[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: New to the List and pulsed DC-VTTC (fwd)



---------- Forwarded message ----------
Date: Fri, 05 Oct 2007 09:39:33 -0400
From: Shad Henderson <shenderson@xxxxxxxxxxxxxxxx>
To: Tesla list <tesla@xxxxxxxxxx>
Subject: Re: New to the List and pulsed DC-VTTC (fwd)

Hi Martin, 

You're tinkering in an area that I've been working on-and-off in for
quite awhile. :)  Welcome aboard!

I've made some excellent progress, and hopefully solved one of the big
technical snags that's driven me crazy.  

My test-bed setup is in bits and pieces right now (busy rebuilding an
engine for my VW bus), but here's the general idea of it. 

A MAX913 comparator senses the primary current waveform via a few turns
of wire around the primary (like a grid coil, only 3-4 turns).  It feeds
that signal to a TC4420/TC4429 complementary pair of mosfet drivers,
which drive a half bridge of IRF740's.  

A 555 timer sets the pulse frequency (60-600 pulses per second) and the
pulse length (75-300uS) by toggling the "inhibit" pin of the MAX913.  

The mosfet half bridge is fed by a set of +/- rails.  For the way I run
it, it's *very* important that the DC supply have it's midpoint grounded
(center tapped supply, full wave rectifier in my case).  That give me +
and - voltage with respect to ground.  

The filament has to be floated by the *positive* bias supply (usually
just past cutoff for that tube), *not* grounded!  The bias supply must
be referenced to ground.  This places the filament at some positive
voltage with respect to the grid.  Then the grid is directly-coupled
(with a series resistor to limit grid current) to the half bridge
output.  No coupling capacitor. (that screws up operation if you drive
the grid positive)

What happens is the grid sits at 0v when the circuit is idle, and the
cathode sits at some positive bias voltage.  This works well, because
the tube only cares about the voltage difference between the elements.
With the cathode at +170v, and the grid at zero volts, the tube sees
this the same as if the grid were at -170v and the cathode at 0v.  The
tube only cares that the grid potential is less than the cathode's
potential to limit the current flow. 

When the circuit is switching, the grid sees a square wave voltage equal
to very nearly the positive and negative rails.  In the case of my
testbed, that's +/- 50v.  The cathode is biased at +50v, so the tube
"sees" the grid swing 100v total.  The *voltage potential* between the
cathode and grid swings between 0v and -100v.  It doesn't matter that
the grid is actually seeing +/- 50v, because the cathode is already at
+50v.  During the tube's "on" time, the grid is at +50v, and during it's
"off" time it's at -50v.  It's only the difference in potential between
the two that matters.    

I really like the above setup, because it eliminates the need for a
transformer to isolate the half bridge from the grid.  It also is
self-tuning, so the switching of the tube follows the frequency of the
primary.   Power throughput is varied by controlling the pulse rate,
pulse duration, voltage on the grid, and the bias voltage.  Because the
duty cycle is so low, it's possible to drive the grid *hard* positive
(tube almost in full saturation) without exceeding either the grid or
plate dissipation ratings.  There is some "hard" switching involved when
the 555 timer interrupts the MAX913, but the tube doesn't care about
that.  Also, the output of the MAX913 *will* look very chaotic on a
scope.  Took me awhile to realize that I was latching it sometimes high,
sometimes low, while the 555 was running.  That doesn't matter at all,
because the TC4420/TC4429 driver chips only care about transitions from
one logic level to another.  Wether the signal stays high or low during
the interrupted periods, one or the other chips is turned on, but not
changing state, the mosfets see no drive signal, so no harm no foul.  

Hope it helps spark an idea for ya!  

Shad H. 




On Thu, 2007-10-04 at 07:59 -0600, Tesla list wrote:
> ---------- Forwarded message ----------
> Date: Thu, 04 Oct 2007 14:14:55 +0200
> From: Martin Damev <mdamev@xxxxxxxxxx>
> To: tesla@xxxxxxxxxx
> Subject: New to the List and pulsed DC-VTTC
> 
> Hello together,
> 
> My name is Martin Damev. I'm an 32-year old
> coiler from Zurich, Switzerland. I've been
> into coiling for more than ten years but my
> recent interests are  specifically the  VTTCs.
> Having a nice collection of transmitting tubes
> from 1 watt to 100 kW RF output, and some good
> knowledge about RF transmitters and industrial
> RF heating it was obvious to build a nice tube coil.
> Here is some data of my VTTC:
> 
> oscillator:  Hartley, parallel fed (plate RF choke: 6.8 mH,
>                    3 nF /20 KVDC coupling cap. from plate to tank)
> tube        :  BBC (now ABB) T 380-1 glass transmitt. triode
>                    with 5V / 15 Amps filament and 380 watts plate
>                    dissipation.  Very  sturdy graphite anode.
>                    Tube often used in RF generators for plastic-
>                    welding.
> primary :   20 turns of  #18 AWG special high-temp.
>                   stranded copper wire on a 8 inch plexiglas
>                   former. Taps at 12.5 and 16.7 percent for
>                   feedback adjustment (connected to ground in
>                   a Hartley osc. circuit).
>                  700-1100 pF of primary cap. Adjustable with a
>                  vacuum capacitor.
> secondary: 3.0 inches diameter. Winding length: 16.5 inches
>                    with #26 AWG magnet wire. 10.5 mH inductance.
>                    No topload.
> 
> Tests with 3000 volts filtered DC gave hot and intense
> "brush" discharges, actually plasma flames of approx.
> 3.5-4 inches length. Measured  DC Input Power was
> 1200-1300 watts.
> These tests proved the excellent efficiency of the
> Hartley design. But I wanted long sparks and not
> a plasma burner...
> I have known  John Freau's excellent experimental work
> with pulsed VTTC for quite some time.
> He reported very good results with the now widely used
> AC-fed staccato-controlled VTTC.  Attempts to pulse a
> DC-fed VTTC were not so promising, except for the tetrode
> version, which proved to have an excellent efficiency.
> Could a similar efficiency be achieved with a triode instead of
> a tetrode ?
> I really wanted to run my VTTC as grid-pulsed oscillator with
> high plate voltage. Grid-pulsing an oscillator is often used to
> regulate the power of large RF generators for industrial
> RF heating and welding. Luckily I got a complete grid-pulsing
> unit from an 12 kW-generator. Driven by a signal generator
> it produces an rectangular output voltage  of  -850 volts (adjustable)
> with any duty cycle and frequency.
> 
> I connected the the grid-pulsing device into the the grid circuit, and
> built a DC supply from 0...6200 volts DC with a 5 microfarad/6.3 kV
> filter cap.
> The results were impressive:
> 
> Spark length with filtered DC (5660 V) was at least as high as with
> corresponding AC (4000 volts), the sparks being approx. 16-18 inches).
> Power consumption was 20..30 percent  compared to VTTC operation
> with 4000 VAC(no staccato).
> 
> Experimenting with the VTTC in this mode (grid-pulsed, filtered DC)
> is real fun. You can easily change the spark's appearance by changing
> the pulse repetition frequency and/or the duty cycle.
> I found that in my setup the spark needs approx. 1.5 - 2 milliseconds
> to grow to full length. Shorter "on"-times of the oscillator give thin
> and weak sparks, longer "on"-times give bright,fat and hot sparks and
> of course higher power consumption.
> Best efficiency was achieved with low rep. rates, not more than 20...25
> pulses/sec.). With 16 pulses/sec. DC input was not more than 200 watts
> for 16-18 inch sparks.
> 
> A lot of measurments will follow. I feel that there is some potential to
> pulsed DC-fed VTTCs. I will post some more my ideas and theories to
> this mode of operation  soon.
> 
> Martin
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
>