Flyback transformer core heating

[MrCrowley]

Hello,
I'm designing dc/dc flyback converter 10-42 to 3.3V/3A. I need efficency more then 75%. So on my prototype board I have about 50-60% and transformer core is heating up to 90 deg. I am using toroidal amorphous iron-based core with air gap. Length of middle line is 50.24mm, Magnetic permeability is 258, amplitude of induction Bm=1500 Gauss. Primal winding is 11 turns, secondary is 6, supply winding is 7 turns. So inductace is about 14uH in prim and 4uH in sec.
The question is why core is heating up? I was trying to use different frequencies from 60kHz to 800kHz. The transformer is not in saturation mode. It is working in discontinuos current mode. Everything seems to be good but its heating up. Why? I thought it can be losses in hysteresis or Eddi's current. But both of them will increase with rise of frequency and peak primary current. I was trying to reduce peak current from 4A to 1-2A, and freq to 60kHz, but losses were quiet same. Winding of several wires in parralel also didn't help. What is the reason of this heating and how the problem can be solved? Thanks.

 

[bala0x07]

Hi MrCrowley,

The core usually heats up due to saturation, but you have stated that the core is not saturated
The core may also heat due to heat transfer from the coil

Have you clear about the coil turns vs frequency and Inductance factor relationship
Have you selected proper core ?
800KHz is really a huge frequency and the core must be able to handle that

 

[FvM]

Did you read core loss curves for the used material? Are you talking about a tape wound or powder core? Can you specify the exact core type?

Do you have each winding uniformly distributed around the toroid? Can you clearly distinguish between core and windings losses (e.g. due t skin and proximity effect)?

Most 10 W fly-back converters are fine with small ferrite cores. Specific reasons why you don't use it?

 

[MrCrowley]

Coil is rather cool. Heat is produced by core. I am limited in choice of core. Core is 25um amorphous iron-based tape wound. Windings are distributed uniformly around.

 

[FvM]

For which switching frequency did you calculate Bm=0.15T? If you keep the output power and change the frequency, Bm will change as well in DCM. That's the probably the reason why the the core losses don't change much.

µeff of 258 sounds still high for an energy storage inductor respectively fly-back transformer. Try with larger air gap (and increased number of turns, if necessary).

 

[MrCrowley]

Bm=0.15 T is absolute maximum for that type of core (from vendors site). Any chance to do converter on this core? It is rather problematic to get another one for me...

 

[MrCrowley]

Sorry, not 0.15 T, but Bm=1.5T

 

[FvM]

Can you please tell the complete core geometry?

 

[MrCrowley]

It is 12.0-8.0-4.5 (in mm) - D-d-h. And effective cross-section is 7.2mm^2.

 

[FvM]

The peak flux is too high, around 0.9 at 60 kHz. Need more turns.

 

[MrCrowley]

I was trying to increase number of turns in two times(from 11/6 to 22/12) but it doest't really help.

 

[FvM]

O.K., there are too conditions to be fulfilled:
1. Number of turns according to ΔB for any kind of transformer
2. Air gap respectively µeff according to DC magnetization for fly back and storage inductors

You probably also need to increase the air gap, I didn't yet calculate.

 

[MrCrowley]

Ok. I will try to increase num of turns one more time. Can you please tell me which formula are you using to calcultate ΔB?

 

[CataM]

The peak flux is too high, around 0.9 at 60 kHz.

What would be an acceptable peak magnetic flux density ? 0.4-0.45 ?

Can you please tell me which formula are you using to calcultate ΔB?

He used standard magnetics formula, found in every standard power electronics book. Or you can simply ask google.