Benjamin's DRSSTC3

[Benjamin Lockhart]

Hi all,
After having problems with the driver in my large freewheeling coil, I decided that I wasn't happy with the overall design. It's too heavy and difficult to set up. So I built a whole new coil.
The plywood base is 16.75 inches in diameter and the whole coil stands about 5 feet tall

It's based heavily on Loneoceans DRSSTC3, and just so happens to be my 3rd coil build.
The specs are;
   8x36 inch chicken wire and plywood toroid (it's a cheap way to build), with 3.25x14 inch smaller toroid below
   6.3x27.5 inch secondary with 28AWG wire (it's actually closer to 29AWG)

   8 turn primary of 1/4 inch tube with 3/16 inch between turns and 0.8 inch clearance from secondary, 0.162 coupling, compressed air cooled, but I may need to upgrade to water cooling, I was melting primary supports! Edit: this has been done.

   365nF 15kV 260A rms MMC using dawncaps, it stays completely cold. Javatc says the primary surge impedance is about 7.5 ohms and I think that number has quite an effect on the behavior of spark growth. Also I think that the primary and secondary geometry is more important than the exact coupling, however you should still try to maximize coupling for a given geometry. Play around with the primary outer diameter in javatc, you will find that there is a sweet spot that gives the most coupling. The exact diameter depends on other parameters as well.

   UD2.7C

   800A OCD for now, primary current peaks around 1000A
   130uS on time
   CM300du-24nfh full bridge, 680V DC bus voltage
   Approx 60kHz operating frequency with 7% primary detuning.


   4x 3300uF, 450V bus capacitors in doubler for 3300uF total, not shown are 2uF snubber caps across each IGBT brick.


   Onboard bus precharge circuit consisting of a 12 ohm resistor that gets bypassed by a 50A relay powered by the 24V SMPS. The SMPS has a 1/4 second turn- on delay which is enough for the caps to almost fully charge before the relay closes.

Here are some build pictures.



I don't know why some pictures are sideways?

Here is a tin can almost melting at half voltage.

And it makes 7 foot sparks! (edit: now 8.5 feet!)

The sparks really are incredibly bright!
What do you all think?

[davekni]

What do you all think?

Looks like a very clean build with great performance!

Approx 60kHz operating frequency with 7% primary detuning.

Would have guessed you'd want a bit more than 7% for long arcs.  Arc length is already good.  Top load size helps with minimizing detuning, and coupling helps maintain performance in spite of secondary frequency drop.  That may be why 7% is sufficient.  Probably best for music.  Would be interesting to see if arc length could be any longer with a bit more detuning and a bit longer on time.  That's only if you want to risk pushing coil's limits.

[Benjamin Lockhart]

I just ran the coil again and everything went well.
I made a shielded 10 gauge power cord for it and put a strike protection wire around the coil base.

/>

/>

/>

[Benjamin Lockhart]

I've got an interesting update now. The coil has hit 8.5 foot (2.6 meter) sparks! with a new toroid and the same tap point. I think it's around 10% detuning maybe a bit less, which still plays midi very nicely! These are video frames of an 8.5 foot ground strike! This is 3.7x the winding length while playing music which is amazing for a regular non pulse skipping DRSSTC!

For reference the specs are, 800A ocd, 120us on time, 680V bus, 8" x 36" toroid, 0.365uF MMC, 7.3 ohm primary surge impedance, 60kHz primary fres. Otherwise it's exactly Loneoceans DRSSTC3 specs.





I made a nice skeleton toroid because I was tired of the chicken wire bending out of shape. It's 8" x 36", the same size as the old one, with 15 rings of 3/8" aluminum fuel line and a welded aluminum frame.

[Michelle_]

Nice toroid

[Benjamin Lockhart]

Here's the video. Longest strike at 42 seconds.

[Jj]

Super results! Does the OCD trigger much?

[Mads Barnkob]

Congratulations on your very well-built coil!

So its free-wheeling, but not pulse skipping? What is the maximum on-time for that MIDI? You must be pushing it hard!

[Benjamin Lockhart]

Thank you! The OCD does trigger most of the time so I could even raise it a bit higher than 800A but I'd rather it live a long time.

The on-time was limited to 120us for that run! The duty cycle limit on the midistick interrupter was set to 8% I'm not actually pushing things that hard. That is about 7 RF cycles at 60kHz and the on-time fits within the first notch of the primary current envelope.

It's not freewheeling or pulse skipping, just a regular UD2.7C with the UVLO fix Davekni suggested using a 1.25V shunt regulator.

[davekni]

I've got an interesting update now. The coil has hit 8.5 foot (2.6 meter) sparks! with a new toroid and the same tap point. I think it's around 10% detuning maybe a bit less, which still plays midi very nicely! These are video frames of an 8.5 foot ground strike! This is 3.7x the winding length while playing music which is amazing for a regular non pulse skipping DRSSTC!

Very impressive results!  Beautiful top load too.  (I rarely get more than 2m when tuned for good MIDI performance.)

[Benjamin Lockhart]

Thanks Dave, I believe my relatively large toroid (1.3x secondary length), and high-ish coupling (0.162), combined with a primary surge impedance that's not too high or low, and a relatively high bus voltage for the size of the coil, are all responsible for good performance.

I still haven't seen a single racing spark and I think that's because the primary impedance isn't too low, as well as the big toroid keeping the topload voltage lower

I realize I'm generalizing here without actual numbers other than my coil's specs, feel free to correct my thinking here or add your thoughts!

It does seem like the most efficient mode of DRSSTC operation is transient with the on-time equaling one beat in the primary current. Of course, the coil needs to be optimized to run well with that on-time.

[davekni]

Thanks Dave, I believe my relatively large toroid (1.3x secondary length), and high-ish coupling (0.162)

Yes, I'd guess those two are the most important factors.  Built my DRSSTC just before finding this forum.  Didn't have a good idea of what to target.  Great work!

[Jj]

Thank you! The OCD does trigger most of the time so I could even raise it a bit higher than 800A but I'd rather it live a long time.

The on-time was limited to 120us for that run! The duty cycle limit on the midistick interrupter was set to 8% I'm not actually pushing things that hard. That is about 7 RF cycles at 60kHz and the on-time fits within the first notch of the primary current envelope.

It's not freewheeling or pulse skipping, just a regular UD2.7C with the UVLO fix Davekni suggested using a 1.25V shunt regulator.

I’ve never been able achieve even the same result with long on times compared to short. My coil which is of similar size to yours, but runs much lower at 45kHz has made a best of 1.7M with 120us and a 5 Ohm primary impedance. At that point it had some pretty violent racing arcs and flashovers. Reducing the coupling to .15 didn’t help. So I thought I’d try longer on times again. I increased the primary impedance to 9 ohms and ran all the way up to 320us. The power draw doubled, but streamer length decreased to 1.3m at best. I ran the primary from so low it would hardly break out to something around fpri =fsec with complete notches. Also tried raising the break rate.  Would love to know how those that have success with long on times don’t it! My setups secondary impedance according to Javatc is around 72kOhm which I take it would be considered on the high side, maybe that is the reason?

[flyingperson23]

I get good (enough) results with long on times by running low bus voltage and pumping tons of power into it without caring about effeciency

[Benjamin Lockhart]

Yeah, if you care about efficiency, as in max spark length per input power, then short on-times are the best way. The good performing long on time coils I have seen usually use on-times up above 1ms at least. and they draw tons of power.

The big benefit with that is getting long arcs at low pulse rates or even single shot.

[Jj]

Ok, thanks for the comments about longer on times, has cleared it up for me.

300us is about the third notch for my setup, so even by that point, which isn’t considered “long on time” the efficiency is very diminished. I like the idea of efficiency, but am only really concerned with it as it relates to IGBT life. I haven’t yet found the limits of the 1200V mini blocks with 500A and 600V which I’m happy about, though it probably is time to upgrade them.

[Benjamin Lockhart]

I also think that the benefit with really long on-times comes when you're trying to get insane spark length for the coils size.

For example high enough primary current with a really low impedance primary will cause racing arcs. A lower rate of current ring up should reduce the risk of racing sparks I think. Then you can increase the input power by turning up the on-time to grow the sparks over time.

I don't think you're hitting the racing spark limit yet, so more primary current would be beneficial.

[Benjamin Lockhart]

I just had another successful run! Here are two videos.
I still don't know why strike rail hits sometimes trigger the 2 second timer in the UVLO (under voltage lock out) circuit of the UD2.7 though?

/>

[davekni]

I still don't know why strike rail hits sometimes trigger the 2 second timer in the UVLO (under voltage lock out) circuit of the UD2.7 though?

Strike rail to ground current is flowing through some path that couples enough into electronics to cause unintended operation.  My first guess would be SMPS generating 24Vdc for UD2.7.  SMPS's are often sensitive to interference.  It may be shutting down momentarily.  UVLO may be operating correctly in response to losing voltage.  Or could be direct interference with UD2.7 UVLO circuitry.

[Admiral Aaron Ravensdale]

Completly nice build with exreme power!
When I build with a friend our BIG coil with CM300 we had also trouble until we used a grounded housing for the UD2.7.
After that we got also issues with the E-Stop because of interferences with the IC whats also could be fixed with a grounded housing.
In  the first pictures it seem you run it completly open.

CM300 are  really durable this coil runs since nearly 6 years