SKP DRSSTC = aliexpress driver + medium coil + large bricks ;)

[Maju]

Hi there!

This post is the first in my life where I share my experiences with others regarding high voltage

I have quite a lot of experience with classic coils, I made one SSTC educationally, but this is my very first DRSSTC. I thought it was worth sharing my impressions as my project contains some "brave" things.
If you feel like saying "you're doing it wrong, stupid Noob!". Feel free! Every remark is valuable!   





"OK CHUMS, LETS DO THIS"


My assumptions:

• Single phase power supply - 230V, 16A (later I want to make additional and optional 3 phase power supply)
• As small as possible for the given power
• Everything done with as much margin as possible, especially the inverter
• SKP feature is a must!

During construction I fell ill with a disease known as "a good upgrade is not bad" and "it's never perfect enough" before I tested my previous solutions I changed them... So my TC have evolved as the project evolved and for now my technical data looks like this (I hope it's mostly the final one  ):

• Full bridge of: 2x 2MBI400U4H-120-50, 1200V / 400A
• DC BUS caps: 4x 6800uF / 450V 2s2p for 6800uF / 900V - there's no to much bus capacitance, right Guys? (I upgraded DC BUS caps 3 times for this bridge only...)
• Snubbers: 2x 3.3uF + 2x 2uF for 2x 5.3uF
• Power: single phase 230V / 16A with voltage doubler
• Secondary: 16x42cm, wire: 0.2mm
• Toroid: alu. tube 12x46cm
• Primary: flat, copper tube 4.7mm, 15 - 30uH (depend of MMC capacity and efects)
• MMC: (so far for testing) old Russians paperoil caps K41-1a 100nF / 10kV 1s2p for 200nF / 10kV or 1s3p for 300nF / 10kV
• Rez. freq: 65 - 70kHz
• Primary current: 600 - 800A (depending on the obtained results)
• Driver: SKP DRSSTC driver from aliexpress like this: https://highvoltageforum.net/index.php?topic=561.0 with many upgrades... Yeah, yeah I didn't know about profdc9 driver  Besides, I was curious how this driver works and I had it at hand.

Everything looks quite typical so far, but let's focus on the bridge and frequency for a moment.
Lets take a look on the datasheet of 2MBI400U4H-120-50:


If you add up the third column with the max switching times you get about 3.1 centuries. That's a lot... Especially when you need to drive these mammoths to 70kHz.
To sum up, I had an IGBTs not tested in DRSSTC by anyone, with switching times bigger than an acceleration of container ship, working at a fairly high freq.
I didn't see any reason why this shouldn't work.

But! Before you doubt people and all that is alive, take a look at the oscillograms:

(yelow: G-E voltage, purple: bridge output, blue: primary current)



IMO it's not bad. A lot of phase lead is needed but there is not even a shadow of voltage spikes in the output. It's suprise me a lot!
So I actually have a fairly powerful inverter that I drive very fast but I'm planing to get 600 maybe 800A max in SKP mode. I don't think any more current would be needed. Already at this value, I will probably exceed the permissible power of one phase and I will make my neighbors despair.

One more funny thing about the bridge, it's not the initial version ...of course, and this element was upgraded during the development of the project

...exactly three times

My bridge is like Pokemon, evolves:


Legend:

• 4x 30n60 TO-247 IGBT mounted on PCB from aliexpress, I used it for testing this exotic driver at the very begining - it's always better to blow up TO-247 than bricks - the shards are lighter
• 4x IXGN60n60C2D1 ISOTOP minibricks from aliexpress - this was supposed to be my final bridge
• "The Big Mama Bridge" because bigger is better

Pre-assembly:


I drive both IGBTs with two GDTs made by 3E25 42 / 25mm. I highly recommend cores made of this material, waveforms are perfect, no dropouts, perfect squere wave even at low frequencies!
Of course, they don't look like this on the screenshots because IGBTs works under load and the waveforms are distorted by the parallel inverter output and CT measurements.

Now MMC:


Those capacitors are robust, but they have high dielectric losses and high ESR, which makes them extremely hot! The effects also leave something to be desired, I compared them with the MKP10 in my classic tesla and they give about 10-15% worse results.

And now something for masochists and how to NOT to make the primary coil:


It took me 6 hours to thread the pipe through these holes!

And the most intresing think, THE DRIVER:


I could write a book about this driver, but maybe not in Chinese. This alone is an interesting thing that this miracle drives my two fat 2MBI400s without any problems. This is a very good and a very bad think at the same time. It has many pros and cons. So I made a lot of improvements there, some were necessary.

Let me know if you want to know anything about this toy, maybe I will write a separate topic in the form of a small documentation?

-------------------------

Now you say "where are the lightnings?" So far I've tested my Tesla with 2nd version of the bridge, and managed to easily reach 110cm at 400 amps, 200-350us and 200bps. It works great! But still I have feeling this is a lot for mini-blocks from aliexpress and their time would be counted in this configuration. That's why "The Big Mama Bridge" was created and it is waiting for the first tests as a Tesla Coil instead of an induction cooker 

In the next posts I will send some photos, maybe a short video from the first tests on IXGN60n60C2D1.

[davekni]

Looks like a great build!  Not sure why you are worried about the 2MBI400U4H-120-50 bricks.  Worst-case timing is slow, but typical is fine, and your scope traces look good.  You have sufficient phase-lead, as the bridge output switches before the opposite IGBTs turn on.  (Optimum might be with a bit less phase lead, but it is better to err on the side of more lead rather than less.)  I expect your MMC will fail before anything else.  Recommend changing to polypropylene caps.

3E25 material looks fine as long as it doesn't get hot for any reason.  Shouldn't get hot in GDT or CT use.  That material has a low Curie temperature, 125C.  Otherwise it's 6000 initial permeability is great for this use.

If you've figured out details of the Chinese driver, that would be of interest on this forum.  Many people have tried using such drivers with limited success due to lack of documentation.

[Maju]

Thanks @davekni for response!

I am waiting for tests in a bigger room, with a large bridge, new MMC and no neighbors  . I took a risk and ordered polypropylene GTO capacitors from aliexpress (of course), I wanted something different than joining dozens of small ones e.g. MKP10, as I did in my SGTC. We'll see how it works.

I will continue to experiment with MMC capacity. With the voltage doubler, the current in this coil grows incredibly fast! The oscillograms from the previous post are with 200nF on the board, with 300nF 800A I reach in the 4th cycle. Such a rapid rise in current is good for the standard DRSSTC, for SKP it seems to me redundant, and even dangerous.


Meanwhile, as I promised a few frames with discharges when runing on the bridge with IXGN60n60C2D1, 400-450A in SKP mode, ~300us on time.

Distance from the breakout point to the bed in a straight line was 110 cm.



Still work in progress!

P.S. I will write something about this expotic driver. Maybe it will be useful to someone.
P.S. 2: How can I add photos to a post without adding them in an attachment? 

[ritaismyconscience]

Looks like someone else bought the same board as me.
I had some success using the board. have you figured out how to get phase lead working. I kludged together some circuit with an inductor and a resistor and it kind of worked.

(haven't posted here in quite a while because busy with college)

[Maju]

Hi!

@ritaismyconscience, yes. I implemented a phase lead circuit based on UD2.7C with a few modifications (I use a solid inductor) with great effect 
You can see phase lead in the first post mounted on the universal board.

External phase lead circuit is a must with this driver! Without it the life of the transistors hangs by a thread 

I've tested simpler solution as you - variable inductor in series with resisor and even that improved waveforms, but I did not fully trust this solution. After all, we do not know what parameters the integrated circuit has at the FB input and whether it will be stable. Furthermore waveforms still wasn't perfec.

I will describe this problem in detail in a separate post for this driver.

[ritaismyconscience]

It would be nice if you could share the schematic for the phase lead part. Using a simple inductor and resistor seems to have a quite a few issues. The main issue is that it seems to be dependent on the current going through the coil, so at low currents it might work fine but it might fail at higher currents. It works acceptably well for me though because the current reaches the OCD limit pretty fast so I don't have to worry about this that much.

[Maju]

Here is my phase lead circuit.

The coil has a fixed value that gives a constant phase lead. The adjustment is made possible by two 10k multi-turn potentiometers which regulate the delays separately for the falling and rising edges.
L1 and R1 have to be selected experimentally, in my case 100uH + 68 Ohm + 1:800 CT works perfectly with my sluggish bricks.

[Mads Barnkob]

Hi Maju and welcome to HVF

Very nice coil you have built there, I am not sure if you looked at my DRSSTC1 for inspiration, if you did, I am sorry for not deleting the pictures of my primary coil made in the same way. 10 holes in acrylic with a copper coil that locks itself up for every 5 cm of winding was... a pain.

The gate voltage waveforms are slightly round and I would expect some noticeable heating from switching losses in the IGBTs. Since you have absolutely no ringing, lower that gate resistance! (You did not mention what value you are using, but it seems too high)

Bridge copper sheet work is very nice, good solution to distance the large caps and give some room to work on, I like it.

Once you have some proper sparks going and not blowing the bricks up, I will it to my list of Tesla coil proven bricks at your frequency/current settings

[Maju]

Hi!

Thanks @Mads for your replay!
It's a very good idea to make on you home page huge red, warning baner: "DO NOT TRY THIS AT HOME!" for primary in that style
This one is not the first. I've made exactly the same in my classic SGTC few years ago! Your experience, and even mine have taught me nothing ... 

But it looks nice and is very stable.

Now I still have a lot things to do with my coil.

Recent I received caps for my new MMC. Whole has 220nF / 18kV. Why so high voltage? I didn't find datasheet for them, so my trust is limited.





I've made a quik test with new configuration (new MMC + large bridge with 2MBI400U4H) on a very small and safe OCD settings.

...This coil is a wild!

My room was to small... I received thick, continue discharges into the ceiling (130cm) - and I still haven't achieved my OCD goal - so it's just the warm up!





Thank you for your advice about gate resistors. I have 5,7 Ohm (ON) and 1 Ohm (OFF). I will lower it for sure, probably to 4,7 Ohm. After this short (~1min) run my bricks become slighty lukewarm.

If I understand correctly, IGBT control is a pure compromise. They need to be driven fast enough to minimize losses, and turned ON slowly enough to minimize voltage transients.
As my bricks have a very high allowable power disipation (2kW per brick) I felt confident and choose the smaller voltage spikes with higher losses but I could exaggerated a bit 

I'm exited and can't wait to start testing on higher Amps in bigger room or outside if the weather will permit.

I have plans to compare two modes and achieved the intended OCD levels:

• higher OCD + high BPS + standard DRSSTC pulses
• lower OCD + low BPS + much longer pulses in SKP mode.

I will post it!