Mains Fullwave CWDRSSTC

[Phoenix]

Hello

Here is my new current project, it is a CWDRSSTC which runs on full wave rectified mains voltage without big DC bus capacitors. There are already such types of SSTCs out there which are usually called SRSSTC (series resonant SSTC), but they normally run from a fixed oscillator or secondary feedback which is not optimal in terms of simple usage and softswitching for low switching loss.

My Idea was to use a common DRSSTC driver, in my case the UD 2.7. I modified it a bit so it can run CW without overheating any components. Initially i thought that I might have to use a mains synchronizer circuit which always enables the UD when the mains sine voltage is around 30V, but as it turned out the primary current ringdown is so slow that the current never drops close to 0A and there is always feedback.

In terms of DC bus capacitance there are only 40µF of snubber capacitors on the bridge. These capacitors are of course very important since pulling 100kHz directly from mains at such currents would create massive transients which would kill everything. But with the snubber capacitors I am seeing beautiful waveforms like in a normal DRSSTC with large DC bus capacitance. The snubber capacitors alone are very effective of keeping the 100kHz away from the mains supply, but additionally I am also using a big mains filter which is normally used for VFDs. The only thing that might create a bit of harmonic distortion is the dynamic current consumption of the spark which fluccuates at around 1kHz, but so far I have not seen anything that would alert me on the mains voltage with my oscilloscope. It is barely visible that anything is going on.

Another very important aspect is the primary current ringdown via the IGBT Freewheeling diodes. In a normal DRSSTC there is a very big DC bus capacitance which absorbs all this energy with basically no voltage rise. But since I only have 40µF here, this voltage rise can become dangerous. Thats why I added a very powerful varistor to the bridge in order to clamp the voltage.

The whole setup is getting powered by my 20A 3ph variac.

Here are the primary and secondary coils. The resonant frequency of the primary is around 90kHz and the secondary is around 155kHz:


Here is the MMC:


Here is the bridge:


This is the output of the fullbridge and the primary current:


Here you can see how the phaselead is tuned to ZVS (yellow Vge, blue Vce, purple primary current):


Here are the waveform envelopes (yellow bridge output voltage, purple primary current):


Finally, I have a first test video. The input power is around 6-7kW and the input voltage is 250V RMS. It is already really scary so I am a bit afraid to go to the full 400V, I might have to make the tuning less aggressive so it does not become to powerful My goal is to achieve around 13-15kW for short periods of time.

Greetings
Phoenix

[AstRii]

Hello,
Thanks for sharing the project. The output is really scary! What is your peak primary current?
How did you pick 40uF capacitance? Did you calculate how much energy you need to store for each mains cycle or was it just a blind pick?
Don't you have any problems with overheating of primary or secondary?
My secondary gets warm with 8% duty cycle, i can't imagine it's temperature in mains synced CW mode.
How loud are the sparks? Usually CW coils are much quieter.

[Steve Ward]

Nice switching there!

That looks like its ready to run with more space around it. 

Also, i noticed the mains line inductance was significant enough to require some decent bus , especially if your OCD just stops the oscillation at peak line current. 

[davekni]

Nice project!  I've thought about changing my SRSSTC into something like that.  This is more motivation to do so.

I'm guessing that the ripples in current and voltage on each line voltage half-cycle are due to an expanding arc causing the initially-150kHz secondary frequency to drop closer to the 90kHz primary.  Have you zoomed in to the scope captures to see how frequency varies at the start of each line half-cycle?  That (or something else) is adding more interesting randomness to your arcs.  My SRSSTC makes relatively steady thick arcs.  (Not as thick as yours, being only 3kW.)  For my SRSSTC, arcs tend to start on the side of the toriod and then rise to the top as the air heats.  (That's when there is no breakout point.)

Your coil would be perfect for colored arcs using salt-solution breakout points.  I've had the best luck with SrCl2 (red) and ammonium borate (light green).  Of course, sodium makes a bright yellow too.  (Toasting a hotdog as a breakout point is fun, but tends to scatter bits of grease around.)  It is also fun to use thin steel wire (ie. small springs) or bits of aluminum window screen as sparkler breakout points.

Temperature rise measurements on the windings would be interesting to see.  My SRSSTC secondary gets warm at 3kW.  Heat is mostly due to RMS current.  Increasing power may not increase RMS current that much if secondary voltage doesn't increase much.  Most of the secondary current is resonant, so depends more on secondary voltage than on arc current.

[Phoenix]

Thank you for all your nice replies 

@AstRii:

The peak primary current was up to about 60A now. I chose the 40uF because i have some of those nice 20uF snubbers laying around that fit perfectly on the 62mm IGBT modules. I ran some simulations and saw that 40uF is more than enough to filter the 100kHz, so I was happy with my choice.
The primary gets warm pretty quickly, but not too hot actually. This is of course due to the short run durations I am currently  doing, if I let the coil run for longer the primary would get way hotter. I havent noticed any significant heating from the secondary so far. The sparks sound pretty loud and angry because of the 100Hz modulation.

@Steve

I definetly have to move the coil to the floor, letting it strike the ceiling is a dangerous idea. I think most of the inductance right now is coming from the variac running the bridge. Depending on how much output voltage is set, there is a lot of inductance in series with the line. If I increase the voltage further, the series inductance becomes less.

@davekni:

So far i have only measured the primary frequency and I could not see any significant changes in frequency during the halfcycle, but it would be interesting to see how the secondary resonant frequency changes during the halfcycle. I definetly have to also measure the secondary the next time, leaving the scope probe laying around should be enough to pick it up.

I will try this flame coloring idea, thanks for the idea. I did something like that a long time ago with my motstack, so i have a lot of nice Sr, Li, Ba, Cu and K salts here to try.

You are right, i also noticed that the primary current almost stays constant with higher input voltage, so the losses in the primary are quite comparable between 1kW and 7kW input power. The primary gets warm pretty fast, but this has not been a problem yet since my runs have been pretty short so far.

[davekni]

Forced air flow across the primary would likely allow continuous operation.  (Or thicker litz primary wire.)

Have you tried feeding this with a three-phase bridge rectifier input?  I'm curious how different the arc behavior would be when the bus voltage modulation is less (and at 300Hz).

[Phoenix]

@davekni:

Yes, I have tried to supply the coil with rectified 3 phase before. The arcs get less loud since there is less ripple on the bus voltage and they tend to get thicker and more bushy.

I measured the current that flows into the bridge today which was very interesting. I measured the power factor with an oscilloscope that can do power analysis and it turned out to be 0.95 which is really good.

Here is the input current before it gets filtered by the mains filter:


Here is the input current after it got filtered. It can be seen that all the ~100kHz components are gone and only the low frequency distortions generated by the arcload remain:


This is the actual current drawn from mains by the variac:


The current drawn from mains is actually pretty nicely shaped compared to the usual very sharp current waveform that gets drawn near the peak voltage when a full bridge rectifier charges big electrolytic bus capacitors.

[davekni]

Nice power factor!  Anything that allows current to even somewhat follow voltage makes much better power factor than diodes into electrolytic caps.  I haven't measured power factor for my SSTC.  Likely that good, as I don't have the spikes caused by your more-dynamic arc behavior.

[Phoenix]

Thank you, I was also very happy about the good PF 

Today I went up to 13kW. The variac is still far from maxed out, but its starting to become crazy. If i want to go further, I have to do it outside in the garage where I can open up the large door. The basement is filling up with nitrogen dioxide faster than i feel comfortable with. I don't think breathing it in at such concentrations is a good idea.

[davekni]

Yes, NO2 is an issue, so do run outside.  I suspect O3 (ozone) is even more of a hazard.  Smell is not a reliably way to determine ozone levels.  Ability to smell ozone varies by 1000:1 among different people.  Smell sensitivity drops with exposure too, at least short-term.

[Mads Barnkob]

Hi Phoenix!

Very nice project, reminds me very much of my own failed project with the same idea.

Another very important aspect is the primary current ringdown via the IGBT Freewheeling diodes. In a normal DRSSTC there is a very big DC bus capacitance which absorbs all this energy with basically no voltage rise. But since I only have 40µF here, this voltage rise can become dangerous. Thats why I added a very powerful varistor to the bridge in order to clamp the voltage.

I am wondering if the large varistor is really the only difference between our projects? https://kaizerpowerelectronics.dk/tesla-coils/kaizer-drsstc-iv/

I had problems with line interference and actually damaged the voltage meter in my variac from it 

[davekni]

I'm guessing that the varistor isn't needed except for fault conditions.  The big difference between this and Mad's project is the lack of a staccato controller.  It is only the sudden turn-off at the end of an enable pulse that creates the large energy dump back into bus capacitance.

[Phoenix]

@David:

Yes, I think you are actually right that most of the smell is ozone. I noticed that the smell in the basement always goes away after around 1.5 days. This is probably due to the O3 decaying back into O2. NO2 does not decompose under normal conditions like that I think.

You are right, this varistor is there in order to prevent overvoltage by the energy dump from the resonant circuit in case the coil gets suddenly turned off at a large voltage. You can see the effect of this here:



@Mads

Thank you

Yes, our projects share many similarities, that is true. But there are also some differences. My primary resonant circuit impedance is even higher than yours I think. The other difference is that you are doing staccato with your setup but my UD is constantly enabled to run CW. But i might try staccato in the future, it seems like a very nice idea.

If you want to try this project again in the future, here are some reasons why I think that you are getting interference:

There seems to be a lot of stray inductance between your snubber capacitor array and the bridge. Its better to directly mount the snubbers to the busbars. I would also use a lot more filter capacitance than 10uF, i would at least use 40uF like i did. I think another reason might be that you did not tune your bridge to ZVS. The forced discharge of the CE capacitance by the IGBT itself creates a lot of interference and transients on the bridge. Try to add more gate resistance and phaselead in order to achieve ZVS. You can verify the ZVS by simultaniously probing the Vce and Vge of one IGBT and checking if the Vce drops to ~0V before the Vge hits ~5V.
I would also add a mains filter directly before the bridge rectifier. It really helps a lot which can be seen from the scope pictures in my previous post.

I really like your Idea of building a QCW like that and it should work. I have had primary currents of up to 80A now and so far no problems with interference. I also probed the voltage coming from the variac and there are no transients at all.

Greetings
Phoenix

[davekni]

If I am calculating correctly, it looks like a significant (>30%) of the energy being dumped into the 40uF bus capacitance is from line inductance (mostly variac inductance).  At 90kHz and 44A peak, I'm calculating 0.305J in primary energy.  40uF charging from 50V to 160V is absorbing 0.46J.  Including losses, total stored energy is above 0.46J, perhaps closer to 0.6J.  Energy in line inductance may be as high as that in the primary.

No matter the energy source, if it is possible for oscillation to stop mid-line-cycle, the varistor is a good idea.

[Phoenix]

Thanks for the calculations!

Yes, it would make sense that there is also some energy dump from the variacs. One whole variac has an inductance of around 800mH when the whiper is at the 0 position. So depending on the whiper position, there might be a lot of inductance in series.

[Phoenix]

Here is a nice image of a big plasma flame discharge with my DSLR at 15kW. The discharge is around 110cm long. It is really interesting to see how the thermal turbulences force the discharge upwards.



Here you can open the full resolution image for zooming: https://i.ibb.co/k1VKzrX/cw-1.png

[Phoenix]

Another one:




Here is the full resolution:
https://i.ibb.co/gzj8tKw/cw-2.png
https://i.ibb.co/zxt6D2M/wallpaper-2.png

[Phoenix]

Here is a slowmotion video i took of the discharges, it looks really fascinating:

[davekni]

Yes, fascinating!  The growth of zigzags in the arcs looks much like what is seen at the end of QCW arcs, and perhaps even more like my low-frequency QCW.  Even normal DRSSTC arcs develop such evolving zigzags when ran at high pulse frequencies.  The big difference here is the continual evolving change without frequent discontinuities.

[Uspring]

Beautiful video and illustrative. Comparing the video to the stills, it looks like the photos suggest a much fatter arc than in the video. It probably looks like this because of the quickly changing paths the arc takes, while it is meandering.