> If you look at the tank circuit of these amps, you see a pi network that is
> usually no different than any other amp, except the coil is small turns wise
> because it is for 11 meters (and close to 10 meters as well). So what kind of
> "filtering" are you talking about?
If the pi network is not different than, say, a 30L-1 or an SB 200 or any other
amp with a pi network, why would harmonics be any more of a problem than with
other amps?<
Firstly, if you do the sums for a Class AB tube amplifier with the usual Q
values, you'll see that you don't meet either the FCC requirements for
harmonics (46dB) nor the Radio Regulations requirements, which are 4 dB
tighter at 50dB. Which is why the manufacturers these days use a Pi-L network.
Not that anyone worries too much about existing equipment, which from memory,
is, at least as far as the Radio Regs are concerned, 'grandfathered' until 2012
anyway. The solid state CB amplifiers, besides being pretty awful on IMD,
usually have minimal harmonic filtering. In fact, you can, in a solid state
push pull amp, have reasonable IMD, but a 3rd harmonic only 13 dB down. So this
is why you need decent low pass filters. Now some transistors, presented with
the input impedance of a filter at the 3rd harmonic, happily oscillate. The
Swan 100MX had a tendency to do this on 15metres with some antenna loads.
Biasing the 12 volt transistors is fun anyway. From a DC viewpoint, you'd like
a constant current source of bias that reduced as the transistors got warm and
the beta went up. But a constant current source is ideally an infinite
impedance, and as soon as you apply RF, the rectification then alters the bias
point. So in practice, you have to end up with a constant voltage, very low
impedance bias source, which has to fall in voltage as the die temperature goes
up at approximately 2mV/degree C plus a bit more to allow for the increase in
beta as temperature goes up. Add to this the fact that the RF saturation
voltage of the 12 volt transistor isn't that negligible, so the available
voltage swing is not quite so big as you would hope - quite likely down to
about 20 volts peak to peak at the best - and you can start seeing why the CB
amps are so dreadful. At 100 watts, with 20v p-p, which is about 6 volts rms,
the load impedance is down to around 0.36 ohms, so losses start to bec
ome a major problem. So this why people go to 24 or 48 volts and even better,
to FETs, which tend to be a bit more linear anyway. You can use a constant
current source of bias with pure Class A, but the bias must not change when the
RF drive is applied - this is done in integrated circuits with driver stages on
board for linear cellphone applications, but not above about +8 or +9dBm output
power.
To avoid the instability problems, you need to look at the S parameters at the
harmonic frequencies and do either a Linville or a Stern stability criteria
calculation when loaded with the impedance that the filters present at those
frequencies, as well as an all phases VSWR of whatever your worst case VSWR is
at the operating frequency. At least, this is what proper designers do, not the
CB amp makers, most of whom, I suspect, have never even heard of Linville or
Stern or stability criteria! The exercise also tells you how much margin you
have on the stability. There are stability criteria equations for tubes, too.
73
Peter G3RZP
_______________________________________________
Amps mailing list
Amps@contesting.com
http://lists.contesting.com/mailman/listinfo/amps
|