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

Re: OLTC update - primary IGBT loss



Original poster: "Paul Nicholson by way of Terry Fritz <twftesla-at-qwest-dot-net>" <paul-at-abelian.demon.co.uk>

Terry wrote:
> I wonder if coupling in three spread out rings hurts the waveform
> on the secondary.

I wrote:
> No sign of a problem.  The FT of your beat envelope shows only the
> normal amount of higher mode ringing.  No sign of HF modes from
> the primary either, so no racing arcs from this cause!

On a closer look, there is quite a large initial excitation of higher
modes, but these are decaying quite rapidly due to low Q.   So overall
the energy is quite low on the FT, but it's concentrated in the first
cycle or two of the bang.  The resulting initial high dI/dt near the
coil base might eventually give an excessive vertical gradient across
the lower turns.  

Terry wrote:
> I hooked up the 60kV Jennings probe to the OLTC tonight:
> http://hot-streamer-dot-com/temp/OLTC09-03-02.jpg
> http://hot-streamer-dot-com/temp/OLTC09-03-01.gif
> http://hot-streamer-dot-com/temp/OLTC9-3v.CSV
> http://hot-streamer-dot-com/temp/OLTC9-3i.CSV 

Ok, I got the model lined up.  I've setup a single turn dummy primary
to have the same k and Lp as your system exhibits. I've had to do
three 'fiddles': An extra 7pF to the topload to allow for the Jennings,
and some extra Rp and Rs to line up the model's Q factors with results
of tcma analysis of your ringdowns.  The rest of the model comes 
entirely from your geometry.

The resulting time domain response of the model compared with your
waveforms is in 

  http://www.abelian.demon.co.uk/tssp/tmp/OLTC09-03a.gif

The match of the higher mode components is quite poor, Q, F and phase
errors are showing up here - thus the fine detail of the waveform is
not properly modeled.  This may be due to the coating you have over
the secondary which will modify the internal C in an unmodeled way,
and may also be due to my 'virtual primary'. Still, it's not a
problem for this exercise so I won't investigate further.

The primary Q is very low at these small voltages.  That's going to be
an embuggerance later on - if Q depends strongly on Vpri then it will
be tricky to compare below/above breakout waveforms.  To get around
this we'd have to characterise the primary Q/Vpri function, as we've
already done, but more accurately.  Then we can make proper allowance
for the intrinsic primary Q as the voltage goes above breakout.

Incidentally, the sec Q is up a little from 45 to 53, and the tuning
seems a little funny because the Fsec has dropped due to the probe
but Fpri remains the same.

Returning to the issue of higher mode excitation, note your initial
-ve going Ibase transient peaks at roughly half the peak Ibase.
Compare that with say Marco's Thor system.

 http://www.abelian.demon.co.uk/tssp/md110701/

or some of your earlier solid state gap tests,

 http://www.abelian.demon.co.uk/tssp/tfss270501/

In these and other similar cases, the initial transient Ibase is
only around 1/10th or less of the peak Ibase.

So I'd say the higher modes are quite high, but just look weak on the
FT because they decay quickly.  

Note that this higher mode excitation is not a result of your
switching, but is intrinsic to the geometry of the resonator - your
OLTC dual resonator has a rich and bright tone with a quick decay - a
bit of a tin can sound if you could hear it.

The animation is in 

  http://www.abelian.demon.co.uk/tssp/tmp/OLTC09-03a.anim.gif

and you can clearly see the initial transient ripple.  It doesn't seem
to excite much voltage along the secondary, but that's a little
misleading, because it is the dV/dx that causes the breakdown. 
The voltage gradient animation is

  http://www.abelian.demon.co.uk/tssp/tmp/OLTC09-03a.grad.gif

which shows the radial and vertical E-field strengths along the
surface of the secondary.  Now you can see where the potential (excuse
pun) for breakdown comes.  If your secondary is going to break down at
some power level, it will be in the bottom 10% of turns, I'd say.

All in all, I'd say the OLTC looks reasonably manageable for some
breakout tests.

> I am wondering if this is useful to you

Oh yes,

>  and what you would like to see?

Breakout!  But from a well defined surface, eg a small sphere or
toroid.  I have to model its surface field so it must be on the
axis of the coil.  I guess it will take some experiment to find a 
low voltage breakout from a suitable small ROC object, bearing in mind
the Jennings ratings.  I'll bet you could push those ratings a bit
though :)

Question is, can you obtain a suitable breakout in a single-shot mode?
Because I'd like to look at the waveforms for a single shot, then for
the second shot of a pair, then the 3rd shot of a triplet, etc. 
--
Paul Nicholson
--