Why not just twist together some multi-filar turns and use all the sirface area you can for cooiling?

I  use standard ETD54  type  SMPSU cores for 400 Watts at 475kHz and 700 Watts on 137kHz and they barely get warm.  I think the ferrite type is EC90, but the cores are quite old, and may be EC85.    Those are just standard transformer winding used for isolation and impedance changing.   

Andy  G4JNT

On 27 February 2016 at 17:36, Paul Nicholson <[email protected]> wrote:

On again.

 Freq: 477690.00000 Hz
 Chars: 15
 Symbols: 0.5 seconds
 Code: 8K19A
 Duration: 496 seconds, 00:08:16
 Repeat: XX:00 XX:20 XX:40

I rewound the transformer, the primary windings are the braid
of coax and the secondary is the series connected coax cores.
FETs running a lot cooler now and the only thing getting
warm is the coax.   DC in 27V at 3.8A and getting the same
antenna volts as yesterday with 150W in so that's a lot better.
The cause of last night's high unloaded current was a shorted
secondary turn - the insulation had melted.

I'm a bit worried about using coax like this, the inner will be
at the temperature of the braid plus its own temperature rise.
I expect a short from primary to secondary when the dielectric
of the tightly wound coax coil softens.   I'd like to find some
flat braid with high temperature insulation.

The toroid I'm using is this one

 http://uk.rs-online.com/web/p/ferrite-rings/2120803/

which I picked because I found a MF PA design on the web which
used it.  It's the wire that's producing the heat, not the core.

--
Paul Nicholson
--