o-scope data

[Garyf]

I’d like to post some scope traces and propose some interpretations for confirmation or correction....

The coil is a smaller version of my earlier DRSSTC attempt: now 3.5 inch X 15 inch secondary, Fs=120khz, primary 9 turn pancake, MMC total=.33uf. Driver is loneoceans 2.7, bridge is loneoceans 80mm full bridge plus.  4000uf w/ twin 2.2uf snubbers.  Currently my interest is measurement and understanding so I’m running at lower voltages: max bus voltage from variac 200v.  Full bridge IGBTs=20N60A4D with Icm=280A. OneTesla interrupter.  Fig 1


Fig 2 Typical GDT output:


Fig 3 Primary voltage (~200v) and primary current(~100A).  I think this is a good illustration of the action of the anti-parallel diodes.  Without the diodes, the voltage at the IGBTs might greatly exceed the bus voltage.  The anti-parallel diodes clamp the voltage across the tank to a few volts over and under the bus voltage, saving the IGBT if max Vce would be exceeded (in this case it probably wouldn’t).
Note the skipped beat after the 6th pulse.  I wasn’t sure what this was when I first saw it, but a post on 4HV.org explained that when the burst is complete, the energy reverses from going in to going out, requiring a phase change.


Fig 4 Secondary voltage (uncalibrated stub antenna) and secondary current (CT, ~1A)


Fig 5 Secondary voltage and primary current at interrupter max, 12 inch streamers.


Some of the traces above are second order exponential decays.  I have some MATLAB code I once wrote to calculate damping and Q from curve fitting selected portions of an exponential decay.  In Fig 5 there are two different areas of decay behavior. The first part has a fast decay I believe associated with energy loss in the streamers and has a Q of about 15.  After the streamer disappears, the much lower loss is from corona and the Q goes up to about 60.  In Fig 4, the voltage is lower and the Q is more consistently high (60-70).

Fig 6 Primary voltage compared with the secondary voltage.  The frequency measures 119 khz.  After the interrupter pulses end, both tuned circuits are in free decay. If they align like they do here, that is a useful indication that the primary and secondary are tuned to the same frequency, at least for this streamer loading. I compared this at different bus voltages and found no separation in frequencies, regardless of streamer length, which surprised me.   Primary and secondary seem to phase lock together.


A ground strike is another matter, immediately quenching the secondary voltage. Fig 7, secondary voltage and primary current.  I worried about the spike shown in primary current at the instant of arc, but it turns out to be a common mode issue from the RF because I don’t have differential input to the scope.  The arc is a release of thousands of watts for a microsecond or two.


Loneoceans has suggested that if I want high performance with long arcs using TO-247s, rather than pushing one bridge to the max I should add a second full bridge with its own matching MMC, both driving the primary in parallel.  I may try that soon.

[Mads Barnkob]

I like your write-up on these different measurements on a DRSSTC inverter/secondary circuits, results explained in simple terms and yes the sudden phase change of the inverter voltage always puzzles people the first time until they find the explanation

I am actually a little surprised that you do not have higher amplitude switching spikes on your inverter output voltage when you are only running it at 200 VDC, but maybe the CE junction capacitance just falls fast on your IGBTs, it properly since you are using small and fast TO-247 IGBTs, this effect is much more pronounced in larger IGBT bricks.

What kind of antenna did you use for the secondary voltage measurements? I never tried this Terry Fritz model: https://deanostoybox.com/hot-streamer/TeslaCoils/MyPapers/planant/waveant3.html

[Garyf]

I started by noticing that my unconnected oscilloscope probes were picking up several volts from just sitting on the table.  I have an inexpensive emf meter I opened up to see how they measured the E and H fields.


The e field pickup is just a small flat plate (the H sensor is a 47 mh choke).  So I imitated this with a small square of aluminum foil clipped to the oscope probe.  It probably forms a voltage divider with the probe capacitance, similar to Terry's.

The secondary current is measured with a CT on the cable grounding the secondary base.  I think the proof that this works is that the two signals are 90 degrees out of phase.

It's uncalibrated and not as professional as Terry's solution but it gives me a reference.

The emf meter indicates a pretty high e-field gradient, it has to be at least 6 feet away from the coil and it only takes a foot or so to go from overscale (1000V/M) to zero.  I wonder if its possible to use two measurements a known distance apart to estimate secondary voltage at the topload without a calibration?

[Uspring]

I wonder if its possible to use two measurements a known distance apart to estimate secondary voltage at the topload without a calibration?

The field depends in a complicated way on your surroundings and the distance from the topload, so having two measuring points won't really help. But calibrating the antenna isn't difficult, since you are already measuring the secondary base current. Check out the help in javatc regarding the "Effective Shunt Capacitance - Ces". Therein you'll find the equivalent of the equation

Vtop = Ibase / (2*pi*f*Ces)

This holds only for toploads not loaded by arcs, so you must calibrate at low voltages without breakout. When later running at full power, the arcs should point away from the antenna to keep the field not too much disturbed by the arc.

[Garyf]

Mads, i didnt catch it live, but i viewed your video today...this chronicle of your long journey will be a resource and inspiration for coil builders of the future!

Uspring, this seems straightforward, so i inserted a capacitance meter between ground and the secondary base..168 pf.  Measuring 2.5 amps with my CT gives 25 kv which seems low.  ?

[Uspring]

Your measurement adds the capacitance of the secondary winding to the top load. This will make the voltage too low. Instead use javatc to calculate Ces. Javatc is very good at this.

[Garyf]

Put in all the numbers and got 13pf for Ces. That is roughly 100kv per amp of base current, which seems about right..  Thanks!
To be clear, Ces is the capacitance of the secondary and topload to ground without the turn to turn self capacitance?