they are good.
Maybe the transformer is hand wound.
Probably this is a bread board with long wires. (we just don't know)

Yes I agree with you current sharing is not the problem, I have used the same diode ( U16C40C ) even in higher current smps and it just worked fine.

The Board I am working on is a well developed PCB ( I am not doing nothing on bread board ), and even the transformer is also professionally made its not hand wounded.

We have have made around 50 such SMPS out of which 30 got failed and all have the same problem, diode , mosfet and PWM IC got failed.

I will try to upload the waveforms photo as soon as possible.
Can you guys tell me which are the important location where I should check the waveforms.

 

[ronsimpson]

U1 pin2 Vcc please test if the voltage is too high.
U1 Pin2 to Pin1 should have a 0.1uf capacitor on the supply. Capacitor must be close to IC.

 

[ravilite]

Can PWM IC damage the mosfet , if yes then what are the possible reasons

 

[kubeek]

It could drive the mosfet with more than Vgsmax. What type is that IC?

 

[ci139]

if the event is not 100% - perhaps it's just the figure for particular mosfet / diode time to live variations

 

[ronsimpson]

Question:

Can PWM IC damage the mosfet , if yes then what are the possible reasons

Answer:

U1 pin2 Vcc please test if the voltage is too high.

Very likely the voltage is OK. I did not design the transformer so I don't know. The supply of U1 would have to be more than 20 volts to kill the MOSFET.

Most likely the MOSFET shorts G-D and sends 100s of volts to U1. So the MOSFET kills the IC. I think the diodes are dead first.

 

[kubeek]

Also, does the diode have a heatsink? How big is it? What type is the mosfet?