Hi Mal and all,

While it would be great if we all had linear amplifiers, I believe that
most modes of interest can be accommodated with a switching amp.

Of course, CW, RTTY, and all forms of FM and FSK are fine.  For PSK,
you could simply feed the signal into your switching amp.  If properly
designed, it won't be damaged when the drive amplitude decays to zero
and then comes up in the opposite phase.  This will, however, generate
some spurious signals.  If they are unacceptable, you can generate a
constant envelope signal that is PSK compatible.  Suppose you want to
send PSK at MS100.  To produce a phase inversion, you shift frequency
up (or down) by 50 Hz for 10 milliseconds.  You then return to the
original frequency for the remaining 90 ms.  This signal can be received
with normal PSK Rx gear, and will appear to be less than 1 dB weaker
than true PSK (for the same PEP).  But, since you can now drive the amp
into hard saturation, you should come out ahead.

Steve Olney's FDK is a very promising mode; it appears to be several dB
better than QRSS; check out http://www.qsl.net/vk2zto/ExpModes/fdk.htm .
But I believe that few have tried it because it needs a linear amp.
When you send FDK, the level of each tone must be 6 dB below saturation.
Instead, you could use a variant where the two tones are sent sequentially,
rather than simultaneously.  This costs a little in performance, because
some equipment drift and change in the propagation path occurs during that
time interval.  But, for a given amp, you are launching twice as much
energy into the air, which should more than make up the difference.
And, a switching amp is now no problem.

I am sure that new modes which are even more robust will be developed.
But I believe that a constraint of constant envelope is not onerous,
and will be the choice of most, if not all, designers.

Just my opinion.

73,

Stewart KK7KA

To: "rsgb" <[email protected]>
Sent: Sunday, December 10, 2000 10:59 AM
Subject: LF: amplifier design

Hello All

The method of using an HF frequency, dividing down then using high power
fets in class D or E then a LPF is not totally satisfactory and only permits
cw mode.
Keying this type of tx is also a problem as described in recent discussions
on the reflector plus the problems suppressing the odd harmonics 3/5/7 th
etc because of the square waves inherent in the design. Some of these
problems are being investigated and rectified but it still leaves a non
linear output and suitable only for cw or rtty.
This approach is used by most and  was a quick fix to get most of us going
on 73 and 136 khz  but maybe the time has come for a more traditional
approach like a linear  transvertor used in conjunction with an HF
transceiver and a high power linear amplifier with appropriate bias
switching for class C for more efficiency on cw.
This would enable us to use any mode generated by the hf transceiver plus
any other mode generated by a computer ie psk 31 and other psk variations.
It also solves all the keying problems.
The disatvantage is of course that with a linear approach you only get 50%
efficiency compared with 90% with the switched mode system.
I am currently thinking of progressing along these lines and building
something suitable when time permits.
The recent discussion on this media about SUPER amplifiers is very
interesting and am awaiting the big thump from one of these on 73/136 soon.
de G3KEV