[Amps] Operational Input Impedance

[Jim Thomson]

Date: Fri, 22 Dec 2017 14:11:13 -0700
From: Steve London <n2icarrl at gmail.com>
To: amps at contesting.com
Subject: [Amps] Operational Input Impedance

<Several years ago, I modified my Alpha 76PA, replacing the 8874's with a 
<pair of 3CX800A7LG's, purchased from Alpha. These were new, Chinese made 
tubes. Other than that, I know very little about them.

I have never been happy with the input SWR - it's in the 3:1 to 4:1 
range, with about 25 watts of drive (and 500 watts output). The input 
network is the standard Alpha 76 2:1 unun. The 95 ohm swamping resistor 
pack has been removed from the circuit. Despite the high input SWR, the 
amp seems to work satisfactorily (when using the transceiver tuner to 
"fool" the transceiver into putting out adequate power).

Today, just to make sure the input network wasn't fowled up, I pulled 
the tubes and put 22 ohms and 50 pf across the cathode pin to ground, to 
simulate the published specs for a pair of Eimac 3CX800's. Using an 
antenna analyzer, the SWR was close to the expected value - 1.2 at 3.5 
MHz, rising to 1.8 at 21 MHz. Nothing like the 3 to 4 SWR observed 
during the operation of the amp with the Chinese 3CX800's.

So, this begs the question: What is the actual input impedance during 
operation ? How would one go about measuring it with something like an 
antenna analyzer or VNA ?

73,
Steve, N2IC


##  An UN-UN is not a tuned input.  You need a PI net to achieve any amount of fly wheel effect.
Henry radio tried the  1 : 1 un-un trick with their  8K ultra..which was a total disaster.  The 3CX-3000A7
tube they used already has a 50 ohm input Z.  SWR sucked, and henry said NOT to use the auto tuner
in the xcvr, go figure. 

##  GG triodes are only driven on aprx half cycles, like typ  210 deg conduction in class AB.   The Z
is sky high during the remaining 150 degs ! 

##  You cant put a VNA or anything else in there to measure the input Z either, it never works.   Tube has to 
be driven, and the more drive applied, the lower the input Z becomes.   That explains the difference between
your VNA and when actually driven.   

##  heres what I have found to work, if installing conventional PI net tuned inputs.   I use  GM3SEKs PI
and PI-L  spread sheets to calculate the C1, L, C2 values.    The total tube C  has to be subtracted  from the
calculated C2 value the software spits out.   The software was made for a kw PI net.  It can also be used for
tuned inputs.   More often than not, the C1  + C2 values ultimately used are a bit off from the spread sheet. 

##  another method I use is to install a small pair of air variables like arco compression trimmers, and padded
for the lower bands  + the coil value that the spread sheet spits out.   I measure the values of the caps +  coil
with a B+K  875-B   digital LCR meter.   Caps  tweaked for 1:1  swr, when amp driven to full output.  Ok, then
power down amp, leave tubes installed.    Put your VNA, or  MFJ-259B  on the INPUT of the amp, then  with TR
input relay activated, put a 0-100 pot  or 0-500 ohm pot  between  cathode and chassis.  Tweak pot for 1:1 swr. 
Then measure the dc resistance of the pot.   Say its  36.5 ohms.  Thats what  your tuned input is matching, that,
plus the total tube C from cathode to chassis. 

## Ok, at this point the PI tuned input works dead on... at least on one freq.  By varying the C1, L, C2 values,  you
are now tweaking the Q  of the tuned input.     You can get extremely close, by entering the desired Q on the
spreadsheet!   Beware, the Q on the spreadsheet is split into input Q....and output Q.     When the various handbooks
talk about an input Q of 2, its actually the input Q.   Output Q will typ be aprx 1.2     Total Q  will be 3.1     
3.1 to 3.5 is the typ Q value you feed into the software.     When the spreadsheet spits out the values for C1, L, C2,
it  will also  spit out the input Q...and  output Q.    The input and output Q  always total the desired total Q   you enter
very early one. 

## same deal with KW  PI net... or PI-L networks.    Old books said to use a Q of 10.  New books say to use a Q of 12. 
When the old books said to use a Q of 10, thats actually the input Q....and the output Q is typ 2.  Total Q = 12..which is
what the newer books use. 

##  Your  2:1   un-un will probably get you pretty close to the pair of tubes input Z,  but the  50 pf Tube C from cathode to
chassis is what will screw up your swr...esp on the upper HF bands.   The only way to tune that out is a coil between chassis
and cathode, to cancel out the 50 pf of tube C.     Problem then is,  that value of coil will be different for each band.   Which is
probably another reason the swamping  resistors were installed.   

##  You cant do this on a table top amp, but on my hb amps, I was not prepared to install 9 tuned inputs.    I instead used 
a pair of  64-2180 pf broadcast variables.... both padded on 160m only.  And a tapped  4uh coil.    4 uh used on 160m, 
2uh used on 80m, etc, etc.   6:1 jackson bros ball drives, which come in a 4 inch diameter, and also 3 inch and also 2 inch.
Skirts are calibrated 0-100  over the  180 deg arc.    Set tuned input bandswitch for desired band.   Then tweak the 2 x caps 
for flat swr.   Repeat for each band.  Write it all down.  After that, its just dial up by the numbers.   swr is always 1:1  across
each of the 9 HF bands.  

Sri for the diatribe.   Tuned inputs over the years have caused a lot of grief for folks. 

Jim   VE7RF