RD16HHF1 power curve flattening...some

John <passionfruit88@...>
 

Having read with interest the information provided by Erhard Haertel (DF3FY) on ubitx.net regarding the changes to the finals circuit I decided to try to improve the flatness or lack thereof of the power curve over frequency on my uBitx unit.
 
I therefore performed his proposed changes (final's transformer, cap across drains, feedback resistors) with one exception: the feedback resistors between drain and source of the final was changed from 220 Ohms to 820 Ohms instead of 1K Ohms.
 
I got the following result at 13.8V, with idle current at 250mA per final, RV1 unchanged at approx 60% as before:
 
Freq.      Power     Total Current
 7Mhz        18W         2.8A
14Mhz        16W         2.7A
21Mhz         9W         1.5A
24Mhz        11W         1.9A
28Mhz        11W         2.2A
 
Hmmmm...not bad but still that dip at 21 and not that high on the high frequencies.
 
But the good news was that I would get power improvement all the way to the end of the RV1 pot and it looked like no compression was occurring. Good news as this means that I have spare power for frequency compensation.
 
Using the LTSpice simulation done by Jerry Gaffke (thanks Jerry) with a replacement of the finals with the RD16HHF1 model found on the internet and I got some interesting pointers (reality is always different, but I find this is a good starting point).
 
I simulated capacitors in parallel with the 22 Ohms emitter resistors in the pre-driver and driver stages alternatively. Since I had "free gain" left by turning RV1 this meant that none of the stages were limiting at 60%, therefore I could increase the gain anywhere in the chain, including at RV1 for that matter.

LTSpice gave me a flat curve with caps of 150-200pf in the pre-driver stages, and more importantly a rapid drop in power above 40Mhz which I hope may prevent V-UHF oscillations of the power train (linear amplifier specialists can comment on that point).
 
So I experimented with various caps values across the R87 and R88 resistors and found that I would not get more gain above 330pf.
 
Results as per above conditions (RV1 back at 60%):
 
Freq.      Power     Total Current
 7Mhz        18W         2.8A
14Mhz        16W         3.0A
21Mhz        10W         1.6A
24Mhz        11W         not recorded
28Mhz        14W         1.9A
 
Not perfect but at least now 10W on the high bands, and above 15W at and below 14 megs.
 
One minute key down then "finger heat test": the finals new transformer is a little warm but not much. The filters the same. The finals and the heat-sink (thermally coupled to the metal case) are fine.
 
So the question is: can the filters handle that kind of power over time? Short of doing and LTSpice simulation is there an easier way to calculate the current and therefore the flux density in the filters cores to check if we are getting to saturation or not? 
 
No instability and oscillation detected with a separate radio, but again I don't have a spectrum analyser. 
 
So in summary this is what I have in the circuit now: Finals as RD16HHF1s, feedback resistors R261/R262 as 820 Ohms, Tranformer 2T primary / 3T secondary on BN43-202, 33pF across the output transformer primary and 330pf across R87/R88.
 
I haven't tried capacitors across the driver stage emitter resistors (R941/942/911/96), nor from the high to the middle side of RV1 for the following reasons: I wanted to leave the drivers alone and not modify the way they drive the finals (impedance change), and at RV1 it meant that the capacitor value would need to change with the adjustment of RV1, although this could be the simplest way to correct the power curve. 
 
Interestingly enough for the units with the IRF510s, LTSpice reports a flat power curve with 800pf caps across these four drivers resistors....it could be just theory...and stability could be an issue as we reduce negative feedback. Is anyone game enough?
 
Happy hacking,
 
73, John (VK2ETA)

Join BITX20@groups.io to automatically receive all group messages.