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

Re: 8-9 RFI noise thoughts.



Hi Alwyn,

At 12:10 AM 8/23/99 +0100, you wrote:
>Hi Terry,
>
>There may well be plasma physics effects that contribute to RFI, ie the
>negative resitance, but I
>assume the major effect is simple the speed that the gap closes.  Presumable
>this produces a wave with a fast rise time and hence harmonics up to UHF. I
>belive someone suggested the SG raise time was a few ns.  The effect of the
>reflections is to reinforce particular frequencies. The analogy is similar
>to plucking a
>string.

That IS an interesting thought!  If the gap fires as a simple step impulse,
there should be all the odd order harmonics generated which may be why we
see so many frequencies.  I can easily do an FFT and the frequencies I can
see, but the 100MHz plus stuff gets lost on my system.  Unfortunately, that
is were the vast majority of the signals are.  The frequency response of
the plane wave antennas could be increased to microwave frequencies but the
scope required is out of my grasp.  There aren't too many people who would
want to get a really nice 1GHz+ digital scope near anything to do with a
Tesla coil...

>I can see how your ferrite rings could reduce the initial rise
>time and hence some  RFI. But given they saturate at a few amps (as others
>have
>noted)  how effective can they be.  Maybe the loss and propagation effects
>(waves travel very slow in ferrite) are more significant than the
>inductance.

Perhaps the high frequency signals, that the ferrites are there to stop, do
not have that much current behind them (although I bet they do!).
Apparently, the ferrites are doing something good but it is hard to say
what.  Perhaps they add just a little inductance or loss, which is all that
may be needed.  The models do suggest that even a small series inductance
can have a dramatic effect on the level of the high frequency signals.  But
that also assumes the effects are electronic in nature.

>When I was taking about a quarter in from the ends of the coil I was
>referring to the voltage wave not the current wave, just explore the coil
>with your voltage probe or an insulated scope probe.
>
>I never did respond to your last comment about the 1/4 wave theory.  I
>misunderstood the statement "the secondary can be modelled as a helical
>antenna" to mean it was finally agreed to be a 1/4 wave effect.
>
>You may be convinced by the following experiment. If you connect a sweep
>generator to the base of a Tesla coil vie a piece of terminated coax and
>monitor the amplitude you will notice a series of resonance spikes having
>the approximate relationship in frequency of 1, 3, 5. etc.. Possibly the
>relationship is not accurately 1,3,5 because the propagation velocity is
>slower near the grounded end due to the higher self-capacitance and or
>because the propagation velocity varies with frequency.

There is no doubt that standing waves can be set up in Tesla coils at a
number of harmonics.  What I would dispute is that there are significant
propagation and phase shift effects along the length of the coil.  If the
length of wire were a factor, and the current had to travel the length of
wire in the secondary, then the current would be delayed 90 degrees along
the coil's length.  However, it is easy to make Tesla coils with wire
lengths other than what the 1/4 wave propagation / wire length theory would
suggest.  In fact, a given coil can operate over a very wide frequency
range with ease given different top loads.  I suggest that the current at
the base of the coil and the current at the top of the coil are
magnetically linked.  This linkage simply overwhelms the effects of pure
wire length propagation.  If the wire were unwound and in a long straight
line, then it would act as a simple 1/4 wave antenna.  However, by winding
it up in a close wound coil, the currents in the wire are locked together
magnetically.  In a fairly similar fashion, the self-capacitance of the
coil is also locked.  Thus, the secondary system acts much more like a
simple lumped LC network rather than a 1/4 wave transmission line.

I used to have a long coil that had LEDs in series with the winding at
every inch.  It was fun to hit the various harmonics and setup up node and
anti-node patterns on it.  However, that device could not detect the phase
along the coil.

	There are computer programs now that can predict a coil's resonant
frequency with top load based on physical dimensions.  They do not depend
on wire length at all.  The programs calculate the self-capacitance of the
coil and the capacitance of the top load as a physical structure in space
(with a ground plane).  This capacitance is then combined with the measured
inductance to arrive at a resonant frequency.  The programs can get within
5%.  These programs are based on the voltage distribution along the coil's
length as an in-phase sine wave.  The current is a cosine wave that is
delayed in amplitude but not in actual phase (I need to find a better way
to explain that...).  Basically the current is maximum at the base of the
coil and is some lesser value (like 40%) at the top of the coil but still
in phase.

One fun thing I have never thought of, till you mentioned all this, is to
change the computer program to use harmonics instead of the fundamental
frequency in the calculations.  It is a simple addition of a 3x, 5x,... in
the secondary voltage distribution that would do this.  Then the program
should be able to find the harmonic frequencies too!  Even more
interesting, is that E-Tesla3 can plot voltage and field stress plots (with
Excel).  It will be very interesting to see what the voltage distribution
is with a top terminal in place!!  I will run these and post the results.
I will also check the results against my coil.  I'll do the bare coil and
then with top load.  I may not be able to do all this till the weekend.  My
little nieces have turned my Tesla coil lab into a Barbie horsy farm while
my back was turned :-O 

A bare coil's resonant frequency can be very accurately predicted by
calculating the inductance with wheeler's formula and the self-capacitance
with Medhurst's formula and calculating the frequency from
Fo=1/(2piSQRT(LC)).  I have never known anyone to be able to repeatedly
calculate and predict a range of coil's resonance frequency based on wire
length with any accuracy.  Propagation delay and other "things" are always
blamed for the errors that occur.  However, the Wheeler and Medhurst method
has no exceptions!


>A particularly surreal effect that I discovered was this. I wound a 10in
>long coil  on an 18in piece of 1.5in diameter plastic water pipe using 0.5mm
>wire. I mounted the coil above a sheet of aluminium to act as a ground
>plain. I connected the base of the coil and ground plain to a signal
>generator that had a maximum output of 20v pk to pk. When the frequency of
>the generator was at one of the resonant frequencies of the coil there was
>sufficient voltage strength to light a small florescent tube if it was put
>close to the coil. At the higher resonate frequencies it was possible to
>observe the standing waves along the coil as a series of nodes at which the
>tube would light brightly. The Q was about 150 so tuning is tricky.
>Presumable the lumped view can not explain this effect.

I am not surprised by your fluorescent lamp experiment.  If you put 20 Vp-p
into the base of a coil with a Q of 150, then the top of the coil should
have 3000 Vp-p present (20*150).


>The same coil could have been used to demonstrate the relatively slow
>propagation down the coil directly by applying a square wave to one end and
>observing
>the delay to the signal at the other. Despite the fact that one end of the
>coil is only 4ns away from the other the propagation of the wave along the
>coil axis is never the less hundreds of times slow than its free space
>velocity. The now old fashioned analogue TVs contained a delay line that was
>inserted in the luminance signal path to compensate for the delay in the
>colour signal path. It consisted of a long helical wound coil. An other
>example is many oscilloscopes contain a delay line in the signal path to
>allow observation of the trigger event. In some cases the delay line is a
>piece of coax the centre conductor of which is helical wound.

Delay lines differ from Tesla coils in two important respects.  Delay lines
are very long thin coils.  This allows the bottom end to easily become
de-coupled magnetically from the other end.  The coils in delay lines are
also capacitively coupled along their length by the outer shield or by many
small capacitors where Tesla coils are coupled to ground through a large
space charge region roughly in the shape of a sphere.  Thus, the delay line
can truly delay the signal since its construction is very close to a
classic transmission line with its parameters adjusted to emphasize delay.

>It has been suggested that because the phase of the current flowing in the
>base of the coil is in phase with the current in to the top C this
>invalidates the above argument. At resonance the voltage at the top of the
>coil will lag the bottom by about 90deg, the current in to the top C will
>lead the voltage at the top by 90deg, hence no phase shift of current from
>bottom to top. If the coil is resonating in its times three mode the voltage
>at the top will lag the bottom by 270deg, hence the current into the top C
>will lag by 180deg. This can be easily verified.

I have probed along coils at resonance driven by a signal generator.  One
has to be sure to use a properly terminated antenna or the capacitance and
coax loading will mess up the phase of the measurement.  I use a short 50
ohm antenna (cell phone or scanner type), a length of coax, and a 50ohm
terminating resistor at the scope end.  Although the amplitude of the
signal along the secondary definitely rises and falls along the length, the
phase of that signal stays in phase.

>Perhaps somebody with a big coil would like to modify it so that the primary
>operates at approximately three or five time the resonate frequency of the
>secondary. It should be possible in a darkened room to observe the rings of
>corona round the secondary coil. Or alternatively the hot end can be
>connected to the ground plain in which case the frequency need only be
>approximately doubled. And it would avoid the effect being swamped by corona
>from the top. You can than take a pic and post it.

Such and experiment should be possible.  But I don't think it will prove or
disprove anything.  There are voltage nodes and anti-nodes created but both
theories support that...

Many thanks for your thoughtful and interesting comments on all this.
Personally, I think the lumped parameter theory is rock solid.  If there
are holes or disagreements, I want to be sure to get them solved.  Nothing
is worse than to have a great theory with only a few exceptions... :-))

Cheers,

	Terry


>Regards  Alwyn
>
>
>snip..............