From: Paul
Date: Sat, 14 Sep 2002 21:09:46 +0100
Subject: Re: [TSSP] Top Voltage
Hi Terry,
> Nope. that was just the bare toroid without the ball.
And without the rod too? If so, that gives us
Measured Modelled Error
f1: 37.801 kHz 37.788 kHz 0.0%
f3: 113.550 kHz 114.107 kHz +0.5%
f5: 171.773 kHz 174.683 kHz +1.7%
Nice set of measurements on the breakout. The breakout
thresholds were
Vfire: Measured Modelled
#1 Small Ball 119 VDC 80V
#2 Medium Ball 191 VDC
#3 Big Ball 226 VDC 130V
so the actual breakout appears some 50% to 70% higher than predicted.
We'll have to look more closely at this.
> They also still almost always go negative
There's something strange here. You did some calibration runs on
09/03 with the Jennings on Vtop and a current sensor on Ibase,
to get
http://hot-streamer.com/temp/OLTC09-03-01.gif
Chan 1 (yellow) is Ibase, and chan 2 (blue) is Vtop * 0.001352,
with the data files in
http://hot-streamer.com/temp/OLTC9-3i.CSV
http://hot-streamer.com/temp/OLTC9-3v.CSV
If you plot these, you'll see that the polarity of your windings is
such that negative going zero crossings of Ibase coincide with
positive going peaks of Vtop.
Now, you're showing breakout in
http://hot-streamer.com/temp/PaulArc/020913/Tek00001.gif
http://hot-streamer.com/temp/PaulArc/020913/Tek00002.gif
http://hot-streamer.com/temp/PaulArc/020913/Tek00003.gif
consistently occuring at a negative going Ibase zero crossing,
therefore a positive Vtop peak.
Consider the sinusoidal component of the fibre probe trace
(yellow) which shows the current charging the fixed cap of
rod+sphere. The polarity of this suggests that you have
the probe signal inverted somewhere. Probe current should
lead the voltage waveform, not lag. For example, consider
the probe trace just before the breakout spike. Voltage is
on a rising positive slope, so current will be flowing *into*
the rod+sphere cap. Your probe signal shows this as negative
current, ie you are reporting current flowing *into* the topload,
not out of it.
Sure looks like your breakout is occuring on +ve half cycles and
that the fibre probe is inverted somehow.
> The breakout point was very consistent and easy to adjust.
Fantastic. A stable test setup is the best thing we can hope for,
the rest is up to us!
> I went back in with the plane wave antenna and caught one of
> the Big Ball arcs:
> http://hot-streamer.com/temp/PaulArc/020913/Tek00005.gif
> Much different than:
> http://hot-streamer.com/temp/PaulArc/020913/Tek00004.gif
> I suspect the fiber optic transmitter just saturated it all
> into a blob
No, the difference is expected. The two traces are quite
consistent. You must have had the E-field probe below the toroid,
and we see the drop in Qtoroid (remember, the field probe responds
to charge Qtoroid not Vtoroid) as some of its charge shifts into
the streamers. The nice thing about this small-sphere breakout is
that the streamer charge is largely shielded from the E-field
probe by the toroid, so that (unlike full breakout from the
toroid) we can make some use of the E-field probe. On the face
of it, according to this E-field probe signal, roughly 50% of
toroid charge has shifted into the streamer - I suspect you've
got a large vertical offset on the chan 1 trace? The step fall
in Tek00005.gif should be about equal to the integral of the
current pulse in Tek00004.gif.
> You can see how the single pulse drained energy from the system
There's a drop of amplitude of around 5-10% (need CSV data for
accurate measure). That means that either energy has been lost
from the system and/or the effective Ctop took a step increase.
We expect both to occur and we need accurate data files in order
to separately quantify the two effects. I think 10uS/div looks
like the best compromise (need to see the first 80uS of the event,
but in enough detail to integrate the spikes).
To summarise, it looks like your OLTC with rod and small sphere
terminals is working very well for studying breakout. I think
you should continue with the 0.1 ohm current shunt in the probe.
Don't worry about the ringing - it's the area under the curve of
the pulse that's important, and that won't be affected too much
by the ringing. Wonder if the Jennings can be equiped with a
divide by two or by four front end? I can only have two traces,
so I'd prefer to see fibre probe on ch1 and Ibase on ch2, and not
use the E-field probe for the moment.
Cheers,
--
Paul Nicholson,
--
Maintainer Paul Nicholson, paul@abelian.demon.co.uk.