My experience might not be helpful, but for what it's worth:
I've used the G3SEK spreadsheet for two amplifiers, both with pi
networks (not pi-L). The first was a class-C amplifier with a pair of
813's. I found that the calculated value for the loading capacitor on
160 was inadequate, but what seemed to be wrong was the calculated
inductance. I ended up increasing the inductance to get proper loading
and good efficiency on that band. I used K=2.0 for class C.
The other amplifier was a 4CX1000A. Here the issue was that the
impedance shown by the load line was much lower than what would be
produced by a K of 1.5 - 1.7 as suggested for class AB (it was closer to
a K of 2). I went with the load line value and the amplifier seems to
load properly with reasonable efficiency.
On 3/11/2014 8:44 AM, Jim Garland wrote:
Hi all, I've built a tank circuit using the on-line Pi-L network
calculators by VE3OZZ and also G3SEK. These are both based on the equations
published in an article in August 1983 QST (by W5FD). Althought the G3SEK
calculator is somewhat more sophisticated (it corrects for tube and stray
inductances and capacitances), both calculators give about the same answers
for 80m and 160m. I'm finding that the predicted values for C1 and L1 are
very close to what I need to tune the amplifier, but the predicted values of
C2 are far lower than what is required to load the amplifier properly. I'm
wondering if there could be an error in the W5FD formulas, and if anyone
else has experienced the same problem? (If there is an error, it probably
wouldn't be noticed on the higher frequency bands, because the load cap
would most likely have enough tuning range to compensate for the error.)
I've computed the Pi-L network values over a range of plate impedances. (My
amp uses bandswitched L1 and L2, so those values don't change.) What I find
is that as the plate impedance increases (e.g., tuning the amp at a lower
power level), the equations predict that Q goes from 10 to 18, C1 doesn't
change, C2 increases only about 5 percent. In other words, according to the
on-line calculators, tuning to a lower power requires a minor tweaking of
the load capacitance, but that's all
At 3.5 MHz, for my amp, the equations predict a load capacitance of 1057 pF
and on 1.8 MHz, a load capacitance of 2057 pF. I'm finding that, in
practice, those predicted values are more than 1000pF too low.
Here are some details of the actual tank circuit:
The design plate impedance is nominally 720 ohms (2500V@2.0A, with k=1.7)
<mailto:2500V@2.0A,%20with%20k=1.7)> , Q=10, and I've computed network
values for a range of plate impedances from 720 to 1440 homs. The actual
tank circuit is:
80m: L1=8.4 uH, L2=3.8 uH
160m: L1=16.3uH, L2=7.4uH
I'm using two paralleled 1000pF doorknobs for a plate blocking capacitance.
The plate choke is 225 uH, bypassed at the base by 7700 pF. The safety RF
choke is 470uH, with an 18 ohm DC resistance. The tune and load caps are
30-240pF and 33-1000pF air variables padded with doorknobs, as required.
The tank seems to tune smoothly, with no heating or quirkiness. THe only
problem is that I need much more C2 capacitance than the formulas predict.
At this point, I'm at a loss to explain the discrepancy, other than
wondering if there's an error in the formulas somewhere.
73,
Jim W8ZR
_______________________________________________
Amps mailing list
Amps@contesting.com
http://lists.contesting.com/mailman/listinfo/amps
--
Vic
_______________________________________________
Amps mailing list
Amps@contesting.com
http://lists.contesting.com/mailman/listinfo/amps
|