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

Re: Bang the rocks together harder lumpophiles



Hi Terry and all.

At the input to the coax you have a discontinuity just before the reflection
90deg after it 0deg inpractice the  reflection must be spread out a little.

As I have said before its very hard to interpret the time domain signals
from impulses or edges. Try digitising it and FFTing it with mathcad then
you have the impulse response which is the transfer function.
For those with out such luxuries it  common practice to perform sinewave
tests on a linear system because superposition holds i.e. you can treat all
the sine waves independently.  It is not true to say that such tests are not
applicable to transients. Transients are just a collection of sine waves. An
impulse has a flat spectrum and an edge has a 1/f spectrum.

I am surprised that your coil characteristic impedance (CI) is so high have
you confirmed it by
measurement and at what frequency.  Perhaps its high by the same
factor the wave is slow compared to a standard TM. The CI will be frequency
dependent for the same reason the propagation velocity is frequency
dependent i.e. the turn to turn coupling.  You cant use the  standard
equation to calculate it just like the propagation time.

It not a matter of what is realistic its a matter of  taking  measurements
on parameters like propagation delay.  This can then be used to build a
model which you can then compare in a realistic case. As you have said the
standing wave completely dominates the generator signal so its impossible to
make measurements on it.  The trick of all experiments is setting them up so
you can accurately measure what you want.

I assume you have been using a standard TM model which is not applicable, no
turn to turn coupling. Try the one I will send you its only got ten elements
and only coupling between the stages unlike  a real coil that is coupled
along its whole length but it should show some dispersion and slow
propagation.

Regards Bob

-----Original Message-----
From: Tesla List <tesla-at-pupman-dot-com>
To: tesla-at-pupman-dot-com <tesla-at-pupman-dot-com>
Date: 19 April 2000 13:58
Subject: Re: Bang the rocks together harder lumpophiles


>Original Poster: Terry Fritz <twftesla-at-uswest-dot-net>
>
>Hi Bob,
>
>At 10:43 PM 04/17/2000 -0400, you wrote:
>>Hi Terry,
>>
>>When you say the phase is consistent along the modelled coax do you mean
>>there was no phase shift?
>
>Once the standing wave is set up,  It's magnitude overwhelms the driving
>signal and it acts like "Malcolm's ruler" in that the coax's signal phase
>is basically zero along the length of the coax.  It took a real RF guy like
>yourself to bring that situation to light for me!  However, that phase is
>90 degrees out from the generator.  I am not sure the exact point of the
>phase shift but I imagine it is close to the base termination of low
>impedance.  It is, in fact, needed for power transfer to take place.  I ran
>through that model fairly quickly so perhaps I misinterpreted something?
>
>>
>>Yes you can mount the coil vertically preferable say one diameter above
the
>>ground plain.  But your making it difficult to get a short ground
connection
>>for termination's and probes at the top end but I think it will be OK up
to
>>a perhaps a meg. Good practice is to  have the probe tip on the signal
point
>>and take the 0v connection vie a lead to the common ground point.
>
>The probe "tip" has a ground reference back to the scope that is grounded
>to the common point.  The TEK probe wire is 2 meters long so perhaps that
>was affecting something??  I want to keep the secondary configuration
>comparable to a typical Tesla coil or all this looses the "golden" meaning.
>
>>
>>The isolator is good practice.
>
>That's a darn neat thing!!
>
>>
>>You suggest you should drive the coil from a very low impedance because
the
>>Q is high why? what are you trying to measure?  NOT PROPAGATION DELAY I
HOPE
>
>As the voltage builds on the low loss coil, the currents reflected back to
>the generator can trigger it's over current function.  The isolator stops
>this while also providing a low Z termination at the base of the coil.
>
>>
>>How you drive the coil depends on what you are trying to measure.
>>
>>If your measuring  propagation time,  the impedance looking into the
>>isolator from the coil must match the coil characteristic impedance and
>>output of the coil must be terminated in the to coil characteristic
>>impedance, as I have already said.
>
>The characteristic "equivalent" impedance of my coil is 31,000 ohms.  I
>don't not think adding a 31K source and load impedance would be realistic
>for a "real" Tesla coil.
>
>>
>>If you are trying to make the coil resonate it should be terminated in an
>>impedance that is low with respect to its characteristic impedance which
may
>>hundreds of ohms so 4.7 is fine but it depends on what you are going to
>>measure.
>
>My 10 ohm resistor provides reliable signals strength above the noise and
>is 1/3100th the characteristic Z so I should be fine.  For simply finding
>the resonance, this resistance seems to be very insensitive.
>
>>
>>If you your performing a propagation experiment the probe input C (X10),
>>even at 1mHz is 15kohm so it want effect the termination significantly
>>provided your characteristic impedance is less than 1500ohm.
>>
>>If your doing resonance experiments think of it as a 10pf  top load.
>
>The probe impedance makes a "BIG" difference.  I try to use a 2.5pF 10Meg
>TEKP6009 probe I borrow from work.  I really need to get a TEK P5100 for
>myself since I can use it often.  Simply changing probes changes the
>measured signals drastically but somewhat predictably.
>
>>
>>Yes you can drive the coil with an impulse or noise why? what are you
trying
>>to measure?  but you will need a spectrum analyser or transfer function
>>analyser to work out whats going.
>
>If I can get "any" model to reproduce what I see.  I will be able to do
>wonderful things.  Neither the lumped or typical T-line model works well
>for this subtle stuff.  The equivalent simple T-line model does not exist.
>However, a multi element lumped model should be able to reproduce the
>situation simply by virtue of the computer being able to crunch through the
>math...
>
>>
>>I suggest you stick to the propagation test with continuous sinewaves (at
>>least initially) at the resonant frequency of you coil assuming  you are
>>trying to measure the delay at the working frequency of a coil.
>
>But the square waves give a great deal more information!  The added fun of
>the step response excites and amplifies the differences in the coil's
>characteristics.  This provides a much more strict standard that or
>"little" models have to meet.  The goal is to come up with a model that
>provides the right answers all the time.  Rest assured that the "right
>answer" looks very much like a transmission line, but it will take a
>complex lumped model to use.  Sort of beautiful isn't it ;-))
>
>>
>>As can be seen by reading the large number of posts there's a lot coilers
>>doing all kinds of things.  Few appear to be well set up propagation
>>measurements. A critical point to the interpretation of many of the
>>experiments is their not using continuous sine waves and with various
input
>>and output termination's this coupled with the high dispersive proprieties
>>of the coil makes a very complicated response and very difficult to
>>interpret.  I suggest they start with a standard propagation test
(correctly
>>terminated input and output) and when they can show the delay or 90deg
phase
>>shift at the resonate frequency their ready to move on.
>
>Most anyone can do these tests.  However, the low impedance probe seems to
>be a rather critical factor and they are not cheap!  It is really important
>to reduce seeing the "probe" and increase seeing the "coil" in these tests.
> With the coil having only 9.3pF to start with, the probe has to be very
>special.  Dropping a 16.6pF probe across the coil really messes things up.
>Perhaps you know if a simple very high Z probe adapter.  I think we can
>increase the coil voltage to get a meaningful signal to a much higher Z
>probe...
>
>>
>>But I am totally perplexed as to why with so many examples of standing
waves
>>with no phase shift that its difficult for so many to accept its true for
a
>>Tesla coil and presumable most on the list are engineers or at least the
>>ones with scopes and things. You must have done those experiments at
school
>>with the lycopodium powder and tubes what about the string expeiments with
>>bits of paper. Can somebody explain it.  What's special about a Tesla
coil.
>>I am starting to feel its a religious believe and I am a heretic. Just do
>>the propagation measurement  with a matched  terminated coil as I have
>>described.
>
>I would submit that Tesla coils deal in the world of fast transients.
>Standing waves are an over simplification of a system that defies a simple
>continuous time model.  A coil that is driving from a sine generator is not
>the same as a coil driven by a spark gap.  That nasty transient response is
>a very key factor!  A simple RLCG T-line does not explain anything.  Us
>lumpophiles have neat computer toys that have drilled the practical things
>to a fine science using just an inductor and two capacitors.  However, even
>"we" know that it can't be that simple and we are surprised "that" lumped
>model works so well.   Thus is the reason we are turning our vast resources
>to the "T-line side" (the dark side ;-)) to help you poor guys out in
>explaining the finest details.  You know your are right (and so do we), but
>you don't have the tools to explain it.  Rest assured, "we" have the tools
>and we will drill this thing down even to the transmission line theory in
>search of the "simple" truth.  Richard Hull thought this quest would drive
>us nuts.  But he underestimated the lengths we would go to get there ;-))
>
>BTW - The C component of a secondary's T-line model appears to be
>'negative' for the first 1/3 of the coil's length.  Ain't that sweet!! ;-))
>
>Cheers,
>
> Terry
>
>
>snip...
>
>