DRSSTC design questions + project progress

[Vaclav]

Hi Vaclav, the picture, gate 4, has a lot of noise I think remove the 2k's that you have added and add 2k from IGBT gate and emitter and see if that helps.

I added 2k across gate and emitter. --> signal didn't change
I even go down to 1K --> nothing changed
Removing the 2K to gnd. --> no change

I make this test only with on driver. Should I do this with all 4 at the same time?

also try adding incrementally resistance say 2k 2w temporarily across 1/2 of the bridge output to the center tap of mains multiplier capacitors. The idea is to add a mild resistive loading to your bridge output to the center tap of the capacitors to damp transient noise generated by emi.

I don't understand you completely. What do you mean with "center tap of mains multiplier capacitors" I have 2x 3.3mF in parallel to smooth the rectified sine.
I tried adding 2,5K from one inverter output to the negative rail. do you mean that?? --> nothing changed with this resistor.

I turned up the dead time to get this signals:




When I rise the dead time the noisy part goes more to the right and gets smaller.

I think the new dead time is much too high about 1,5-1,7µs (frequency is 82KHz):


The inverter output looks now like this (with the 75W bulb as load):
The edges are far too bad I think.

Without the bulb the inverter signal looks like this:


rising/falling edge (without bulb):



rising/falling edge (with bulb):



Could it be a problem that I use tow slightly different IGBT bricks?
FF400R12KT3: https://www.infineon.com/dgdl/Infineon-FF400R12KT3-DS-v03_00-EN.pdf?fileId=db3a304412b407950112b4345f49601d
FF400R12KE3: https://www.galco.com/techdoc/eupc/ff400r12ke3_dat.pdf
The FF400R12KT3 is a lite bit faster.

I really have no clue how to make the signal better or what the exact problem is.

[Teravolt]

it looks good. I agree 1.5uS is a little to much, how about 1uS, is it adjustable? hear in the US we or I have 220vac with a center taped ground and typically builders use a full wave capacitive doubler to boost the voltage but I'm looking at your setup closer and I am assuming you have 220Vac with no center tap and you are running rectified mains and the big caps are your filtering and energy storage. have you tried firing up the bridge with a isolation transformer and variac? in a answer to your question about adding some dampening resistance at the output to a center taped junction ignore that. I was able to reduce the over shot on my bridge by adding non inductive resistance to a stiff reference. In your case I am not shure where to do that or even if it is necessary. have you tried adding the primary circuit and using the feed back yet? when you do start at a low mains voltage and monitor as raising the voltage watch for to much overshoot and cross conduction

In a note for your future builds I would have put the ADuM1200CR next to the UCC ic instead of separating them with a twisted pair and use a TTL A-drive and B-drive signals in coax, ie BNC. About using 2 different half bridges as long as your dead time covers the slow brick I think you will be ok. keep the pictures coming if you have questions. just remember that you are doing something that tesla enthusiasts don't usually do but because of its complications is done in the commercial arena all the time.

[Vaclav]

I made some further test and went down with the dead time to 1µS.

I added 2k across gate and emitter. --> signal didn't change
I even go down to 1K --> nothing changed

I noticed that adding the resistor damps the spike. But not that much
IDK why I didn't see this last time...

I make this test only with on driver. Should I do this with all 4 at the same time?

I made the test with all 4 at the same time and noticed that the spikes on the gate signal for the high side IGBTs are much worse.
Than the spike on the gate of the low side IGBTs.
White high side gate (White is stored in the oscilloscope normally they are 180° phase shifted)
Yellow low side gate


I tried adding a resistor in series to the gate. At first I thought I don’t need them… now I think that was pretty naive.
I only have enough resistors to do this with 2 gates at the same time.
Each gate has 2,5Ohm in series.


These is the new voltage directly at the gats:
High side IGBT:




Low side IGBT:

Ignore the straight line at ~-18V I forgot the stop the oscilloscope.



Inverter output has now some flat part at 0V:




I will buy some smaller resistors 0,5Ohm and 1Ohm and redo the test with all 4 IGBTs.

[Teravolt]

If you want to try more dampening going from each half bridge to ground say 1000 to 220 you will have try it your self. the last wave form is much better and when you get the spikes and overshoot under controle you might slowly decrease your dead time to something less to close the time at zero volts. you should be able to see on the scope as you are addjusting. Things may change when you add your resonant primary and feed back so go slow and monitor safely if you can whyle you see what happens when you turn up the voltage to see if the spikes get bigger.

[Vaclav]

I added now 2,2Ohm in series to my gate
New signals looking like this:

"Worst" gate signal of all 4 signals:

Edges:

Edges in detail:


Inverter output:
The slope at the end is still there I think it's ok.

Edges:

I can't do anything against the sinusoidal part in the rising edge.

All my test until now were made with my "high voltage" DC power supply (max. 600V,6A,1000W)
I made all test with up to 325V.
I made the first test run with main power via isolation transformer.
One driver has failed and cause short circuit on one 18V rail.
I switched off the supply to all the control voltages (5V and 4x18V).
And it happened the way it had to... I blow another IGBT brick... (I can get a used on for 50€ at eBay, so it’s not that much of a problem)
I started digging around and noticed that the problem does not came from the driver and not from the controller.
The problem are the ADUMs the make a spike at the output if the low voltage supply is off before the high voltage side does.
This spike will occur at all 4 ADUMs at once and the drivers short the complete inverter.
As a workaround I will ad 0,5F at my 5V supply rail. I tried all my available big capacitors and measured that ~30mF are enough.
0,5F is the next bigger cap I find at the shop I'm used to buy from.


I've already finished winding my current transformers.
I will mount them on a brass rod.


I'm on holiday now. I' ll be back in about 3 weeks.

[Teravolt]

you might be able to find where the spikes are comming from by finding out what edge or operation they are synced to. you also may have stray inductance in your connections that are feading into the driver

[Netzpfuscher]

It's more simple the datasheet has a table. It's normal that you get a high signal, if you switch off the Vcc.

[Vaclav]

Finally, I'm back with some progress.

It's more simple the datasheet has a table. It's normal that you get a high signal, if you switch off the Vcc.

Wow I feel like a huge idiot because I don't have seen this...
I chanced the ADuMs to this one ADuM1285BRZ.
After that I had a similar problem but it came from the fiber optic receiver.
I solved the problem with adding a 0,5F cap across the 5V supply and a better under voltage detection for the 5V supply (with OPV).
New under voltage detection:


Old one:

Vaclav

The problem is now completely gone.
And the the Inverter with function generator as feedback signal and lightbulb as load is running perfectly.


I build the primary coil, I made the high of the coil adjustable so I can change the coupling between primary and secondary.


I also finished the current transformer assembly.


The secondary and the topload is completed.

Secondary was planned with 1800 windings, I got about ~1616
I got some small bubbles in the varnish of the secondary.
I used this varnish: https://www.ebay.de/itm/Wilckens-Yachtline-DD-Hartlack-750-ml-Farbauswahl-2-K-Polyurethan-Bootslack/321632791504?ssPageName=STRK%3AMEBIDX%3AIT&var=510526219327&_trksid=p2057872.m2749.l2649
I measured the resonant frequency of the secondary coil. ~86.75kHz

My IGBTs support a maximum current of 800A peak.
So I designed my MMC for max 800A peak.
I used Mads MMC calculator (http://kaizerpowerelectronics.dk/calculators/mmc-calculator/)
Idk which value I should use for the on time and the BPS.
I even don’t know if I have understood BPS and on time right.
BPS = Basically the Interrupter frequency??
On time = time the coil is on --> on time of the Interrupter signal??
I want to use a Musical interrupter later.
To play different sounds the musical Interrupter changes the frequency of the Interrupter signal? --> different BPS
With the given 200µS on time and 200BPS get these two MMCs.
I think that I will use these capacitor: https://www.mouser.de/ProductDetail/Cornell-Dubilier-CDE/940C20P1K-F?qs=KAyk88b%2FVhdRi0pubeTg4g==
Which one is better?
0.4µF capacity (results in ~5 windings Primary)
0.5µF capacity (results in ~4 windings Primary)
My primary has only 5 windings so I think the 0.5µF would be better because I have more adjustability. And the 0.5µF Is cheaper.

0.4µF:

0.5µF:

[Vaclav]

I think my question went down in my long post.

BPS = Basically the Interrupter frequency??
On time = time the coil is on --> on time of the Interrupter signal??
I want to use a Musical interrupter later.
To play different sounds the musical Interrupter changes the frequency of the Interrupter signal? --> different BPS

I have really been looking for an answer for a long time but found nothing that makes me really 100% sure.

Which of the two options is better?

0.4µF capacity (results in ~5 windings Primary)
0.5µF capacity (results in ~4 windings Primary)
My primary has only 5 windings so I think the 0.5µF would be better because I have more adjustability. And the 0.5µF Is cheaper.

Thanks for all the help, I certainly couldn't do this project without some help.

[dexter]

BPS = Basically the Interrupter frequency??
On time = time the coil is on --> on time of the Interrupter signal??
I want to use a Musical interrupter later.
To play different sounds the musical Interrupter changes the frequency of the Interrupter signal? --> different BPS

Yes for all

For your MMC the.0.5 one might suffer when playing music as higher BPS translate to higher rms current. Increase tbe BPS in the MMC calculator and see how much the Temperature rise per cap increases. If you plan to use the coil for sort-ish periods of times (giving the MMC time to cool) it will work fine.

Another thing to consider is impedance matching of the primary and secondary... The primary capacitor plays a role in that and there is a formula but i can't find it right now (on the phone)

[Vaclav]

Thanks for the quick answer.

For your MMC the.0.5 one might suffer when playing music as higher BPS translate to higher rms current. Increase tbe BPS in the MMC calculator and see how much the Temperature rise per cap increases. If you plan to use the coil for sort-ish periods of times (giving the MMC time to cool) it will work fine.

What is a good max BPS for music? I think 1500BPS should be enought. Or should I go for more BPS?
I will recalculate the MMC in relation to the new BPS.

Another thing to consider is impedance matching of the primary and secondary... The primary capacitor plays a role in that and there is a formula but i can't find it right now (on the phone)

The primary side should have approximately the same resonance frequency as the secondary side.
As far as I know, the primary circuit should have a little more inductivity than it must have in the resonance case.
So if I use a bigger MMC the inductivity will be smaller. For the same resonance frequency.

Is it now better to have more or less turns on the primary side, or is that completely irrelevant in the case of one turn more or less?
Because I would then choose the MMC accordingly.

[dexter]

What is a good max BPS for music? I think 1500BPS should be enought. Or should I go for more BPS?
I will recalculate the MMC in relation to the new BPS.

depends on the music...
Keep in mind there are pauses and the BPS is not constant so the RMS current will vary quite substantially.
I wouldn't go as far to overdesign the MMC for 1500 BPS but i'd go ~10-20% over the minimum values for normal operation.

Is it now better to have more or less turns on the primary side, or is that completely irrelevant in the case of one turn more or less?
Because I would then choose the MMC accordingly.

read this
https://highvoltageforum.net/index.php?topic=113.0

[Vaclav]

thx for the tips!

I think I will go with the 0.4µF MMC.
If I want higher current, I can change it later to 0.5µ or higher.

I made some simulations (primary oscillating circuit coupled with secondary oscillating circuit).
I got ~80µS to transmit all energy to the secondary side. Therefore I will go for ~80µ on time

I chose 1000BPS. Probably my rectifier will not be able to provide enough energy, but I can change that later on.

I think I will go with this MMC, I don't want to increase the RMS current of the MMC, maybe I will cool the caps active or go with a little bit less BPS.

[Mads Barnkob]

Designed for 1000 BPS, with only RMS current being higher than MMC specifications, it is all fine! You are not going to push it that hard continuously anyway, not even with MIDI if its edited correct.

The simulated 80us transfer-time does not mean that you can not run longer on-times, longer on-time will still produce longer sparks as you keep pumping energy into the grown spark channel.

[Uspring]

Vaclav wrote:

I made some simulations (primary oscillating circuit coupled with secondary oscillating circuit).
I got ~80µS to transmit all energy to the secondary side. Therefore I will go for ~80µ on time

That holds true for a perfectly tuned SGTC. In a DRSSTC you keep supplying energy to the primary coil. Since that is not immediately transferred to the secondary, there will always be some energy in the primary. Have a look at Mads docs on his TCs. He shows primary current scope shots and all show some primary current, i.e. energy in the primary tank as long as the burst persists. Once you stop the burst, most of the primary energy returns to the bus caps via the IGBTs reverse diodes. Actually a sad side effect of stopping the burst.

[Mads Barnkob]

Vaclav wrote:

I made some simulations (primary oscillating circuit coupled with secondary oscillating circuit).
I got ~80µS to transmit all energy to the secondary side. Therefore I will go for ~80µ on time

That holds true for a perfectly tuned SGTC. In a DRSSTC you keep supplying energy to the primary coil. Since that is not immediately transferred to the secondary, there will always be some energy in the primary. Have a look at Mads docs on his TCs. He shows primary current scope shots and all show some primary current, i.e. energy in the primary tank as long as the burst persists. Once you stop the burst, most of the primary energy returns to the bus caps via the IGBTs reverse diodes. Actually a sad side effect of stopping the burst.

I got the best oscilloscope shots showing this from my small DRSSTC, as a very long on-times is less and less "necessary" on a large coil with higher peak currents, there you get the spark length more in a manner comparable to a SGTC, but for low peak currents, longer on-time gives better reults.

http://kaizerpowerelectronics.dk/tesla-coils/kaizer-drsstc-ii/

[Netzpfuscher]

A long ontime is better for MIDI modulation. It gives you more resolution for the MIDI volume. Lets say you have only 4 full cycles in a interrupter period, you can only get 3 different volumes on the midi voices.

A good midi interrupter should always have a duty cycle limiter. It should lower the ontime for high pitched notes or transpose them down.

[Vaclav]

Thanks for the informative posts.

I bought the caps for the MMC.
36 of this one: https://www.mouser.de/ProductDetail/Cornell-Dubilier-CDE/940C20P1K-F?qs=KAyk88b%2FVhdRi0pubeTg4g==

Vaclav wrote:

I made some simulations (primary oscillating circuit coupled with secondary oscillating circuit).
I got ~80µS to transmit all energy to the secondary side. Therefore I will go for ~80µ on time

That holds true for a perfectly tuned SGTC. In a DRSSTC you keep supplying energy to the primary coil. Since that is not immediately transferred to the secondary, there will always be some energy in the primary. Have a look at Mads docs on his TCs. He shows primary current scope shots and all show some primary current, i.e. energy in the primary tank as long as the burst persists. Once you stop the burst, most of the primary energy returns to the bus caps via the IGBTs reverse diodes. Actually a sad side effect of stopping the burst.

I got the best oscilloscope shots showing this from my small DRSSTC, as a very long on-times is less and less "necessary" on a large coil with higher peak currents, there you get the spark length more in a manner comparable to a SGTC, but for low peak currents, longer on-time gives better reults.

http://kaizerpowerelectronics.dk/tesla-coils/kaizer-drsstc-ii/

Good to know, untill now I thought you can't pump more energy in the secondary after the 80µS.

A long ontime is better for MIDI modulation. It gives you more resolution for the MIDI volume. Lets say you have only 4 full cycles in a interrupter period, you can only get 3 different volumes on the midi voices.

A good midi interrupter should always have a duty cycle limiter. It should lower the ontime for high pitched notes or transpose them down.

I will pay attention when I designe/select the MIDI interupter.
But first, the coil has to work.

[Vaclav]

I'm finaly back with some huge progress.

I had some major EMC problems...
Due to these problems even the digital signal to the gate drivers were coruppted.
The EMC problems were caused by my badly designed gate drivers.
(There were some shootthrougs in the output amplifier --> because of these high pulsed curents I got huge problems)

To get rid of the problems, I redesigned the complete control electronics and built a new gate driver.

Summary of the new designe:
-Every signal leaving the shielded enclosures has got some solder pads for EMC filters. (I didn't need them at the end).
-Every low power signal cable is now twisted pair (I planed to use standart ethernet cabels but I mixed up the pin out... so I had to crimp my own cabels)
-The Isolation ICs for the digital signals, betwene the gate driver PCB and the control unit, are now directly on the driver PCB.
-I built a 5V powersupply with two seperated rails one for 5V digital and one for 5V analog. (I don't think that was really necessary)
-I've added a microcontroller to monitor the ON-Time and duty-cycle of the Interrupter signal.

I've attached the new schematics and the microcontroller programm.

MMC:


Here are some pictures of the first test:

The test was done with 100V bus voltage and 150µ-200µ ON-Time. (I do not know the exact ON-Time)
Coupling should be roughly 0.13. (this is the recommended value of JAVATC)


Pictures of the second test with 200V bus voltage:

If someone is interested in my documentation for school, I'll post it here. (but it will be in German)
Maybe I'll post the KiCad data later, I'm just to lazy to sort the files out.

Greetings
Vaclav

[Mads Barnkob]

Good to see that you have it all running and making sparks. Good first and second tests at lower voltages, I am looking forward to see you run the coil at full voltage and maximum on-time

Did you get some optimization done? Like that very long primary lead, that could be much shorter. But overall your setup looks very clean and well designed.

I can read German, so I would like to see you documentation if you still want to share it.