[Lowfer] The real filter deal!

[Lyle Koehler]

My approach is to use a power limiting circuit (a resistor in series with
the DC supply) that limits the power input to 1 watt, tune the antenna and
any matching networks for maximum field strength, and keep the final
amplifier out of reach of any scope probes :-)

This non-purist approach might not work at high power, where weird waveforms
could spell disaster for the final transistors, and may not extract every
last fraction of a dB from the transmitter. But that's how the LEK beacon
has been operating for the past few years.

Ringing in solid-state power amplifier matching networks is discussed in QST
for October 1978, page 34; and January 1980, page 55. The problem is very
evident in single-ended amplifier designs where the output transistor
essentially looks like a switch that is either open or closed. With the
complementary-pair final, one transistor is always conducting, so the
ringing is not as evident. However, the effect is still there, and
efficiency can be lost even though other nasty effects such as
self-destruction of the output transistor(s) does not occur (at least at the
1-watt level). A Zener diode across the input of the filter is often used to
suppress the spikes caused by the ringing. However, as Roy Lewallen points
out in the October '78 article, it isn't necessarily the Zener action but
the large capacitance of the diode that does the job. I've used capacitors
across the filter inputs in single-ended HF finals to tame the waveforms and
improve the efficiency. This trick didn't seem to be as effective (or
necessary) at LF, but it might be worth investigating further. I've sent
copies of the referenced articles to Bill Ashlock. Perhaps his PSpice models
will reveal whether the capacitor technique is useful when driving something
like a high-Q loop.

Lyle, K0LR