[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Re: [TCML] A s.s. "ring bridge"

Steve (& all)-

Glad you see some value in it! As to the C.M. chokes, simulation of 2 mH/coil toroidal chokes that I have on hand shows coil current starting at 0, with only 150 mA p-p Fr ripple, and increasing to 1.6 A at the end of 200 us. The 1.6 A is from the mains & so flows oppositely in the 2 coils & thus cancels out. The 150 mA would be additive, for 300 mA total per choke.

As to your simulation, I'd be happy to send you my SiMetrix file if that would be useful. "sxsch" extension.

It should be noted that this scheme is amenable to expansion: just add pairs of transistor/capacitor/snubber sections (plus isolated drivers, of course) ad infinitum.

I just laid out an assembly and got it all--not including the driver transformers--comfortably within an 11" diameter circle. I'd plan to mount it all on a disk (likely using a commercial serving-tray, for convenience) and put the drivers directly below on another tray.

If I do go ahead with this, I shudder at the MOSFET cost: some $54 each! Have to keep that from the wife...

Next thing I'll be doing is simulating the pretty large capacitances of the MOSFET in my PNP/NPN driver circuit: some 25,000 pF input capacitance.


Steve Ward wrote:
Hi Ken,

Good to see you're still thinking about this stuff!

Something this complex would require me to run the simulations myself to
really get all the details of whats going on.  But i had just a few concerns
anyway, which perhaps you already considered.  The main thing was with the
CM chokes, i was curious just how much current they needed to handle in CM,
and if this amount of current would impose some nasty consequences for
actually building something?  I realize they provide brief isolation, but my
instinct says this is a possible issue that may require some brute force.
Anyway, if a CM inductor is to have good DM (differential mode) properties,
it usually implies a gap-less core of ferrite or laminated steel so that
there is very little leakage between the 2 coils (that is, their mutual
inductance is large).  Now, this inherently implies that in the common-mode,
it cannot handle much flux in the core as an inductor, because the lack of a
gap.  Perhaps im getting excited over nothing...

Other than that, i must say its pretty clever.  I'll try to get it going in
my own simulator sometime and see what the real trade-off is for all that
extra voltage gained by the clever arrangement of parts.


On Fri, Aug 21, 2009 at 10:07 PM, Ken or Doris Herrick <kchdlh@xxxxxxxxx>wrote:

I'd like to solicit opinions from the solid-staters on a scheme I've been
toying  with.  See http://drop.io/kch_ring_brg for schematics.

Ring-bridge.jpg shows the simulation drawing.  It incorporates 6 power
transistors, MOSFETs or  IGBTs, plus 6 capacitors, all connected into a
ring.  (In the freebie-simulation, I have to use switches rather than
transistors.)  Each capacitor is charged from the mains via an isolating
"common-mode" choke, and it is kept isolated from the others, for charging
from the mains, while the interposed transistors are off.  Each transistor
is to be driven via an isolating transformer.  Feedback to start & maintain
oscillation is taken from the return lead of the Tesla coil's secondary and
the primary is untuned.

The primary is connected to opposite sides of the ring.  In the drawing, it
connects to the "emitters" of S2 and S6.  Half the transistors (S2, S3 and
S5) are turned on during phase 1 of the drive and the others, during phase
2.  Because the capacitors are inductor-isolated from the mains, each set of
3 becomes effectively connected in series during alternate half waves of the
drive signal.  In that way, +/- (3x300V - the various IR drops) becomes
connected across the primary during each Fr cycle.

The diodes D9, D10, D14, D11, D12 and D13 perform the function of clamping
each reverse-Vce transistor voltage to 600V peak, while at the same time
obviating an over-voltage problem should the transistors' turn-off times

In the simulation it works like a champ.  In the hardware I would plan to
use ST STE40NK90ZD MOSFETs and CDE 942CDW2K capacitors--paralleled with
close-connected electrolytics--configured generally as shown in the other
drawing, RING-BRG.jpg.  (Sorry about the low-res.)

RING-BRG.jpg shows what basic hardware configuration I have in mind.  Its
main feature is a scheme for keeping stray inductance to a minimum by the
use of coaxial copper braid surrounding each tubular capacitor.  The
charging chokes connect anywhere convenient, of course, but the clamping
diodes are to be connected closely to the ends of the respective braids as
shown: to a "+" at one end and a "-" at the other.  The capacitor leads are
similarly connected.  I don't show the electrolytics, which would be
connected as closely as possible to the CDE capacitors.  It's my feeling
that by properly configuring the braid-ends and the clamping diodes and by
conforming the braids closely to the o.d.s of the capacitors, I can
effectively minimize the stray inductances in the ring of diodes and
capacitors.  And, with this configuration the MOSFET connections at the same
time can easily be kept very short.  I would add varistors across the
MOSFETS for additional insurance.

I show a simple transformer-isolated PNP/diode gate-drive scheme but I
suppose the rather high gate-capacitance of the ST MOSFETs might require a
more robust drive.  I've developed a floating-supply drive with a PNP/NPN
pair that might do it.

The circuit has the advantage of safely allowing use of the 900V MOSFETs
while delivering perhaps +/-800V, and a hundred A or so during the
pulse-burst, into the primary--straight off-line from the 115V mains.  I
plan to use MOSFETs rather than IGBTs because the Fr's of my two secondary
coils are 100 and 120 KHz--perhaps a little too high for reliable IGBT
switching, right?

This scheme is along the lines of the one s.s. coil I managed to get going
some years ago (see the photo at http://drop.io/pat_ck_dblH).  That was
pretty satisfactory until it (was caused to) quit for good, and I'd look for
similar performance--with much less complexity--out of this new arrangement.

So...will anyone shoot it down?  If it looks feasible I may or may not give
it a shot (maybe too old...rocking-chair beckons), but I would hope someone
else might have a go.

Ken Herrick

P.S.  I'm to have a small "Gadget Freak" piece in Design News magazine
soon, about my T.c. work & with a pic of the 1-&-only coil.  Don't know if
it's just on-line or also in the printed mag.


Tesla mailing list

Tesla mailing list