A friend mentioned a fascinating thought:  He is contemplating a future project of a high-voltage squirt gun to make water-guided arcs.  That got me thinking about options for squirting water from the top of a Tesla coil and ways to avoid getting the base wet.  Also thinking in general about water streams and conductivity.  Surface tension of water usually makes thin streams (ie. from squirt gun) break up into drops.  Series of drops would be more conductive than plain air, but require rather high voltage to bridge all the gaps.  Adding a little surfactant to the water reduces surface tension, so should help make a more continuous stream.  High molecular weight water soluble polymers (tiny amount) would further help keep stream connected.  That combination of surfactant and trace of polymer are the key ingredients of soap bubble solution.  Many surfactants are sodium salts, so would color the arc yellow.  Down side is that any additives to the water make protecting the coil base more important, as mist or splashes would be more conductive and leave residue rather than evaporating completely.

Another variation would be to have a water stream arch spaced away from a Tesla coil to form a strike target.  Might be safer for the coil.   But I could also imagine arc causing water splatter when it hits stream.

Has anyone tried arcs guided by water streams?  It is a new item for my list of some-day projects.

For years I've pondered the trade-offs between GDTs (simple) and isolated gate driver chips (more complex, including isolated supplies for each gate).  For my new QCW experiment platform, decided on a compromise: GDT with simple output buffer.  Rising edge behavior is standard GDT output through damping resistor.  Falling edge is a local PFET shorting Vge.

This is intended for IGBTs with no internal gate resistors (all TO247 devices I've seen and some smaller bricks).  For these IGBTs, there is no need for negative Vge as long as gate drive impedance is low.  In normal GDT drive, Vge falling edge speed is limited by GDT leakage inductance.  With buffer, GDT has light load on falling edges, so is fast.  For high frequency CW applications where gate power is a significant concern, avoiding negative Vge reduces power by ideally 4x.  Reality here is about 3x.

Posting this under "DRSSTC" category, though it could be useful for induction heating or other applications.

Buffer schematic.  One needed for each IGBT (or parallel set of IGBTs):



R1 is the normal rising-edge damping resistor for GDT output, 9.4 ohms in this example.  R1 has a series diode D1 to prevent conduction when GDT output is negative (when Vge is zero).  PFET M1 (IRFZ24) provides low-impedance shorting of Vge when GDT output is negative.  D2 limits PFET Vgs to spec +-20V.  Would likely work fine without D2.  R2 provides damping for PFET Vgs and limits current through D2 during any GDT undershoot.  I use this circuit with +-19V into GDT.  D3 clamps Vge under any anomalous conditions, especially ESD during construction and handling.

Here's my first H-Bridge using four of above buffer circuits and two GDTs (one per half-bridge).  Circuit is simple enough to construct on raw double-sided copper-clad board and dremel-tool cuts.  IRFZ24's are surface-mounted on one side and remaining parts on other side.  R1 is an array of 1206 SMD resistors here to handle power of continuous 500kHz operation.  Power is significant even though much less than it would be with +-19Vge drive.  IGBTs here are parallel pairs of FGH75T65SHDTLN4, which are in 4-lead TO247 packages (Kelvin emitter lead).  Power ECB is per my low-parasitic-inductance tutorial with bypass caps on back side.



Only issue so far with actual use is that these IGBTs have roughly the same turn-on and turn-off times, at least in my testing.  With this buffer's very fast turn-off (~20ns from UD output to Vge falling), turn-on delay (dead time) can't be set low enough.  Reducing R1 below 9.4 ohms causes more overshoot on Vge.  That would be OK for low duty cycles, but increases gate drive power dissipation for CW cases.  If excess dead-time becomes too problematic, I'll need to make new GDTs with even lower leakage inductance to allow lower R1 values.

Below are a bunch of scope captures.  Including all these for a second purpose, as an explanation of H-bridge output triple-transitions.  Ideally, I'd get rid of triple-transitions by reducing dead-time.  Only other option is to increase phase lead.  However, at ~480kHz, additional phase lead causes IGBT turn-off at high current, since current slew rate is so high.

For all below scope captures:
Ch1 (black) is H-Bridge output, differential between two outputs, 20V/div.
Ch2 (green) is H-Bridge (primary) current at 50A/div.
Ch3 (red) is Vge at 5V/div.
Ch4 (cyan) is one output from UD-like driver.

Near end of a small burst at +-190A:









Notice the triple-transitions in H-bridge output.  Initial transition is at IGBT turn-off, while that IGBT pair had been conducting current.  A bit later current reverses polarity, pulling H-bridge output voltage back in the opposite direction.  As voltage gets about half-way back, opposite IGBT pair turns on, pulling H-bridge output voltage to desired new state.


Start of burst where current is low:









Notice that above low-current captures have no triple-transition.  H-bridge output transitions only after opposite IGBT pair turns on.  That's because there is insufficient current to cause a voltage transition at turn-off.  Below is a capture at mid-current, enough current for output transition at IGBT turn-off, but barely enough to cause a hint of triple-transition:




Finally, here's a capture at the end of a burst where residual primary current is causing H-bridge output transitions.  GDT input is at 0V (both GDT inputs high in my case rather than more normal both-low).  IGBT Crss (collector-to-gate capacitance) causes small swings in Vge.  Positive swing is limited by PFET threshold voltage.  PFET must be selected to have threshold voltage well below IGBT's threshold voltage.

Purchased a cheap string of 1080 LEDs:
https://www.amazon.com/dp/B0B8JSQFMB?ref=ppx_pop_dt_b_product_details&th=1
(BTW, amazing how prices bounce around.  Was $24.99 a couple weeks ago.  Now $55.99.  When searching earlier, many strings were ~$50.  Purchased this one because it was cheap.)

String runs on +-30V provided by a small supply/controller, drawing +-200mA.  String is wired 9S120P, with 60 of each 120 connected reverse polarity.  Allows effects such as fading from one half to the other.  Down-side is that "steady-on" mode is slightly under 50% duty cycle, not as bright as when one half or the other are lit.  255Hz chop frequency, 100us dead time one direction, 140us the other direction.

LEDs are all blue.  Phosphor converts blue to red, yellow, and green.  Less saturated color than with normal LEDs.  Advantage is that identical forward voltage allows direct paralleling.  Appears to have no per-LED series ballast resistors.  Ballast is provided by thin copper interconnect, total around 27 ohms for the string.  Wiring keeps all LED voltages matched:


I plan to make a higher-power supply, steady-on mode only.  About +-38V increases current from 200mA to 400mA.  Efficiency of red and blue drops as current increases.  Unfortunate that DC powers only half the LEDs.  200mA DC would be better than 50% duty cycle of 400mA.  However, green slightly increases efficiency as current increases.  Not sure if phosphor is non-linear, or if the change in blue wavelength with current matches phosphor better.  Yellow efficiency is roughly linear with current.

Made LED strings specifically for use in AC electric fields such as near Tesla coils.  Idea is to replace more fragile fluorescent tubes.

/>
Standard LED replacements for fluorescent tubes work OK, but are lower voltage and higher current than is ideal for Tesla coil use.  Most emit light on only one side.  So I made LED strings as an alternative, with different colors on the two sides.  They are visible under HV power transmission lines, and of course brighter in the higher-frequency field of Tesla coils.

One option is to connect pairs of LEDs anti-parallel.  Made a couple small ones that way previously.  However, each LED gets only half the total current (either positive or negative half-cycle).  To improve current efficiency, these strings use signal diodes in a bridge rectifier connection for every 20 or 24 series-connected LEDs.  That way full current passes through each LED.  Pitch is one LED per 5mm (one per 10mm on each side), for 100 and 120mm long sections.

One feature didn't work very well:  A goal of sectioning the strings was to mimic behavior of fluorescent tubes where sliding one hand up the tube prevents the section between hands from glowing significantly.  My issue appears to be insufficient capacitance from LED strings to outside of plastic tube (hand), especially for my SMD version.  LED ECBs are only 7mm x 120mm inside a 16mm OD tube.  Next version needs to have some metal closer to tube walls (and perhaps larger-diameter tube) to increase capacitance.  Not sure how to best accomplish this while minimize light blocking.  The three commercial LED tubes connected in series worked well in this regard, but in large steps of their ~300mm tube length.  I think their opaque back side appears to have metal just under plastic outer layer, so plenty of capacitance.

If there's any interest, I'll post ECB files.  SMD version uses 2835 LEDs and SOT23 series-pair diodes (MMBD1203).  Includes anti-parallel diodes across every 3 LEDs to avoid any reverse-current in case capacitance from hand introduces AC to middle of the string.  Not at all sure such is necessary.  One could populate just the two diode packages, one at each end, that form the rectifier bridge.  (BTW, this is how direct-replacement LED tubes work.  One diode pair at each end of tube rectify incoming AC from ballast.)  Through-hole version is for 5mm LEDs.  Diodes are same, still SMD.  I plan to make an updated SMD version with multiple test-pad holes for the internal DC rails for adding external wires to increase capacitance.

I'd be happy to send MMBD1203 diodes to anyone who needs them.  Saw a good EBay deal for left-over Fairchild parts originally purchased from Mouser.  I now have ~6k parts and no need for that many.

I've been looking through TO247 IGBTs available through normal electronics distribution, focusing on fast parts suitable for phase-shift QCW.  Also looking for large internal diode to handle pulse-skip.  Ended up purchasing FGH75T65SHDTLN4 ($3.88 each, a bit less in quantity):
https://www.arrow.com/en/products/fgh75t65shdtln4/on-semiconductor
Cheap because Arrow is closing out this discontinued part.  83 parts left there and 450 at Rochester Electronics (a bit more expensive) last I checked.

650V, 75A at 100C case (both IGBT and diode), 300A pulse, 4-pin TO247 (kelvin emitter connection).

FGH75T65SHDTLN4 has only a preliminary data sheet from 2018.  Short time in production.  I can't find any difference between this and the same part without 'N' character, FGH75T65SHDTL4 (from 2015 and still active).

A very similar part FGH75T65SHD-F155 appears to have the same IGBT die paired with a faster diode die, but diode is rated only 50A at 100C case.  It's available from Mouser for $6.55 each.  Would be better if any hard turn-on switching is needed.  This part is in the more-typical 3-pin TO247 package.

No personal experience yet.  Just posting what I purchased after some data-sheet searching.

Other fast IGBT suggestions?  Even though I've already purchased these parts, it would be interesting to see other recommendations.

Looking for ways to measure surface tension of surfactant/water mixtures, I ran across capillary waves as a measurement technique.  Read a paper about a smart-phone app for that using phone's vibrator, flash, and camera.  It has a few drawbacks, however.  Key one for me is that I don't own a smart phone

Since I had some small 25mm diameter off-axis parabolic mirrors, decided to make a small telecentric Schlieren setup to view capillary waves in plastic petri dishes.  A stepper motor driver generates sine wave current feeding a small E-core coil.  Coil field vibrates a permanent magnet taped to petri dish, generating capillary waves in water inside dish.  LED pulses synchronously with sine wave, imaging capillary waves at the same phase each cycle.  (Red LED to match gold-plated mirrors.)  A 2mm pitch scale below the dish provides a measurement reference.  Due to telecentric optics, scale is viewed at the same magnification as waves.  Frequency is manually adjusted to make 2mm wavelength.  Measured frequency is used to calculate surface tension using formula (edited from Wikipedia):
        surface_tension = density * frequency^2 * wavelength^3 / (2 * PI)

The entire setup, measuring water at 242Hz:


Side view to show optical path.  LED reflects down to lower mirror, forming parallel beam up towards upper mirror.  Upper mirror images parallel beam to small rectangular aperture (hole in black paper) above LED.  Lens just past aperture (behind wood frame) images the object space (parallel beam) onto a finite-distance screen.  (Would otherwise image at infinity.)  The aperture is reverse-imaged at infinity by the upper mirror, which makes the setup telecentric.


With pure water (low viscosity), the image is quite clear as shown above.  With detergent-water, viscosity is higher, which damps waves.  It still works, but is a bit harder to view and adjust accurately to 2mm wavelength.

Even though the result is fairly simple, this project took much longer than I'd expected - especially trying different ways to vibrate the petri dish or vibrate an object within the dish.  Advantage of vibrating the dish is avoiding another surface that needs cleaning between testing different surfactant solutions.  Final goal is to optimize surfactant solutions for making bubbles.  Perhaps became more of a project than this goal warrants.  Measuring bubble film strength may be more relevant than surface tension.

Hopefully this isn't wasted redundant work.  I've seen many builds with good GDT construction, using both halves of each twisted pair.  However, all the GDT construction guides I've found use a single primary wire.  Below is construction of a half-bridge GDT using two twisted pairs from CAT5 cable.  This is easier to see in pictures.  Extension to full-bridge isn't difficult.  Use four pairs, with four paralleled windings for primary, one wire from each pair.  (BTW, CAT5 isn't necessary.  If starting with single wires, twist pairs together, and use those pairs exactly as you would with pairs extracted from CAT5 cable.)

Start with a suitable high-permeability ferrite core.  (Some options listed at the end.)  The core I'm using here works, although the shape is not typical.  Pay attention to winding technique, not so much to core shape.

Wind two (for half-bridge) twisted pairs around the core.  More discussion about the number of turns to follow.  I'm using 4 turns here.  Mark the starting end of all four wires for later identification.  Here I've "marked" starting ends by length (short), with the tail ends left longer.  Other (preferred) options are to add bits of tape to the starting ends or strip insulation from the starting ends only.





Untwist the pairs almost to the core:



Twist each wire (each winding) with itself all the way back to where the pair twisting starts.  Don't leave any significant loop area of untwisted wire:



Pair one winding of each pair together for the primary.  I've chosen the lighter-color wire of each pair for simplicity, white and light-blue.  Most important: connect the two starting ends (short ends or stripped ends or however you marked them) together, then the two tail ends together.  If pairing is swapped, driver can be damaged by the shorted load.  (Test at very-low duty cycle initially just in case of error.)  The remaining winding of each pair is for an IGBT.  Starting end of one IGBT winding is gate.  Starting end of other IGBT winding is emitter of other IGBT.



Since I hadn't used tape to identify starting ends at the beginning, I added tape now.  Then cut the tail ends to length and strip.  Connect the two primary tail ends together:



The reason for the twisting is to minimize leakage inductance.  Leakage inductance slows down gate waveforms and causes overshoot and undershoot and generally-sloppy gate drive.  Twisting forces the wires to remain close together with little loop area between wires for magnetic field to slip through.  Best to maintain this pairing all the way to the driver for primary and all the way to gate and emitter terminals of IGBTs (or FETs) for the secondaries.  Avoid excess loop area when adding gate series resistors.  Keep the emitter wire adjacent the resistor to minimize loop area.

Now for a bit about cores and turns.  I'll add a second post on measuring cores and finished GDTs, a bit more advanced topic.
Toroid shape is generally preferred, but E-cores work if ungapped (no air gap or spacer between the two halves).
Most important two parameters are:
     Core material (reasonably-high permeability and saturation flux density).
     Core cross-sectional area.  Picture the area of a core slice inside one turn of the GDT winding.
Iron and other compressed-powder cores never work well.  Most (but not all) ferrite materials are OK.
Low-frequency EMI suppression cores are reasonable.  That is what I used above.  Most larger EMI cores are low-frequency, so workable.  This includes common-mode chokes found in power supplies.  (Remove existing windings.)  Such EMI materials include:
3C11, 3E6, 3E12, 3E10, 3E15, 3E25, 3E26, 3E27, 3E65
More ideal ferrite materials are generally designed for switching power supply transformers etc.  These include:
PC40, PC200
N27, N30, N35, N41, N49, N51, N72, N87, N88, N92, N95, N96, N97
T35, T37, T37, T38, T46, T57, T65, T66
3C90 through 3C97

Concerning cross-sectional area, more is better, within constraints of fitting the GDT mechanically into the build.  The core I used above is 28mm long, 14mm ID, 28mm OD.  The ring is 7mm thick (0.5 * (OD - ID)).  So cross-sectional area is 7mm thick * 28mm long = 196mm^2.  If using a more-typical ring toroid, the formula includes a factor of PI/4:  Area = length * 0.5 * (OD - ID) * PI / 4.

In general, look for cross-sectional area to be at 50mm^2 or more.  A bit smaller is fine for high-frequency Tesla coils.  Larger for low-frequency coils.  10 turns is usually plenty.  Excess turns increases wire length and therefore leakage inductance.  The above example is 4 turns.

(All my GDTs have either 2 turns or 3 turns.  That works with large area cores and careful measurement to make sure its enough.  A few more turns is generally safer.  Too many turns causes subtle issues due to leakage inductance.  Too few turns is more catastrophic, possibly damaging the driver and/or IGBTs when the core saturates.)

Here's a great picture from Mads of a full H-Bridge GDT constructed this way:
https://highvoltageforum.net/index.php?topic=1856.msg13969#msg13969

A couple other posts with images that aren't quite as obvious.  For this first, look at the upper GDT wound with CAT5 cable including jacket:
https://highvoltageforum.net/index.php?topic=588.msg3779#msg3779
And this 3-turn GDT from my bridge tutorial:
https://highvoltageforum.net/index.php?topic=1324.msg9886#msg9886

About 15 years ago I built a couple simple continuity checkers, one for home and one for work.  They apply low voltage to avoid forward diode junction conduction and light an LED when resistance is below about 10 ohms.  Power is from a single 18650 LiIon cell.  I find them very convenient - probing an ECB to see which side of a bypass capacitor is ground, checking for hidden solder bridges under parts, etc.  However, these draw enough quiescent current that an on/off switch is necessary.

Last weekend I made two new continuity checkers drawing only 10uA quiescent.  Also reduced threshold resistance to between 1 and 2 ohms.  No on/off switch to forget to switch and battery life of 5-10 years between charges.

Normal people would use a micro-power opamp for such a circuit.  Being abnormal and fond of discrete circuity, mine uses three transistors instead:



Dremel-tool cut circuit board:



Finished continuity checker.  Most of the volume is the 18650 LiIon cell.  Board and probe tip are taped to the cell.





The circuit can easily be made in reverse, using PNP transistors and an NFET.  (Actually, one of the two I made is reversed.)  Power supply is reversed too.  I prefer the version shown here, as one probe is the negative supply which is battery case.  Makes accidental shorts of the battery less likely.

Still in the planning/simulation stage, but time to seek input/suggestions.  This is an induction heating driver powered directly from rectified 240Vac line.  No bulk capacitance, so induction power will track line voltage.  Output will be ferrite-transformer coupled to work coil, for both isolation and to reduce voltage.

    Up to 10kW (for use on a 40A 240V circuit).
    ~100kHz frequency (perhaps 80kHz minimum, ~200kHz maximum)
    Resonant capacitors on primary side.  (Transformer handles resonant current.)
    IXYH24N170CV1 24A 1700V TO247 IGBTs, 4 total, 2 parallel pairs.
    SCT2H12NY 4A 1700V SiC FET for gate drive to IGBTs
    GD10MPS17H 10A 1700V SiC diodes across IGBTs (which have no internal diodes)
        (ZVS circuit ideally has no reverse IGBT voltage, but slow turn-off requires diodes for momentary conduction.)

My thought for the LARGE output transformer:
    4:1 ratio (with the idea to match voltage from 60V direct-drive ZVS systems).  Would love to learn more about inductance and frequency of typical work coils for <= 10kW power.
    Two sets of large U93/76/30 U-cores arranged to look like an even larger E-core set.
    4-turn primary made of 90mm x 0.2mm copper foil wound as four layers around E-core center.  Leads (copper pipe or bus-bar) exit one end of the E-core set.
    1-turn secondary, 90mm x 0.5mm copper, over (around) primary.
    Windings spaced radially to allow for axial forced-air cooling.
    Each work coil can be soldered to its own 1-turn primary if desired.  Core is separable, so new secondaries can be slid into place.
    90mm x 0.5mm secondary winding can be extended as parallel-plate leads in order to locate work coil farther from transformer without adding significant additional parasitic inductance.

For circuitry, here's my modifications to the simple ZVS oscillator in order to accommodate larger forward drop of HV devices while keeping Vge low voltage close to 0V:


Any feedback and suggestions would be appreciated.  Likely to be some time before this project gets to the top of my list.

Also thinking of a variation w/o ferrite transformer for driving my SSTC.

Finally having some success with a fixed-frequency HFSSTC.  Unlike most coils, gate drive is from a crystal oscillator (with amplification), not feedback from the drain circuit.  Frequency is the lowest ISM allocation at 6.78MHz.  Otherwise this is the same as other class-E circuits.

Power is somewhat limited so far, about 800 watts.  Hope to get a little higher.  With a fixed frequency, the coil must be tuned to achieve class-E operation before starting an arc.  Gate-drive frequency does not track increasing arc capacitance.  Larger arcs pull the coil farther out-of-tune.  The FET sees much more reactive power than real power.  To keep the FET within voltage and current ratings, FET reactive power can't get too high.  That makes real power even more limited.

I'm experimenting with a 4S2 (NiZn) ferrite for adjusting primary coil inductance.  Can't get more than about 1% frequency range (2% inductance range) before the ferrite gets too hot.  The clean (and more complex) solution to a fixed-frequency HFSSTC would be to use a PLL to start at a higher frequency, then lock to the desired fixed frequency once the arc is large enough.

Here's my simulation schematic.  The 3.3V pulse generator V3 simulates the crystal oscillator.  Amplifier stage is ZVS feeding a GDT.  GDT leakage inductance is the ZVS resonant inductance, with gate voltage opposite phase to GDT input due to this tuned leakage inductance.  FET is a 17A 1200V SiC NVH4L160N120SC1.


Breakout is a carbon rod following Steve Ward's example.  The arc wanders around the rod top and upper sides.  Not sure if this is due to the lower-than-typical frequency, or just that the breakout doesn't get hot enough for thermionic emission.  Any thoughts here?

/>
BTW, the backdrop in the video is a white wall with bright room illumination.  I've manually set camera exposure low to avoid arc wash-out.

Edit:  Fixed one schematic error.  I'd lowered breakout capacitance and forgot to add in the corresponding secondary coil capacitance, now as C5.