First DRSSTC

[Aranox]

Hello everyone.
I am Jonas, a electronics engineering student from Germany and I started with my first Tesla coil build. I used the Google sheet provided by Kaizer Power electronics to check what coil size, h/w ratio and wire will get me to the golden 50k ohm impedance. The parameters I used to wind the coil are these:

0.22mm wire diameter, 0.08mm spacing and 1500 turns on a 90mm tube. The lenght of the windings is ca 450mm.
The topload is a toroid from a 150mm aluminium tube with 550mm outer and 240mm inner diameter.
I measured the resonant frequency at 160Khz.

The coil spacing wasn't inteded to be that big but after winding it on a lathe it turned out a little bigger than I wanted. The google sheet says the impedance now is only 41k ohms I hope that will still be ok.

I got a bunch of MACMIC MMG300D120B6HN IGBT bricks from work. I hope that I can use these at 160Khz your expertise would be really appreciated here. I also still need a tank capacitor and a driving circuit.

I see that most people use the Steve Ward driver with a GDT. Are high side gate driver IC's like a IR2184 a bad idea or is the extra isolation of a GDT that important? I would also rather use a microcontroller like a pi pico to control the outputs, so that I can check the IGBT temps and implement some safety features like a max duty cycle. Is this a bad idea or is there another reason the Steve Ward driver uses only logic gates and opamps?

Thanks for your help and all your great projects on this great page.

[flyingperson23]

GDTs are used because they're generally easier than direct drive and work fine. The Steve Ward driver was created many years ago before microcontrollers were cheap and powerful like today. Some microcontroller drivers exist, but they're generally more complicated to do the same functionality, especially if you want to do things like digital phase lead. Microcontrollers are also a lot more sensitive to emi than an analog driver. Duty cycle limiting is usually done by the interrupter, but is not too hard to implement on an analog driver. The ud2.9 pcb on a stickied post here does have a connector for a thermal switch. The 50k impedance is not as strong of a requirement as it seems on the kaizer website, anything between 25-75k should work fine.

[Mads Barnkob]

Hi Jonas and welcome to HVF

The magic 50k is not mentioned as end-goal, I do mention that experimentation is needed with larger coils and the tables on my website does range from 37-61k. The focus of the guide is to provide best practise measures, that will give you a working system, without knowing all the why's.

IGBTs is something you have to calculate, use the IGBT part of the guide to calculate your maximum frequency, it all depends on many other variables like on-time, current and your willingness to kill them fast or nurse them to live forever.

I would recommend using the open source UD3 ( https://github.com/Netzpfuscher/UD3/wiki ) instead of making your own from scratch, contribute to UD3 git instead. For the classic drivers, free PCB designs and schematics for the mentioned UD2.9 can be found here: https://highvoltageforum.net/index.php?topic=353.msg2401#msg2401

[Aranox]

Thank you for all the info you posted. 

Your IGBT guide was very helpfull and after checking it out it seems the IGBT bricks I have should work fine. I guess its better to use a known good driver instead of designing one myself for the first coil. Looking at the UD2.9 it think I will start with this board.

I am unsure why there are 2 optical receivers on this board, is it just as a alternative? Sadly I havent found a readme or something that gives a little overview on the design.

I have some Toslink optical receivers (Cliff FC6842032R) at hand that are a lot cheaper than the other fiber optic receivers. Since this is a smallish coil I think that the 10 meters of maximum fiber lenght should be fine as safety distance. What do you think? Would you advise against using Toslink as a little cheaper alternative?

[flyingperson23]

Yes, there are two common standards for drsstc optical connection, the board gives you the option of using one or the other. I'd recommend using the IF fiber, the connectors are pretty cheap. There's a good explanation of the ud2.7 here https://www.loneoceans.com/labs/ud27/
The 2.9 is just the 2.7 with a little extra logic to do pulse skipping.

[Late]

Thank you for all the info you posted. 

Your IGBT guide was very helpfull and after checking it out it seems the IGBT bricks I have should work fine. I guess its better to use a known good driver instead of designing one myself for the first coil. Looking at the UD2.9 it think I will start with this board.

I am unsure why there are 2 optical receivers on this board, is it just as a alternative? Sadly I havent found a readme or something that gives a little overview on the design.

I have some Toslink optical receivers (Cliff FC6842032R) at hand that are a lot cheaper than the other fiber optic receivers. Since this is a smallish coil I think that the 10 meters of maximum fiber lenght should be fine as safety distance. What do you think? Would you advise against using Toslink as a little cheaper alternative?

Ive been using toslink in my DRSSTC I and havent had any issues. I have 5 meter cable for it. For reference heres my coil https://highvoltageforum.net/index.php?topic=2550.msg18705#msg18705. 5 meters is more than plenty for even bigger coils. For my new build DRSSTC II ill be using 10m but even for that 5m is sufficient. Projected arc lenght of the second coil is 2.5-3m. Heres my post about its progress https://highvoltageforum.net/index.php?topic=2802.0.

UD2.9 is a pretty good choice for a simple driver that doesnt need programming and can be ran without needing computer. If you want i have my improved version of the UD2.9 in the SMD format with bord size identical to the classical UD2.7. Heres the link to it https://highvoltageforum.net/index.php?topic=2685.msg22521#msg22521. Currently ive only bench tested it but there it seems to work great. Havent had the time to test it on a coil yet. But its basically updated design replacing outdated or expensive parts.

[Aranox]

Ive been using toslink in my DRSSTC I and havent had any issues. I have 5 meter cable for it. For reference heres my coil https://highvoltageforum.net/index.php?topic=2550.msg18705#msg18705. 5 meters is more than plenty for even bigger coils. For my new build DRSSTC II ill be using 10m but even for that 5m is sufficient. Projected arc lenght of the second coil is 2.5-3m. Heres my post about its progress https://highvoltageforum.net/index.php?topic=2802.0.

UD2.9 is a pretty good choice for a simple driver that doesnt need programming and can be ran without needing computer. If you want i have my improved version of the UD2.9 in the SMD format with bord size identical to the classical UD2.7. Heres the link to it https://highvoltageforum.net/index.php?topic=2685.msg22521#msg22521. Currently ive only bench tested it but there it seems to work great. Havent had the time to test it on a coil yet. But its basically updated design replacing outdated or expensive parts.

That is great, I still have a bunch of toslink receivers laying around and would love to use them for this. I will modify the board and put the receivers on it. Next step will be, getting some good MMC Capacitors. I still hope that i can find some new old stock of FKP's on ebay.

[Aranox]

Sadly I am not able to import the UD2.9 or the UD2.9SMD project into Kicad 8 without breaking everything. Not sure what CAD the UD2.9 SMD was made with but the Kicad project file of the UD2.9 breaks when I edit it, a lot of the net calasses are just wrong after the import, Kicad tries to fix the libraries but I guess it isnt working. Would you recomment me to just do the layout again from scratch, or does anyone have ideas of how to import the project in kicad 8? I could also just get the UD2.9 PCB and the solder wires directly to the place where the fiber receiver conencts and do my toslink stuff on a separate board.

[Late]

Sadly I am not able to import the UD2.9 or the UD2.9SMD project into Kicad 8 without breaking everything. Not sure what CAD the UD2.9 SMD was made with but the Kicad project file of the UD2.9 breaks when I edit it, a lot of the net calasses are just wrong after the import, Kicad tries to fix the libraries but I guess it isnt working. Would you recomment me to just do the layout again from scratch, or does anyone have ideas of how to import the project in kicad 8? I could also just get the UD2.9 PCB and the solder wires directly to the place where the fiber receiver conencts and do my toslink stuff on a separate board.

For MMC I suggest using dawncaps from aliexpress. They are nice amazing pulse caps. In my DRSSTC I i used 2 of 220nF 3000VAC in paralel for total 440nf and haven't had any issues.
As for the driver it was drawn in easy eda. You could edit it in browser or import it into kicad.

[Aranox]

In my DRSSTC I i used 2 of 220nF 3000VAC in paralel for total 440nf and haven't had any issues.
As for the driver it was drawn in easy eda. You could edit it in browser or import it into kicad.

Thanks a lot, I imported it and the schematic worked, only the associated footprints were lost. I think I will do the layout again and use THT components for all the caps, resistors and diodes, this way I can use the stuff I have laying around and maybe mill the PCB myself.

I just have one question about a part i dont understand. What does the MBR0530 do in the circeled area? Could you reccomend a THT diode as replacement? The other 4 MBR0530 are just there to recitfy the AC voltage of the current transformer that trips the over current protection right? Is any shottky with similar voltage and current rating ok there?

[davekni]

What does the MBR0530 do in the circeled area? Could you reccomend a THT diode as replacement?

These diodes serve two purposes.  First, they clamp feedback CT voltage to a reasonable range for comparitor input.  CT output will typically reach tens of volts, far too much for comparitor input.  Other function is to set initial comparitor input voltage before each enable pulse starts.  ~16uA current forward-biases D6, making comparitor - input voltage slightly below + input voltage.
I've seen UD2.7 variants using 1N4148 signal diodes for these.  Works fine for clamping.  Comparitor can easily handle the higher forward voltage of standard silicon diode.  Initial comparitor input voltage differential will be much higher, however.  Typically still works, as long as first half-cycle of primary resonance generates enough CT voltage to switch comparitor input.  I suggest using 1N4148 diodes and adding a resistor in parallel, around 2k ohms.  That resistor will reduce initial comparitor input voltage differential to ~32mV, around what UD2.7 would have depending on temperature.  (Existing UD2.7 design has a steep temperature coefficient of initial voltage due to forward drop of schottky diodes at very small currents.)
 I vaguely recall seeing THT versions using 1N5819 here.  Diode capacitance is ~3x higher than for MBR0530, causing excess delay.  Delay could be compensated for with phase-lead for low frequency coils, but that's not ideal.  1N4148 will have lower capacitance than even MBR0530, so reduce delay by ~150ns relative to UD2.7, a beneficial feature.

The other 4 MBR0530 are just there to recitfy the AC voltage of the current transformer that trips the over current protection right? Is any shottky with similar voltage and current rating ok there?

Yes, 1N5819 or most other such diodes should work fine here.