My First DRSSTC

[sjsimmo]

Thanks for the guidance

I just finished doing the calculations, and got the following values for my coil at 350A peak, 100uS on-time, 1000BPS:
- Fmax1=1290kHz
- Fmax2=984kHz
These values seem about 5 times bigger than what I was expecting. Am I doing something wrong, or are these IGBTs invincible?


Thanks again,
-sjsimmo

[Mads Barnkob]

1000 BPS is really extreme to calculate with, if you want long sparks I think its better to use longer on-time to transfer more power and maybe maximum 500 BPS, a 1 kHz tone is not that pleasant to listen to anyway

I looked at this datasheet: https://www.mouser.com/pdfdocs/FGY75N60SMD1.pdf

Did you extrapolate "Figure 15. Switching Loss vs. Collector Current" to get the real mJ losses for the higher current? Or where did you get those low values? Just by eye it looks more like it should be 10-30 mJ (before derating)

[sjsimmo]

I'm confused now, that is the datasheet I was using...


The calculated current at switch on is 11.6A, and switch off current is 22.7A. These values are both below the minimum current shown on the graph of figure 15. extending the graph gives a switching loss of about 0.25mJ and 0.4mJ at 80 degrees celsius respectively, or am I reading the graph wrong? 


These values can then be multiplied by 0.2, giving the final values that I wrote down.


The reason for using 1000BPS for the calculations, is because I want to midi modulate it eventually, and I figure that 1kHz should be a sensible limmit... (or is this actually too high?) . As for on-time, I might make that a bit longer, at 200A anything above 80uS has little effect on spark length, but I suppose at 350A this time will grow (although by how much, I'm not sure, since I can almost double bus voltage still.)

[Mads Barnkob]

I'm confused now, that is the datasheet I was using...


The calculated current at switch on is 11.6A, and switch off current is 22.7A. These values are both below the minimum current shown on the graph of figure 15. extending the graph gives a switching loss of about 0.25mJ and 0.4mJ at 80 degrees celsius respectively, or am I reading the graph wrong? 


These values can then be multiplied by 0.2, giving the final values that I wrote down.


The reason for using 1000BPS for the calculations, is because I want to midi modulate it eventually, and I figure that 1kHz should be a sensible limmit... (or is this actually too high?) . As for on-time, I might make that a bit longer, at 200A anything above 80uS has little effect on spark length, but I suppose at 350A this time will grow (although by how much, I'm not sure, since I can almost double bus voltage still.)

You are right, I jumped the first step there.

I will try your numbers in a excel spreadsheet I got on my PC at home, I will take a look at it tonight.

Edit: I used a bit more conservative extrapolations on the graphs, noticeably 6V CEsat and Ton 30mJ, Toff 20mJ, (derated by 0,05% on-time and 0,2 for pulses) and I am getting 144 kHz

[sjsimmo]

Thanks for doing the calculation  

It's interesting to see how much effect changing those values had on the maximum frequency, especially since they're all from reading the same graphs. I guess it's a pretty big guessing game once we go outside the values shown on the data sheets...

So should I proceed to turn up the current until I either get 350A, or an explosion? (going by our two wildly different results...)
Out of curiosity, how much can an IGBT overcurrent failure affect other components? Is it likely for example, to kill the UCCs?

I'm also a little concerned about the fact that my gate waveform still has about 500 nS rise/fall times. Whilst this is a lot better than before, I'm not sure what effects it will have on reliability at higher peak currents. Should I be concerned about this?

EDIT: I've attached a scope shot of my gate waveform (1uS/div)

[Fumeaux]

EDIT: I've attached a scope shot of my gate waveform (1uS/div)

You might already know about this page: http://www.richieburnett.co.uk/temp/gdt/gdt2.html
If not, it looks like your waveform is overdamped so i guess you should reduce the inline resistance.

The rest of the topic goes over my head so that's that.

[sjsimmo]

Yeah, I've seen that page. It's a really good resource 


I've previously tried reducing the gate resistance from 6.8R to 3.3R with no effect (it's still using the 3.3R). This resistance change was before I doubled up the UCCs which had dramatic improvements, resulting in the waveform in my previous post. Previously it was more of a sawtooth wave. I'd think that needing less than 3.3R would be very unusual, considering I'm using a TO-247 IGBT full-bridge.
One thought that occured to me is that maybe I need to run 3 UCCs in parallel, but this also seems a bit ridiculous, given the IGBTs that I'm using.

On a side note, the biggest spark so far is now 68cm to a grounded object with 300A peak current, 240VAC input and ~50uS on-time. (increasing on-time causes the OCD to fire constantly, may need tuning?)

[sjsimmo]

So I had my first IGBT failure recently 

One of my IGBTs has always noticeably run warmer than the others (for unknown reasons), and I suspect this to be the cause of the failure. The coil had not been run in over a week, and so had been packed up since the last full power run. I ramped the coil up to 240VAC input. Everything seemed normal for about 5 seconds. However, before i could turn the interrupter from 10uS on-time up to 50uS on-time, one half of the bridge failed short circuit (Gate, Collector and Emitter now read ~0 ohms on both IGBTs), blowing up the 3A slow-blow fuse in my variac. Neither of the IGBTs have any sign of physical damage on the package. Is it normal for IGBTs to fail without any visible damage, and when not operating at full-power?

Thanks again for your help, everyone

[Hydron]

The 75N60s were fine hard switching 225A (their rating), I suspect you could get away with a bit more than this with resonant switching but haven't tested them at higher currents than this. They won't be as resilient when grossly overloaded as a brick, but if you pop a set then they are also cheap to replace!

Edit: I wrote this reply based on the first page, didn't see page 2. Comments still stand but it sounds like you've already done some testing to failure

[sjsimmo]

Sorry that I haven't given any updates on my DRSSTC for a long time (more than a year, as I see now...),
but here is one final post, just to show the completed coil and modifications made since my last post.

After the IGBT failure, I decided to complete the coil with a fiber-optic midi interrupter, as I suspect that the failure was caused due to the Arduino used as a make-shift interrupter crashing. I also replaced 30v TVS that was used to limit voltage on the output of the feedback CT with the Zener diode/Schottky diode configuration used in the UD1.3B, in order to reduce the possibility of phase shift due to transformer saturation.

This is the layout of the driver circuitry, in its final housing (including LV PSU and fiber-optic/Indicator PCBs, which were not previously made):



This is The Midi Interrupter with its housing open:



And Finally, The Coil with the primary secondary installed:



In addition to this, I found out why Steve Wards UD1.3 includes a RC timeout circuit for the feedback signal. If it is omitted (as it was in my original design), the circuit will sometimes start up in a state in which the interrupter cannot start an oscillation. The same thing could occur if the coil lost feedback due to no voltage on the bus etc. As a result, I made the following modification to my driver circuit:



As a result of all these modifications, the coil now operates reliably with 350A peak primary current, and produces sparks up to ~60cm. I am not entirely sure why longer sparks cannot be achieved anymore, but I suspect it is due to the faster current ring-up time and lower on-times that have resulted from slightly less phase shift in the feedback CT. The OCD triggers constantly if I increase on-time past ~40uS at 240VAC input, despite careful tuning. This behaviour may be due to the low impedance of the primary coil, and high impedance of secondary coil (~2000 turns, rather than the generally recommended 1000).

Here is a video, showing the coil operating in midi mode:

/>
Thanks again to everybody on the forum who gave advice throughout the project 

[Zipdox]

Sorry that I haven't given any updates on my DRSSTC for a long time (more than a year, as I see now...),
but here is one final post, just to show the completed coil and modifications made since my last post.

After the IGBT failure, I decided to complete the coil with a fiber-optic midi interrupter, as I suspect that the failure was caused due to the Arduino used as a make-shift interrupter crashing. I also replaced 30v TVS that was used to limit voltage on the output of the feedback CT with the Zener diode/Schottky diode configuration used in the UD1.3B, in order to reduce the possibility of phase shift due to transformer saturation.

This is the layout of the driver circuitry, in its final housing (including LV PSU and fiber-optic/Indicator PCBs, which were not previously made):



This is The Midi Interrupter with its housing open:



And Finally, The Coil with the primary secondary installed:



In addition to this, I found out why Steve Wards UD1.3 includes a RC timeout circuit for the feedback signal. If it is omitted (as it was in my original design), the circuit will sometimes start up in a state in which the interrupter cannot start an oscillation. The same thing could occur if the coil lost feedback due to no voltage on the bus etc. As a result, I made the following modification to my driver circuit:



As a result of all these modifications, the coil now operates reliably with 350A peak primary current, and produces sparks up to ~60cm. I am not entirely sure why longer sparks cannot be achieved anymore, but I suspect it is due to the faster current ring-up time and lower on-times that have resulted from slightly less phase shift in the feedback CT. The OCD triggers constantly if I increase on-time past ~40uS at 240VAC input, despite careful tuning. This behaviour may be due to the low impedance of the primary coil, and high impedance of secondary coil (~2000 turns, rather than the generally recommended 1000).

Here is a video, showing the coil operating in midi mode:

/>
Thanks again to everybody on the forum who gave advice throughout the project 

Can you share the code for your interrupter?

[sjsimmo]

Of Course 

I have attached the Arduino code to this post. It should be noted however, that whilst the code works, it is quite badly written, and I would recommend to anyone considering using it that there are better alternatives out there (The OneTesla interrupter code, for example).

Some notes About the functionality:
  The code runs on an ATmega328, with a 16Mhz external crystal, more or less simulating an Arduino Uno.
  The Interrupter is only monophonic, and takes the melody line from Midi Channel 1 when in midi mode.
  Changing between Manual frequency/Midi mode requires the microcontroller to be turned off and on again.
  The interrupter only responds to Midi "Note on" and "Note off" messages. Pitch bend, All notes off messages etc. are ignored!

Atmega328 Wiring (Using Arduino Pinout names):
  RX: Midi input
  TX: Output to fibre optic transmitter
  A0: Mode Set: Manual Frequency (LOW signal)/ Midi Mode (HIGH signal)
  A1: Pulse width set (0-100uS)
  A2: Frequency Set (Up to 500Hz when in Manual Mode/lowers pitch by  0-2 octaves when in midi mode)
   2: Mode indicator (Outputs HIGH when in midi mode)

EDIT: I have also added the toner transfer compatible PCB layout for the interrupter (as a word document), in case anybody wants to try and create the PCB for themselves. There is however no silk screen or schematic, so you'll have to trace the PCB and guess where components go.

Let me know if you have any questions 

[Mads Barnkob]

Thank you for coming back and sharing your success and your design/code.

Despite your simple MIDI code, it does sound pretty good.

Are you using 2 in series, 2 parallel strings of 0.15uF/2kV 942C capacitors in the MMC? I looked through the thread and could not find the MMC mentioned anywhere. Since you say its a low impedance system, I assume you are using 0.15 uF in the resonant capacitor where I only used half of that in my small DRSSTC.

What did the resonant frequency end up at? I would like to add it to my IGBT list at the bottom here: http://kaizerpowerelectronics.dk/tesla-coils/drsstc-design-guide/igbts/

[Zipdox]

I've been working on writing a polyphonic interrupter for atmega328p/arduino uno too. I managed to do it at one point but it was out of tune. I'll rewrite it without using interrupts.

[sjsimmo]

Zipdox:
Sounds cool. Let us know how it goes, if you get something working

Mads:
The MMC is 2 series 2 parallel 0.1uF/2kv 942c capacitors (942C20P1K-F), so 0.1uF effective capacitance. I have considered reducing it by perhaps adding another set in series, for 3 series 2 parallel giving 0.067uF effective, and also making a new secondary with 1000 turns and a larger toroid, but I'm happy enough with performance at the moment anyway, so I haven't got around to doing it.
The resonant frequency ended up been somewhere close to 160Khz. The primary current figure of 350A peak is going purely off the OCD setting, as I haven't actually been able to measure it directly at full power due to interference crashing my USB oscilloscope. The OCD was showing correct behaviour at 100A, so I presume it to also be correct at the full 350A (a somewhat dodgy assumption, I know).

[Zipdox]

Zipdox:
Sounds cool. Let us know how it goes, if you get something working

Mads:
The MMC is 2 series 2 parallel 0.1uF/2kv 942c capacitors (942C20P1K-F), so 0.1uF effective capacitance. I have considered reducing it by perhaps adding another set in series, for 3 series 2 parallel giving 0.067uF effective, and also making a new secondary with 1000 turns and a larger toroid, but I'm happy enough with performance at the moment anyway, so I haven't got around to doing it.
The resonant frequency ended up been somewhere close to 160Khz. The primary current figure of 350A peak is going purely off the OCD setting, as I haven't actually been able to measure it directly at full power due to interference crashing my USB oscilloscope. The OCD was showing correct behaviour at 100A, so I presume it to also be correct at the full 350A (a somewhat dodgy assumption, I know).

I wrote a sketch that plays a C-major chord on pin 8. Try it out (on a speaker, not your coil please) and see if you can make it work with midi.

Code: [Select]

#include <MIDI.h>

#define samplingRate 62500


float midiNotes[] = {8.18, 8.66, 9.18, 9.72, 10.3, 10.91, 11.56, 12.25, 12.98, 13.75, 14.57, 15.43, 16.35, 17.32, 18.35, 19.45, 20.6, 21.83, 23.12, 24.5, 25.96, 27.5, 29.14, 30.87, 32.7, 34.65, 36.71, 38.89, 41.2, 43.65, 46.25, 49, 51.91, 55, 58.27, 61.74, 65.41, 69.3, 73.42, 77.78, 82.41, 87.31, 92.5, 98, 103.83, 110, 116.54, 123.47, 130.81, 138.59, 146.83, 155.56, 164.81, 174.61, 185, 196, 207.65, 220, 233.08, 246.94, 261.63, 277.18, 293.66, 311.13, 329.63, 349.23, 369.99, 392, 415.3, 440, 466.16, 493.88, 523.25, 554.37, 587.33, 622.25, 659.26, 698.46, 739.99, 783.99, 830.61, 880, 932.33, 987.77, 1046.5, 1108.73, 1174.66, 1244.51, 1318.51, 1396.91, 1479.98, 1567.98, 1661.22, 1760, 1864.66, 1975.53, 2093, 2217.46, 2349.32, 2489.02, 2637.02, 2793.83, 2959.96, 3135.96, 3322.44, 3520, 3729.31, 3951.07, 4186.01, 4434.92, 4698.64, 4978.03, 5274.04, 5587.65, 5919.91, 6271.93, 6644.88, 7040, 7458.62, 7902.13, 8372.02, 8869.84, 9397.27, 9956.06, 10548.08, 11175.3, 11839.82, 12543.85};

typedef struct Notes {
    int onTime;
    int offTime;
    int hasBeenOn;
} Note;

Note getNoteTimes(float pitch, float dutyCycle){
  float period = samplingRate/pitch;
  Note finalNote = {round(dutyCycle*period), round((1-dutyCycle)*period), 0};
  return finalNote;
}


//volatile Note newNote1= getNoteTimes(261.6, 0.1);
//volatile Note newNote2 = getNoteTimes(329.6, 0.1);
//volatile Note newNote3 = getNoteTimes(392.0, 0.1);

volatile Note playingNotes[] = {getNoteTimes(261.6, 0.1), getNoteTimes(329.6, 0.1), getNoteTimes(392.0, 0.1)};



MIDI_CREATE_DEFAULT_INSTANCE();

void handleNoteOn(byte channel, byte pitch, byte velocity){
   
}

void handleNoteOff(byte channel, byte pitch, byte velocity){
   
}


void setup(){
    pinMode(8, OUTPUT);
    MIDI.setHandleNoteOn(handleNoteOn);
    MIDI.setHandleNoteOff(handleNoteOff);
    MIDI.begin(MIDI_CHANNEL_OMNI);

  cli(); //stop interrupts
  TCCR2A = 0; // set entire TCCR0A register to 0
  TCCR2B = 0; // same for TCCR0B
  TCNT2  = 0; //initialize counter value to 0
 
  OCR2A = 0; // = 16000000 / (freq*prescaler) - 1 (must be <256)
  // turn on CTC mode
  TCCR2A |= (1 << WGM21);
  // Set CS01 and CS00 bits for 64 prescaler
  TCCR2B |= (1 << CS20);
  // enable timer compare interrupt
  TIMSK2 |= (1 << OCIE2A);
  sei(); //allow interrupts
}

void loop(){
//    MIDI.read();

}

ISR(TIMER2_COMPA_vect){

  bool shouldPlay = false;
 
  for(int i = 0; i<(sizeof(playingNotes)/sizeof(playingNotes[0])); i++){
    playingNotes[i].hasBeenOn++;
    if(playingNotes[i].hasBeenOn > playingNotes[i].onTime){
      playingNotes[i].hasBeenOn = -playingNotes[i].offTime;
    }
   
    if(playingNotes[i].hasBeenOn >= 0){
        shouldPlay = true;
    }
  }

  if(shouldPlay){
    PORTB |= 1 << 0;
  }else{
    PORTB &= ~(1 << 0);
  }





 
//  byte ^= 1 << bit_to_flip;
//  byte &= ~(1 << bit_to_clear);
//  byte |= 1 << bit_to_set;
}