Tardief's DRSSTC

[Mads Barnkob]

Thanks!

Could you tell me some more about the snubber capacitor on the IGBT?
http://kaizerpowerelectronics.dk/tesla-coils/drsstc-design-guide/snubber-capacitor/
Here you asumed that the busbar inductance is 200nH. How should I calculate that?
Do you have some recommendation for the IGBT protection?

I put this new snubber capacitor calculator online only a few weeks ago, if even that long: http://kaizerpowerelectronics.dk/calculators/snubber-capacitor-calculator/

You should ultimately measure the inductance at your resonant frequency, or similar, with a LCR meter. You could also just use the high / low estimate rules, as seen in the calculator page.:

Estimated high inductance snubber capacitance = Peak current / 100
Estimated low inductance snubber capacitance = (Peak current / 100) * 0.5

With a good clean, short, sturdy bus layout, you only need snubber capacitors to protect the IGBTs, TVS diodes are not needed then.

[Tardief]

It seem obvious, but i want to be sure. When you made the calculations with the result fmax=56kHz for the IGBT is it the same for half and full bridge or it concern to only 1 piece?

[Mads Barnkob]

I assume that you are referring to the IGBT part of the DRSSTC guide.

Since its mainly estimated around the currents through the switch, there is no difference from a half- to a full-bridge, if we adjusted the voltage to have the same current go through, each switch would see the same current. 1000A going through a IGBT, to coil to ground in a half-bridge is the same as 1000A going through a IGBT, to coil, through another IGBT.

As current is the same for components in series, the advantage of a full-bridge lies in the double voltage rating and thus also possible to lower the currents that has to be switched. So a full-bridge is vital in high frequency inverters with bricks.

[Tardief]

I can't thank you enough. 
Hope we can buy the parts soon and start to build.

It semm thath we will use CM600HA-24H.
In the IGBT selection part of the guideyou use CM600DU-24H.
I have some questions about the difference.

Here there is a 0.02 multiplier. What is this number?


We didn't find this information for the CM600HA. Can we use that what you used?


Here you used 0.081. In the CM600HA that is 0.03.Can be there that much difference?

(Sorry for the big images I dont know how to make them smaller )

[Mads Barnkob]

The CM600 model I have was a custom made version that optimized for faster turn-off, I can not really recall to be honest, but overall it is very much the same as other CM600 IGBT half-bridge bricks.

As I tried to write in the guide, we make an assumption up against datasheets with hard switching specifications, but the DRSSTC is soft switching at a very low duty cycle, so at 200uS on-time and 200 BPS it corresponds to 4% duty cycle (or 0.04 in a equation) and since that is for square wave signal, we half that figure since a DRSSTC burst is a damped sine wave ocsillation, so we end up at 2% (or 0.02 in a equation)

Yes, you can use the same switching loss energy curves, we are already assuming so much and extrapolating curves that it will do just fine

0.081 is the thermal impedance Zth(j-c) read out from the curve showing this against the on-time, you have to find the value that fits your duty cycle from that curve. This coefficient is highly dependant on the package the IGBT comes in, so this is important to get from the right datasheet. But this is always in the datasheets.

There is a thread about my DRSSTC guide, I would really love if you replied to that with parts of the guide that is hard to understand or need futher explanation. https://highvoltageforum.net/index.php?topic=38.0

[Tardief]

Hi,
We found a really cheap seller on ebay who sells a CM600 for 25$.
What dou you think? It seems to cheap for me? Maybe its used or fake?

[futurist]

Probably used, exactly what you should look for
I'd go for it

edit:
Just check if it's 600V version, not all are rated for 1200V
For example http://www.ebay.com/itm/IGBT-Modul-MITSUBISHI-CM600DY-12NF-High-Power-Switching-Use-600V-600A-/263115142944

[Tardief]

Thanks.
http://www.ebay.com/itm/NEW-1PCS-CM600HA-24H-CM600HA24H-MITSUBISHI-IGBT-MODULE-/230805167980
I forogot the link in the previous comment.

[futurist]

It's a single IGBT module, you'll need 4 of those for fullbridge

[Mads Barnkob]

I only see them at 50$/piece, so that is 200$ for a full-bridge, that is about what I payed for the used bricks I have.

They do write that they are new, it is from Malaysia, company called IGBT wholesale, feedback on ebay looks fine... its hard to say, but they are much cheaper than competition.

Import taxes might also add to the total price.

[Tardief]

It's worth a try. Hope it will be good.

We are planning to buy the driver and the midi controller from Eastern Voltage.
Here are the ones we thought:

Driver:  http://www.easternvoltageresearch.com/universal_drsstc.html 
Controller: http://www.easternvoltageresearch.com/plasmasonic_mod.html
They are pretty expensive but seems good. These two items are enough to controll the coil?



We will make the secondary coil tomorrow.

[futurist]

Nice winding rig!

I would suggest you build your own driver, that one is really expensive
For start, use interrupter from oneTesla, you just need 2 pots, arduino nano and optic transmitter - total cost 20$ at most

If you really don't want to bother with soldering and testing driver PM me, I have one spare board

[Mads Barnkob]

Winding/varnishing rig and pipe looks good!

I would not advise anyone, except if you are in a hurry and money is not a issue, to buy a EVR product, which is 99% based on the Steve Ward universal drivers which was further developed by Finn Hammer and Gao Guang, among others.

Gao got both MIDI controller and UD2.7 for sale on his website, as kits, so prices are also completely different.

UD2.7: http://loneoceans.com/labs/sales/ud27/
Midi: http://loneoceans.com/labs/sales/midi2/

there is also the onetesla midi interrupter: http://onetesla.com/products/kits/interrupters-all.html

I got both Gao's and onetesla, except I did not put Gao together yet

There are also other alternatives around on other forums/websites, heck, if you want you can have the ExpressPCB files to etch a single sided(all DIP) UD2.1 that I did a layout for. Still has some issues like components doesn't fit completely and needs a little "compacting" or "strecting"

[Tardief]

Thanks both of you.
 Eventually we decided at Loneocean's driver.

We already have some questions.

We have a disagreement in the team.
   -Put the rectifier, precharger, voltmeters etc. in a box away from the coil near to us. In this case we would need  about 10m cable in the DC side.
    -Put all these things under the coil. No unnecessarily long DC cable, but we will need a buch of cable for the feedbacks(Voltage, relay status etc.).

What dou you think?

The other question is about the ferrit in the feedback circuit. What we need to pay attention? High permeability and that's all?

[Mads Barnkob]

I made all my coils to be supplied by AC, most of the time from a variac.

I got two different ways of measuring the input current / voltage, I got my large variac with a 3 phase voltage/current meter in ( http://kaizerpowerelectronics.dk/power-supplies/3-phase-13a-variac/ ) and a Merlin Gerin PM700 3 phase power analyzer instrument ( http://kaizerpowerelectronics.dk/power-supplies/merlin-gerin-pm700-power-meter/ ) that I can use for power factor readout, so I know the DC voltage / current from calculations, so I never ran any wires back for instruments. There is only the AC cable and fibre optic cable.

I usually just used the same cores as I used for the GDT, they do however only need to be physically big if its needed because of the conductor you have to get through it or insulation to the inverter voltage is a problem. You properly already know the datastream GDT guide? http://thedatastream.4hv.org/gdt_index.htm

[Tardief]

We almos done with the winding.


We had a little accident.
We wantedf a second layer, so we put the epoxi into to fridge. It became so hot that the frige melted.  Stupid idea...
So no second layer now.



How did you attached the secondary coil to the base?
And what is you experience about bus voltage 564V vs 680V?

[Hydron]

A few notes:

- 564V vs 680V is not critical if you are running 1200V IGBTs. Assuming you have some control over the bus voltage up to that point (variac? SCR rectifier?) then I'd go for the configuration that gives you the highest possible voltage, as you'll have more adjustment room. If you do not have control, then I'd start off with tests using a DC power supply, then the lowest mains voltage possible (rectified 230V AC is approx 325V DC), and finally move to the full voltage once everything works well. A SCR controlled rectifier for voltage adjustment would be extremely useful, and will not add significant cost/size to the project, so I would strongly suggest building one if you don't have a variac.

- I also would suggest a loneoceans driver, or (even better), a UD+ driver from Philip Slawinski (see http://classictesla.com/pslawinski/udplus.html )

- Ferrite choice for current transformer is not critical, most will work, including anything suitable for a GDT. Gate drive transformer requires more design effort, but is not difficult, and Mads' link is very useful.

[Mads Barnkob]

I have only been using 564VDC from 3x400VAC, but I would switch to up to 800VDC any day if I could Most of the components are already rated for the higher voltage, its simply the problem of making/finding a 3 phase PFC power supply for that voltage range.

We made a large diameter nylon bolt with split pin, this mounts through a hole in the primary base plate, the split is inserted on the under side. It is loose and attached to a string, so that the bolt can disappear into the secondary coil in order for it so stand flat on ground when not mounted on the base plate.

Takes two people to mount the coil on the plate, but that has more to do with the weight of the secondary coil itself, it is very close to 40-50 kg.

[DanniL]

This rectification how can 680VDC?

[Mads Barnkob]

It is basically 3 parallel single phase voltage doublers. Instead of having 400VAC fullwave rectified to 564VDC, you have the midpoint/neutral to get 2x 325VDC ~ 650VDC across the capacitors.

I am unsure about how this balances in regard to load on each phase or neutral.

I tried a quick simulation here: http://www.falstad.com/circuit/circuitjs.html?cct=$+1+0.000005+10.20027730826997+50+5+50%0Av+128+192+64+128+0+1+50+320+0+0+0.5%0Av+128+192+176+128+0+1+50+320+0+2.0943951023931953+0.5%0Av+128+192+128+256+0+1+50+320+0+4.1887902047863905+0.5%0Ad+240+128+240+64+1+0.805904783%0Ad+272+128+272+64+1+0.805904783%0Ad+304+128+304+64+1+0.805904783%0Ad+240+256+240+192+1+0.805904783%0Ad+272+256+272+192+1+0.805904783%0Ad+304+256+304+192+1+0.805904783%0Aw+128+256+240+192+0%0Aw+240+128+240+192+0%0Aw+176+128+272+176+0%0Aw+64+128+176+112+0%0Aw+176+112+304+160+0%0Aw+272+192+272+176+0%0Aw+272+176+272+128+0%0Aw+304+160+304+128+0%0Aw+304+160+304+192+0%0Ag+64+304+80+384+0%0Aw+64+304+48+208+0%0Aw+48+208+128+192+0%0Ar+592+64+592+256+0+10000%0Ac+400+160+400+64+0+0.001+-319.1533318626577%0Ac+400+160+400+256+0+0.001+318.94065644304817%0Aw+48+208+176+336+0%0Aw+176+336+352+240+0%0Aw+352+240+352+160+0%0Aw+352+160+400+160+0%0Aw+304+256+400+256+0%0Aw+400+256+496+256+0%0Aw+496+256+592+256+0%0Aw+304+256+272+256+0%0Aw+272+256+240+256+0%0Aw+240+64+272+64+0%0Aw+272+64+304+64+0%0Aw+304+64+400+64+0%0Aw+400+64+496+64+0%0Aw+496+64+592+64+0%0Ao+21+64+0+4099+1280+0.1+0+2+21+3%0A