I would be glad to oblige.  See the attached PDF for a full schematic.

It is a pulsed DC output at 60 Hz.  For 1/2 the cycle, the capacitors are charged.  For the other half, the power supply is in series with the capacitors, and it delivers double the voltage.  This probably contributes to the sparks being a little more wild.

Also, I have included a template for making the wheel for the angle grinder.  One of the hardest parts is balancing the wheel because it is rotating so fast.  This will get you close, then you put it on the angle grinder and grind off the edge until it is round using a file.

Watch the video at 0.25 X speed.  It looks cool that way.  Next time I'm going to take a video using a high speed camera.

Dan

High voltage capacitors are hard to make.

The problem is that if there is air in the capacitor, you get this effect called "partial discharges."  Basically its tiny little sparks between the layers of dielectric that slowly eat away at the dielectric through oxidation because the air is ionized because of the high electric field.  The capacitor will be fine for a while, but after awhile it will arc through.  Capacitors are often filled with oil or wax to prevent this.  A vacuum needs to be applied for the air bubbles to be removed as well so there is no air left in the capacitor.

That said, I made a capacitor out of stacks of aluminum roof flashing and wrapped them in 3 mm thick multiple layers of polyethylene painters tarp and it continues to work.  I put the stack in a vise and clamped on it very hard to try to push the air out as much as I could.  Furthermore, I carefully trimmed the roof flashing to have rounded corners so there would no points that serve to concentrate the electric field.  But I used a lot of painters tarp, had very thick layers, and only got 26 nF capacitance.  It probably would have been cheaper had I just bought a MMC when I got done with it, but like a lot of things I just like to try things out.  My guess is that the dielectric you have is thin because the capacitance is high.  The breakdown field is going to be at best 20 kV/mm and probably a whole lot less once partial discharges start.   If you're using aluminum foil as your conductive layers, its probably crinkly, and the creases will act to concentrate the electric field.   So I'm guessing your capacitor may not last long.

You could try to get mineral oil and fill the pipe with that, or maybe even vegetable oil would have a high enough dielectric strength.  Heat the oil to get as much water and air out of it as you can before pouring it in, pour in it in to make sure no air gets in the pipe, and then use a plastic screw perhaps as a plug so that no air is introduced into the pipe when you seal it.  If you use oil, you may just want to use paper insulation instead, and let the oil be the dielectric.  But after all of that trouble, I think most opt for just putting together a MMC if that is an option.

Dan

I was reading and trying to understand what skip pulse operation is of a DRSSTC.  My understand is that ordinarily when the overcurrent detection occurs, the transistors are shut down when the next zero current crossing occurs, and the current in the half/full bridge is allowed to ring down by circulating through the freewheeling diodes.  With skip pulse operation, instead of having the transistors shut off until the next trigger comes from the interrupter, it allows the current to ring down until the current has dropped below the overcurrent level, and then the transistors to switch back on to allow the current to build up again, and this continues until the interrupter shuts off the pulse.  I could not find a schematic for a skip pulse operation, so I decided to study it myself.

I was playing around with the schematic of the universal driver to try to implement this (if this is indeed what skip pulse is).   The schematic I included rearranges a few things in the universal driver to do this but hopefully keeps the part count about the same.  In the old universal driver, the D-type flip flop U7A has its set input along as the interrupter pulse is active, unless it is shut off by ANDing with the output of the other flip-flop U7B, which is driven low if there is an overcurrent condition.  Each time the signal from the feedback transformer transitions from low to high, it would clock in a low signal, but this is overridden by the interrupter pulse (assuming it is still active and there's no overcurrent condition).   Once the interrupter pulse is over or overcurrent occurs, the feedback from the comparator clocks in a low signal at the low-to-high feedback transformer transition, shutting off the transistors.

I changed this by only applying a short negative pulse to the set input when the interrupter pulse transitions from low to high, the short pulse being generated by the network of U8C, so that the AND gates U5B/U5C allow the transistors to clock on.  Instead of the data input to U7A always being low, it is high as long as the interrupter pulse is active and there's no overcurrent condition.  Jumper J16, if it is closed, forces flip-flop U7B low which shuts off the AND gate and so the next low-to-high transition of the feedback transformer shuts off the transistors, and the transistors stay off until the next rising edge of the interrupter pulse, which is the usual non skip-pulse operation.  If jumper J16 is open, the transistors stay off only until the overcurrent condition is ended, and on the next low-to-high transition of the feedback transformer, the transistors are enabled again, and if overcurrent keeps happening, this cycle should continue until the interrupter pulse is shut off.  However, I kept the circuit that shuts off the transistors in case it doesn't otherwise occur from the interrupter pulse ending.

Anyways,  if someone could help me figure out what functionality embodies skip pulse DRSSTC I could work that into the design, in a way like I have modified the driver above.

Thanks,

Dan

Could you create a "dummy gate load" to connect to your GDT which consists of a 10 nF capacitor in series with a 2.5 ohm resistor to try to see if it can charge and discharge a capacitor with a low rise/fall time?  You could also attach the dummy load directly to your driver as well and see if it works differently to see if its the GDT.  This won't model the Miller capacitance but it might help you troubleshoot the driver/GDT setup.

I was looking at using a potentiometer with a capacitor to adjust phase lead.  I attached a simulation of the circuit, as well as a simulation of the conventional inductor circuit.  It seems to perform similarly, except that the phase lead decreases with frequency rather than increase of course.

Dan

Do the IRF540/IRF9540 get hot?  Being packed so tightly I am concerned that there is little opportunity for cooling.  I left them staggered in my layout so a heatsink could be added.

I was wondering how the TL3116 compares to using a 74HC14 as a comparator with hysteresis.  The TL3116 allows the capacitor in the hysteresis feedback path to be adjusted.  How important is this?  I was looking at the feedback circuit and it seems that the TL3116 can be replaced by a 74HC14 if one does not need to adjust this capacitor.   Take a look at the schematic to see how I changed it.  The whole schematic is attached as a PDF.

Dan

Hello,

I have added another board to the PCB pack I am working on, and fixed a few minor things.

I wanted to make a version of the universal driver that requires minimal or exotic proprietary parts because through-hole parts are harder to find, so I mixed and matched UD2.7C and UD1.3B parts, keeping the phase lead, but taking out the TL3116/LT1016 type op-amp and putting back the 74HC14 so the hysteresis is not adjustable.  Also, I made the UVLO a LM311 comparator rather than a specialized chip.  The only more specialized part left is the TC4427/MAX4427/IR4427/UD27423, and TC4427s are available from many sources.  I removed the older OPF2412T module from the board, and added 4N28 optoisolator control in a region that is separated and insulated from the rest of the board.  I found a minor flaw in the interrupter which I have fixed.  I have attached a copy of the schematic, and an archive of the now eight boards.

Dan

You really need more space

Is there any other way to adjust the phase lead than monitoring the bus voltage?  That is mainly what I need to do.   I am just looking for transients on the bus.

Dan