[600MRG] Fwd: Class E Keying

[Hans Summers]

Hi Frank (et al)

This topic is of particular interest to me since I am experimenting with
key-shaping of a CW PA on the workbench at the moment. Mine is broadband
(all HF) and not Class-E but I think my comments apply equally. My
questions/comments for discussion:

1) Why key the power supply?

http://njdtechnologies.net/wp-content/uploads/2016/02/GW3EUP-keyed-pa.gif
 and
http://www.its.caltech.edu/~mmic/reshpubindex/papers/buckwalter.pdf
I am not clear on what the point is, of keying the power supply. It appears
to me, better to key-shape the amplitude of the PA input. This is at a much
lower level, requiring smaller, lower power devices and simpler circuits.
What am I missing? To be sure, the shape might not be ideal... but that
brings me to my next point:

2) RC key-shaping isn't very good... is it?

An ideal envelope shape is the raised-cosine. Now:
http://www.its.caltech.edu/~mmic/reshpubindex/papers/buckwalter.pdf
Look at figure 4. it's typical of what you get when you use a simple
Resistor-Capacitor envelope shaper. The capacitor charge/discharge through
the resistor is an exponential rise/decay. On key-down, the front of the RF
envelope has a nice shape. But on key-up, the sudden exponential decay is
far from ideal and will create a click. I'm sure the quoted 10ms decay is
better than a hard on/off keyed envelope (rectangular) but still, it isn't
ideal.

3) PWM sidebands?

Finally on the subject of using PWM to shape the envelope, surely that has
the potential to introduce sidebands on the output waveform, at the PWM
frequency and its harmonics. Yes filtering will help but... will it be good
enough?

4) My project:

What I am exploring is adjustment of the bias on the driver stage. I am
using an 8-bit Digital-to-Analogue Converter (actually a shift-register and
R-2R network) to control the bias level to that driver stage. In my circuit
It allows me to vary the final power output from zero to full power in 256
steps. A capacitor smooths out the steps too. The DAC is programmed by a
microcontroller to create the key shape. It is possible (and not very
difficult) to quite closely approximate the ideal raised cosine keying
envelope. I mean on both key-up and key-down. Yes, now you need a 74HC595,
some resistors for the R-2R network, and a microcontroller - but all these
things are very cheap (your project may already include a microcontroller),
I bet they cost less than the lumpy MOSFET that's being proposed to ramp
the PA Power - and the result is a real shaped envelope!

My purpose is a future PA module QRP Labs kit, that can be used with the
Ultimate3S QRSS/WSPR transmitter kit http://www.qrp-labs.com/ultimate3 and
other future QRP Labs kits. One simple cheap IRF510, I have not tested it
on 600m. The closest I tested it at is 160m. With 13.8V power supply I get
14W output at 75% efficiency.

Comments very welcome... I feel sure I must be missing something(s) !

73 Hans G0UPL
http://www.qrp-labs.com
http://www.hanssummers.com