>> The plate swing probably will be slightly less than the DC voltage,
>> for a reasonable efficiency amplifier. Otherwise it will be even less and
>> the amp will be underloaded. So figure that the voltage goes from 0.2 to
>> 1.8 x the DC or something like that, at the plate. The experts here
>> probably know the correct figure. Then on the other side of the blocker,
>> you only have the RF swing without the DC component. So the cap passes the
>> RF through, and blocks the DC, right?
>That is true only for a properly loaded amplifier. In a misloaded or
>mistuned PA,. or a greatly over-driven PA, the peak RF voltage can
>greatly exceed the supply voltage on each half of the RF cycle!
/\ During the Grate Parasitics Debate, Tom Rauch made the same claim in
an attempt to put the blame for bandswitch arcing on operator error. As
a result, someone measured the worst-case tank mistuning peak V. in a
SB-220. The measurement was 3600V-pk - which is c. 24% higher than the
supply V. Nacherly, Mr. Rauch switched to ignore-mode. [note: the
SB-220 bandswitch has a 5000V. breakdown.]
>I can demonstrate, with an amplifier biased to the low edge of class
>AB, that peak voltages many times the supply rail can be
>produced. Switching supplies work on this principle.
>If there is not a very fast load-fault circuit, the PA should have a
>spark gap of some form on the anode side of the tank.
/\ Most do. It's the Tune-C.
>In any amplifier biased much less than class A, and unless I had a
>load-fault trip that would work faster than heat would damage the
>component that arced, I would include an intentional spark gap of
/\ Fairly good idea - providing the spark-gap has a 50 - 200 ohm low-L
resistor in series with it.
>73, Tom W8JI
- R. L. Measures, 805.386.3734,AG6K, www.vcnet.com/measures.
FAQ on WWW: http://www.contesting.com/FAQ/amps
Administrative requests: amps-REQUEST@contesting.com