[Amps] NFB vs IMD question

[Larry Benko]

Jim,

I believe you are measuring the IMD of the K3 or at least seeing results 
that are influenced by the K3.   I have made IMD measurements of the K3 
and they are not that great and were the best at the 30-40W level.  I 
built a simple fixture for IMD testing that used 2 separate transmitters 
and had IMD3 of better than 60dB.  See 
http://www.w0qe.com/Technical_Topics/imd_testing_of_amplifiers.html

NXP has an app note describing how much better the source IMD needs to 
be in order to measure the output IMD to a certain level of 
uncertainty.  Do a search for "NXP RF transmitting transistor and power 
amplifier fundamentals" and it is table 3.1.  This paper is better than 
any other doc I have seen for transistor high power amps.  It discusses 
many other transmitter topics such as bias level etc. vs IMD.  This 
paper does NOT have the usual NXP ECOxxxxx prefix and has a 1998 date.  
Other testing I have done shows that to have really good IMD you need to 
significantly under drive the amp.  Some 300W Freescale LDMOS parts were 
great at the 100W level and OK at 150W.  Beyond 150W the IMD began a 
constant rise.  Unfortunately backing down the output level kills the 
efficiency.

I thought the same as you about increasing negative feedback. However 
there is just not enough excess gain to make much of a difference if the 
goal is increasing IMD3 from 30dB to say 40dB. Simulations in LTSpice 
unfortunately confirmed the measurements.  A big downside of negative 
feedback is the reduction in input impedance.  To really drive the IMD 
lower you would need another gain stage within the negative feedback 
loop.  Source degeneration is a good idea and it increases input 
impedance while reducing gain.

These are my opinions and are at least backed up by measurements and 
simulations but I am not a high power amp expert.

73,
Larry, W0QE

On 8/7/2013 7:46 AM, Jim Thomson wrote:
> A buddy is building the 250w SS  HF amp you see in the last few arrl books.  Its almost finished.  It only requires aprx 7-10 watts of drive to the pair of VRF-151G finals.
> He also ran a series of tests on his K3....and found that min IMD  occurs with 30 w pep out.    It increases  above or below 30 w pep.   Its very clean
> at the 30 watt level.   The original plan was to use a 5-6db pad between the K3 and the hb 250w SS amp.  Upon close examination of the 250w ss amp,
> it occurred to me that if the NFB was increased by 5-6 db, the 5-6db pad between k3 and SS amp could be eliminated.
> I think the base resistor values on the SS amp can be increased to do this.   Resulting IMD from the 250w SS amp...on paper, should drop a bunch.
>
> The 250w SS amp has pretty good IMD as is, at 250w pep out...and a lot better at 225w..and better still  at 200w.  The bias on the SS amp could also be
> tweaked a bit to improve it some more.
>
> OK, for every 1 db increase in NFB on the SS amp,  how many db will the IMD3-5-7-9-11-13 drop ??     I have no info on this anywhere.
> Is it going to be a 1 to 1 deal, or does the IMD drop several db for  each 1 db increase in NFB ?   What about a tube amp,  does it follow
> the same pattern as a SS amp  re:  NFB  vs IMD.  ?
>
> On paper it appears this scheme should work.    What we are after is a squeaky clean  180-250 watts pep out.   The K3  only puts out
> 100w pep.....and the IMD at 100w pep out is nothing to brag about....its lousy.
>
> The thought here was we could kill about 3 birds at once.   Increasing idle current on the K3 final stage plus its mating driver is not appealing
> due to heat issues, plus minimal IMD  improvement.   Ditto with the 250 W  SS amp.   Now whether the 250w SS amp  will handle the 30w of drive,
> and no 5-6 db pad....and just relying on a 5-6 db increase in NFB is another issue.   Any thoughts would be much appreciated.
>
> later........ Jim   VE7RF
> _______________________________________________
> Amps mailing list
> Amps at contesting.com
> http://lists.contesting.com/mailman/listinfo/amps
>