Some work with Toobs

["Tesla list" <tesla-at-pupman-dot-com>]

Original poster: "sundog by way of Terry Fritz <teslalist-at-qwest-dot-net>" <sundog-at-timeship-dot-net>

Hi All,

   I've been working with my vacuum tubes as time allows.  I got my 
horizontal bipolar VTTC running nicely, then promptly dismantled half of it 
to work with some experiments.  (ain't progress wonderful?)

   One of my experiments is to run the tubes in grounded grid 
configuration, and use a mosfet for cathode switching.  Similar to the 
staccato mode, but better in some ways.

   For most tubes, this requires either a full wave rectifier on the 
filament (and usually drops the filament voltage below it's needed level), 
or a center tapped filament transformer.  I'm working on finding a way to 
hook up a bridging transformer to provide a centertap without 
re-winding.  6.3V center taped transformers are hard to come by, and using 
a variac to run a 12v CT tranny at 50V input isn't practical for most people.

    Anywho!  Back to the meat & potatoes.

  Starting with a 6SN7 tube (30W dual triode), I put an IRF510 between the 
cathode and ground.  The 6SN7 has an indirectly heated cathode, eliminating 
the need to centertap the filament transformer.  Using my signal generator 
on the FET, and a 8kohm, 25W dummy load between the plate and plate 
transformer, I took measurements with my O scope.  It worked great.  The 
FET does see a high-ish voltage because it's capacitively coupled to the 
plate, especially when the cathode is open-circuit to ground.  But, the 
passed current is for all practical purposes, the same as the plate 
current.  Adding a tuned tank circuit, and a secondary, i was able to put 
filtered DC on the plate and feed the FET a chain of square waves at 
~50-80hz.  I got output (barely, it was a dim corona) running the tube in 
such a mode.

   So, I tried it on an RS1026 triode I have.  I put the 8kohm, 25W dummy 
load in series with the primary and removed the tank cap.  I grounded the 
grid and put the IRF510 between the filament tranny CT and ground.  Worked 
good.  Plate voltage dips while the FET is on, and pops right back up when 
it's turned off.  Given that the fet may see highish voltages, but low 
current, there is not so much danger of it burning up whilst switching off 
under a load.  And, unlike the triac used in staccato mode, you aren't 
waiting on a zero crossing to let it turn itself off.

    I still have to refine the drive circuit some.  I don't think a gate 
transformer would be that bad of an idea, to isolate the FET from the 
driver.  Some zero-crossing detection circuitry would be nice also, to keep 
the tube switching efficiently.  I have yet to do any high powered tests 
that generate sparks, but those will be coming along.

    Now, something interesting.  Usually grounded grid mode is used in 
class AB or B amplifiers.  There's always some current flowing in the 
plate.  This idle current is controlled by a resistor in series with the 
cathode connection to ground, usually 4k-8k.  A pair of RF chokes are in 
the filament leads at the tube, and a signal is fed to one side of the 
filament via a capacitor.  The cathode must have a DC ground, but it can 
float at an RF voltage within reason.  This isn't a good way to operate a 
tube for a VTTC, or audio modulation, because of the high drive required 
for it.  Though a multi-stage amp (12AX7 driving a 6SN7 driving an 811A 
driving an 833A...)  would do it.  But, that'll give you a fair amount of 
steel to lug around, and the need for negative feedback in the amplifier 
circuits, etc.   doable, but a headache all the same.

   I'm thinking of using a high value resistor to maintain a negligible 
amount of plate current all the time, also to help provide a ground 
reference for the cathode.  May not do much, but maybe it'll help curb the 
voltage rise on the cathode when disconnected from ground.

    Now, I have yet to try this on a "popular" tube.  The 6SN7 is a small 
tube used in RF oscillators, etc.  Only 30W.  The RS1026 is not designed 
for zero bias (grounded grid) useage, and will attempt to pull over an amp 
of plate current at 4kV (!!!).  An 833, 811, or 304TL may not behave this 
way.  They *should* kick over to full available plate current when the 
cathode is tied directly to ground, with only the current drop limited by 
the resistance of the FET (low enough not to matter).

   A curious note here... Due to switching the FET through it's non-linear 
region, it dissipates more heat than zero crossing switching.  But, even on 
the RS1026's, moving only ~1A of plate current, they should have no problem 
with excessive heating.  The voltage on the plate clearly shows the FET 
switching off under a load, because you can see the curve as the plate 
voltage rises (FET resistance increases), until the mosfet is all the way off.

   If I can get this working properly, it'll do away with grid leak 
networks. (my least favorite pain in the keister).

  Some other thoughts I want to try out...

    Use a low-turn primary and largeish tank cap (you've basically got a 
4-6kv, 300mA-1A transformer available, dependant on tubes), and wind an 
inductor to provide some plate resistance for the tube. Shunt that with a 
resistor to prevent parasitic oscillations, and switch the tube on and off 
at 120hz to 4khz with a filtered DC plate supply.  I want to see how well 
it mimics a disruptive coil (with crappy quenching).

  Another, is to use an IGBT in series with the plate supply.  you change 
up a good sized pulse capacitor, then shut the FET off, isolating the plate 
supply.  You then use two tubes in push-pull configuration, and fire them 
in opposites, draining the energy out of the pulse cap through a tightly 
coupled, low turn count primary.  This is using the tubes in a pseudo pulse 
duty.  Plate currents will soar, so it'll be necessary to keep the 
switching at Fres, and limit how long the tubes are firing for each "bang" 
(ie, IGBT controlled pulse cap charge-discharge cycle).  Graphite plates 
are a must for this (or high quality tubes that utilize a copper anode that 
*IS* the plate, such as the GU10A and similar), because of the high heat 
inertia and that they don't go gassy when you really abuse the 
plate.   This type of duty is literally murder on the tubes.  If you trash 
your tubes (burn up the plates, overstress and burn out the filament), 
please don't come screaming at me.  This mode of operation can be as 
dangerous as running a resonant cap on an NST.  But, a tube switching 4-5kV 
at 3-4A for 50-100uS or so at 300khz or so *does* sound like a hoot to me! :))

   Time will see.  Unfortunately, I don't have as much time to tinker with 
this as I'd like to have.

Comments, questions, etc welcome!

-----------------------------------------------------------------------
Shad (Sundog)
G-5 #1373
"Ever stop to think, and forget to start again?"
-----------------------------------------------------------------------