I wasn't suggesting actually going as high as 0.3Tesla as a working value
for B, this is too close to saturation, but did say that a figure of 0.2T
seems Ok for many applications.  Furthermore, what I forgot to say in that
last email is that when carrying a sinewave (as is the case in my 700W
switch mode Tx as the transformer comes after the tank circuitry) a core
seems to be able to carry significantly more power than when used with
squarewaves such as in a simple SMPSU.   Intuitively this seems as if it
'ought to be true' but I wouldn't like to justify the statement
mathematically.  It is the case ,though, that an ETD49 sized core in 3C85
material (or it may have been 3C90, can't remember) is rated at around 400
Watts at 100kHz in the data sheet, but survives 700W of sine at 137k quite
happily.

Wonder where your formula comes from?  40mT sounds extremely low and surely
means that the number of turns needed (51 turns for the example at 50 ohms)
will lead to a point where copper losses become quite significant.    I've
designed several switch mode PSUs over the years in the 50 to 200 Watt
region,  all around a Bmax of 0.15 to 0.2T operating at 100kHz usually with
3C85 and 3C90 material.  In none of them have the ferrite cores ever
appeared to be stressed at all, most run cold.   I may have been using cores
that are bigger than really needed, though, as they are in plentiful supply
in my junk box and make winding the bobbins simpler!

Must also admit, I've never had to consider linearity.  In simple SMPSUs the
waveform is hard switched, and in the 137k Tx the transformer was only ever
carrying CW, harmonic levels of which were dictated by the filtering action
of the tank LC arrangement with its loaded Q of 6, rather than any generated
by non-linearities in the core.

Andy  G4JNT

I must admit that I am not too happy with your suggestion for
0.2 to 0.3 T. Ferrite manufacturers may in their catalogues mention
such values of B, these values are, however, for low frequencies,
typically 10 kHz.
As for the loss components in ferrites, eddy current loses are mainly
frequency dependent and  hysteresis losses are mainly dependent on
magnetization.
In my suggestion to Dick, I used a figure for B of 40 mT. This figure

comes

from a formula I have been using for many years. My formula is fairly
conservative with regard to core losses and it also ensures that the
intermodulation caused by the nonlinearity of the magnetizing current
is negligible.
My formular for a recommended safe B is:
B  equal to 15 divided by the square root of the frequency, B being in
milliTesla and F in MHz..
This formula is in acordance with recommendations of e.g. the
ferromagnetic material producer AMIDON and, I think,
also with the pratice of designers of solid state wideband power
amplifiers.
73
Niels, OZ8NJ+