[Amps] Using 813 tubes on 17 meters and above...

[Jim Thomson]

Date: Sat, 2 Dec 2017 10:12:45 -0500
From: Peter Bertini <radioconnection at gmail.com>
To: amps at contesting.com
Subject: [Amps] Using 813 tubes on 17 meters and above...

<I'm getting serious about building an 813 amp as winter project and need
<some input from other OT's who've played with these tubes on 17 meters and
<above.

Plans are to go with three or four tubes in parallel.  Power transformer is
rated 1500va and I'll be near 2.8 kV plate voltage, in grounded grid.

Plate load for two is around 3500 ohms, so I'll be around 1750 or so with
four of the beasts in parallel... which is doable using a Pi L on 80 and
160 to keep the load capacitor size reasonable, and then going over to a Pi
Net for the upper bands.  I was more concerned with the combined plate
capacitance, which is going to be close to 60 pF  ???  That is value is
over the design value for a Pi Net with a reasonable Q at that load
impedance.  I am still very curious how others have made out using two or
three tubes in parallel on 17 meters and above. I've heard, anecdotally,
that tapping the plate tune cap down on the tank coil helps, but I've not
seen that in any published literature.  Any problems with using Amidon
T255A-2 core for a combined shorted-turn inductor for 80 and 160 meters on
the low Z side of a Pi-L,  and another core for the 160 meter inductor on
the plate side of the Pi L?

Any advice?  By the way, I am going this way because I have plenty of NOS
RCA tubes and sockets, and I can always change over to a more modern tube
in the future.

73

Pete

##  Are u sure there is  15 – 20 pf of stray C between the anode and the  grounded screen and
also grounded grid ??   You  are using a tetrode in a high- mu  triodr config, with both screen +
grid bonded to chassis. 

##  Download  GM3SEKs  excellent fee  PI and PI-L  software.  Then you can introduce a tiny bit
of uh anywhere between the anodes..and the main C1  tune cap.    Typ this tiny bit of uh, like  .5 uh,
is inserted between the  plate block cap and the main C1  tune cap.   As others  have noted, the tube
anode to grounded grid + screen C  and this extra coil, form a step down L network.   This transforms the 
plate load Z...down to a much lower value.... low enough that a practical PI or PI-L can be designed. 

##  To get 1.5 kw out,  you require 2.5 kw dc input...assuming 60% eff.   With 2.8 kv under load,  you require
893 ma of plate current,   That equates to a plate load Z of aprx  1844 ohms. 

## I would not mess  with a PI-L on 160 or 80m.   Use a plane PI on all bands, and just pad the load cap on 160M.   
If you use a Torroid, use the type 2, which are red.   T-225-2A...and you require  THREE of em.   Difference between
the A and the 2A suffix is... the  2A is an inch thick, the A is only .5 inch thick.   Im talking about when the torroid is
laying flat on the  desk, the thickness is from table  to  top of torroid.   With 3 of em, you end up with a mess that is 
2.25 inchs OD  x 3 inch long.... and thats before you start winding.    Its no light weight.  IMO, you may as well use
a piece of  airdux for the job, like 12 gauge, close spaced, for the 160 + 80m portion of a 160-10m PI net.  Then use
heavier gauge for 40m, and tubing for  20-10m.   

##  PI-L is a pita.  You end up with that extra L  coil that follows the C2 load cap.  It has to be shunted out when using the
higher bands.  If the L coil is a 160 and 80m affair, then  half of it has to be shunted out when using 80m.   Its one more
thing to have to switch, which complicates the bandswitching to no end.   It also puts the load cap at a higher peak V
point, since the load cap on a typ PI-L is 200-400 ohms....and not the 50 ohms you have on a plane PI net.   Using the
software you will see that the PI-L is not all its cracked out to be.  The load cap value will reduce quite a bit, but the
main coil between the tune and load cap will end up larger.   You will see at a glance.  The software is very easy to use.
It can also be used for designing the PI tuned input.   The only thing u cant do with the software is to try and match two 
identical impedances, like  50 ohms to 50 ohms, say on a tuned input. The  best it will  do is 50-49 ohms...or  49-50 ohms.

##  when playing with the GM3SEK soft ware,  you will fast find what works, and what doesnt.   But you really need a 
digital lcr  meter to build any hb amp.   Without it, you are  dead in the water.  You need to be able to dial in the C1, L,
C2 cap on any tuned input..and ditto with the exact values for any kw rated   PI or  PI-L network components. 

##  U also need the lcr meter to actually measure the stray C  from anode to the chassis.  This will tell you the exact value
of anode to  grounded grid + screen.   Do that measurement with all four tubes installed in their sockets..and all grids
and screen well bonded directly to the chassis.   Ok, now that you know the actual C  with 4 tubes plugged in, then
you can input that value into the software.   Along with the other parameters, like loaded Q, freq, plate load Z, etc. 

## You will see that on the 12 and 10m bands, it all goes to hell, since your typ min C1  tune cap value is still  too big for
10M.    But by increasing the loaded Q,  it will work, but a sky high loaded Q  on 10M is not what we want.  High loaded
Q just means narrow banded, has to be re-tuned every time u qsy, and high circulating RF current through the coil, and
also the bandswitch, or roller contact, if a roller coil used.   By inserting somewhere between  .2 and  .7 uh, between the
plate block caps and main C1  tune cap, the  Q can be reduced  a bunch.  That drop in plate load  Z will be displayed
in software, but u have to play with several vales of loaded Q..and also several values of that small coil b4 main PI net. 
This trick has been around for decades in the commercial world.   One caveat, the tiny coil used has to be the same size tubing
as what you are using for the main 10M tank coil, it has a lot of current flowing through it.  You will also see that the small
coils effect diminishes  rapidly when below 10m band.   You will still see a very small effect on 12m, but it will vanish  on 21
mhz.  The software will depict the transfomed Z on each band.   On 10m, where the issue is, your 1844 ohms  will be transformed
down to something like  500-1000 ohms, that a practical  PI can be designed around.  

##  I have done the extra coil by winding a coil, and also just tapping into the main tank coil..with the hot side of the tube cap,
both methods work.   But u need to know the exact value of coil u are inserting b4 the tune cap...and the exact value of coil, 
between the tune and load cap.    Once u are happy in software, build it, it will work every time.    To dry test it, wire a temp
resistor, equal to the original calculated plate load Z of the tube, say  1844 ohms, between anode and chassis.  Then with mfj-259
on output of amp,  tweak the tune and load caps  for flat swr on the MFJ.   If that test works, it will work, with  temp resistor
removed,  B+ and fil V applied..and drive applied.     

http://www.ifwtech.co.uk/g3sek/


Later... Jim   VE7RF