Heavily improvised DRSSTC

[Grzegorz2121]

Hi!

For few years i wanted to build a SSTC.
Recently I have managed to get most of the components required to build a SSTC.
My first test (a bit rushed) resulted in the blown IRFP460...
Interesting fact: Only one MOSFET failed and it failed open! I know! A miracle!
That try wasn't fruitless, for 2 seconds the SSTC managed to work.

Im still in process of finishing my coil and buying rest of the parts to make sure i have everything to finish my project.

But wait, you will ask why i posted it in the DRSSTC section!
Because I want to spice up my coil a bit... Yes I will be making now a DRSSTC.

Please forgive me now because i will try doing stuff that is just horrible sinful and wrong.
But please for the love of our Nikola don't try yelling at me. Motivation is the resource that i lack more than money.

I know that i have made few mistakes in my construction but i will try it anyway to see what i can get away with.
I know that my first mistake was to power the coil straight from the 230V without proper low voltage testing.
Now Im planning on testing it more thoroughly and in steps.

I don't remember all the parameters of the coil that i have planned but I can say now that:
Power section will use full bridge configuration with 40N60's in TO-247 + ultrafast freewheeling diodes in TO-220 + TVS's
As the main MMC i want to use 940C, 4*0.068uF for ~ 200nF tank capacity with 2kVDC, 100Apeak, 8Arms each
(I don't want to go over those MMC ratings, I will limit current for that, i think 2KVDC should be enough with 400Apeak, 8Arms)

The secondary with the topload will have ~ 75kHz freq. It is ~1m tall, 10cm diameter pipe. I made few mistakes while winding, but i hope it will work.
Of course it will show in action

I know that SSTC is much simpler but i think i call make a DRSSTC if im careful enough.

I hope nobody will banish me for that heresy, but a man has to learn somehow right?
Also Im not that stupid to repeat the same mistake twice. If i see something behaving in a wrong way i will redesign that.
Please just don't scream at me. (Other forums aren't that kind)
I will post further info and plans tomorrow.

[Mads Barnkob]

We will not show the door to someone that showed enough effort to build and fail, you have gotten enough things right to end up with a working system, not the greatest, but a working one.

If you need PCB designs for bridge, interrupter or drivers: https://highvoltageforum.net/index.php?topic=353.0

Your secondary coil and MMC are in proportions to each other, but the 40N60 TO-247 bridge is not powerful enough to drive the rest of the system to its capability.

For experience with TO-247 IGBTs see here: https://highvoltageforum.net/index.php?topic=236.0 and in order to design a bridge with a proper impedance to match your secondary, see here: https://highvoltageforum.net/index.php?topic=113.0 DRSSTC inverter vs. secondary impedance inequality discussion

A variac is recommended for low voltage testing, but you can also just turn on-time to 0 us on your interrupter for a DRSSTC and get full voltage on the DC bus before turning up the on-time and get some current flowing.

Please do post pictures in your next update.

[Grzegorz2121]

Thanks for the words of optimism! I seriously lack optimistic people in my life!

Unfortunately I'm very limited by my budget. A coil that just works is good enough for me at the moment.
I might upgrade (or design a second coil) later when i get more experience and money.

I'm now struggling with few calculations for few circuit components.
I have picked most of the components already but not all of them. Since i don't want to make a order that would be incomplete
i want to double check if everything is going to work (with some safety margin).

My first problem comes from the current transformer.
I have few ZnNi (2cm in, 3cm out diameter, 1.5cm height) toroids.
My calculations show that they would saturate when working as CT.
They might be wrong so please correct if i'm mistaken.

Since L=(B*A)/I
B*A=I*L
A=1.5cm*0.5cm
B=0.2T (typical max saturation for ferrites)
I=(B*A)/L
30 turn inductor on that core has 0.45mH so 1 turn induktor should have N^2 less
(the permeability turns out to be 200, one core before had it closer to 2000... the same material different result, i have no idea how it works)
0.45mH/(30^2)~0.5uH
The maximum peak current that that core can handle is I=30A
Since the DRSSTC's work around few hundred of A i want to make sure I have at least 1kA CT current capability
Maybe i need to get some lower permeability cores? Or i need to use different equations for that?
I have seen your DRSSTC's use similar size toroids. Can you help me explain that?

I want to use the steve's universal driver but without the output mosfets.
I have thought that TC4422AVAT and TC4421AVAT might be good enough alternative.
They are in TO220 packages and they can handle up to 9A. Not only that would be cheaper than those mosfets but the schematic would be easier.
I think they are good enough for those 40n60's in TO247.

I was thinking about the MMC bank again. My primary is around 0.09mH my secondary resonates around 75kHz
so I was thinking about buying ~6 0.047uF 2000kV 942C caps. It comes to cost and ability to experiment with different configurations. (also the Mouser stock is the reason)

I don't want to repeat the same mistake twice. So this time i will increase the voltage slowly. (and the interrupter on time)
I haven't got fiberoptic cable so I'm planning on using a twisted pair with metal grounded shielding.
I also will make a interrupter using my arduino. Im good at coding in C and that would allow me to control my coil more precisely.

I have a theory why my previous coil have failed.
My driver had a fault where a sufficiently high signal would switch the 74 inverter again. That could cause heating issue. I will use 3.3V zeners on the input now.

The second theory about that fault; high voltage spikes in the bridge. I think snubber caps would solve that. I have MUR1860 already on every fet but i think they might not be fast enough. (The bridge is build, i just need to remove schotkeys and replace IRFP460 with IGBT's). Can I use WIMA caps for that? The 942C would be perfect but expensive. I was thinking about using WIMA caps because they are 10x cheaper but they don't list their current limits or their internal resistances anywhere. I think they are worth a shot because with a correct value they wont pass too much current at 100kHz and the spikes are often at higher freq so they would still perform their role.

[Mads Barnkob]

Since you are going with the Steve Ward UD, why not use the same kind of current transformers? A high permability core around 4000-6000 AL cascaded for 33:33:1 so you end up with a 1:1089 ratio which is roughly 1:1000.

As long as you use 942C capacitors, they will serve you good if you treat them under their DC voltage peak rating and watch their temperature if you are pushing beyond their RMS current rating.

I use fibre due to galvanic isolation, I do not want to put myself on the line if there is a fault in the driver/bridge/coil/whatever flashes over. You can find cheap fibre optic wire on ebay, its more the TX/RX parts that are expensive.

It is hard to say what happens when you just switch the power on, full voltage, but I think it is more a matter of uncertain state of the gates, this is why it is important to have separate power to the driver and to the bridge supply. You always turn on driver first and turn it off as last thing. All other systems inbetween.

There is plenty of good WIMA capacitors, I got some specifications collected on some of them: http://kaizerpowerelectronics.dk/theory/good-mmc-capacitors/

[Hydron]

Since you are going with the Steve Ward UD, why not use the same kind of current transformers? A high permability core around 4000-6000 AL cascaded for 33:33:1 so you end up with a 1:1089 ratio which is roughly 1:1000.

One thing to remember is that you only need 3 cores to get the two outputs (needed for UD2.x) using the cascaded method - the primary core (with the high current wire running through it) can be shared with both the ZCS and OCD output cores - simply pass one of the ~33 turns on the primary core through each of the secondary ones.