[Prev][Next][Index][Thread]
Vacuum Tube Tesla Coil
From: David Sharpe[SMTP:sccr4us-at-erols-dot-com]
Reply To: sccr4us-at-erols-dot-com
Sent: Sunday, September 14, 1997 10:34 PM
To: Chip Atkinson
Subject: Vacuum Tube Tesla Coil
Chip, ALL
I wanted to post some preliminary information on a Vacuum Tube Tesla
Coil (VTTC) that I have been dinking with for the past 6 months. The
unit is novel in that it is a true push-pull system based on the
Hartley Oscillator. It is also novel in that the power supply allows
for powering the tubes with AC (as normal) as well as "level shifted"
mode which is used in most microwave ovens sold.
Specifications
Circuit Push-pull Hartley Oscillator
Transformer MOT, out of 700W oven, no shunts removed
Tubes 811A or 572B / T160L (prefered)
Fil Power 6.3VAC, 4A x 2
Peak AC voltage 2.4kV
Power Control Variac, 0-140VAC, -at-10A
Fusing 150VAC, 20A fast blow
Primary 10.5" OD(10" sch40 PVC),30 Turns #10 THHN wire
tapped at CT and 4 turns either side of CT
L ~ 244uH
Tank Capacitor 5500pF+2200pF series Fabmicas
total ~ 1470pF (measured)
Resonator 6.625 OD (6" sch40 PVC),18.25" long, close
wound #24, 18" long
L ~ 30mH, Fo ~ 270kHz
Grid leak bias C (2) 1200pF, 10kVAC barium titanate
(2500pF) to grid of each tube from grid taps
Grid leak bias R 1 5kohm, 50W rheostat / tube, + 2.5Kohm, 25W
Bal Pot to Ground
R ~3.0 - 3.5 kohm from grid to ground
Spark Length Observed 10"+ AC powered (1000W Input)
13"+ "Level shifted" powered (1400W Input)
Meters Primary VAC in
Total AC current in
Peak kVDC to tubes
Grid current (each tube)*
Plate current*
*DC Average responding
External measurement taps
Plate current (through 10ohm non-L R)
Grid bias tap (between GLB R, C and tube
grid current metering circuits)
The push-pull circuits have typically been avoided by coiler's due
to the perception of much increased circuit complexity. However,
push-pull offers some very compelling advantages:
1. Double the voltage swing across the tank circuit versus
single-ended designs, resulting in higher coupled voltages into
resonator, with a given input voltage (AC or DC).
2. No dead time in the oscillation, the tank circuit is powered
during each half of the RF alternation. The circuit appears
to tolerate significant loading without circuit detuning (much
improved power oscillator frequency stability).
3. Reduced EMI/RFI generation, harmonic reduction (with balanced
tube conduction).
A multitude of meters was installed on the system to attempt to develop
a clearer understanding of the operational parameters of AC power
oscillator operation versus DC. There was quite a bit of personal
frustration in that all analysis of Class C Power Amplifiers/Oscillators
to date are based on DC POWERED ( :^C ). AC powering of the circuit
completely changes the way the circuit should be viewed and analyzed.
Level shifted powering of the VTTC produced some quite unexpected
results. A momentary regression...
Level shifting in the MO consists of the MOT, a 2kVAC (typical)
capacitor, and a 12kV,500mA (typical) high voltage diode, allowing
the capacitor to charge to the peak positive potential during the
negative AC mains alternation. When the positive alternation occurs,
the instant the HV diode switches off impresses peak DC on top of the
sinusoidal wavefront from the mains, allowing the tubes to "INSTANT ON"
AT MAXIMUM SATURATION CURRENT OSCILLATION POWER. This dV/dT risefront
is
rather fast (about 0.5uS) and is controled by the switching time of the
HV diode. The result is the modulation envelope does not appear half
wave modulated RF envelope, but has a steep nearly vertical rise front,
exponentially decreasing to the AC wave, then the current decreases to
normal maximum saturation current predicted by various sources
(I use Termann's Radio Engineering).
Peak current of up to 1A (40mA per filament W) is predicted by Termann
for thoriated tungsten power grid tubes. With the normal AC powering,
plate current of up to 1.25A peak is observed (2.5A/2 tubes). With
Level shifting, currents approaching 2A peak have been observed on my
oscilloscope (Tek 2213 60mHz analog). The asterik by the metering
above is a note, DC average meters will not produce accurate results
when calculating wattages, and true powers. It is required that RMS
currents and voltages either be calculated or measured when calculating
true power (watts or joules). This condition is true and valid for ANY
TESLA SYSTEM, PERIOD!!!!!! As a good example, the plate meter
is indicating 330mA DC average plate current, but because of the large
crest factor (approaching 6:1) would result in a large error in
measurement, if dc average values were depended on. Measuring with a
oscillioscope, video recording and then manually digitizing and
importing to a spread sheet and calculating true RMS values will provide
meaningful results, regardless of signal envelope.
As an example, if apparent DC voltages and currents are used to
calculate Power based on previous published information, I had
divergence of as much as 60% from what was measured using my
opto-isolated wattmeter!! That was a huge wake up call for me. And a
VTTC system is SIGNIFICANTLY more controlled (confounding variable wise)
then a impulse driven (spark gap) Tesla Coil.
The discharge currents are very high, if 22AWG magnet wire is used as
a takeoff point, current is high enough to cause copper to glow yellow
hot! Arcs to a grounded target will flame like a NST!
The MOT is the weak link, even with fan cooling of tubes and MOT,
the MOT in level shifted mode at full power will be too hot to touch in
30 secs! At low powers (50-200W), the VTTC operates lamps and the
resonator E field can be mapped (it looks like 1/4 wave antenna E
distribution).
Further work includes installation of a significantly higher peak power
level shifting power supply, and long term construction of a 3-5kW
level shifted, push-pull system using 4-400C tetrodes as hard switch
tubes. As for myself, if any further VTTC are built, they will be
push-pull Hartley Oscillators.
Regards
DAVE SHARPE, TCBOR