[Amps] Glow, gettering, and voltage

[Jim Thomson]

Date: Sun, 08 Jan 2017 14:51:18 +0000
From: Manfred Mornhinweg <manfred at ludens.cl>
To: amps at contesting.com
Subject: Re: [Amps] Glow, gettering, and voltage

But I think that you are wrong in a few points here, Jim, although most 
is correct:

> IF  you run the same po in both settings, the plate current will be a
> lot less in the higher B+  setting. Your plate load Z will have 
> skyrocketed.

Only if Cathy re-tuned the tank circuit for optimal efficiency at the
higher voltage and the same power. If instead she kept the same settings
for the tank circuit, then the plate load Z will remain the same, the
RF voltage on the plate will also remain the same, but being centered on
the higher supply voltage the tubes work farther from saturation,
and thus more linear but at less efficiency.

&&&& Lemme  re-phrase that.   tune and load caps re-tweaked for
max PO .    Say tuned for 600w out in both the cw and also ssb position.
Loaded Q  will double if u try and run 600w  out when in the ssb position..
assuming tune + load caps  tweaked for max po in both cases..and drive
tweaked if required.  

The actual load Z on the plate depends on the antenna impedance and the 
PI tank values that have been set. Not on the supply voltage and desired 
power! But of course the supply voltage and desired power define what 
the best load Z would be.

&&&  Plate load Z =  loaded B+   /  1.7 X  loaded plate current. 
Assume a 50 ohm load.   The maths come out correct. 


And there is one mistake in this:

> With higher plate load Z, the tank coil has to have twice the uh. The
> tune + load caps have to be reduced to one half their values.  Since
> that is not done, your  resulting tank circuit Q  will have
> doubled..... like from 12  to 24.   Sky high loaded Q = lousy
> efficiency.   Lousy eff = higher plate dissipation.

The mistake is that a very high loaded Q will cause poor efficiency due 
to increased loss in the tank coil, and to a lesser extend perhaps the 
capacitors and switch, but not in the tubes! Except for the fact that if 
this is taken to an extreme, the tubes needs to deliver more output 
power to supply the tank's losses in addition to the desired output 
power, and this may, or may not, incur in higher plate dissipation.

&&&&  On a 4-1000 amp years ago, with a roller coil, I tried
increasing the uh a bit at a time..and each time re-tuning the tune
+ load caps for max po... with the same 130 watts of drive.   Of course
the tune + load cap values are reduced a bit each time this is done. 
The PO on the wattmeter kept on increasing a bit each time..until
it levels off.   All  as I was doing was reducing the loaded tank Q 
in small increments.   If you double the tank Q, you also double the
circulating current through the roller coil, and both caps.   The return for
the entire mess is through the C2 cap..closest to the cathode...on the 
PI tuned input.   The C2 cap on the tuned input has to handle a fair whack
of total current...like applied drive  +  return current. 

&&&  I installed a RF ammeter in series with the PI net coil on one of
my  HB tuned inputs.   When Q is increased, sure enough, the current through the
coil also increased.   With  200 watts applied,  RF current on the ammeter was aprx 4-5A. 
This was done on 160 + also 80m.   HB  tuned input was a 
simple PI, consisting of 17 turns of 6 ga solid cu wire, wound on a 1.5 inch ID.  Both
caps  were just a pair of broadcast variables.  Both caps padded only on  160m.   Coil was
aprx 4 uh on 160m, then tapped for each band.  Each cap was a 4 x section, with 17-540
pf per section, and all 4 x sections strapped in parallel.  

&&  for interest, I ran 58 u from C2 cap of tuned input  to wattmeter....then the dummy load.
wattmeter on input side as well.   With 200w applied, and caps tweaked for flat swr,  saw 195
watts out...except on 15m... where it was only 160 watts out.  The fix was to increase the uh, by
increasing the tap by just .25 of a turn.   Then re-tweak both caps.. which were then slightly less. 
Output side  rose from 160w  to 195.    And that’s using 6 ga solid cu and broadcast caps, not
puny 20 ga magnet wire on a tiny torroid / tuned slug, and broadcast caps, not silver micas. 

&&&  Moral of the story is... with a simple PI net on the KW output side... dont use  too  high a
loaded Q.   8-10 is plenty.   Ditto with the input PI net, typ  2-3 is plenty.   An input Q of say 5, like
you saw in old eimac notes is clearly way too much.    How many folks have actually measured the
PO of a simple tuned input ?   Eye opener for sure.   here u think the tube is hard to drive, when actually
u are losing some power on the tuned input assy.   If a bandswitch is used on the KW output Pi net, the
increased RF current through it, when using high loaded Q  can easily cook and overheat it, depending
on band used, typ the upper bands will suffer from this effect.   A lot of bandswitchs are marginal to begin
with on the upper bands. 

&&&  Those CCS current ratings you see for the typ model 86-88 style switches are  DC /  60 HZ..not RF ! 
They have to be de-rated for RF use...esp on 20-10m upper bands. 

&&&  Similar problems occur when re-using the 10m position on 12m...and ditto with using  15m on 18M band.
Q goes way up. 

Jim   VE7RF. 
   


Manfred