From: "Terrell W. Fritz"
Date: Sun, 15 Sep 2002 19:24:35 -0600
Subject: Re: [TSSP] Top Voltage
Hi All,
I was interested in where the losses are in the OLTC system so I worked it
out. Using the best guess model and using the Lee and Cee values:
http://hot-streamer.com/temp/PaulArc/020914/00006.gif
I was able to model the system pretty well:
http://hot-streamer.com/temp/PaulArc/020914/00006a.gif
http://hot-streamer.com/temp/PaulArc/020914/Tek00006.gif
I then found the instant power into R1 and R2:
http://hot-streamer.com/temp/PaulArc/020914/00006b.gif
Finding the peaks to 52uS and summing them up...
Primary Secondary
13292 301
8323 1826
3024 4587
92 6732
-------- ----------
24731 13446
Knowing the integral of sin^2 = pi/4 (it really is a decaying sine but...) and
the time 1/(2 x 38500) we get a factor of 1.02e-5 to convert the above to
Joules.
Primary = 0.252J
Secondary = 0.137J
So we start with 0.94 Joules in the primary caps.
0.252 Joules is lost in the primary (mostly IGBT loss).
0.137 Joules is lost in the secondary
That leaves 0.551 Joules in the system at peak power. Yesterday, I got 0.5725
Joules for the peak secondary energy so things are fairly close.
So "if" the streamer really is just basically a big single shot event at peak
voltage, the Q's of the primary and secondary have less importance than if the
streamer is a prolonged event. It certainly appears that the streamer can
consume 0.213 Joules of energy over just one cycle.
So increasing the secondary Q of my OLTC my not have a dramatic effect and
Antonio and Paul suspect. Primary Q is much more important and that keeps
getting better as power is increased (to some extent since the IGBTs are
showing resistive effects at 3X their rated current ;-))
So it appears a TC should do the following:
1. Have a high coupling value to transfer energy to the secondary very fast
before losses can eat it.
2. Primary Q must be kept as high as possible.
3. Secondary Q must be kept high too but this seems to have lesser importance.
4. The streamer seems to use a lot of energy very early when the peak voltages
are reached. Large top capacities may store this energy better than the coil's
intrinsic capacitance and allow it to be delivered faster. Large top loads
also increase secondary Q.
Of course, this is all for a single shot event. Streamer formation may
(probably does) need other factors too. but this indicates what it seems to
take to get them started.
One very odd thing is that the voltages here seem to confirm at about 170kV.
However, many times the 1" brass ball does not breakout!! That is about 5X
(Paul gets 2.5X) the voltage is should take... So breakout voltage is going to
be tricky....
Cheers,
Terry
At 09:22 PM 9/14/2002 -0600, you wrote:
>Hi All,
>
>Being a little frustrated with the fiber probe tonight, I whipped out the
"good
>o'l trusty" plane wave antenna. I hooked the pinger to the coil with a TEK
>6015 HV probe and got the plane antenna calibrated (1V = 88.816kV). I then
set
>the firing voltage at 200VDC which is where the small ball sometimes breaks
out
>and sometimes does not this particular night...
>
>http://hot-streamer.com/temp/PaulArc/020914/Tek00006.gif
>
>I figured out the current directions tonight too and decide just to set things
>so "they look right". I know what winding direction the secondary winding is,
>but with a one turn primary.... Aruugh!! Darn Pearson! "Positive output
>for electron flow in direction of arrow"... Of course electrons flow in the
>"wrong" direction... I think Ben Franklin did that one... Oh well... I
>decided just to "make it look right" :-)))
>
>So, here is some real fun!
>
>For the input:
>200 VDCfire
>Cpri = 47uF
>ECpri = 0.94 Joules
>
>Secondary peak voltage (no breakout):
>164kV
>Csec = 35pF
>ECsec = 0.471J
>
>Secondary peak current(no breakout):
>1.68 A
>Lsec = 0.492H
>ELsec = 0.694J
>
>I can see having a bunch of loss from primary to secondary, but the secondary
L
>and C energies "should" match. But Paul's work tells us the L is really less
>and I assume C is really more in the resonant case. so I used:
>
>1/2 x C x V^2 = 1/2 x L x I^2
>and
>Fo = 1 / (2 x pi x SQRT(LxC)
>
>Fo=38300Hz
>
>as "two equations and two unknowns" (I almost hurt my brain here! :o))) to
>find:
>
>C = 42.57pF
>and
>L = 0.4056H
>
>Ldc = 0.492H and Cdc = 35pF
>
>So Leq/Ldc = 0.8244 and Ceq/Cdc = 1.216. I may have the eq or ee or whatever
>terms mixed up. Have to review Paul's paper for the proper terms... But
>perhaps these numbers match what they should be as sort of an independent
>experimental confirmation.
>
>So with the new values, I get a secondary energy of 0.5725 Joules. So out of
>the original 0.94 Joules, 61% made it to the secondary's peak power. 39% "got
>lost" even with a fast k=0.25 ring up.
>
>I also compared a breakout to no breakout energies (165kV vs. 131kV) and
found
>a breakout event consumes 0.213 Joules in this case. That is about 25% of the
>original primary energy and 1/3 of the available secondary energy for just a
>bare minimum breakout. Assuming the event time is 100nS and the voltage is
>150kV, we get a current of 14.2 amps. That suggests that just a "little"
>breakout is a 2,000,000+ watt instantaneous event!!!
>
>Also note that the 1 inch diameter sphere's radius is 1/78.7 of a meter. That
>suggests that the breakout voltage is 3MV / 78.7 = 38100 volts.
>HAHAAHhahaaohoho.... Seems to be taking 150kV!!!!
>
>So that's what I know tonight... Many more questions than answers... I guess
>if this stuff was easy, everyone would be doing it :-))) I did order up that
>nice P5205 differential probe. So at least I got a new toy coming :-)))
>
>BTW - Paul mentioned about divider caps for the Jennings probe. They
>apparently "did" make "one" to go to 120kV. I guess a 3pF 60kV shielded cap
>that goes in series with the probe. I think Jennings made like "two" and the
>recipe died with Joe Jennings there... The present staff at Jennings could
>probably look the details up in their "paper room" (realizing, they hate
me...)
>but trying to find one, I would think would be impossible...
>
>Cheers,
>
> Terry
>
>
>
Maintainer Paul Nicholson, paul@abelian.demon.co.uk.