Re: VTTC controller - First Light

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

Original poster: "K. C. Herrick by way of Terry Fritz <twftesla-at-qwest-dot-net>" <kchdlh-at-juno-dot-com>

Shad (& all)-

I too look fondly on tubes although it's been upwards of 50 years since
I've designed anything with them.  But a comment or two:

1.  Consider using an untuned primary and then deriving the FET/tube
excitation from the secondary's return current, as I do with my s.s.
coil.  That makes the coil self-tuning.  You feed the current through a
capacitor whose reactance is small at Fr, to ground.  Take the small
capacitor voltage, now shifted 90 degrees w/ respect to the current, and
couple it to a pair of back-to-back diodes to ground via 1K or so.  Then
amplify the diode voltage & drive the FET(s) with it.  Voila!...a
feedback oscillator with the secondary's Fr as the sole
frequency-determining element.  You need enough gain in the amplifier so
that noise will start the oscillations going.  Or else, configure it as a
weak internally-fed-back oscillator that will become swamped with the
secondary-signal as soon as you turn it on.

2.  I'd characterize the FETs more as just switches rather than as part
of a divider.  Just semantics...

3.  Sounds as if your FETs are turning >off< quite rapidly since the
excursion to -300 takes only ~200 ns.  They don't turn on fast enough.  I
suppose you have checked the gate waveforms themselves?  You could use
N-channel FETs, of course, by merely 1:1-transformer-coupling their
drives.

4.  If it's the case instead that it's the return to -300 that's slow,
then they're, of course, turning >on< slowly.  In any case in a push-pull
situation, I'd think slow-fast, or fast-slow, would be good since that
would afford a dead-zone between 1 tube turning on & the other off & vice
versa.  But during that dead zone, the primary is going to ring something
fierce--if untuned & unless clamped.

5.  I should think you could use a totem-pole configuration of 500 V
FETs, driving the grid & connected between 0 and -300 V, to avoid the use
of the series resistor.

6.  You mention using 2 MOTs in series:  I hope they're insulated for it.

7.  Fundamentally, you'll have the same problem as I do with my s.s.
coil:  Not enough power contained  in the first few cycles of excitation
to generate the longest sparks.

8.  I don't see an advantage to tuning the primary; you'll not get any
more through-put of power thereby.  What's the point, I wonder?

Ken Herrick


On Mon, 11 Nov 2002 12:51:35 -0700 "Tesla list" <tesla-at-pupman-dot-com>
writes:
 > Original poster: "sundog by way of Terry Fritz <twftesla-at-qwest-dot-net>"
 > <sundog-at-timeship-dot-net>
 >
 > Hi All,
 >
 >     Aside from some glitches to iron out of the design and
 > implementation,
 > it works, and surprisingly well.
 >
 >      After testing the controller extensively, I decided to fire up
 > the
 > tube coil and see what happened.  The worst was I could smoke $24 in
 > FETs,
 > and some cheaper chips in the controller.
 >
 >    Quite to the contrary, I got a nice purple fuzz on the breakout
 > point.
 >
 >    The setup -
 >      3" PVC secondary, wound with 26ga wire, Fres is ~390khz (best
 > output)
 >      Primary is 2 layer, helical, with 14 turns per layer and
 > centertapped
 > for push-pull.
 >      Power supply is 2 MOTs in series feeding raw AC to the plate.
 >      The controller is a TL494 based PWM, with a pair of TC4420 FET
 > driver
 > powering 2 -500v -7A FETS
 >      The grid supply is a 120/480v CT transformer with diodes on the
 > output
 > to give me -340v (with a filter cap) for each FET.  With the
 > dropping
 > resistor in plate (to limit current to the grid), I get about -300v
 > on the
 > grids.
 >
 >      I'll post waveforms and whatnot on my site as I get time.
 >
 >     The skinny on this system...  The P channel FETs act as part of
 > a
 > resistive divider, feeding the reference voltage to the grids.  When
 > the
 > FETs turn on, the voltage on the reference (and grid) drops to zero
 > volts,
 > turning the tube on.  The fet turns off, and the voltage comes back
 > to
 > -300v.  At -300v, the tube is *hard* off, and I have to crank the
 > variac to
 > 140v to even begin to get any power out of the tube.  At 4kV on the
 > plate,
 > there is no current flow to the tube.
 >
 >     Switching -  The voltage drops from -300v to 0v in about 1.6uS
 > (not a
 > typo), then climbs back up to -300v in about 150-200nS (sometimes as
 > long
 > as 300nS for low frequencies).  I attribute this to poor "off" gate
 > switching of the FET, as the drive signal is very sharp.  Still, the
 >
 > opposite tube turns on while the first has around -170 to -200v on
 > the
 > grid, which has the tube "mostly shut off". Still, it's burning up
 > some
 > power there, and the matter needs addressed.
 >    Surprisingly, the FETs don't see that much kickback with a
 > standard RL
 > choke at the plate of the tube.  The kickback never surpassed the
 > grid bias
 > voltage, so the FETs were in no danger.  Even running the system
 > *way* out
 > of tune, there was no damage to the fets or tubes in any way.
 >     An aluminum enclosure to shield against RF for the drive
 > electronics
 > will be forthcoming, as will better adjustment of the TL494's
 > operating
 > frequency.
 >
 >
 >     Now, the issues I ran into -
 >       Poor FET turn off.  I think a lowish-value resistor to ground
 > from
 > the gate should help dump the capacitance once the FET driver turns
 > off,
 > giving me a much sharper return to -300v on the grid, and reduced
 > FET
 > heating (which isn't even an issue at the moment).
 >      The tubes refused to pull more than 500w, even with 0v on the
 > grid.  I
 > will substitute a pair of 833A's for the RS1026's to compare, as the
 >
 > RS1026's should be pulling ~4kW on their own. For now, I think the
 > tubes
 > are fine, and my filament tranny's ground and MOT's ground
 > connections are
 > suspect.
 >      Power resistor heating - The dropping resistors in the divider
 > get
 > *hot*.  Especially under the CW duty they were under last night.
 > Resistors
 > are current 3kohm 25w aluminum sinked, mounted to a 4x6" heatsink.
 >      RF kickback into the FET causing spurious triggering -  There
 > was a
 > *nasty* ~8khz screeching from the spark at times, which I narrowed
 > down to
 > spurious triggering of the FETs.  RF hardening should help eliminate
 > that
 > nicely, as will a better FET output waveform.
 >
 >
 >     Once I get the tube power situation under control, I'll be ready
 > to run
 > real full-power (multi kW) tests on the coil.   It's *incredibly*
 > nice to
 > be able to run a coil out-of-tune, and not fry anything.  Being
 > out-of-tune
 > will matter worse at higher current levels, as the tubes will have
 > to
 > dissipate the energy, but they're tough enough to take the overload
 > (within
 > reason) without cooking, as FETs do.
 >
 > ToDo
 >     Try 833A tubes in place of the RS1026's.
 >     Get sharper turn-off on the FETs
 >     Add a 555 timer to enable staccato/pulsed mode to the TL494
 >     Re-configure the tank circuit to utilize a tank capacitor and
 > run the
 > controller as a grid-leak replacement (single-ended)
 >     Finalize the electrical design and get PCB's etched for easy
 > construction/population.
 >
 > All in all, I'm not surprised that it works, but I am surprised that
 > it
 > works as well as it does, with so few problems.  I'm in the process
 > of
 > drawing up schematics for the whole thing (TL494, grid-switching
 > circuit,
 > etc), which I'll post as I complete them.
 >
 >    Smile everybody, there was a new flavor of TC born in my garage
 > last
 > night. :)
 >
 > Shad
 >
 >
 >
 > -----------------------------------------------------------------------
 > Shad (Sundog)
 > G-5 #1373
 > "Ever stop to think, and forget to start again?"
 > -----------------------------------------------------------------------
 >
 >
 >
 >
 >