[Amps] RF Chokes

[Peter Chadwick]

So what inductance should the shunt feed RF Choke be?
Take a 'classic' choke - the old National R-175, used at the kW DC input in AM  days. It's 145 microhenries, so with something around 2kV rms across it, and its reactance of about 3.2kohm at 3.5MHz, there's 625mA of RF in the choke. Plus the DC, of course. Assume that the choke Q is 100 (probably being generous here) and we have 12.5 watts dissipation in the choke.
I have a 4 'pie' choke, reputedly used in the NCL 2000. That's 950 microhenry, and issipates a couple of watts. But the problem is the series resonance - in the way I have it mounted, that just happens to be at 21.5MHz.
Some transmitters I've seen used under 100 microhenries, but had no series resonances in their operating range.
So what's the best to use?
The lower the value, the better the chance that there won't be a nasty series resonance, but there's more tuning capacity needed, and more current through the bypass cap at the HV end. Incidentally, I seem to remember that Rich has made some comments about the value  of the bypass capacitor, which obviously has to handle the current. It seems that you need an extra choke between the bypass cap and the HV supply if you're really going to keep the RF out of the supply, and thus the mains. I know Alpha in some of their amps switch the choke, which is one approach. But especially when you're covering 160 as well as 10, it's obviously not straightforward as to what choke to use. I've tried the tack of a solenoid choke (possibly ex ART13 - solenoid, about 1/2 inch diameter, about 4 or 5 inches long, 95 microhenry) in series with the NCL 2000 choke. Series resonance for the pair is 24MHz.
Question: how far away from the operating frequency is acceptable for a series resonance? The resonances seem to have quite high Q.
Measurement method is to remove the tubes, use a scope probe as that gives about the same capacity as the tubes, connect a load resistor of the calculated laod value between plate and ground and feed a signal generator in (via a return loss bridge) at the pi output. Tune the pi for max volts across the load resistor consistent with max return loss. Vary the sig gen frequency, and you can see frequencies where there's very sharp dip in plate volts - down pretty well to nothing. Which I feel is a delicate test for the series resonance.
It does seem, prima facie, that a lot of the chokes in use are rather low in inductance for ideal operation.
73
Peter G3RZP