[Amps] 400 watt Polyfet HF amplifier

[Manfred Mornhinweg]

Hi John,

 > Polyfet has a 2014 technical bulletin on their website, TB239.

Please allow me to throw a bucket (or at least a few cups) of cold water 
on that design:

- According to the gain curves shown, that amplifier is pretty linear up 
to a PEP of roughly 300W at 2MHz, 250W at 15MHz and 230W at 30MHz. The 
400W rating is for deep saturation, not for linear service.

- The claimed efficiency of 66 to 70% is for the saturated condition, at 
400W. Another graph shows efficiency at 1dB compression, and goes from 
66% and nearly 400W at the low frequency end to 58% and 290W at 30MHz.

- The IMD is rated at 300W, and is barely -28dB for IMD3 at 16MHz. It's 
much better at the low extreme of the frequency range. And there is a 
reason for this...

- At 400W the rated IMD3 is clearly too bad for ham use.

Now some technicalities:

- The bifiliar feed choke is completely inadequate, having far too high 
leakage inductance. It acts like a bifiliar choke only at the low end of 
the frequency range, degrading effectively into two separate chokes at 
the higher frequencies, so that the two drains become uncoupled in that 
range, making the amplifier run in something close to current-mode class 
D rather than in class AB. This explains the drastic worsening of IMD as 
the frequency increases beyond the proper range of that choke.

- This bifiliar choke is drawn with the wrong phasing in the schematic.

- The bypass capacitors C11-C16 are poorly placed. The loop formed by 
the FET, the bifiliar choke, and these capacitors should be absolutely 
the shortest possible. It would have been easy to place them better.

- This amplifier has no common-mode feedback whatsoever, and almost no 
common-mode gate loading. Of course there is no common-mode drain 
loading either. For these reasons I would suspect that it is extremely 
prone to common-mode self-oscillation. This is an often overlooked 
problem in push-pull amplifiers.

And a minor thing: Strapping together the two cores of the output 
transformer isn't a clever thing to do. It would be better to leave some 
distance between them. Magnetically it makes no difference, but 
separating the cores gives a thermal advantage by exposing much more of 
their surface to cooling air.

After so many bad points, I would also like to stress a good one: This 
amplifier shows a good layout with the bifiliar choke soldered directly 
to the drain connections, and the output transformer shoved out of the 
way of the choke. Many designers don't get this right.

Manfred


========================
Visit my hobby homepage!
http://ludens.cl
========================