[Amps] Designing the Cleanest Linear with RF Negative Feedback

[Tom Cathey]

Hi Marv,

The 75M mono-bander:

Well, it looks like your original driver from that Hughes amplifier with the 
6146 driving the 4CX-250 is still the best way to go. Thanks for clearing up 
the stage gain question.

The reason I figured 50 ohms would be OK for the OPA-695 is cuz the data 
sheet showed sample circuits using 75 ohm resistors terminating both the 
input and output. Did you interpret the 200 ohm minimum load for  better 
distortion from the data sheets or is this your opinion from experience?

I was also thinking of a stepped-up link coupled parallel tank  from the 
'695 to the driver grid.  That would give some voltage gain, but if I use 
the 6146 pre-driver, it will not be needed. Still, maybe it's a good idea to 
use it and be able to reduce the gain of the 6146 and make it more stable 
there, dunno.

BTW, for my other project, the 10-40M linear...  I have a  3CX-3000A7 triode 
and socket. I was thinking of putting that in GG,  driven with a pair of 
grid driven 4CX-250's in class A.  Then wrap about 20db of neg feedback from 
the 3X3 plate to the driver grid.  And then drive the 4CX-250 with a similar 
OPA-695 / 6146 circuit in class A. Good for a clean 1500W, I'm hoping. 
Maybe I will not need to neutralize the 4CX-250, having only 0.25pf G-P 
capacitance, thus avoiding interaction with the NFB loop at the same point. 
Does this sound like a reasonable approach for a GG final looking for total 
system cleanliness?

73,
Tom, K1JJ

---------------

Hi Tom,
   A couple of points for you to ponder.

   A single OPA-695 won't make 5V RMS.   However, it's capable of a voltage 
gain of 10X, 20dB power, so you only need a single stage to go from 5mW to 
500mW.   A pair of them connected differentially will output 5V RMS, and 
provide some 2nd harmonic cancellation as a bonus, but, they probably 
shouldn't drive less than a 200 ohm load for best distortion numbers.

   Or... a single '695 could be followed by a +/-15V rail part, a THS3001 
(or a more recent part from that family) for instance, to output an easy 5V 
RMS... or 10V from a differential pair.  Again, this scheme may not be happy 
driving into 50 ohms.

   Note: Both of these op-amp approaches only exhibit exquistely low 
distortion on the low bands.  At higher freqs., their distortion numbers 
rapidly deteriorate.

   As the 4CXwhatever grid is a high impedance point, a tuned matching 
network might be better fit than a wideband transformer for single band 
operation.

   Your proposed 20dB negative feedback around the driver-final means that 
you will require more grid drive, say somewhere around 50V RMS to the 
driver, to maintain the same overall power out.  The op-amp pre-driver may 
not provide enough oomph in this instance since a 10-1 voltage gain, 
employing a tuned matching network sourced from 200 ohms, would imply a grid 
impedance of 20K ohms.  Where's that free lunch?  :-)

73 & Good evening,
   Marv  WC6W