Flyback transformer heating in 3842 based laptop charger

[kapurk]

I have a strange problem. I am working on an ac power adapter design for a 90W laptop charger based on the 3842 controller. I have 2 identical boards from different sources. One of them is the original chinese pcb (average quality laminate) and the other is (supposedly) a higher quality FR4 pcb which is a clone of the chinese pcb. THe problem is that the flyback transformer in the chinese pcb does not heat up beyond 80degC but the same transformer in the clone board heats up to 120degC or more. Both boards have identical parts. I have even tried swapping the parts from one to the other but the transformer always heats up in the clone board. I have checked the switching waveforms on both boards and they are identical.
I have gone to the extent of cutting the boards in half to mix and match the primary section from one with the secondary section of the other. The problem seems to be on the primary/auxiliary winding side because the transformer always heats up with the primary side from the clone board. I have spent a whole week trying to figure out the reason by modifying the snubber, changing the diodes, mosfet, ... but the heating does not stop. At the same time no matter what I do to the chinese board the transformer does not heat up on that board. Note that it is not just one transformer that I am talking about. I have tried this with a whole set of identical transformers and boards.
Any ideas? Anyone?

 

[Orson Cart]

Are the frequencies the same for the two cases? and the peak current in the primaries in the two cases?

Orson.

 

[kapurk]

Are the frequencies the same for the two cases? and the peak current in the primaries in the two cases?

Orson.

Yes, all measurements and waveforms appear identical. Both are running at the same frequency (80kHz) and being tested at exactly the same load (19V 4.9A) on the same test jig.

 

[PRATIK TUPE]

I have a strange problem. I am working on an ac power adapter design for a 90W laptop charger based on the 3842 controller. I have 2 identical boards from different sources. One of them is the original chinese pcb (average quality laminate) and the other is (supposedly) a higher quality FR4 pcb which is a clone of the chinese pcb. THe problem is that the flyback transformer in the chinese pcb does not heat up beyond 80degC but the same transformer in the clone board heats up to 120degC or more. Both boards have identical parts. I have even tried swapping the parts from one to the other but the transformer always heats up in the clone board. I have checked the switching waveforms on both boards and they are identical.
I have gone to the extent of cutting the boards in half to mix and match the primary section from one with the secondary section of the other. The problem seems to be on the primary/auxiliary winding side because the transformer always heats up with the primary side from the clone board. I have spent a whole week trying to figure out the reason by modifying the snubber, changing the diodes, mosfet, ... but the heating does not stop. At the same time no matter what I do to the chinese board the transformer does not heat up on that board. Note that it is not just one transformer that I am talking about. I have tried this with a whole set of identical transformers and boards.
Any ideas? Anyone?

can u remove that Chinese flyback transformer....if so then just remove it & go to the renowned local transformer manufacturer he will get all the required details from it and make the exact prototype of it for u....did u check for the losses of both may that will solve ur problem....

 

[Orson Cart]

If the output rectifying didoes are markedly different (recovery time) this could cause extra currents in the transformers and therefore heating (and core heating due to high di/dt at fet turn on)
You would also see this in the mosfet and output diode if this was the case.

It seems unlikely the pcb is the culprit (i.e. more capacitive or lossy) you would see something in the drain waveforms if this was the case (i.e more ringing at turn off).

Ferrite does conduct and get hot over time if you put volts across it so check the "hot" pcb is not doing this.

Orson.

 

[kapurk]

The problem started with locally manufactured transformers. At first I thought it was the local transformers that were bad but it turned out that it was something to do with the board. Both transformers heat up with the FR4 board but not with the chinese board!

- - - Updated - - -

Hi Orson, thanks for thinking out of the box. the diodes and mosfet can't be the culprits because as I said in my first mail I have tested the boards by interchanging all parts but the results remain the same - the FR4 board causes heating but the chinese board does not.
Your idea about the ferrite seems implausible because the ferrite core does not touch any conducting areas on the board - it is infact raised about 1 mm above the board because of the transformer bobbin pins.
Any other ideas?

 

[mtwieg]

It's plausible that the two boards might be dissipating the same amount of power, but the thermal conductivity of the PCB might be the issue. If that's not the case, then there absolutely must be some observable difference in the voltage or current waveforms between the two of them. Conservation of energy and so on.

 

[Orson Cart]

any progress on this one? We would be interested in solving it if you don't mind sending the bits to NZ?
Orson