[Amps] Designing the Cleanest Linear with RF Negative Feedback

[Tom Cathey]

Thank you Marv. You da man!

That answers a lot.  I'll ask you just a few more, please, and then let you 
be...  :-)

I've got a few of 4-400A's, sockets and chimneys here. I was hoping to 
substitute one for the 4CX-350 driver. I think the IMD figures are not as 
good, but running it in deep class A with just voltage needed to drive the 
4CX-3000 might work OK?  -62V on the grid, just like the 4CX-350, appears to 
idle a 4-400A into class A too.

I will look at the OPA695 RF op-amp you suggested. A friend also  suggested 
I  could take the 10 mw out of the FT-1000D and build a Norton push pull 
amplifier with  some 2N3375's he has. A pair would do about a watt class A. 
Maybe that would be a good driver for the 6146, since only 100Mw is needed. 
[keeping with the light duty rule]

I'm glad you supplied the specs for bias, drive, etc. I can see the 
advantage of running everything light in the drivers. Then let the final 
make the IMD and try to control it best with NFB. I would imagine that a NFB 
cathode resistor in each early voltage stage would help if no loop is used 
back there.

So, bottom line is I plan to have one watt, class A,  at -55db 3rd order, 
available to drive the 6146,  which drives the 4-400, then the 4CX-3000. I 
would like to go for more like 20-25db NFB. [Trying for -50 to -55db 3rd IMD 
total]  Do you think a second 6146 would be required to generate enough gain 
to do this, or am I just asking for trouble with stage stability from too 
much gain and NFB phase problems?

73,
Tom, K1JJ

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

Hi Tom,
  In answer to a couple notes:

  20 feet of coax should pose no problem as long as both ends are 50 ohms.

  There may be some modern devices that perform better than those ancient 
TRW things I suggested.  I'm not really current with those parts.  Texas 
Instruments has some great Op-Amps which, at around 100mW are not quite as 
powerful as the old hybrids, sport really fine distortion numbers at 4Mhz. 
Take a look at the OPA695.

  I don't know much about the insides of the FT-1K.  A look with a spectrum 
analyzer would be better than my speculation.  :-)

  Running the tubes "lighter" improves the numbers at a faster rate than 
negative feedback.  NF nets about a dB distortion reduction for each dB of 
NF.  Reducing the stage power 3dB reduces the 3rd order products 
(theoretically) 9dB -- the cube!   The higher order products diminish at 
progressively higher rates.

  Since you running at a relatively low frequency, you might be able to 
implement a feedback loop around three stages.  The phase shift is the 
limiting factor.

  To answer the questions about the 10KW Hughes amp:
     That drawing is from more of a marketing than engineering document.  It 
does contain a few errors & some (obvious) omissions.  I added the ground 
symbol to the final autotuning sense ckt. myself before scanning.

  Some numbers:
     Grid bias:
      4CX250R:   -62V
      4CX1000D:  -340V  (adj for 500mA resting plate current)
    A resistive feed to the bottom of the grid tuning coils, as Karl-Arne 
noted, is valid to feed the bias.

    Signal levels:
      Input: 100mW / 2.25V RMS
      4CX250R grid:  22V RMS      (this illustrates how light the 6146 is 
being run!)
      4CX1000D grid:  210V RMS  (this illustrates how light the 4CX250R is 
being run!)

    Power Out:
       10KW avg.  14 KW peak
       better than -40db IMD @ peak power

   Negative feedback: 12dB

   A resistive bias feed to the bottom of the grid tuning coils, as 
Karl-Arne noted, is valid to feed the bias.


73 & Good morning,
  Marv WC6W

P.S. -- Steve T. posited in another post that there was little intrinsic 
difference between a vanilla 4CX350A and the 4CX350FJ.  The data sheet 
claims 10dB lower distortion at equal power.  That's a lot.  I wonder if 
Eimac uses a striped cathode aligned grid scheme, as in the 4CX600J & 
4CX1500B, in the '350FJ?







*


___________________________________________________________________
Try Juno Platinum for Free! Then, only $9.95/month!
Unlimited Internet Access with 250MB of Email Storage.
Visit http://www.juno.com/value to sign up today!