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

Re: Snubbing an IGBT in a sstc



Original poster: "K. C. Herrick" <kchdlh@xxxxxxx>

Steve (& all)-

Comments interspersed...

KCH

Tesla list wrote:
Original poster: "Steve Ward" <mailto:steve.ward@xxxxxxxxx><steve.ward@xxxxxxxxx>

Hi Ken

Sorry, i forgot to post about your snubber idea earlier, since when i
first saw the email, the link didnt work yet (it works now though).
Yeah, we need to give Chip enough time to do his most-appreciated work.
I think this snubber design is something i will have to simulate for
myself and get a good understanding of its operation and limitations,
so i will try to comment on that when i have had the time to do so.

But id like to address some of your other comments below:


It might be thought that in doing so, the magnetic flux entering the
secondary is diminished during this process since the two primaries'
currents oppose...and that is exactly so.  However:  If a) one can
draw current from the mains up to the limit imposed by the circuit
breakers; and b) one can draw that current without exceeding the
capabilities of the driving transistor(s), then...what's the
problem?

I would think that the circuit breaker shouldnt mind.  The power isnt
being wasted as you said by the snubber, but its being recycled back
to the main supply.  The mains line shouldnt see this excessive
current being "pushed around" within the circuit, just as my mains
doesnt need to supply the 1500A kicks that my largest coil requires.
Agreed.

 There's relatively little in the circuit to dissipate power
since the snub capacitor's charge is recycled.  So the major fraction
of the mains power must be going into the secondary; ...where else?.

Primary resistive losses might surprise you!  But i agree, seems most
of the power *should* go to the secondary provided the primary side
isnt lossy for some reason.
I'm using heavy copper-braid for the (short) leads to the primary coils, and those "coils" are made from 0.011" Cu sheet-stock with 6 1/2" i.d. and 12 1/2 " o.d. so their resistance, both dc and ac, will perhaps be pretty low.


And that brings me to a "philosophical" question:  Given the above,
wherein lies the advantage of a resonant primary in a s. s. t.
c.?  If criteria a) and b) above were to hold for the resonant case
on the one hand and for the non-resonant on the other, then--where's
the advantage of the resonant primary with its added
complication?

What do you do when you arrive at only 1 primary turn on your system
and you still want more power?  You cant just up the voltage since
higher voltage silicon costs lots of green.  The resonant primary
allows you almost endless power from your inverter, provided it can
supply the current.
"...provided it can supply the current. " That's the kicker: I'm running from plain-vanilla 115 V mains so am limited to 30 A absolute rms max. My simulations run to 40-60 A rms at 1% duty-cycle with 2-turn coils so I figure I'm already going to be pushing it, untuned.

BUT, if your application is satisfied with a 1 or 2 turn, un-tuned
primary circuit, then there is no reason you need a resonant primary.

I personally dont find the tuned primary to add complication, except
for your case, where you are using IGBTs that are too slow for the
driving frequency.  But, i do feel each system has its place... i
still have a big "untuned" primary SSTC in my garage ;-).  Its used
for producing massive flames of plasma.
My 1-&-only prior sstc, untuned, produced very satisfying fat, bushy sparks at upwards of 5 ms duration and from all around the 6" x 24" Landergren toroid (utilizing no breakout-point). But...they were only 30" or so long. And I hadn't seen any others' images of their t.c.'s that produced similarly-appearing sparks so I'm wondering if it was the spark-duration that caused the difference. The present simulation implies that I won't be able to achieve that kind of duration with this design, however...unless I'll be satisfied with a rep-rate of 1/s or so!


 Why not use a non-resonant one, take feedback from the
secondary, and have the thing always in tune?

Yep, that works very well.  I should mention that sometimes secondary
feedback can be spoiled by having the arcs jump to ground.
I remember that I had that happen once or twice. The sparks actually jumped to the MOSFET drains. Needless to say I was horrified, but no damage ensued. That current is relatively low so the only damage might be carbon-tracking in the secondary. But even then...unlikely since the sparks issue from the outer areas of the wire (where, in my case, there's only thin varnish in the way).
  Im not
sure what exactly happens, and ive seen that some coils do not exhibit
any problem with this, but its something to be careful of.  Test
ground sparks at very low power first (maybe a few inches of spark at
a low input voltage) to see if they effect your feedback loop too
much.  Thats an added bonus to primary feedback (which of course,
requires the tank capacitor then), you cant mess up the primary
oscillations from what the loosely coupled secondary is doing.
Hmmm...don't follow that.


It might be thought that the flux from the resonant primary could be
greater since its inductive reactance is largely "nulled out" by the
capacitive reactance, so more turns could be employed, carrying the
same current.  Flux varies as current x turns.  But one still faces
the limitations of a) and b); if they're the same in both cases, then
does that matter?  It might seem, in fact, that the added primary
turns would be detrimental, diminishing as they would the turns-ratio.

But wait...  If you have higher current through MORE inductance
(turns) at the same frequency, then that implies your voltage is
higher too (and it is, often in the kVs)!  So the V/turn wouldnt be
diminished here (not to say that i buy into the need for high turns
ratio, my coils just happen to have them for other reasons,
particularly a low operating frequency).
Agreed. But one is still limited by how much rms current can be drawn from the mains, & peak current thru the transistors.

Anyway, i think you are right that the output current and frequency
are what matters, but you must realize the constraints that an untuned
primary will put on the output current.
Which are...?
  And since its coupled to the
secondary, the primary impedance will appear to be greater than just
that of its own inductance at whatever frequency you are driving it.
But perhaps not by a whole lot, with a k of only 0.2 or so. I don't remember the relationship.


Enlighten me, someone.  Perhaps, in my old age, I'm becoming
forgetful--of basics I should long ago have learned.

Im only 21 and i forget things all the time ;-).
Yeah, and I'm 79--as of yesterday, as a matter of fact--and have quite possibly forgotten more than you've yet learned. Be sure to stick with the learning...at the necessary expense of the Tesla-coiling. But coiling involves learning too, of course, tho perhaps not with very much remuneration as an object.

I will try to look at your design soon (when school permits me to have
time for myself) and let you know what i think of it.
How about a few more interested parties throwing in their two-cents' worth?...

Steve

KCH