my first DRSSTC is not working properly

[zanikk]

Hello.
I'm making my first DRSSTC and I'm having some problems.

here are some pics and details
/im using one tesla interrupter with a 250us/
coil





full bridge: CM200DY-12H



MMC: 6 strings in parallel of 2 in series Cornell Dubilier , 0.45uf


well , the first problem is that i have a strange waveform coming out of  the bidge , the current wave form looks good for me but the voltage looks very strange, I have  already tested the bridge with a fixed square wave on the feedback input and i got a really nice square wave on the output so i think the bridge works well.





 The second problem its that i'm geting  5/7cm sparks when i put the ground cable near the topload (at 50v dc) , i tested with 100v and 200v dc but eny sparks jump out of the breakout point

/>
Thanks for helping

[davekni]

For the voltage waveform, I'd guess it's distorted by inductive voltage drop in the ground.  If you can ground the scope probe at the IGBT brick and measure the bridge output voltage near by on the same brick, you'll likely see less distortion.

My first guess for low performance is mismatched primary-to-secondary frequency.  But that's entirely a guess.

Hopefully you'll get more responses from others here who have built more similar coils.

[Mads Barnkob]

Hi Zanikk and welcome to HVF

First off, it looks like a really well built coil and setup overall, you have done your home work.

What is the resonant frequency of your secondary + topload? What is the resonant frequency of your primary tap point + MMC?

It looks like you are using some regular 3 pin current transformers used for maybe 50 Hz mains current sensing? What are those two black current monitors? Since you say it works at low voltage, but at higher it stops working, maybe you are saturating those current transformers?

Another simple test, have you tried to reverse the polarity of the primary wires or feedback transformer?

Low voltage testing can be misleading due to IGBT output capacitance being relative to C-E voltage, the higher voltage the lower output capacitance, so you will always see higher switching spikes at low voltages, yours looks a bit over damped so its actually a slow charging instead, just guess work.

[zanikk]

Thanks for the tips, the same thing happened to another guy , so i think im going to ground the scope .
The resonant frequency of the secondary + topload its 71.4khz and the resonance of the  primary +MMC its 69.2khz.
The current transformers are CST206-1A (100:1) ,i saw a guy on youtube that use those ct at 100khz with 350vdc on the bridge so...
(The primary circuit still oscillating at 200vdc)
if I reverse the feedback CT the coil starts oscillating at 9khz and I don’t see eny current on the primary  .
Could be something with the gdt? I saw a peak voltage of 10v and im running it with ucc37321/2 at 15v peak ,im just thinking that im running the igbt at linear region , thats bad for the igbts i think.
Yesterday i had tried to make a new gdt (1:2:2:2:2) , but seems very difficult to wind , could i use two gdts (1:2:2), one for each half bridge, instead of one gdt for the full bridge ? I apreciate your help

[Mads Barnkob]

I would not put my money on those current transformers (even though they are free from many kinds of power electronics). I attached the datasheet, for 70 kHz you are looking at some 40-50% derating of the V/uS rating and it is only specified up to 110Arms (~150Apeak) which is way too low for a DRSSTC, a coil like yours is able to push some 500-800 Apeak.

GDTs are very easy to make with CAT5 network cable, you can see how I did + calculations here: http://kaizerpowerelectronics.dk/tesla-coils/kaizer-drsstc-i/ stick with 1:1:1:1:1. Which leads me to another question, what kind of driver are you using? If you only have UCC driver chips on the output stage that is too weak for driving bricks!

Are your bricks heating up? Since you suspect linear operation only? Do not fear too much about IGBTs getting damaged from that, they can dissipate a lot of power. If you can touch the heat sink there is not a problem.

[klugesmith]

Thanks for sharing the CT datasheet, Mads.

The volt-seconds limit (ET maximum) is about core saturation, which gets worse at low frequencies.  [edit] Maybe also core loss & temperature rise, which gets worse at high frequencies. At any frequency you can reduce ET by using a smaller burden resistor.  Same primary & secondary currents, smaller voltages, so less magnetization & loss in the core. [\edit]

What do veteran coilers have to say about Rogowski coils for getting quantitative current waveforms?
No saturation, because no iron.  I guess that for the narrowband, high frequency of a TC,  it's easy to get plenty of sensitivity, and you could even skip the voltage integration step (so scope waveform is proportional to di/dt).

The wikipedia page about Rogowski coils starts out with a definition that's too narrow.  Coil does not have to be helical with return winding along the axis; that's just a popular way for DIY'ers to make them from coaxial cable.  I have some multilayer PC-board Rogowski coils, made in a waste section of a circuit board for work. Would love to send one to a competent coiler for experiments. Not the same as the stock picture attached here.

[Uspring]

Probably the biggest disadvantage of a coreless CT is the requirement of a much lower burden resistor R due to the reduced inductance. In order to get a low phase shift between input and output current R << Reactance of L.

[Hydron]

I've had a quick play around with Rogowski coils, though my aims were more to make something that could fit around a TO-247 leg. It quickly became apparent that I needed to put in more time into it than I had available though, so not much to report other than it's not necessarily easy to shield one from E-fields. Gotta be possible though, as there are commercial devices out there designed to do exactly what I want (at a crazy price), sadly without much info from patents, teardowns etc. Like you I also intend to make a PCB one at some point - for the lazy there are even Altium design files for one floating around on the TI website (as part of an app-note of theirs).

As for those CTs, I haven't crunched the numbers but if they are getting pushed too hard then you can always make a cascade using another ferrite core with say 10 turns as a primary, then run one of those turns through the small CTs so that the ratios multiply to get 1000:1. If you keep the same V/A sensitivity (by increasing the burden R) you won't gain any margin in saturation on the small ones, but you will reduce the likelihood of parasitics ruining your day, so you likely can live with a lower sensitivity and thus a lower chance of saturation.

[zanikk]

I would not put my money on those current transformers (even though they are free from many kinds of power electronics). I attached the datasheet, for 70 kHz you are looking at some 40-50% derating of the V/uS rating and it is only specified up to 110Arms (~150Apeak) which is way too low for a DRSSTC, a coil like yours is able to push some 500-800 Apeak.

GDTs are very easy to make with CAT5 network cable, you can see how I did + calculations here: http://kaizerpowerelectronics.dk/tesla-coils/kaizer-drsstc-i/ stick with 1:1:1:1:1. Which leads me to another question, what kind of driver are you using? If you only have UCC driver chips on the output stage that is too weak for driving bricks!

Are your bricks heating up? Since you suspect linear operation only? Do not fear too much about IGBTs getting damaged from that, they can dissipate a lot of power. If you can touch the heat sink there is not a problem.

Sorry for my late reply i took some time out of this project , I was using an old daniel McCauley's driver (bad idea) it uses ucc in parallel at the output for the gdt and sn74hc14 for the input signal(also it dosen't have fase lead).
I changed the 1:1:1:1:1 GDT to a 1:2:2:2:2 GDT and i got 25cm sparks with 50v at the input, the problem i had was that the interrupter and the coil was not isolated (i was using a coaxial cable to connect the interrupter with the coil) also bad idea ,when the spark came out of the break out point the interrupter turned off immediately and the coil started oscilating in like a continuous mode¿?,(i got a really hot and red arc from the break out point like those videos of jacob's ladders) really scary  . I know that is very bad for the igbt and the mmc so immediately i turned off the power of the coil .
Well that killed my driver and luckily my interrupter still works , i couldn't fine the problem with the driver (i replaced all the ic's and its still having some problems), so im switching to the ud1.3 , i looked at the schematics and i think that i could add the fase lead circuit from the ud2.7 also im going to add fiber optic to the interrupter and make new current transformers for better feedback.
What do you think?
DRSSTC UD1.3.pdf

[Mads Barnkob]

If you want to use phase lead compensation, the simplest upgrade from Steve Wards UD1.3 is his UD2.1b, it is not publicly linked from his old website, but if you know the URL it is all documented, check it out here: http://www.stevehv.4hv.org/leadcomp/