Measuring Streamer Characteristics Discussion
Edited/Updated: October 20, 2004
Date : Wed, 29 Sep 2004 12:16:42 -0600. Subject : Measuring Streamer characteristics
Original poster: "Bob (R.A.) Jones"
Hi all, It would be nice to be able to visually monitor how a streamer
grows but what we really need is its electrical characteristics.
Suppose we took a toroid and drilled a hole in its outer edge then mounted a
round ended metal rod in the hole such that it protruded an adjustable distance
from the toroid surface and insulated from it.
Next we add a fiber optic current probe between the rod and toroid to monitor
the current and add an antenna to monitor the toroid voltage. Perhaps if
the antenna is mounted on the opposite side to the break out point it would only
be weakly effected by the streamer
Using this set up it should be possible to monitor the current flow in to the
streamer directly and adjust the break voltage.
Perhaps a target can also be set up and its current flow to ground monitored to
check that a strike has occurred.
Bob.
Date : Wed, 29 Sep 2004 19:59:20 -0600. Subject : Re: Measuring Streamer characteristics
Original poster: "Jim Lux"
At 12:16 PM 9/29/2004 -0600, Tesla list wrote:
Such an experiment has been conducted by Greg Leyh during testing of
Electrum (which had a top load large enough to put a person inside)..
This WOULD give very useful data. One would also need a fast digitizer,
because each streamer is different, so you need to capture single shot data.
Date : Wed, 29 Sep 2004 20:25:26 -0600. Subject : Re: Measuring Streamer characteristics
Original poster: Terry Fritz
<teslalist@twfpowerelectronics.com>
Hi Bob,
Greg measured the streamer current for the Electrum directly from inside
the sphere (he was INSIDE it while it was running!) with a hand held
scope. I have a top load sphere with fiber probes built into it. This is
how we came up with the 220K +1pF per foot of capacitance model for a
streamer load.
http://hot-streamer.com/TeslaCoils/MyPapers/modact/modact.html
http://hot-streamer.com/TeslaCoils/MyPapers/smallar/smallar4.html
http://www.lod.org/electrum/electrumspecs.html
http://www.lod.org/electrum/electrumfull.jpg
Cheers, Terry.
Date : Fri, 01 Oct 2004 15:43:42 -0600. Subject : Re: Measuring Streamer characteristics
Original poster: "Bob (R.A.) Jones"
<a1accounting@bellsouth.net>
Hi all,
Nice work Terry and Greg very interesting.
I notice from both streamer current profiles there is an initial triangle
pulse of current approximately at the peak positive top load voltage.
The pulse has a duration of about 5us I assume this is the initial break
out. Given the different sizes of the two system the pulses are very
similar.
In the case of Electrum there are then three bursts of oscillation at
alternate voltage minima of the top load while the current builds to an in
phase value with the top load current.
Which means its 90deg out of phase with the top load voltage.!!! I suspect
the burst of oscillation is when the streamer connects to ground but then
why the difference in top load current and streamer current.?????
In the case of Terry's papers the second one seems to support the
oscillation idea when the streamer connects to ground.
Assuming the first paper is a streamer to air, in one cycle the streamer
current develops in to an in phase (aprox) current with top load voltage.
A visual estimate of the predominantly resistive impedance is 240kV/0.8A =
300k . The streamers were believed to be between 12 and 24 inches long.
Of cause a better value could be obtained from processing the data points.
Unfortunately we don't know if the streamer was growing during the
successive current cycles. Its length may have formed in the first current
pulse.
Some cams have exposures times of a 1us and many have 10us. Unfortunately
most don't have a sync input.
Even with minimum exposure times of 100us provided you could sync the start
of the exposure to a particular peak of the output voltage and the was some
repeatability form streamer to streamer it should be possible to determine
of the there is significant streamer growth during repeated cycles.
Assuming no sync is available perhaps a bar graph display could be
synchronized to the coil and just take video at the shortest exposure with
the bar in view with the streamer. Occasionally the start of exposure could
occur on a particular cycle of interest. Probably a very tedious experiment.
Perhaps the coil could be synced to the cam. This could be very simple for
those with the right kind of coil.
Thinking about the above presumably the guys with variable burst width SSTC's
can answer that if they have not done so already.
Bob.
Date : Fri, 01 Oct 2004 15:45:57 -0600. Subject : Re: Measuring Streamer characteristics
Original poster: "Bob (R.A.) Jones"
<a1accounting@bellsouth.net>
Hi all,
Some additional observations and some brain storming.
> oscillation idea when the streamer connects to ground.
>
> Assuming the first paper is a streamer to air, in one cycle the streamer
> current develops in to an in phase (aprox) current with top load voltage.
> A visual estimate of the predominantly resistive impedance is 240kV/0.8A =
> 300k . The streamers were believed to be between 12 and 24 inches long.
> Of cause a better value could be obtained from processing the data
points.
That's about 192kW peak required to grow a 12 to 24in streamer!!!!
Lets assume the duration was 25us the energy was 192,000 * .707*.000025 =
3.3J.
As cross check I checked what the primary bang J was.
0.5*16700*16700*2.7*10**-9= 0.377J !!!!
Obviously a problem some where. perhaps my calcs or interpretation.
I tried the Frau formula on 0.3777J * 120 and got 11.4in
What I was trying to get a feel for was what the input power had to be to grow
the spark In my proposed SSTC the power is going to be more like 7kW or less
(during the burst) This is much less than the above. So I assume after break
out and initial stream growth from the energy stored in the top load there would
be very little additional streamer growth from the power input. If that
level of power is required to grow a 12/24in streamer then a low
peak power (300V, 100A) SSTC must rely on getting the power relatively slowly in
to the top load and letting the stored energy do the rest after break out. It
may also mean that a low frequency(50kHz) system has an inherent disadvantage
because it is supplying the power over a longer period and hence the total
energy will be higher. i.e. if the streamer is growing from cycle to cycle
the longer the period the greater the energy required to maintain the streamer.
Does any one know if you can form a long streamer on the first peak or does it
always require several cycles? Bob.
Date : Fri, 01 Oct 2004 16:13:34 -0600. Subject : Re: Measuring Streamer characteristics
Original poster: Terry Fritz
<teslalist@twfpowerelectronics.com>
Hi Bob,
The actual streamer current is usually seen as a jagged edge on the current
graphs. This picture shows them well:
http://www.lod.org/electrum/sphere10us.jpg
There is always an initial very short but very powerful "burst" when the
spark gap first fires composed of MHz to GHz noise that swamps the
equipment. It often saturates the fiber optics creating longer apparent
pulse widths. also, when current is delivered, the top voltage may drop
drastically very fast.
In ground strike pictures like this:
http://hot-streamer.com/TeslaCoils/MyPapers/smallar/Image149.jpg
The powerful RF noise form the discharge simply wipes out the measurement
equipment and any scale factors are lost. Your just seeing radiated noise
into the equipment and grounds. One could conclude by the voltage and
discharge time what the current "might" be. But don't trust the scale of
the graphs at those moments.
I hope that answers your question. Plane wave antennas are far more
capable in many instances:
http://hot-streamer.com/TeslaCoils/MyPapers/planant/waveant3.html
Cheers, Terry.
Date : Sat, 02 Oct 2004 12:41:08 -0600. Subject : Re: Measuring Streamer characteristics
Original poster: "Finn Hammer"
<f-h@c.dk>
Tesla list wrote:
>Original poster: "Bob (R.A.) Jones"
<a1accounting@bellsouth.net>
>
> Hi all,
>
>snip
>
>Does any one know if you can form a long streamer on the first peak or does
>it always require several cycles? Bob.
All,
I can say, that with my Solid State Disruptive TC, 1 bang doesn`t bring much
neither in length nor light.
I am not the analytical type, so I don`t have a chart over bangs/length but to
get to the max. streamer length you need several bangs. I saw the same
behaviour in Steve Connor`s OLTC at the Derby Teslathon, but then, he built my
controller and my coil is a 1/2 scale model of his, basically. Cheers,
Finn Hammer.
Date : Sat, 02 Oct 2004 16:32:26 -0600. Subject : Re: Measuring Streamer characteristics
Original poster: "Dr. Resonance"
<resonance@jvlnet.com>
We looked at streamer formation once with a high speed camera (film type) at
it usually takes 150 to 200 uSec to get average peak length. This was over
12-15 cycles of 3 uSec burst peaks. Lots of ion formation is required to
grow the spark.
Dr. Resonance
>
> >Original poster: "Bob (R.A.) Jones"
<a1accounting@bellsouth.net>
> >
> > Hi all,
> >
> >snip
> >
> >Does any one know if you can form a long streamer on the first peak or
does
> >it always require several cycles?
> >
> > Bob
>
> All, I can say, that with my Solid State Disruptive TC, 1 bang doesn't bring
much neither in length nor light.
> I am not the analytical type, so I don`t have a chart over bangs/length but to
get to the max. streamer length you >need several bangs. I saw the same
behaviour in Steve Connor`s OLTC at the Derby Teslathon, but then, he built my
>controller and my coil is a 1/2 scale model of his, basically. Cheers,
Finn Hammer.
Date : Sat, 02 Oct 2004 22:47:56 -0600. Subject : Re: Measuring Streamer characteristics
Original poster: "Bob (R.A.) Jones"
<a1accounting@bellsouth.net>
Hi Dr
> Original poster: "Dr. Resonance"
<resonance@jvlnet.com>
>
>
> We looked at streamer formation once with a high speed camera (film type) at
> it usually takes 150 to 200 uSec to get average peak length. This was over
> 12-15 cycles of 3 uSec burst peaks. Lots of ion formation is required to grow
the spark.
>
> Dr. Resonance
>
That's interesting can you post the pics or do you have link to them? Do
you have any more info to go with those fugures such as streamer length, and
bang energy or is that a general observation about big and small streamers
What do you mean by "3 uSec burst peaks" ? Regards, Bob.
Date : Mon, 04 Oct 2004 07:47:21 -0600. Subject : RE: Measuring Streamer characteristics
Original poster: "Steve Conner"
<steve.conner@optosci.com>
>This is
>how we came up with the 220K +1pF per foot of capacitance model for a
>streamer load.
Terry (and others)
Could you explain this further please? It never seemed right to me. Richie
Burnett and I have seen a lot of weird things in our SSTC work, and the best
explanation we can think of is: As a streamer gets longer, its equivalent
shunt resistance to ground must go down, not up. So I don't see how a model that
says "X ohms per foot" can be correct.
I'm trying to do some experiments similar to Terry's and see what comes up
http://scopeboy.com/tesla/experiment/
Steve C.
Date : Mon, 04 Oct 2004 13:09:21 -0600. Subject : RE: Measuring Streamer characteristics
Original poster: "McCauley, Daniel H"
I believe the model assumes the streamer is already connected to ground.
Therefore, a longer streamer will have a higher impedance than a shorter one.
Dan.
Date : Mon, 04 Oct 2004 13:10:26 -0600. Subject : Re: Measuring Streamer characteristics
Original poster: "Paul Nicholson"
<paul@abelian.demon.co.uk>
Steve Connor wrote:
> So I don't see how a model that says
> "X ohms per foot" can be correct.
I agree - we'd expect the effective load resistance to
fall as the streamers grow.
Here's a different standard to use for load impedance.
This is based on the idea that a properly working
coil (whatever its dimensions) should be loaded down to
some low value of Q factor by the streamer load.
If we specify a desired loaded Q factor, we can give
the load resistance in terms of the characteristic
impedance of the resonator, ie
Rl = load resistance = Ql * Zo
where Ql is the loaded Q, and Zo = sqrt( L/C), with L and
C being the equivalent reactances for the resonator.
If Ql falls down to as low as around 6 as a result of
streamer growth, the loading is such that the remaining
stored energy of the resonator is dumped into the load
in only around one more cycle. Therefore we might conclude
that 6 is about the lowest it will ever go.
Thus it would make sense to test TC models with a load
resistance of 6 * Zo to exercise the case of maximum
streamer loading. (Whether that degree of loading is
actually achieved in the real coil depends of course on
lots of factors such as toroid size, bang energy, BPS, ...)
So much for the resistance, what about a parallel
capacitance to represent the charge stored in the streamers?
Perhaps the simplest way to estimate a load capacitance
is to assume that the streamers reduce the resonant
frequency by some percentage. A nominal value for
heavy loading might be, say, 10% reduction of frequency
which would be achieved by around 25% increase in the
total effective capacitance.
As a worked example, my CW coil has L = 84mH and C = 55pF
so Zo = sqrt(84e-3/55e-12) = 40k ohms. A heavy streamer
loading would therefore load the thing to around 6 * 40k,
or 240k ohms. The parallel load capacitance would be
around 0.25 * 55pF = 14pF.
Using this rule of thumb is sure to give you a reasonably
realistic load model appropriate to the size of the system.
If we apply this recipe to one of Terry's coils [tfltr45],
we have L=70mH, C=40pF, so Zo=42k. Then the load model is
252k || 10pF.
A figure of 1pF per foot of streamer is a reasonable order of magnitude for an
estimate of charge storage. Terry might
expect a couple of 5 foot streamers. Paul Nicholson.
Date : Mon, 04 Oct 2004 13:11:05 -0600. Subject : Re: Measuring Streamer characteristics
Original poster: Mddeming@aol.com
In a message dated 10/4/04 9:52:59 AM Eastern Daylight Time,
tesla@pupman.com writes:
Original poster: "Steve Conner"
<steve.conner@optosci.com>
Snipp>>>
As a streamer gets longer, its equivalent
shunt resistance to ground must go down, not up. So I don't see how a model
that says "X ohms per foot" can be correct.
Hi Steve,
I don't follow the mechanism for "resistance must go down". If one
visualizes a streamer as a thin wire parallel to the ground, then
resistance and capacitance to ground are increasing functions of length.
The capacitive reactance will go down as equivalent C gets larger, but I
fail to see a way for the resistance, which is a function of channel
length, to do anything but increase. Actually, a streamer looks more to me
like a transmission line that changes length and diameter cyclically over
time, not an easy item to model. Matt D.
Date : Mon, 04 Oct 2004 13:11:51 -0600. Subject : RE: Measuring Streamer characteristics
Original poster: "Jim Lux"
<jimlux@earthlink.net>
At 07:47 AM 10/4/2004 -0600, you wrote:
>Original poster: "Steve Conner"
<steve.conner@optosci.com>
> >This is
> >how we came up with the 220K +1pF per foot of capacitance model for a
> >streamer load.
>
>Terry (and others)
>
>Could you explain this further please? It never seemed right to me. Richie
>Burnett and I have seen a lot of weird things in our SSTC work, and the best
>explanation we can think of is: As a streamer gets longer, its equivalent
>shunt resistance to ground must go down, not up. So I don't see how a model
>that says "X ohms per foot" can be correct.
>
>I'm trying to do some experiments similar to Terry's and see what comes up
>
>http://scopeboy.com/tesla/experiment/
>
>Steve C. I think that Terry just adjusted the values in the model until
the waveform in Spice looked like the waveform on the real device.
In this model, the resistance is the resistance of the plasma tube along the
leader. In "real life", the resistance would be inversely proportional to the
current, along with a time varying term as the air heats up and cools.
The capacitance (which is the dominant load) is proportional to streamer length,
and, of course, would vary as the streamer grows. 1pF/ft is a sort of empirical
number that seems to work. Bazelyan and Raizer also use
numbers of this general magnitude, with several pages of justification, based on
calculations of the diameter of the leader, etc.
One can literally spend a PhD dissertation worth of work trying to come up with
a model of a growing spark, but given all the other variables, it might not
change the overall results much.
Date : Mon, 04 Oct 2004 19:07:08 -0600
Subject : Frau formular and a streamer model
with L as an output
Original poster: "Bob (R.A.) Jones"
Hi all, Streamer growth is believed to occur over several cycles of voltage.
Now a starting approximation for the energy dissipated in a streamer would be
that it's proportional to length and time. If we assume the streamer grows
linearly:
Then the energy E dissipated for a given length l would be given by:
E = P* l *(0.5*l*t) where P is the power dissipated per unit of length and t is
the time to reach length l
If we rearrange for L and assume a constant for t this would produce an equation
of the same form as Frau.
The next thing to do is generate a circuit element whose output is l. It may be
a none linear and obviously time dependent. I starting point would be a
fixed resistor and the streamer length would be equal to the energy dissipated
times a constant.
Perhaps something like neon light with a striking voltage equal to the break out
voltage and a lower glow voltage. Presumably with a series R and perhaps C.
The total energy dissipation times a constant would give the streamer length.
Bob.
Date : Mon, 04 Oct 2004 20:08:30 -0600. Subject : Re: Measuring Streamer characteristics
Original poster: Terry Fritz
<teslalist@twfpowerelectronics.com>
I think what we really need is "new independent" data... The 220k + 1
pF/foot thing is "six years" old now!!! It is just a first stab at a very
averaged out streamer impedance that was nice to help computer models
work... In those days, computer modeling of Tesla coil operation was
considered as a joke :o)))) Now, we can't live without
it!! Hehehe!! That one we did win ;-))
But there are others now that have some very nice equipment to work with
that can make "brand new" data!!!
Cheers, Terry.
At 10:12 AM 10/4/2004, you wrote:
>Steve Connor wrote:
> > So I don't see how a model that says
> > "X ohms per foot" can be correct.
>
>I agree - we'd expect the effective load resistance to
>fall as the streamers grow.
>
>Here's a different standard to use for load impedance.
>
>This is based on the idea that a properly working
>coil (whatever its dimensions) should be loaded down to
>some low value of Q factor by the streamer load.
>
>If we specify a desired loaded Q factor, we can give
>the load resistance in terms of the characteristic
>impedance of the resonator, ie
>
> Rl = load resistance = Ql * Zo
>
>where Ql is the loaded Q, and Zo = sqrt( L/C), with L and
>C being the equivalent reactances for the resonator.
>
>If Ql falls down to as low as around 6 as a result of
>streamer growth, the loading is such that the remaining
>stored energy of the resonator is dumped into the load
>in only around one more cycle. Therefore we might conclude
>that 6 is about the lowest it will ever go.
>
>Thus it would make sense to test TC models with a load
>resistance of 6 * Zo to exercise the case of maximum
>streamer loading. (Whether that degree of loading is
>actually achieved in the real coil depends of course on
>lots of factors such as toroid size, bang energy, BPS, ...)
>
>So much for the resistance, what about a parallel
>capacitance to represent the charge stored in the streamers?
>
>Perhaps the simplest way to estimate a load capacitance
>is to assume that the streamers reduce the resonant
>frequency by some percentage. A nominal value for
>heavy loading might be, say, 10% reduction of frequency
>which would be achieved by around 25% increase in the
>total effective capacitance.
>
>As a worked example, my CW coil has L = 84mH and C = 55pF
>so Zo = sqrt(84e-3/55e-12) = 40k ohms. A heavy streamer
>loading would therefore load the thing to around 6 * 40k,
>or 240k ohms. The parallel load capacitance would be
>around 0.25 * 55pF = 14pF.
>
>Using this rule of thumb is sure to give you a reasonably
>realistic load model appropriate to the size of the system.
>
>If we apply this recipe to one of Terry's coils [tfltr45],
>we have L=70mH, C=40pF, so Zo=42k. Then the load model is
>252k || 10pF.
>
>A figure of 1pF per foot of streamer is a reasonable order
>of magnitude for an estimate of charge storage. Terry might
>expect a couple of 5 foot streamers.
The 220K seems pretty stable, but the 1 or 2 or 3... pF per foot has always
been pretty unstable...
Cheers, Terry.
>--
>Paul Nicholson
