Hybrid Solid State Vacuum Tube Design [LONG!]

["Tesla list" <tesla@xxxxxxxxxx>]

Original poster: Shad Henderson <sundog@xxxxxxxxxxxx>

Hi All,

Please snip unneeded portions when replying!  Terry will Thank You!

With the surfacing of a thread on VTTC's, I felt it was time to spill
the beans on my little pet project.  Though I can't take credit for the
whole thing, I'm surprised that I haven't seen the idea posted here
earlier.  I dub this monstrosity the "SSVTTC", Solid State Vacuum Tube
Tesla Coil.

The idea is very simple, and the circuit is elegantly simple.

A tube needs negative voltage on the grid to shut it off. Most tubes we
use have the grids driven positive to turn the tube on "hard" and get
massive plate current.  The biggest obstacle to running a tube in pulsed
mode is having a sufficient negative bias to turn it off when it's not
switching, and having enough "drive" to pull the grid voltage positive
in the face of grid current.  The grid acts as a tiny plate when driven
positive, and conducts current from the grid supply to ground. (same as
the plate.)

I tried several schemes, and finally settled on this one for it's
simplicity.


I use a half bridge of IRF740's, and voltage doubled 120v mains to give
me roughly +/- 170v for a split rail supply.  The half bridge is a run
of the mill design, identical to an SSTC with a few minor changes.

Instead of a primary, I have a largeish (8.2k, 25W) resistor in place of
the primary as a load.  I also fitted a voltage divider between the
negative rail and the center point.  (bypasses the lower FET.).  The
grid drive signal is taken from the midpoint of the divider.

When the bridge isn't switching, the grid is biased to nearly the full
negative leg, shutting the tube off.  When the bridge is switching, the
grid alternates between the negative leg and the maximum positive
voltage as set by the divider.  This helps control the maximum positive
grid voltage, and therefore maximum grid current.

I use a 555 timer and a TL494 to run the half bridge, generally the 555
gives me pulses at 70-200pps, and an "on time" of the TL494 of
75-200uS.  The TL494 runs at the resonant frequency of the tank
circuit.  (this module is a drop-in replacement for a grid leak
network).   Later mods include primary current feedback for better
switching.  (there's a TC4420 and gate drive transformer hanging off the
TL494, just to clarify)

The half bridge "sees" mostly a resistive load because of the load
resistor.  The few pF in the tube's plate/grid C is amplified, but still
usually less than 500pF, so the resistive components dominate and you
get very clean, crisp switching to the grid.  The IRF740's can handle
magnitudes more power than the grid needs to be driven (an 833A requires
about 30W of drive, and 811A requires about 10-12W of drive), so there's
no shortage of power to drive the tube.  In fact, it's advisable to fuse
the link between the modulator and the grid, because it's surprisingly
easy to exceed the maximum grid dissipation switching from pulsed to CW
operation.  (bypass the 555 timer).

Going to push-pull operation is as easy as building a full H bridge.

I've yet to test this setup on a large coil producing sparks, but I have
tested it on an 811A with an 8.2k 50W resistor between the plate and B+
supply (1kv, 500mA).

The scope showed magnificent clean switching across the resistor, and
with the inherent jitter in the setup (built on breadboard and strung
across my computer bench), I couldn't really see any appreciable
propogation delay in the tube (signal at grid terminal to plate resistor
output).  Then again, even the 833A is content to work at full input
power up to 30mhz, so the tubes will switch just as fast as the silicon
can throw the signal at it.

I left the system running for about 1 hour at reduced plate input power
(tube dissipating about 45W), and at the end of the run the solid state
gear was barely warm to the touch.  The tube presents no real load to
the half bridge.  The tube anode resistor, though, was nearly glowing
;)  Lots of real power there, folks.


Controlling tubes beefier than the 811A should be within the realm of
this controller working off the voltage doubled mains, but for more
cranky tubes (304TL), that need a higher negative bias, either a 120:480
control transformer (must be center tapped!*), or voltage doubled 240v
would be a good solution. Just change the FETs in the half bridge and
re-adjust the voltage divider to give you the necessary positive peak
voltage.

I'm no EE, so I'm sure somebody will beat me to a self-tuning system
that tracks Fres of the primary (not secondary!) to switch the tube at
zero crossings.  Not like the tubes care one whit about it, but it'll
make the primary circuit run smoother.


Things I love about this setup -
  It isolates the delicate solid state from the nasty high RF powered
drive inherent in the tank circuit.  The only link between the two is
the tube's internal C, about 50-300pF coupled back through the
plate/grid.  That's easily managed by the FETs in my setup without any
special protection.
  Tubes withstand over voltage/over current abuse easily.  Tubes in
push-pull, during a cross conduction event, only tend to show a little
more plate color.  Solid state, on the other hand, does an impromptu
impersonation of a hand grenade.  right Terry? ;)
  Tubes are *fast*  I've cranked the TL494 up to it's screaming unstable
limits (about 820khz in my case), and whilst the FETs showed some
heating from poor switchin waveforms, the tubes faithfully replicated
everything.

  By decreasing the voltage of the positive going pulse, you can "ease"
the drive on the tube, for less power throughput in the VTTC, without
reducing tank voltage.  That should open a whole new door for exploring
sparks.  (VTTC's tank circuits float at the B+ supply level).


 At this particular moment, I have the full modulator set up on my
computer bench at home, but I'm in the process of building the "guts"
for the 811A based coil to make sparks with it on.  I needed to wind a
new secondary to lower Fres.  :/  my current 811A's secondary runs
around 500khz, and I can't get really solid switching at that freq with
the circuit built on breadboard (too much C between the parts)

 I'm also designing a much smaller setup running off a TV Sweep Tube
(6DQ6A) that's more portable and hopefully support much higher secondary
Fres without FET troubles.  (crossed fingers).


So in short, any run of the mill SSTC can be converted to run a vacuum
tube.  It provides all the goodies associated with SSTCs and DRSSTC's
(control over everything but quenching), and astoundingly high power
throughputs.  After all, 5-6kW of tube power is easy to obtain, and now,
easy to drive.  Being that tubes tend to be much, much tougher than
solid state, there is less danger of your expensive parts going "boom!",
so more freedom to experiment.   Add to this the ability to run your
VTTC like a spark gap coil (how I designed my circuit to work), and it
becomes extremely interesting.


A note to those that wish to try this...

The grid must be referenced to the cathode!  In most cases, it's no big
deal.  We ground one side of the filament transformer, and that grounds
the cathode.  The midpoint of the half bridge *must* be ground
referenced, or else "odd" things happen.  Using voltage doubled 120v
mains isn't the best way to do this, but the neutral line is "close
enough" in my case to ground.  If you use a control transformer into a
full wave rectifier, the center tap must be grounded to give you +/-
voltages relative to *ground*.  Yes, you could float the cathode and
split rail supply's "ground" to some other voltage relative to ground,
but there's no real reason to do that here.  :)

I hope to see lots of posts about this. :)

Questions? Comments?

Thanks!

Shad


--


Sundog G5-1373
www.thegeekgroup.org
Because the Geek Shall Inherit the Earth!