Hydron's 160mm DRSSTC and topload current measurements

[Hydron]

The 2877 CTs have a bandwidth of something like 200MHz, so will not be the limiting factor. I was proposing to use a 35MHz CT for the base current measurement and a 2MHz CT for primary current (determined by what i have available - 2x2877, a 4100 and a 7800). I would be surprised if the CTs filter anything of interest out with a lower -3dB of 300Hz. I assume that you agree the best way to use 2 extra channels is to measure current?

[Uspring]

Yes, I agree. Primary current being even more interesting than base current, since it reflects the amount of power transferred to the secondary due to arc load and arc detuning.

300 Hz lower bandwidth looks ok. I guess that depends much on the loading resistor for your CT. It should be quite low.

[Mads Barnkob]

Great analysis that you got done there Uspring!

Regarding the phase shift / secondary impedance graph, what can be concluded on the loaded Q/impedance compared to the secondary Q/impedance?

As always the dynamic nature of arcs and our lack of plasma physic applied knowledge to it puts the models in the same regions as the existing ones

[Uspring]

Thank you, Mads.

Wrt the primary tank answers are difficult. Secondary Q affects primary Q but primary Q is also dependent on the location of the secondary resonance. This frequency changes during the burst just as secondary Q. One idea is to design a coil in such a way, that everything is perfectly tuned at the point, when power requirement of the arc is largest.

[Hydron]

Yes, I agree. Primary current being even more interesting than base current, since it reflects the amount of power transferred to the secondary due to arc load and arc detuning.

300 Hz lower bandwidth looks ok. I guess that depends much on the loading resistor for your CT. It should be quite low.

For reference, these are what I am planning to use for the measurements:
http://pearsonelectronics.com/pdf/2877.pdf (x2, for topload measurements)
http://pearsonelectronics.com/pdf/4100.pdf (for secondary base current)
http://pearsonelectronics.com/pdf/7800.pdf (for primary current)

Still playing tourist in Australia right now, so haven't had a chance to look at your hard work yet sorry!

[Uspring]

@Hydron:

These CTs seem fine to me and they look like they have their resistors integrated. There is lot of inductance in the 2877, must be around 25mH.

I've taken a lot of time writing that up, so don't feel pressed yourself.

@Mads:

Using the equation

Qpri = (Qsec/k^2) * (1 - f^2/fsec^2)^2 + 1/(k^2 * Qsec)

we have now some numbers for it at the point of maximum arc load:
Qsec=4, k=0.15, f=70kHz, fsec = 55 kHz (with arc loading), which
puts Qpri to about 80. That looks seriously too large. Either the coil is getting quite out of tune regardless of its low secondary Q or I've made some mistake about the secondary resonance frequency. The equation is quite sensitive to fsec. Hydron, did you measure secondary fres? I took your secondaries parameter from the posted Javatc results.

[Hydron]

So I have some new measurements to share, taken in Jan 2019. While there are some issues with them (less comprehensive than last time, and breakout from a second point on the toroid), I also have characterised the coil a LOT better (see javatc load file - all values should be accurate to within a percent or two based on physical and electrical measurements) and also measured both primary and topload currents.

Files are shared in the google drive folder below:
https://drive.google.com/drive/folders/1bDoH_atL-bCJeF0Slkl4aauANMzLPnha?usp=sharing

I may be able to add to and improve these in the next month or so as I will have access to the coil again (but not necessarily time to do much due to family commitments).

[Uspring]

Hydron, thank you for the new data. Currently I'm in downunders outback, where mobile net sort of lacks. I'll definitely scrutinize what you've got as soon as i'm back (end of July). Can't wait. 

[davekni]

Old thread, but I finally got around to playing a bit with the big_ground_strike data files.  Only partially-successful with non-linearity compensation by adding a bit of extra gain and a bit of added signal-squared to positive samples.  Required different values for the two channels.  Even after tweaking, charge integrals still had bows in the middle.

A few observations and questions:

There is a somewhat-random 0.15V step-function noise that's common to both channels.  Both step positive by 0.15V at the same point, then later step back down by 0.15V, again at the same point.  Any ideas for a cause?  My first thought was some switching-regulator noise in chirp-mode.  But the noise doesn't appear until shortly after the intended signals start.

As was mentioned previously, the ground-strike current spikes are about 1us wide at half-amplitude (FWHM).  Is that normal?  It seems long to me, but it roughly fits with the post-strike current and voltage, which is on the order of 50k ohms.  (I need to look at that resistance more closely.)  I'd thought the ground strike would have lowered the arc resistance well below 50k.  But, 30pF and 50k fit the pulse width reasonably.

The ground-strike current pulse appears to be clamped.  Total charge is usually about half of what's on the top load.

For any future measurements, I wonder if it would be better to use capacitors instead of resistors for current sensing.  Then the scope signal is charge instead of current.  Also avoids possible scope overload at the ground-strike point.

Gratitude to Hydron for the data, and to everyone else who contributed to previous analysis!

[Uspring]

David, I wasn't really successful in compensating non linearities either. My approach was similar to yours. Probably better results could be obtained by some many parameter regression to fit sinusoidal waveforms, but from my point of view, that wasn't worth the effort. Hydrons own method, which is filtering out low frequency parts of the current integral, works pretty nicely and is much simpler.
In Hydrons later measurements the top load current can be integrated without any problems. But if you try to integrate the primary current, which he also measured, the large DC component reappears.

One has to be careful in comparing TC ground arc timings to e.g. to those of the much faster Marx or VDG arcs. The latter have a lot more field strength and therefore their charge carrier generation depends more on field induced electron acceleration with than following ionisation. TC arcs in contrast are much more like your Jacobs ladder arcs, where voltages and currents are roughly proportional in time scales of the voltage period. In that case, conductivity is related to the charge carriers generated by heat or temperature (plasma). Their values change much slower. Field induced avalanches on the other hand can be really fast. In TC arcs both effects play a role, heat is more important near the breakout point, voltage more near the tip of the arc.

[Hydron]

I have a suspicion that any non-linearities and DC component may be down to the scope involved; the unit used for the initial topload measurements and later the primary measurements was one I was given as a reject by the producer (whom I personally know) - while I managed to at least mostly fix it with replacement of some front end parts it could easily still have some minor issues. The second scope doesn't have this chequered history!
Note that I used high bandwidth CTs (Pearson model 2877s) for the second lot of topload measurements - these should eliminate most of the worries about resistors.

Sadly the COVID-19 crisis and associated air travel issues have put paid to my latest attempt to get newer data - I should have been running the coil in question over the last couple of weekends but instead I had to return to the UK early due to my original flight being cancelled.

[davekni]

Uspring, thank you for the clarification about different arc regimes.  My experience is weighted to shorter faster arcs (Marx generator etc.).  Looking a bit more at Hydron's first data set today, arc resistance seems to start around 50k, then drop to ~25k.

In case it's still of interest, here's a couple plots of big_gnd_strike_4.  I've inverted the breakout point current to make interpretation easier, and applied some hand-tweaked correction algorithms.  The integrated waveforms are in units of micro-coulombs (amps * microseconds).


And a zoom into the arc point:


This is the second file with a ground strike (after big_gnd_strike_3), and has the highest voltage at the strike time of the files I viewed.

I hadn't noticed that there's newer scope files.  Any estimate of how long the unintended breakout streamer is relative to the intended one?  Not sure when I'll get to processing the new files myself.

Has anyone measured how much total power is dissipated in the arc before, during, and after a ground strike?  In particular, I'm thinking that the power dissipation after the strike is a relatively small fraction of the total arc power, so may not contribute that much to sound and light generation.  However, the primary current usually ramps up rapidly after a ground strike (at least on my DRSSTC), wasting power in the H-Bridge and primary coil.

[Hydron]

There are a couple of videos in the folder with the new data - maybe enough to give a rough idea of the relative breakout lengths?

[Uspring]

@Hydron:
I had suspected, that the top voltage scope was swapped to the bottom (primary and base current measurement) in your second set of data.
Thanks for mentioning the video. I had overlooked that.

@davekni:
I've calculated arc energy by integrating Vtop * Iarc. The uncertainty here is getting Vtop right. Vtop does not only depend on the net charge deposited in the toroid but also on the fields originating from the charge along the secondary winding and the space charges of the arc. The effect of the secondary winding charge can be estimated by a JavaTC calculation. The value of Vtop I used is calculated from the current going in and out the top load (Isectop - Iarc) and the net top load capacitance, i.e. the bare capacitance of Hydrons top load objects without any secondary coil. JavaTC suggests for the calculation of the top voltage the use of the secondary base current and the Effective Shunt Capacitance (Ces). From the JavaTC file, that Hydron included (a big thank you), it turns out, that my calculation is about 5% low. That it is too low was expected, since the charge in the winding capacitance, that JavaTC includes, adds to the voltage.

The effect of the arcs space charges on the top load voltage is difficult to ascertain, since these charges are distributed in space and depend on length and direction of the arc. Anyway, I presume that effect also to be in the few percent region.

The arc energy calculated from the data seems to be considerably lower, than the energy transferred to the secondary coil from the primary, which indicates, that the unintended streamer is almost comparable in size to that of the one we have data for. Power transferred to the secondary can be calculated by

P = -M*Isec*dIpri/dt

M being the mutual inductance.

However, the primary current usually ramps up rapidly after a ground strike...

The reduction in secondary voltage after a ground strike lowers Isec considerably and consequently limits energy transfer to the secondary (see above equation). The result is that the energy from the bridge has nowhere to go except into the primary tank.