(Flyback) Transformer heatsink?

[daqq]

Hi guys,

I'm designing a flyback power supply, and the transformers' copper goes up to about 90 deg. C. The efficiency is ~90%, which is good enough (though I will investigate on how to improve that), but that's still a few watts (around 5W I'm guessing) of thermal power to dissipate on a very small surface.

One quick fix is to add a heatsink to the offending area, as proposed in the attachment. Basically, a machined alluminum block would be thermally connected to the copper through the outer kapton tape and through a thermal pad material.




My questions are:

1. Is this a good idea? If not, why?
2. Will there be any stray currents created in the alluminum heatsink when it's this close to the winding? If so, how much of an effect will they have?
3. Are there any better solutions to the problem?
4. Is this a problem? 90.deg C is a lot, but the rest of the power supply (transistor, sense resistor) are relatively cool.

Thanks,

David

[T3sl4co1l]

Is that 90C at ambient, or at worst case?

Heatsinking works, yes, but only as well as the total thermal resistance.  None of the materials involved are terribly good, I mean, below the outer layer of wire.

If you can soak it in varnish, it'll be able to do a much better job.

Tim

[daqq]

Thanks for the reply Tim,

Good idea with soaking the thing. The stuff I have at the moment would probably not like the temperature very much.

The 90 deg.C (85 actually) is at about 30 deg ambient.

[Miyuki]

Don't be cheaper and easier to just use higher temperature ratting wire and just keep it hot, when it can be hot in safe way ?
90°C is no issue if it is designed to be at this working temperature. Just must be selected right wire and core material.
Cores and wires are available in temperature classes and can be even over 200°C design for full time at this high temperatures, so 90°C is nothing if you chose right material.

[Zero999]

As mentioned above, 90^oC is no problem with the correct materials.

Is there room on the former for more wire? Can you increase the gauge slightly to reduce the copper losses? Is it feasible to increase the frequency? How about using a different topology than flyback?

[daqq]

Thanks guys for the replies.

Don't be cheaper and easier to just use higher temperature ratting wire and just keep it hot, when it can be hot in safe way ?

Well, the cable and the other parts of the core should survive those kinds of temperatures. Also, price is not much of an object here, and best to be on the safe side...

What's a typical flyback transformer operating temperature?

Is there room on the former for more wire?

Well, not that much, but I'm currently using DIY Litz wire - the primary (48V IN) is made up of 4 strands of 0.4mm copper, the secondary (350V OUT) is made up of 2 strands of 0.4mm copper wire. I'll be experimenting with that as well.

Is it feasible to increase the frequency?

Possibly, but not by much.

How about using a different topology than flyback?

Sadly no, that would be too much of a change. Basically this setup works and if it can be tweaked, I'll try and do so.

[Zero999]

I haven't personally encountered a flyback transformer that runs that hot but it wouldn't surprise me if some small, high power density, switched mode power supplies do.

If cost is no object, then can't you simply buy a larger core? That could get the temperature down and reduce the copper losses, as you can use thicker wire.

[daqq]

If cost is no object, then can't you simply buy a larger core?

Might do that. But at the moment I'm trying to fit into an existing PCB and there's only one size of cores that'll fit there, bobbin included. Also, size is an issue as the whole PCB must be less than 40mm high. The current core with the bobbin included is around 35mm high.

I'm actually suprised at how high the copper losses are - or rather how much of an effect they are having. Oh well, live and learn 

[daqq]

So a bit of an update: I've rewound the transformer. The transformer is a step up transformer.

The original cable was litz wire I made up of 0.4mm thick strands (4x on primary, 2x on secondary). I changed to 0.2mm strands (16x on primary, 8x on secondary).

The DC resistances are pretty much the same (old: 32mOhm and 306mOhm vs. new: 27mOhm and 240mOhm), the other parameters (ratio, (leakage) inductances) are pretty much the same.

The results are very nice, I got a ~20 deg.C reduction in temperature. Now the transformer is happily buzzing at ~60 deg.C. I've also taken T3sl4co1l advice and sprayed in a little varnish into the windings. I don't know if it had any effect, but it did not hurt as far as I can tell. I will have to buy proper higher temperature lacquer/varnish later on.


Now for more parts I'll be needing to buy the wire rather than make it. Any recommendations on a manufacturer or litz wire? I can't find it on farnell nor any other distributor.

Thanks,

David