Here is where i'm at...
This is experiments 101, interim progress.
While waiting for parts, used whats on hand:
Made Q90 a 2n2369A smt part, Better gain bandwith product (ft) gain maintains at higher frequency (also lowered bias)
I installed 6 2n2222A in the predriver and driver. (HFE hold up at high current(
Put 470 pf across R82 ( gain increases with frequency.
Placed 390pf cap from top of RV1 to the wiper (only works if your not maxed out!)
The trick here is couple more rf through RV1 with increasing frequency. If it is all the
way up its hard to get more vis this path.
Put 3.3uH inductor in series with R86 (allows decreasing feedback with frequency)
Turned up the bias on the IRF510s. I'm running at 210ma per device!
Doing this really helped the output power at 20M and up.
Key down at 10W is 2A.
Power set to 10W at 12.0 V 3.6mhz
30M 6W ??? output low pass filter??
17M 7.5w ??? low pass filter again??
This is still interim testing and I have many more things to try.
It does confirm that the performance can be more uniform.
I'm perplexed that 30M and 17M is so low, it may be the low pass filter in use.
The only problem I see with 210ma per device is the no-modulation standing current is .8A
which is not battery friendly but maybe what we need. They do not seen to heat much.
It suggests T11 is less than optimum.
I need to try improved output transformer both 1:1 (2t:2t) and 1:2 (2t:3T) on
bn43-6802 (and bn61-6802). I may also try a different configuration driver
This IS NOT FINAL, net even close. I need to see what it looks like with spectrum analyzer.
The final result will be put up with pictures and explanation of what, where, and how.