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

Re: More Coupling...



Original poster: "by way of Terry Fritz <twftesla-at-uswest-dot-net>" <paul-at-abelian.demon.co.uk>

Barton B. Anderson <tesla123-at-pacbell-dot-net> wrote:
 
> My reply here is to ask (plead) for more coiers to make coupling
> measurements.

Seconded. Remember that theories and software may come and go, but a
measurement lasts forever.

Thanks to measurements from Bert, Bart, and Terry, I've been able to 
make some improvements to acmi. It no longer rounds to integer turns,
it handles fractional turns to a resolution of 1/10th turn. It also
approximates better the spiral winding, by representing it as a number
of segments of concentric circles of intermediate radius. Also, I've
taken the filament radius for larger conductors as being the inner
radius of the turn. Each of these changes gave an improvement in
accuracy, and the results are now summarised in

   http://www.abelian.demon.co.uk/tmp/acmi-compare.html

This version 0.1f of acmi is up on the website,

  http://www.abelian.demon.co.uk/acmi/

We still have some trending in k, but the trends from Bart's and
Terry's readings are opposite. Awaiting further measurements to
resolve this. Primary self inductance errs consistently on the high
side. If this persists, I'll put in a calibration factor.

Terry wrote:

> You will notice that the voltage dV/dL is greatest about 25-30% up
> the coil!!  The old sine function for secondary voltage is long gone
> now...
> The current is like 40% higher in the lower section of the coil
> than at the base!!

Perhaps not quite so high up when a larger h/d ratio is used. But
there is some theoretical and good experimental evidence (from Terry)
that the highest voltage gradient occurs at a point some way above the
base of the coil. Maybe 10% to 40% up. That's for the free resonance.
Add to that the EMF induced from a driven primary and if there's
going to be trouble, it will be in this region of the coil.

Peter Lawrence  <Peter.Lawrence-at-Eng.Sun-dot-com> wrote:

> What bothers me at the moment is all I've heard is "smaller K
> required for the smaller TCs", but no explanation of why - 

It may be that for one reason or another a higher coupling coefficient
might be desirable, but to achieve higher k means raising the primary
which puts its max induced EMF closer to the part of the secondary
which already has the highest stress. Perhaps therein lies the benefit
of a cone primary, spreading the induced EMF more evenly along a
longer length of coil, thereby allowing use of a higher k or more
power before somethings breaks.

To illustrate this idea, I've plotted the profile of the induced EMF
due to a primary current, for Bart's secondary with his flat primary
at k=0.205, and for a hypothetical cone primary with the same k.

   http://www.abelian.demon.co.uk/tmp/bart-emf-profiles.gif

Remember that the profiles shown are the voltage *gradient* and is
just the component of EMF due to induction from the primary. The free
resonating secondary voltage gradient would need to be added to this
(with account of phase) to get the actual voltage gradient of the
driven secondary. It's clear that using the cone primary considerably
reduces the peak EMF gradient in the secondary for the same amount of
coupling. Compare the location of the induced EMF peaks due to primary
drive, with that of the free resonating secondary gradient as measured
by Terry in the second graph in 

 http://www.abelian.demon.co.uk/tssp/pn2510/

Clearly the potential (excuse pun) for secondary breakdown is going
to be increased if these two peaks coincide. 

Note also that the cone primary in the example profiles has a 50%
higher self inductance for the same k factor - of possible benefit to
primary tank circuit efficiency.

acmi has been modified to produce the above EMF data when given the
command line option -profile, which also suppresses the normal output.
eg

  ./acmi -profile < bart.in > bart-gradient.data

Data is output in two columns, turn and relative-voltage-gradient.

Peter Lawrence  <Peter.Lawrence-at-Eng.Sun-dot-com> wrote:

> However, I seem to have had the most racing sparks in the middle
> of my secondary, not at the bottom.

Does your secondary have a modest h/d ratio, say 3:1 or less?
Is it positioned quite high up off the ground, say a third or half
a coil length? Do you operate without a ground plane beneath your
secondary - relying on the building ground? I suggest tentatively that
each 'yes' answer will raise the point of maximum stress higher up
the secondary, although not higher than the mid-point.

Marco Denicolai <Marco.Denicolai-at-tellabs.fi> wrote:
 
> I have measured some time ago Thor's coupling coeff. and reported
> my results at:
> 
> http://www.saunalahti.fi/dncmrc/th_ccoef.htm
> 
> I can't recall if Paul has already used my data or not.

Unfortunately I cannot make use of Marco's coupling data due to the
discrepancy in Thor's secondary inductance. The measured value is
some 7% less than predicted by Nagaoka, and I suspect this occurs due
to mutual inductive coupling between the secondary and eddy current
loops in the metal floor beneath Thor.

Terry wrote:
> ... I don't thing the "err%" column is valid.  With the measurements
> loosing so much precision as they get smaller.  A tiny measurement
> error could make a big difference in the err% number.

Good point. At very low k the induced secondary voltage is getting
very small.

Bart, another quick check you can make - measure your secondary
voltage with the primary circuit connected through the dryer, etc,
but with the dryer switched off, ie no primary current. Let's see if
a background 60hz pickup exists which is enough to affect the  k
values. Depending on the phase of the stray induced voltage wrt the
desired induction from the primary, the background may add or subtract
from the reading. Alternatively, reverse the primary connections and
repeat the +2" and -2" readings.

There are a number of ways to measure k. The ratio of secondary Fres
taken with primary open and shorted can be used. However, the k thus
obtained will be a little different to the k obtained at low frequency
due to the non-uniform current distribution in the resonating
secondary. The value produced by this method may be more appropriate
for setting the actual operating point of the coil, although for my
purposes in establishing the limitations of current filament summing,
the low frequency k is the desirable one to use.  Another method
involves measuring the combined inductance of primary and secondary
connected in series: Measure the inductance, then reverse one of the
coil connections and measure again. From the two readings the mutual
inductance can be obtained. However this method suffers from the
drawback that the result is sensitive to the small difference between
two large quantities, and is thus limited in accuracy. My preference
is for the simple, direct, and accurate method recommended by Terry.

Peter Lawrence  <Peter.Lawrence-at-Eng.Sun-dot-com> wrote:

> I'ld like to some day (maybe over xmas vacation) take some K
> measurements for you guys. I've got helical, and conical-spiral
> primaries, and am about to make a flat-spiral too.

Peter, measurements on these would be most useful. 

Finally, I should admit to an ulterior motive behind acmi. Next year
the software model of the secondary as used by tssp will be extended
to take account of the primary coupling. It is hoped to adopt the 
current filament summing technique as used by acmi, therefore any
experimental validation of this method will be of direct benefit in
the calibration of this part of the model.

Many thanks for your various contributions - from which considerable
progress has been made thus far.

Regards All,
--
Paul Nicholson,
Manchester, UK.
--