By using a parallel tank and a step up transformer, you are making it about the
same complexity (only one less capacitor per tank) than you would with switched
pi-networks. If you choose the relay to have DPDT or DPST contacts, you can
switch both sides of your pi-net into the input with only a couple more
soldered joints. Most relays have more contacts, by the way.
I went through this same analysis a month ago, with a Harris RF Comm RF101
amplifier I am restoring. It lacked an input circuit, and has a pair of 3-400Z
triodes. So the input is virtually 50 ohms, but only part of the RF cycle. I
found a handful of small circuilt board relays made by Panasonic (Aromat) that
were for DC/AC and not really RF. I wired one with coax and BNC spigots, so
that I could measure through my HP network analyzer. The thru loss is low, the
leakage (isolation) between in and out for the relays is 70 dB at 160 meters
and better than 60 at 20 meters (the highest band on the amplifier). The
contacts are rated for more current than I would ever see with 100 watts of
drive. So I built a small vector board, with one relay per band, and a National
tunable inductor and two mica caps for each pi.
Pi network makes a low pass filter too, whereas a parallel LC is a band pass
filter. Not that it matters since most exciters or rigs are already filtered. I
suppose that the only thing I could think of in favor of a parallel LC is that
it has a DC ground whereas a Pi doesn't. But you have to have a blocking
capacitor anyway, to keep the filament voltage off the input.
Just some thoughts. Additional contacts on relays are cheap. Had you been using
rotary wafer switches, then it does become more wasteful to use a pi. But if
you only have single pole relays, then your argument makes more sense. Either
way, you will get some stored energy with your network, and help keep the
driving current waveform sinusoidal and continuous across the cycle.
73
John
K5PRO
> I continue to work on the design/construction of my 4-1000 amplifier. I'm
> working on the input network details.
>
> The 4-1000 has a generally accepted input impedence of 100 ohms. So I am
> planning on using a 2:1 step up unbalanced to unbalanced balun followed by
> a parallel tuned circuit, Q of 2-3, for each band. Taking this approach I
> will need only one relay contact per band to switch each tuned circuit (as
> opposed to two needed to switch a pi network).
>
> In all of my reading and research I haven't seen this approach taken. The
> grid compartment is under construction. Before I go ahead and construct
> the input this way - can anyone come up with a reason this won't work?
>
_______________________________________________
Amps mailing list
Amps@contesting.com
http://lists.contesting.com/mailman/listinfo/amps
|