Re: [TCML] BIG solid state

[Steve Ward <steve.ward@xxxxxxxxx>]

Hi Mike,

On Wed, Jun 15, 2011 at 8:36 AM, Michael Twieg <mdt24@xxxxxxxx> wrote:

> So Steve, was there a specific goal in mind here, or at this point have you
> just gone mad with power?
>

Well, it started off as "what to do with this big PVC tube and spools of
wire".  Over the years i collected parts, and happened to find those MMC
caps for a great price.  More recently there was some possibility of using
coils of this size for a burning man exhibit, and then possibly for Comic
Con in San Diego, but both of these prospects have now fallen through.  But
honestly, I mostly wanted to do it to see it happen.


>
> Anyways, very impressive.
>
> My main question is about the primary control.  You're still using primary
> current feedback, right?


Yup.


>  Does each bridge have its own CT and detect its
> own zero crossing?


Not quite.


>  Or is there one CT which controls all the bridges?
>

Almost.


> Sounds like it would be very important to have identical tuning on all four
> primarys to avoid chaos.
>

Eh, not really as hard as it might seem.  So yes, there is 1 master
driver/controller that takes the *summed* currents from all 4 primary
circuits.  The summing is done by using 4 small CTs (one on each primary
conductor) whose outputs all run through a 5th CT, which steps down the
current again, and feeds this into the controller board.  If the primaries
dont have exactly matching L's and C's and loading, then there is simply a
small phase error and slightly different primary current per coil.  I
measured each bridge output individually and found the currents to match
within 10% or better.  Also saw that the phase error was acceptably small,
meaning some IGBTs might switch a bit earlier than others, but i seem to
recall being able to achieve nice switching waveforms on all the bridges
relatively easily.

I did find that i needed to add bleeders on the bridge outputs to discharge
the MMCs before each "bang".  Typically, the MMC can be left with any
voltage that is less than Vbus.  I suspect due to variations in IGBT leakage
currents, i was getting different voltages left on each MMC, and so when the
system would start up, there would be different starting conditions on each
primary.  This did seem to cause some strange behavior, but adding 1600 ohm
power resistors across the output of each bridge discharges the MMC between
firings, and suddenly everything worked wonderfully.  I suspect the problem
should be even less noticable when making big sparks, as this should help
discharge the system energy as well.  I didnt take very good notes on what
the strange behavior was without the resistors installed, but i seem to
recall that it improved current sharing, and eliminated some issues i was
having with improper switching on the first 1-2 cycles.


>
> Also please post some waveforms if you have some.
>

I do have some, though it might be a few days before i have time to mess
with posting them.

Steve


>
> -Mike
>
> On Tue, Jun 14, 2011 at 11:35 PM, Steve Ward <steve.ward@xxxxxxxxx> wrote:
>
> > Hi Everyone,
> >
> > Just a post about a recent project of mine.  Many hands involved on this
> > one
> > (see credits later).  So far its just a prototype, but results are
> > promising:
> >
> > http://www.flickr.com/photos/kickermagnet/5811280440/in/photostream
> >
> > This is, as far as im aware, the largest Double Resonant Solid State
> Tesla
> > Coil built and working as intended.  The machine stands 15 foot tall (to
> > the
> > top of the toroid).  With breakout point at 16 foot above the ground, its
> > made some 25 foot sparks.  Power input at 200bps was measured at 33kVA
> > (480VAC 3 phase, about 40.5A RMS).  I estimate a power factor of .92
> given
> > the line reactor used to smooth the current peaks charging the 2kJ of DC
> > bus
> > storage.  The secondary is 24" diameter and wound to just over 8 feet
> > length.  The toroid is 96" by 16", and is indeed very ugly.
> >
> > For now, i dont have many photos of the detailed construction (this is
> > something i should get to soon) but you can sorta see whats going on
> here:
> >
> > http://www.flickr.com/photos/kickermagnet/5487415037/in/photostream
> >
> > <http://www.flickr.com/photos/kickermagnet/5487415037/in/photostream>On
> > the
> > lower level, you can see banks of white electrolytic DC capacitors,
> 3300uF
> > 400V X 16 total, wired in series parallel banks, one per each of the 4
> IGBT
> > inverters.  The inverters consist of a full-bridge of CM300DY-24H
> > transistors.  So each pair of half-bridge bricks (there are 8 total, 4
> > H-bridges) drives its own .625uF 20kVAC capacitor, and 2-turn primary
> coil.
> >  There are 4 MMCs and 4 primaries.  The mutual inductance between
> primaries
> > and equal MMCs force current balancing of the 4 drives (each drive sees
> the
> > same load impedance).  This was chosen to avoid some possible issues with
> > directly paralleling modules, such as, what happens if 1 device fails?
> >  Does
> > it take out all the rest?  In this scheme, there is DC isolation between
> > inverter outputs, so one inverter failure should not take out the other
> 3.
> >  And, in fact, if i did lose an inverter, i should be able to short its
> > tank
> > circuit (using its energy storage of the MMC to maintain tuning) and
> > continue operating at 3/4 power.  Hopefully it never comes to this!
> >
> > The coil has a natural resonance of about 45khz, but i found that
> de-tuning
> > the drive down to 40khz or so was best to keep it in tune with large
> > streamer loading.  The primary current peaks at 6000A after about 6
> cycles
> > of excitement.  The coupling coefficient of the system is fairly high, at
> > about .19 (estimated by a 5 cycle beat period that didnt quite notch).
>  So
> > detuning the primary seemed to work well at efficiently transferring
> energy
> > to the secondary in the event of being detuned by streamers.
> >
> > The primary copper is 3/8" diameter to keep the surface area down (which
> > happens to reduce AC losses as the eddy currents are lessened).  I pump
> tap
> > water through the primary to keep it cool, as the RMS current is
> something
> > like 80-100A per primary coil!  Me (tall guy) and friend Mike working on
> > the
> > primary coils:
> >
> > http://www.flickr.com/photos/kickermagnet/5487414563/in/photostream
> >
> > The MMC is made out of giant polypropylene snubber capacitors, rated 5uF,
> > 2500VAC pk, 3600Apk and 80A RMS each.  They make the CDE MMC caps look
> very
> > small indeed.  I think the whole MMC bank (32 of these caps) weighs
> > something like 150lbs.  Total bank is 4 strings of 8 in series, 2.5uF
> > 20kVAC
> > effectively.
> >
> > RF grounding was primarily a 4x20 foot sheet of metal "cloth" (1/2"
> square
> > steel mesh) laid on the parking lot, and also clipped to the building
> > ground
> > rod (about 40 feet away).  Would have preferred an even larger sheet of
> > metal down there, but this was available and quick, and seemed to do the
> > trick!
> >
> > Spark performance ended up being right on target, at 25 feet, with
> > extrapolations from smaller DRSSTCs (that make 10-12 foot sparks).
> >
> > I gotta thank the help of many friends, Jeff Larson for his use of
> > materials
> > and labor on the secondary coil for this beast.  Big thanks to the guys
> in
> > Arcattack, who stayed at my house for a week this january to help me
> crank
> > out this prototype (we actually got it running by the end of a week!).
> >  Terry Blake has been supportive in helping setup and teardown of this
> > system, which isnt trivial, and for being the guy in the suit.  And
> > recently
> > Jimmy Hynes (who got me working with DRSSTCs originally) who visited and
> > instigated the testing of this machine, and helped tune it up to almost
> > perfectly match our expected outputs. And Dave (who isnt on the TCML) for
> > being such an awesome host, letting us run and store this machine at your
> > shop, and dealing with the city and fire department on issues that really
> > dont concern them, thats a story for another time.
> >
> > Steve
> > _______________________________________________
> > Tesla mailing list
> > Tesla@xxxxxxxxxx
> > http://www.pupman.com/mailman/listinfo/tesla
> >
> _______________________________________________
> Tesla mailing list
> Tesla@xxxxxxxxxx
> http://www.pupman.com/mailman/listinfo/tesla
>
_______________________________________________
Tesla mailing list
Tesla@xxxxxxxxxx
http://www.pupman.com/mailman/listinfo/tesla