Author Topic: DRSSTC with litz-wire primary and 40 x TO247 IGBT H-Bridge  (Read 13623 times)

Offline davekni

  • Executive Board Member
  • *******
  • Posts: 3000
  • Karma: +142/-2
  • Physicist, engineer (electronic), and hobbiest
    • View Profile
DRSSTC with litz-wire primary and 40 x TO247 IGBT H-Bridge
« on: October 10, 2019, 06:31:02 AM »
Here's my first attempt at documenting a project.  It is my first DRSSTC - just complete enough to run, but with much enhancement work planned for the next couple years.  A few videos of operation, at the Portland (Oregon, USA) Mini-Maker Faire, and a couple in my back yard two weeks earlier :
   
   
   
   
   

Don't have a good assembled picture of the coil, but here's the parts:
   

Edit:  Now I do have an assembled picture:


MMC, with lots of room for planned expansion:
   

H-Bridge with 0.1mm x 100mm copper foil output leads.  E80-core current transformer is hidden under bump in output foil at right end.  Purple clip in the upper left quadrant is a crude optical probe (LED and resistor into plastic fiber) monitoring one of the high-side gate drive signals.  The pairs of TO220 FETs are the gate drives on floating 18.5V supplies, one PFET and one NFET per H-Bridge switch, driving 10 paralleled STGW60H65DRF IGBTs (hiding under the aluminum bar clamps):
   
The bulk cap array is under the H-Bridge, so not visible.  96 x "470" uF 450V.

Control circuit - in an aluminum-foil lined cardboard box at this point:
   

4-turn primary coil made of 200 strands of 27AWG wire.  Inner turn gets warm at 3kW.  Will need finer litz and/or cooling air directed that way to get 10kW eventually:
   

Bulk-cap circuit:
   

Main oscillator simulation schematic, using voltage-controlled-voltage-sources for gate drive, and FETs  for the H-Bridge because they simulate faster.  The center lower part is mostly a comparator made of discrete TO92 FETs, from M3 on the left through M10 and M11 on the right.  Nodes "vn" and "vp" are the inputs and nodes "v2" and "v3" are the true and inverted outputs.  R2, R3, R4, R7, R11, and C10 are the relevant feedback around the comparator.  No one would want to copy my comparator itself.
     

Current limit is also a bit unusual.  It is fed with a voltage-transformer (L1 and L2) from the oscillator current transformer output.  (My current transformer is 40:1 first-stage, then two 25:2 second stages.  However, one of the second stage outputs is for scoping only.)  The current limit circuit includes not only an immediate shutdown at 3500A peak (2600A in this initial version), but also simulation of the IGBT thermal transient response (C1 and C2 and associated resistors).  That way repeated long bursts will get aborted even if they don't reach 2600A.  (So far, the current limit trips only in my initial primary-only testing, not in real use so far.)
     

Finally, here's a hint of where I hope to go.  This is a sketch of a small part of what will be an array of ~500 TRIACs to switch in more primary tank capacitance as each arc grows.  My goal is to make it behave more like a QCW coil, but hopefully even better with a reverse-chirp in primary frequency to match the secondary frequency change with arc growth:
     

Before this fancy reverse-chirp attempt, I need to improve my MIDI interrupter setup, at least for next year's Maker Faire.  Also need to redo the primary tank circuit connections - going to use spade connectors instead of many paralleled 0.025" square pin connectors - that come apart too easily.  Going to explore detection of power arcs and termination of the drive enable pulse once a power arc forms.  That should make overall operation more efficient.

Hopefully I have the jpg attachments figured out now.
« Last Edit: December 01, 2021, 04:47:41 AM by davekni »
David Knierim

Offline johnf

  • High Voltage Technician
  • ***
  • Posts: 191
  • Karma: +6/-0
    • View Profile
Re: DRSSTC with litz-wire primary and 40 x TO247 IGBT H-Bridge
« Reply #1 on: October 10, 2019, 08:44:38 AM »
oops
same pic repeated!!!

Offline Hydron

  • Administrator
  • High Voltage Engineer
  • *****
  • Posts: 489
  • Karma: +25/-0
    • View Profile
Re: DRSSTC with litz-wire primary and 40 x TO247 IGBT H-Bridge
« Reply #2 on: October 11, 2019, 10:06:32 PM »
Woah, was worth waiting for the new pics. Can't say I'll be copying the comparator bit, but I do like the idea of the OCD having more sophisticated behaviour than a simple on/off threshold.
Are you doing anything in particular to try and force the IGBTs to share current fairly? Or just relying on a low-inductance bridge configuration?

Offline davekni

  • Executive Board Member
  • *******
  • Posts: 3000
  • Karma: +142/-2
  • Physicist, engineer (electronic), and hobbiest
    • View Profile
Re: DRSSTC with litz-wire primary and 40 x TO247 IGBT H-Bridge
« Reply #3 on: October 12, 2019, 05:52:51 AM »
The IGBTs have a positive Vce-temperature coefficient, which helps with current sharing.  The only explicit step I took besides tight packing was to feed Vbus from one end of each half-bridge, and take the outputs from the other ends.  That way the Vbus voltage drop roughly matches the output voltage drop, keeping Vce roughly constant across the array.  I haven't attempted any specific measurements of current sharing.  Would like to check it before increasing max current from 2600A to 3500A, but not sure how.  (3500A still has some margin for sharing errors.  I tested one individual IGBT to 500A, both transistor and diode, at 90C initial temperature.  Rated peak current is 240A - the best I could find in a fast TO247 part.)

The reason for the discrete comparator (besides my habit of designing that way at times) is that it tolerates input voltages to 10V beyond supply rails.  That avoids diode-clamping the current-sense signal, which would make a non-linear load impedance on my phase-lead filter.

The phase-lead filter is a bit fancier than normal, designed for roughly constant time lead (not phase lead) from 50-80kHz.  That covers the range I'm planning with my eventual reverse-chirp primary resonant frequency.  The filter also has the advantage of not amplifying high-frequency noise as much as a single inductor or capacitor circuit.  (I'm currently running about 83kHz primary.  It works better with the arc-length constraint of the Faraday cage with the primary closer to the secondary's 92.2kHz resonance.)

Forgot to include a bit of basic information:
    Primary:  6.72uH (6.95uH including inductance of MMC and interconnect), 4 turns of 200 strands of 27AWG wire.
        222mm diameter of inner turn (diameter at center of wire).  308mm diameter at center of outside turn.
    Secondary:  68.0mH, 75 ohms DC, ~1800 turns of 24AWG wire, 157.5mm diameter, 1100mm high (not counting loose-wound ends).
    MMC:  10s16p of 330nF 1200VDC 630VAC Chinese induction cooker capacitors, 528nF total, half on each end of primary coil.
    Top Load:  194mm minor diameter, 632mm major diameter (826mm outside diameter).

Ignoring the comparator implementation, the circuitry around it is somewhat similar to the UD2.7 design.  A couple differences:  First, the bias point of the current-feedback is centered with no signal present, rather than a diode-drop low.  Second, I have positive capacitively-coupled feedback (as does UD2.7), but also weak resistive negative feedback.  That combination causes the circuit to oscillate with no current-sense present.  I've designed that oscillation frequency to roughly match the primary resonant frequency.  This allows synchronous oscillation to start even with very-low Vbus (5V works), independently of the initial state of the primary tank circuit.  No need for a resistor across the H-Bridge output.  (I do have bleed-down resistors with indicator LEDs across the MMC sections.)
David Knierim

Offline ScottE

  • High Voltage Enthusiast
  • *
  • Posts: 4
  • Karma: +0/-0
    • View Profile
Re: DRSSTC with litz-wire primary and 40 x TO247 IGBT H-Bridge
« Reply #4 on: October 12, 2019, 07:38:47 AM »
I am planning on building a coil with litz wire on the primary and secondary coils. The secondary wire is made up of 100 strands of 0.07 mm wire (about 21 awg). The primary wire is made up of 4000 strands of 0.07 mm wire (about 6 awg) and has outer teflon covering. It should work up to 250 kHz.

I see that you are in Portland and I am quite sure I met you at Westercon when it was in Portland.
I know that you met Jon Hannis from Control H (PDX Hackerspace) at the maker fair. If you go Control H and talk to Jon he can show you samples the litz wire I have.
I am in LA but I will be building the coil at Control H in year or so after I move up to Portland.

Offline davekni

  • Executive Board Member
  • *******
  • Posts: 3000
  • Karma: +142/-2
  • Physicist, engineer (electronic), and hobbiest
    • View Profile
Re: DRSSTC with litz-wire primary and 40 x TO247 IGBT H-Bridge
« Reply #5 on: October 12, 2019, 06:57:55 PM »
Should be a great primary with 4000-strand litz!  With such fine wire, the inner turn may not get much warmer than the outer turns.  Would be fun to see thermal images of your primary after a run.  Mine clearly shows higher temperature on the first turn.

Litz will have minimal affect on the secondary.  Typical secondary Q is already high, so relatively little power loss.  Even if you got infinite secondary Q (zero power loss), it's not that much savings.  No harm in using litz, though.

I'd love to see your coil once your up here.  Haven't been to Westercon, but I think Don Anderson was there with his coils.  He's the one who got me interested in Tesla coils.  We share demo space at Maker Faire here.
David Knierim

Offline Weston

  • High Voltage Technician
  • ***
  • Posts: 132
  • Karma: +11/-0
    • View Profile
Re: DRSSTC with litz-wire primary and 40 x TO247 IGBT H-Bridge
« Reply #6 on: October 13, 2019, 12:41:07 AM »
Cool coil! Its always nice to see people coming up with their own controller instead of directly copying the UCD design. I have a FPGA based controller for a QCW coil and I was considering implementing a similar limit on the total IGBT heating, its nice to see that someone else is already doing it.

Why did you choose to use so many paralleled TO-247 IGBTs? Surplus IGBT bricks are pretty cheap nowdays so I assume they would be cost competitive in addition to simplifying the bridge layout.

How much modeling have you done on the the concept of switching in additional capacitance with TRIACs? I would be worried about dv/dt induced turn on in the TRIACs beyond the general added complexity. QCW based coils typically get around the detuning issue by tuning the secondary above the primary and then having the controller run on the upper pole, so the added capacitance from the arc pulls the coil better into tune. 

Offline davekni

  • Executive Board Member
  • *******
  • Posts: 3000
  • Karma: +142/-2
  • Physicist, engineer (electronic), and hobbiest
    • View Profile
Re: DRSSTC with litz-wire primary and 40 x TO247 IGBT H-Bridge
« Reply #7 on: October 13, 2019, 06:36:13 AM »
The paralleled TO247 IGBTs were for fast switching, both because small IGBTs are rated faster, and the paralleling gives lower net lead inductance.  I would have had to parallel at least two sets of bricks to get 3500A peak.  Now that I'm reading what others are doing, it's clear that the bricks can be pushed to much higher frequencies in ZVS mode than I'd realized.  (The lower inductance does have an advantage still in peak voltage.  These 600V IGBTs still have sufficient headroom at 450Vbus, as there is very little overshoot during switching.  However, if I fry the H-Bridge, I may rebuild with bricks.  It was a lot of work to build a 40-IGBT bridge!

The thermal modeling seemed sensible for pushing devices to their limit.  The zener/resistor array at the bottom of the schematic page makes a current roughly proportional to the square of the primary current-sense transformer output, so proportional to forward-conduction power loss in the IGBTs.  That current charges C1 and C2, which are coupled with resistors modeling the transient thermal impedance from the IGBT data sheet.  I had no idea if anyone else was doing that - hadn't found much information before stumbling into this forum.  (Peak current causes all zeners to conduct, so behaves as infinite power, causing an immediate (next zero-current, rising or falling) end to that burst.  It also leaves the model in its hot state, delaying when the next burst is allowed.

The TRIAC switching will be a challenge!  I haven't found good modeling tools, so am relying mostly on testing individual TRIACs and gate-drive transformers etc.  Had to use NiZn ferrites for high electrical resistivity to avoid excess capacitive coupling.  MMC is split in half, one on each side of H-Bridge output to keep peak voltage to ground lower.  Finally, it's going to be inefficient in MMC capacitor use - many extra ones.  The plan is to build a huge MMC with extra voltage capability, then short out individual small groups of caps at the voltage zero-crossing.  Snubberless TRIACs have one quadrant where they won't turn on.  That allows the gate drive pulse to start early, before voltage zero-crossing, avoiding excessively-tight timing requirements.  Placing the TRIACs directly across capacitors prevents stray inductive spikes from triggering the TRIACs.  I did choose TRIACs with fast enough DVDT rating to handle the voltage slew rate of individual MMC sections.

I have some previous experience in failed TRIAC array switching.  My very first Tesla coil was an attempt to replace the spark gap with a TRIAC array.  Made three sparks before a failure cascaded through the array.  Turned that project into a rotary-gap coil.

I'll need to play with my basic coupled-resonator simulation some more.  (Anyone have a good SPICE model for a growing arc?  I've made a couple guesses, but not likely very accurate.)  In simulation, it seemed that starting on the upper pole didn't perform as well.  Does the ramped voltage make that upper pole work better?  I gather that most DRSSTC designs oscillate at the lower pole, as mine does.  Either way, I expect there's a limit to how far detuning can go, and I'm hoping to add to whatever that limit may be.
David Knierim

Offline T3sl4co1l

  • High Voltage Technician
  • ***
  • Posts: 174
  • Karma: +9/-0
    • View Profile
    • Seven Transistor Labs
Re: DRSSTC with litz-wire primary and 40 x TO247 IGBT H-Bridge
« Reply #8 on: October 14, 2019, 07:10:39 AM »
I don't see that TRIACs can be used for tuning here.  They turn on automatically when exposed to high dV/dt.

A saturable inductor may be more feasible?

Capacitors can be tuned in the same way as well, but you can only do it with lower-Q ceramics, and you'd need a lot of them to handle that much reactive power.

Saturable inductors tend to have a low Q for the same reason.

Alternately, a vacuum variable capacitor (with servo control), with a transformer to match the impedance.  Exquisitely priced. :o

Perhaps a frequency tracking control, and an inverter with enough excess VA capacity to handle the change in SWR, would be easier.

Tim

Offline davekni

  • Executive Board Member
  • *******
  • Posts: 3000
  • Karma: +142/-2
  • Physicist, engineer (electronic), and hobbiest
    • View Profile
Re: DRSSTC with litz-wire primary and 40 x TO247 IGBT H-Bridge
« Reply #9 on: October 14, 2019, 07:01:34 PM »
I purposely selected TRIACs with high dV/dT rating, 1000V/us (BTB16-800BW from ST).  That's well beyond the 400V/us they'll see.  Each TRIAC will be shorting one MMC section, which will have 800V peak at 80kHz, thus 400V/us.

The TRIAC specification that is most risky is "critical rate of rise of on-state current".  These parts, as with almost all similar parts I could find, are specified at 50A/us maximum.  In my particular odd application, the rate-of-rise is controlled primarily by how fast the TRIAC manages to turn on.  (Ideally, it would instantly conduct all the capacitor current at the voltage zero-crossing.)  I made a small 3-cap MMC in a resonant circuit with one TRIAC shorting one stage.  It ended up with ~100A/us, about twice spec.  I've done some other single-shot tests running these to 250A/us, but not repeatedly nor with high die temperature.  It's been fun learning about "critical rate of rise of on-state current", as I had no idea previously what that spec. meant or why that would be limited.  (Turns out that TRIAC gates are in one corner of the die.  It takes time for the primary current to spread across the die.  It starts in the gate corner and spreads.  Too high current immediately overheats the gate corner of the die.  I can see this in the forward drop voltage/time behavior.  Starts at ~20V forward drop, which goes down to normal ~1V over 1-2us.)
David Knierim

Offline Weston

  • High Voltage Technician
  • ***
  • Posts: 132
  • Karma: +11/-0
    • View Profile
Re: DRSSTC with litz-wire primary and 40 x TO247 IGBT H-Bridge
« Reply #10 on: October 14, 2019, 08:46:54 PM »
In what way did the upper pole tuning did not preform well in simulation? The tuning relies on the fact that the capacitance from the discharge pulls the secondary into tune, so you need some sort of model relating power and spark length/capacitance. I recall there being some good posts on 4hv about the tuning method and models for arc impedance. The easiest model is assuming the arc has a capacitance equal to an equivalent length of wire, which is how I tuned my QCW coil.

One thing to note is that when the secondary is tuned above the primary the lower pole has higher gain, so a self oscillatory controller will lock onto the lower pole.

I am still unclear about what advantage the triac based system would have. It seems like a lot of work with a high potential to blow up if anything goes wrong.


Online Mads Barnkob

  • Administrator
  • Executive Board Member
  • *****
  • Posts: 2359
  • Karma: +78/-0
  • Denmark
    • View Profile
    • Kaizer Power Electronics
Re: DRSSTC with litz-wire primary and 40 x TO247 IGBT H-Bridge
« Reply #11 on: October 15, 2019, 12:38:24 PM »
Impressive design and + for effort going your own way!

Regarding spark load models, Uspring shared his version 2 dynamical model ( https://highvoltageforum.net/index.php?topic=670.0 ) this summer, based on the new information from Hydron's spark load measurements ( https://highvoltageforum.net/index.php?topic=117.0 )

Usprings version 1 model, but without pictures as 4hv is broken: https://4hv.org/e107_plugins/forum/forum_viewtopic.php?153922
https://kaizerpowerelectronics.dk - Tesla coils, high voltage, pulse power, audio and general electronics
https://www.youtube.com/KaizerPowerElectronicsDk60/join - Please consider supporting the forum, websites and youtube channel!

Offline davekni

  • Executive Board Member
  • *******
  • Posts: 3000
  • Karma: +142/-2
  • Physicist, engineer (electronic), and hobbiest
    • View Profile
Re: DRSSTC with litz-wire primary and 40 x TO247 IGBT H-Bridge
« Reply #12 on: October 17, 2019, 03:22:22 AM »
Mads, Thank you so much for pointers to so much amazing information!  That's going to take a long time to study.  I did notice initially that some of the top current traces look somewhat like what I measure at the bottom of my secondary coil.

Concerning upper pole tuning (which I've tried in simulation only), what I see is significantly higher primary current and voltage needed to achieve a given secondary voltage, especially after the secondary frequency drops (capacitance increases).  I'll need to learn more about QCW drive circuits - there's probably something I'm not understanding yet.

Yes, the big TRIAC array for dynamic primary tuning is a huge project with a significant chance of cascading failure.  That is part of what makes it interesting.  (Like putting a scope on top of a Tesla coil - risky but rewarding.)  My hope is to extend arc length for my moderate-sized DRSSTC.  May not work, but I hope to find out.  Prabably 1.5-2 years out, as I have other improvements to make to my DRSSTC first, and other small projects etc.

In what way did the upper pole tuning did not preform well in simulation? The tuning relies on the fact that the capacitance from the discharge pulls the secondary into tune, so you need some sort of model relating power and spark length/capacitance. I recall there being some good posts on 4hv about the tuning method and models for arc impedance. The easiest model is assuming the arc has a capacitance equal to an equivalent length of wire, which is how I tuned my QCW coil.

One thing to note is that when the secondary is tuned above the primary the lower pole has higher gain, so a self oscillatory controller will lock onto the lower pole.

I am still unclear about what advantage the triac based system would have. It seems like a lot of work with a high potential to blow up if anything goes wrong.
« Last Edit: October 18, 2019, 06:54:35 AM by davekni »
David Knierim

Offline davekni

  • Executive Board Member
  • *******
  • Posts: 3000
  • Karma: +142/-2
  • Physicist, engineer (electronic), and hobbiest
    • View Profile
Re: DRSSTC with litz-wire primary and 40 x TO247 IGBT H-Bridge
« Reply #13 on: July 11, 2020, 09:36:02 PM »
July 2020 update:  Made a new MIDI interrupter (to replace the Eastern Voltage Research kit that had significant algorithm issues) and increased input power to 4.8kW.  Set it up for July 4th (Independence Day here in US) with four surrounding ground poles with wire extensions bending in at the top.  Distance from breakout on top-load to poles/wires varies from 1.8 to 2 meters.  Here's some video of neighbor kids playing tunes on my keyboard.  This is before sunset, so arcs are a bit faint against the bright sky.  The end shows an after-dark clip when some other people dropped by to play:

/>
Does anyone else run with live music?  I don't see anything about that here or on youtube videos.  (I have no personal musical skills, so rely on friends or audience to play.)

For recorded music, I wrote PC code to process MIDI files into enable pulses, then record the enable pulses onto an MP3 player.  Software processing ahead of time allows multi-pass algorithms and looking ahead as well as behind.  I like that flexibility in deciding how to handle close and overlapping pulses.

For both my live MIDI interrupter and file-processing software, I think results sound better assuming that sound pressure is proportional to the square of pulse width.  That may be unique to my coil, as most pulses are short enough that current is still ramping up, 50-150us.  Otherwise average power gets past my PFC's 4.8kW capability.  (Plan to improve PFC to 10kW before too long.)

Here's a couple more videos, these with prepared music.  They're mostly 3-note polyphonic, except for "America the Beautiful" which is 4-note.  Most songs sound worse with a fourth note, but I'm hoping that more power will improve such, allowing the algorithm to space pulses more closely and combine them into longer pulses when needed.  Notes are chosen as the highest 3 or 4 in the MIDI file (or from the keyboard in the above video).  I couldn't find any better algorithm given my limited musical knowledge.

/>
/>
David Knierim

Offline davekni

  • Executive Board Member
  • *******
  • Posts: 3000
  • Karma: +142/-2
  • Physicist, engineer (electronic), and hobbiest
    • View Profile
Re: DRSSTC with litz-wire primary and 40 x TO247 IGBT H-Bridge
« Reply #14 on: October 02, 2020, 05:44:26 AM »
This is related to the "DRSSTC Tuning for Music vs Big Sparks?" thread:
https://highvoltageforum.net/index.php?topic=1239.msg9160;topicseen#new
However, it's more about my DRSSTC, so I'm adding it here.

Finally getting reasonable strikes to ground poles 3m from breakout tip.  Works better with a short breakout.  I appreciate Uspring's answer in the above tuning thread, but still wonder about arc growth and branching details, especially when comparing to Mads large DRSSTC III.  I'm running high burst rates, from 600Hz to 2kHz, and even explored up to 4kHz a bit.  Here's a set of sequential 30Hz video frames showing arc growth and branching, eventually finding a ground pole 3m away.  (Top of top-load is 2m high.  Breakout point is 2.06m high.  Ground level doesn't show up in the images, however.)


















Burst width is ramping from ~60us to ~130us over the 0.3 seconds of this sequence.  Longer bursts don't add much.  Added arc capacitance drops the secondary resonance frequency below primary resonance frequency.  At the start of each burst, the ZCS control runs at the lower pole.  After the first secondary current peak, the ZCS changes to the upper pole, which is at roughly the same frequency with arc loading as the lower pole was before arc loading.  (About 70kHz for my coil with this detuning level.)  Below is a scope capture of the final burst with my controller set for a longer final burst width:



Green trace is primary current at 500A/div (100x scale).  Red is secondary current at 5A/div (1x scale).  Blue is one side of the H-Bridge output, 500V/div (100x scale).  Secondary current is roughly in-phase with primary current for the first half, then roughly 180 degrees out-of-phase for the second half (upper pole operation).

My biggest surprise is that each burst appears to start out with relatively little arc loading, even at the 2.5kHz burst rate of this scope plot, but then grow very quickly back to where it was from the previous burst.  Hopefully next year I'll get some measurements of top-load and arc currents to understand better.  (I'm building some simple optical-fiber-isolated scope probes for this.)

Anyone have more thoughts on why this high branching?  Resonant frequency and burst frequency are both above Mads DRSSTC III.  His videos appear to show comparatively straight arcs even before strikes to the grounded ladder.  Perhaps branching is too faint to show up clearly in the videos?
« Last Edit: October 02, 2020, 07:23:18 AM by davekni »
David Knierim

Offline Uspring

  • High Voltage Engineer
  • ****
  • Posts: 348
  • Karma: +32/-1
  • Physicist
    • View Profile
Re: DRSSTC with litz-wire primary and 40 x TO247 IGBT H-Bridge
« Reply #15 on: October 08, 2020, 04:24:47 PM »
A nice sequence of shots you show here. It almost looks like they were made within a single burst, probably a result of the high burst rate. They show a good example of arcs retracing previous paths.

Quote
My biggest surprise is that each burst appears to start out with relatively little arc loading, even at the 2.5kHz burst rate of this scope plot, but then grow very quickly back to where it was from the previous burst.
Below is a simulation with my arc model of 2 bursts of a 70kHz coil in quick succession, 250us length and 250us inbetween. The red trace shows secondary current and the green one 1/100 of the primary current. The second burst shows a somewhat lower secondary current at about 100us into the burst, which is due the lower secondary voltage caused by the larger arc loading at this point. But generally there is not much of a difference.




In the model, I've used a warm up time constant of 300us, which is mostly an adjusted fit value to make the arc currents match measurements. A theoretical justification comes from the power consumption of QCW arcs. Measurements indicate a power requirement of about 150 W/cm averaged over its length. Since these arcs are almost stationary, i.e. barely growing, it can be assumed, that the power goes mainly into keeping the arc hot, rather than heating it more.
The heat capacitance of the arc channel is very small, since it is thin, maybe of 1 mm radius, and also there is not much mass in the volume due to the expansion of hot air. This adds up to a cooling off time constant in the region of hundreds us.

It could be argued, that the air does not have time to expand when using a coil such as yours. But that would require a radial expansion velocity at a sizeable fraction of the speed of sound. I don't believe this to be the case.

Offline davekni

  • Executive Board Member
  • *******
  • Posts: 3000
  • Karma: +142/-2
  • Physicist, engineer (electronic), and hobbiest
    • View Profile
Re: DRSSTC with litz-wire primary and 40 x TO247 IGBT H-Bridge
« Reply #16 on: October 09, 2020, 01:33:28 AM »
Uspring,

There's probably several parametric differences between my coil and your simulation.  I do see similarities too, such as changing from lower to upper pole frequency during a burst - clearest in your first burst.  One significant difference:  Your second burst appears to show arc capacitance loading from the beginning.  I'm seeing little loading effect at the start of each burst, even after multiple rapid-fire previous bursts.  The loading then develops over the first few cycles of each burst.

I didn't quite understand your final sentence: "It could be argued, that the air does not have time to expand when using a coil such as yours. But that would require a radial expansion velocity at a sizeable fraction of the speed of sound. I don't believe this to be the case."  I expect expansion is happening close to or above the speed of sound (shock-wave for the first mm or two) - why generated sound is so loud.  Radial expansion us likely in the few microsecond time frame.  I'm not sure how to estimate radial contraction after a burst.  There'll probably be some immediate contraction due to bounce-back, as inertia likely makes the expansion extend beyond pressure equilibrium.  Cooling is probably related to turbulent gas flow given the look of arcs.
David Knierim

Offline Uspring

  • High Voltage Engineer
  • ****
  • Posts: 348
  • Karma: +32/-1
  • Physicist
    • View Profile
Re: DRSSTC with litz-wire primary and 40 x TO247 IGBT H-Bridge
« Reply #17 on: October 13, 2020, 06:05:29 PM »
Hi David,

Quote
There's probably several parametric differences between my coil and your simulation.

Yes, I couldn't find all of the necessary parameters for your coil, so I used a set from another somewhat similar coil.
Quote
One significant difference:  Your second burst appears to show arc capacitance loading from the beginning.  I'm seeing little loading effect at the start of each burst, even after multiple rapid-fire previous bursts.  The loading then develops over the first few cycles of each burst.
That is one of the major problems of the simulation circuit. The arcs start too early and lead to too much loading in the beginning of the burst. A problem, that I haven't solved yet. The situation is much better for QCW coils.

Quote
I expect expansion is happening close to or above the speed of sound (shock-wave for the first mm or two) - why generated sound is so loud.

Yuri Raizer in his book "Gas Discharge Physics" mentions shock waves from high current arcs (10^4 A). In your coil, the max length of the arc is reached after 130 us. That corresponds to about 15 cm extra length on the average for each voltage peak. The energy deposited in this extra length is probably too low to generate hot plasma there, judging from tip voltages of around 100kV and a few pF arc capacitance for that front piece of it. I can't rule this out, there actually might be hot plasma there, but the radius of the discharge would need to be very small, e.g. 0.1 mm. If that is the case, expansion time will only be 0.3 us (i.e. 0.1 mm/speed of sound), which is shorter than voltage rise time of around 3 us. For a larger spark channel, say of 1 mm radius, temperature rise will be much less and correspondingly possible compression.

The situation is not clear cut. My belief is, that with TC arcs, compression does take place, although not into the bar region. It is definitely possible to have loud bangs with sound pressures below 1 bar.

I think, air expansion is a necessary feature for arc growth from TCs. Between voltage peaks there is a time, when there is no voltage drop along the arc. If the plasma is not hot enough to generate free electrons by its heat, which seems likely near the tip, then the electrons generated by avalanches will almost vanish since their lifetime is around 10 ns. Positive and negative ions live much longer, but they don't contribute much the current, since their mobility is almost 3 orders of magnitude lower than that of electrons. Further prolongation of the arc on the next voltage peak is made possible due to the reduced density of hot air. That reduces the breakdown voltage. But then, the density of air is only lower, when it has had time to decompress.

Offline davekni

  • Executive Board Member
  • *******
  • Posts: 3000
  • Karma: +142/-2
  • Physicist, engineer (electronic), and hobbiest
    • View Profile
Re: DRSSTC with litz-wire primary and 40 x TO247 IGBT H-Bridge
« Reply #18 on: October 14, 2020, 06:28:32 AM »
I was thinking that sparks would at least initially have very small diameter, but can't find much data.  I certainly agree that it doesn't require a shock-wave or pressures on the 1-bar order to make loud sounds.  Did find some in a fascinating paper about guided arcs using a combination of laser ionization, Tesla coil, and DC-charged capacitor:

https://www.nature.com/articles/srep40063#Sec13

https://www.nature.com/articles/srep40063.epdf?sharing_token=JDLHHYZVYzCG0d7hlCcCH9RgN0jAjWel9jnR3ZoTv0MZG0hFlwgOOmu-SOkfE9nUFnd8OwFqE3wiSr7C1PWaFFK3t4fVTu04iR3C9ZYbsu3l6JLCWUzq75bgRivVCyQlrenkaniEi3TUJx_TI-uJcQ%3D%3D


And especially this supplement with details about Tesla coil arcs with a very nice detailed image:

https://static-content.springer.com/esm/art%3A10.1038%2Fsrep40063/MediaObjects/41598_2017_BFsrep40063_MOESM292_ESM.pdf

Decided to take a quick image of a simple spark-coil spark to see if I could get some idea of size.  The screw in the images is 32 threads/inch, so about 0.8mm per thread (for size reference to compare with spark diameter):



Spark coil pulses are fairly slow, ~100us, especially from this large ancient coil intended for mechanical interruption.  Haven't scoped the above specific case.  I also don't know if there was much image bloom in this cheap low-res USB microscope image.

I'd be happy to share more details if you want to tweak simulation.  What's missing that is useful?  Probably best for me to edit the initial post with appropriate missing details.

David Knierim

Offline Uspring

  • High Voltage Engineer
  • ****
  • Posts: 348
  • Karma: +32/-1
  • Physicist
    • View Profile
Re: DRSSTC with litz-wire primary and 40 x TO247 IGBT H-Bridge
« Reply #19 on: October 17, 2020, 03:26:32 PM »
Thank you for the links. It's the first time I've seen images from arcs growing from a TC captured at that rate.

Quote
I was thinking that sparks would at least initially have very small diameter, but can't find much data.

Yes, they are thin and your picture is a good example of this.
Hot plasma channels tend to be thin due to the strong dependence of conductivity on temperature. The current will flow, where the resistance is least and that is in the center of the arc. That will cause the heating to be there. Also cooling is very effective initially, since the plasma channel is surrounded by cold air. If the arc lasts longer, than the heat will spread out, which leads to thicker arcs, such as in Jacobs ladders. Short burst TC arcs are thin, QCW ones a bit thicker and more of that will be the case for your proposed project here.
https://highvoltageforum.net/index.php?topic=1268.msg9305#msg9305

Quote
I'd be happy to share more details if you want to tweak simulation.  What's missing that is useful?  Probably best for me to edit the initial post with appropriate missing details.

I've been missing the top load capacitance or alternatively the secondary res frequency (uncoupled). I also need coupling, voltage amplitude to the primary, an estimate to the primary loss resistance and the distance to which your breakout rod is extending beyond the rim of the toroid. You seem to have made some changes to the coil recently, so I don't know, if the parameters here https://highvoltageforum.net/index.php?topic=798.msg5319#msg5319 are still correct.

« Last Edit: October 17, 2020, 04:07:27 PM by Uspring »

Offline davekni

  • Executive Board Member
  • *******
  • Posts: 3000
  • Karma: +142/-2
  • Physicist, engineer (electronic), and hobbiest
    • View Profile
Re: DRSSTC with litz-wire primary and 40 x TO247 IGBT H-Bridge
« Reply #20 on: October 17, 2020, 08:23:11 PM »
I have two secondary-only resonant frequencies in my notes.  The one I think is more reliably is 93.65kHz, which implies 42.5pF on the 68.0mH secondary inductance.  Also have 92.2kHz noted once, but don't recall that test.

Coupling is about 0.140 or 0.141.  I've had trouble getting really consistent measurements of coupling.  Was originally 0.157, but had arcs to the secondary, so had to drop the primary coil position a bit.

I've been using 0.018 ohms for total primary series resistance in my simulations.  DC resistance is lower than that.

Bus voltage is adjustable, in the range of 300V to 400V for all tests so far.  I think it was 330V for the last scope trace I shared.

I've experimented with many breakout points.  The latest one for 3m arcs was an angled foil cone with 20mm of 0.5mm diameter tungsten rod protruding from the cone.  The tip was 60mm above the toroid top and perhaps 30mm beyond radially.  I didn't measure precisely.  It was tiny compared to the toroid size.  Worked better for long arcs than my previous breakout that was around 600mm long.

MMC is larger for this long-arc test, 690nF.  Other parameters are still good, 6.95uH primary including leads, 826mm OD toroid, etc.
David Knierim

Offline davekni

  • Executive Board Member
  • *******
  • Posts: 3000
  • Karma: +142/-2
  • Physicist, engineer (electronic), and hobbiest
    • View Profile
Re: DRSSTC with litz-wire primary and 40 x TO247 IGBT H-Bridge
« Reply #21 on: July 14, 2024, 06:03:07 AM »
Question about DRSSTC coupling factor and racing sparks, prompted by last week's issues with this coil:

It's generally understood that high coupling can contribute to racing sparks.  I've presumed this is due to higher secondary volts/turn induced directly by primary in region of secondary adjacent primary (in addition to voltage due to secondary resonance).  If that's accurate, then a large diameter primary with same coupling should reduce racing sparks.  For same total coupling, primary field is inducing less secondary volts/turn, but over more turns, to achieve same coupling.

Any theory or evidence to suggest other than above?


Reason for my question is a problem running this DRSSTC last week (July 4th, 2024).  At start I had many strikes to ground guard rings around primary.  (Caused at least in part by a couple top load defects - foil tape corners slightly loose.)  Before long these ground ring strikes triggered some racing sparks and then some strikes to primary.  Not sure if racing sparks were triggered due to local ionized air and/or voltage spikes due to guard ring ground connection inductance.  Any thoughts?  Unfortunately no video running.

I pressed down top-load tape corners, lowered primary to reduce coupling, reduced Vbus a little, and increased breakout point length (on top-center of top load).  Coil worked for rest of show, ~75 minutes including a few short breaks to allow primary to cool a bit.

But my primary overheated more than usual, turning silicone rubber tan/brown:



Smelled terrible by end of show.  My initial theory was that early primary strikes damaged litz wire insulation, shorting strands and increasing loss.  Primary Q still measures fine today, ~1500 at 100kHz.  Next thought was that charred silicone rubber was conductive enough to heat.  I can't measure any conductivity using 10kV supply (<100nA).  But that is today at room temperature, not hot during operation.  My final thought is that lower coupling just required more RMS primary current for same total coil power (which was running 8-9kW much of the time).  Any other thoughts?

So I'm planning to make a new primary using 19mm copper tube that I have.  Larger diameter and conical.  Per JavaTC, I can get same coupling with a larger diameter primary combined with higher average primary position relative to secondary bottom due to conical shape.  That's why I'm asking about racing sparks.  Existing coil is marginal for avoiding racing sparks.  It didn't take much to trigger them last week.

Any input is welcome!
« Last Edit: July 14, 2024, 06:49:47 AM by davekni »
David Knierim

Offline NyaaX_X

  • High Voltage Technician
  • ***
  • Posts: 133
  • Karma: +2/-0
    • View Profile
Re: DRSSTC with litz-wire primary and 40 x TO247 IGBT H-Bridge
« Reply #22 on: July 15, 2024, 03:03:26 PM »
Hello~

Sorry for my interruption :P

My final thought is that lower coupling just required more RMS primary current for same total coil power (which was running 8-9kW much of the time).

That make me think of I tried to increase the wireless power transfer charging gap and maintain the transferring rate of power. Use the same pair of Tx and Rx flat-coil with ferrite, litz wire and tune the series capacitors to match the resonant frequency. Then I got the more heating coils maybe 20°C more, the inverter MOSFETs also getting hotter. There was around 2 times Ipeak-peak in the Tx coil compared to low distance charging. I think the total efficiency decrease 6%~8%, the coupling from 0.6+ to 0.3+, and the DC input current increase ~8%.
« Last Edit: July 15, 2024, 03:08:16 PM by NyaaX_X »

Online Mads Barnkob

  • Administrator
  • Executive Board Member
  • *****
  • Posts: 2359
  • Karma: +78/-0
  • Denmark
    • View Profile
    • Kaizer Power Electronics
Re: DRSSTC with litz-wire primary and 40 x TO247 IGBT H-Bridge
« Reply #23 on: July 15, 2024, 09:26:42 PM »
High coupling or bad tuning, is the main culprits of racing sparks.

I have experienced racing sparks on my DRSSTC1, when using it indoor in a regular height room, the close vicinity of a ceiling was simply enough to affect the resonant frequency. I had to add two deck of cards underneath the secondary to get it to run there. Same coil, without any adjusted coupling, ran fine outside in the free on top of a tent, for 3 days in a row.

Environment with close quarters or humid air, has given me the greatest trouble over time :)
https://kaizerpowerelectronics.dk - Tesla coils, high voltage, pulse power, audio and general electronics
https://www.youtube.com/KaizerPowerElectronicsDk60/join - Please consider supporting the forum, websites and youtube channel!

Offline davekni

  • Executive Board Member
  • *******
  • Posts: 3000
  • Karma: +142/-2
  • Physicist, engineer (electronic), and hobbiest
    • View Profile
Re: DRSSTC with litz-wire primary and 40 x TO247 IGBT H-Bridge
« Reply #24 on: July 18, 2024, 07:47:24 PM »
Quote
High coupling or bad tuning, is the main culprits of racing sparks.

I have experienced racing sparks on my DRSSTC1, when using it indoor in a regular height room, the close vicinity of a ceiling was simply enough to affect the resonant frequency.
Interesting.  From a theoretical view I can see tuning being a factor.  For this coil, I've experimented with a very wide range of tuning without any racing sparks.  This was my first case of racing sparks since original build 5 years ago.  (Initially had coupling at 0.157.  Immediately reduced to coupling to 0.141 to avoid racing sparks.)

Quote
Environment with close quarters or humid air, has given me the greatest trouble over time :)
Humidity wasn't issue in this case, but is much more of a factor than I'd realized.  My ZVS driven coil (different project) had a massive racing spark issue this Halloween.  Set it up in my garage as I'd done before.  This time outside weather was rainy.  Worked great in initial testing including after opening garage door.  Left door open for a couple hours until trick-or-treaters started showing up.  Next run was immediately swamped by racing sparks.
« Last Edit: November 03, 2024, 01:33:36 AM by davekni »
David Knierim

Offline davekni

  • Executive Board Member
  • *******
  • Posts: 3000
  • Karma: +142/-2
  • Physicist, engineer (electronic), and hobbiest
    • View Profile
Re: DRSSTC with litz-wire primary and 40 x TO247 IGBT H-Bridge
« Reply #25 on: November 03, 2024, 01:40:36 AM »
Made a new primary for this DRSSTC, from 19mm copper pipe instead of litz wire.  Larger diameter and conical.  (Outer two rings are grounded strike rail.  Primary is inner 4-turn conical coil.)



Coupling factor is 0.16, up from 0.141 with previous primary.  (Had been 0.157 in very initial build with primary positioned higher.  Lowered primary to fix racing sparks after initial testing.)  No hint of racing sparks with this new primary.  Supports the theory that racing sparks are caused at least in part by the induced secondary volts/turn.  Larger primary positioned higher (on average since conical) can have higher coupling without higher induced volts/turn.  Induced secondary voltage is spread across more turns.  Higher total induced voltage but no more (or perhaps less) volts/turn.  Performance is as good as ever:
   
/>
David Knierim

Offline Benjamin Lockhart

  • High Voltage Technician
  • ***
  • Posts: 145
  • Karma: +15/-0
    • View Profile
Re: DRSSTC with litz-wire primary and 40 x TO247 IGBT H-Bridge
« Reply #26 on: November 03, 2024, 02:00:35 AM »
I'm curious if faster current rise in the primary (low surge impedance) contributes to higher volts/turn in the the secondary from higher di/dt across the secondary inductance?

That's interesting that the conical primary can have higher coupling without racing sparks.

Offline davekni

  • Executive Board Member
  • *******
  • Posts: 3000
  • Karma: +142/-2
  • Physicist, engineer (electronic), and hobbiest
    • View Profile
Re: DRSSTC with litz-wire primary and 40 x TO247 IGBT H-Bridge
« Reply #27 on: November 03, 2024, 03:17:00 AM »
Quote
I'm curious if faster current rise in the primary (low surge impedance) contributes to higher volts/turn in the the secondary from higher di/dt across the secondary inductance?
Yes, I think so.  This coil has low impedance on both primary and secondary compared to typical.  (Finished it just before finding this forum or knowing what normal/ideal impedances might be.)  I suspect my low impedance makes racing sparks harder to avoid.

Quote
That's interesting that the conical primary can have higher coupling without racing sparks.
Larger diameter is key to reducing induced secondary volts/turn.  Conical allowed largest diameter turns to be higher to increase coupling.  Increasing diameter alone without changing any other parameters/positions will decrease coupling.  Conical allowed increasing coupling to compensate and a little more.  Hard to separate all factors in this one experiment.  Changing from litz wire to large diameter tubing also increased coupling, as did spreading out the four primary turns farther from each other relative to litz wire primary.  Design goals for this new primary were minimizing volts/turn and maximizing coupling (and keeping higher voltage parts of secondary not too close to primary).
David Knierim

High Voltage Forum

Re: DRSSTC with litz-wire primary and 40 x TO247 IGBT H-Bridge
« Reply #27 on: November 03, 2024, 03:17:00 AM »

 


* Recent Topics and Posts

post im new to tesla coils and i think i screwed up
[Spark Gap Tesla Coils (SGTC)]
noahbmaker
Today at 08:04:14 PM
post Re: Measuring the coherence length of a laser
[Light, Lasers and Optics]
alan sailer
Today at 07:47:52 PM
post Re: Measuring the coherence length of a laser
[Light, Lasers and Optics]
Uspring
Today at 07:34:38 PM
post Re: Measuring the coherence length of a laser
[Light, Lasers and Optics]
haversin
Today at 05:48:16 PM
post Re: Big Coil Build Log
[Dual Resonant Solid State Tesla coils (DRSSTC)]
flyingperson23
Today at 03:20:08 PM
post Re: Measuring the coherence length of a laser
[Light, Lasers and Optics]
Uspring
Today at 01:50:30 PM
post Re: Welcome new members, come say hello and tell a little about yourself :)
[General Chat]
shBLOCK
Today at 07:39:21 AM
post Re: First DRSSTC, Full Bridge PCB & IGBT Selection question.
[Dual Resonant Solid State Tesla coils (DRSSTC)]
Benjamin Lockhart
Today at 04:14:03 AM
post Re: First DRSSTC, Full Bridge PCB & IGBT Selection question.
[Dual Resonant Solid State Tesla coils (DRSSTC)]
davekni
Today at 01:12:38 AM
post Re: Testing and teardown of a commercial induction heater
[Electronic Circuits]
dejuli2
December 11, 2024, 09:58:07 PM
post Re: Push Pull VTTC
[Vacuum Tube Tesla Coils (VTTC)]
myoniwy
December 11, 2024, 09:26:40 PM
post Re: Welcome new members, come say hello and tell a little about yourself :)
[General Chat]
myoniwy
December 11, 2024, 09:08:24 PM
post Re: Measuring the coherence length of a laser
[Light, Lasers and Optics]
haversin
December 11, 2024, 08:05:20 PM
post Re: First DRSSTC, Full Bridge PCB & IGBT Selection question.
[Dual Resonant Solid State Tesla coils (DRSSTC)]
Beggernator.
December 11, 2024, 03:48:40 PM
post Re: APC Back-UPS Pro 1500 Watt Teardown
[Electronic Circuits]
Mr.Cas
December 11, 2024, 05:27:48 AM
post Re: First DRSSTC, Full Bridge PCB & IGBT Selection question.
[Dual Resonant Solid State Tesla coils (DRSSTC)]
davekni
December 11, 2024, 01:41:52 AM
post Re: First DRSSTC, Full Bridge PCB & IGBT Selection question.
[Dual Resonant Solid State Tesla coils (DRSSTC)]
Beggernator.
December 10, 2024, 04:18:23 PM
post Re: Ferrite transformer from Microwave
[Transformer (Ferrite Core)]
davekni
December 10, 2024, 04:05:07 AM
post Attracting (and picking up) nonferrous metals with an electromagnet
[Electronic Circuits]
davekni
December 10, 2024, 02:49:03 AM
post Ferrite transformer from Microwave
[Transformer (Ferrite Core)]
Simranjit
December 09, 2024, 11:37:35 PM
post Re: Measuring the coherence length of a laser
[Light, Lasers and Optics]
alan sailer
December 07, 2024, 09:41:37 PM
post Re: Measuring the coherence length of a laser
[Light, Lasers and Optics]
haversin
December 07, 2024, 07:07:49 PM
post Re: Measuring the coherence length of a laser
[Light, Lasers and Optics]
alan sailer
December 07, 2024, 06:33:32 PM
post Re: Measuring the coherence length of a laser
[Light, Lasers and Optics]
haversin
December 07, 2024, 06:13:56 PM
post Re: Measuring the coherence length of a laser
[Light, Lasers and Optics]
alan sailer
December 07, 2024, 02:40:55 AM
post Measuring the coherence length of a laser
[Light, Lasers and Optics]
haversin
December 07, 2024, 01:07:46 AM
post Re: Single board for SSTC and DRSSTC operation
[Solid State Tesla Coils (SSTC)]
davekni
December 07, 2024, 12:18:49 AM
post Re: Single board for SSTC and DRSSTC operation
[Solid State Tesla Coils (SSTC)]
Simranjit
December 06, 2024, 11:59:05 PM
post Re: First DRSSTC, Full Bridge PCB & IGBT Selection question.
[Dual Resonant Solid State Tesla coils (DRSSTC)]
davekni
December 06, 2024, 11:33:05 PM
post Re: Single board for SSTC and DRSSTC operation
[Solid State Tesla Coils (SSTC)]
davekni
December 06, 2024, 11:28:25 PM
post Re: Welcome new members, come say hello and tell a little about yourself :)
[General Chat]
GaelJaton
December 06, 2024, 09:47:04 PM
post Re: First DRSSTC, Full Bridge PCB & IGBT Selection question.
[Dual Resonant Solid State Tesla coils (DRSSTC)]
Benjamin Lockhart
December 06, 2024, 08:06:27 PM
post Re: 160mm DRSSTC II project | Questions
[Dual Resonant Solid State Tesla coils (DRSSTC)]
Late
December 06, 2024, 07:54:01 PM

Sitemap 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 
SimplePortal 2.3.6 © 2008-2014, SimplePortal