Re: New formula for secondary resonant frequency

["Tesla list" <tesla-at-pupman-dot-com>]

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

Kurt,

Thanks for the updated figures. Let's see how they look now.

Kurt Schraner <k.schraner-at-datacomm.ch> wrote:

> http://home.datacomm.ch/k.schraner/bw_sec.htm

A top class piece of engineering. Clearly the indoor measurements on
a coil this size will not be of use for determining the goodness of
resonant frequency estimators. If you get the opportunity to measure
this coil in a big space then I'd like to see the results.
 
> The b-values were entered just from memory; they have not been
> recorded before. Recognizing the influence of fb on Fres not
> being too critical, I felt doing so justified.

No problem. As you say, fb is something of a small correction.
 
> The updated data look like:
> 
> Coil     Sk-12cm
> turns     921
> h        0.585
> d        0.1212
> b        0.2
> awg      22.6076
> 
> Fres,cal  398.0  Paul's formula
> Fres,exp  368
> Diff       8.1%  cal-exp
> 
> Fres,cal  370.4  Wheeler/Medhurst
> Diff      0.7%    cal-exp

OK, that winding fits now. I've modeled Sk-12cm based on these new
figures. 

Modeled open space f1: 392.1 kHz
          New formula: 398.0 kHz, 

Modeled Confined space f1: 390.0 kHz
              Measurement: 368.0 kHz

I would say that the model and the formula are giving a good
prediction for Sk-12cm in open space, but the formula is unable to
account for your confined environment. Nor for that matter is the
model, a large error of 6% in f1. I'm using your given ceiling height,
which is quite high and should not perturb the frequency by more than
a couple of percent. Did you have a probe or anything near the top of
the coil?

> I confess, a new Fres-experiment (with harmonics recording too!)
> might be opportune

Yes, some more refined measurements would help clarify things. The
overtones are useful for two reasons: First, you can hit f1 by 
accident, as we see with the Wheeler/Medhurst figures, but you
have to get the physics right to get 3 or 4 resonances to match 
simultaneously.  Secondly, if the resonances don't match, the pattern
of errors across the resonances gives useful info about what is wrong.
For eg, if they are all out by a similar amount, then the inductance
is wrong, usually the turns are given wrong. The internal capacitance
of the coil controls the position of the higher resonances to a
much greater extent, so a rising error with frequency indicates
something wrong with coil h/d or h. A falling error with frequency
generally indicates poor definition of the external capacitance return
path.

> The problem of a compltely new measurement is: I might need to
> remove a glued corona-ring from the top of this coil, in order to
> perform the new measurements, which can destroy the coil.

I wouldn't have thought that necessary. No need to damage the coil.
Can you measure the coil upside down, so that the corona ring is at
the bottom and therefore of negligible effect?

> The measurements for this coil have been teaching me, how sensitive
> to the (-capacitive-) environment such tests behave: A 0.5m free
> wiping end of the winding wire, at the top of the bare coil, was
> showing large movements of the resonance peak weaveform-amplitude
> on my oscilloscope, when just pushed to vibration by hand.

Indeed. A carefully controlled environment around the coil is
essential for meaningful results.

> I measured [for Sk-long indoors] the values as follows:
> 
> L =67.6 mH, ---> Measured at 120 Hz
> L =67.7 mH, ---> Measured at 1 kHz

> f1: 147.7 kHz
> f3: 382.45 kHz
> f5: 558.42 kHz
> f7: 709.28 kHz

> The size of my living room, where i tested the coil is:
>  4.93m x 4.48m x 2.84m(=heigth)
> The coils have been positioned in one corner, about >= 1.5m
> distance from the walls, which appears to be in a plausible
> relation to your guess of 2m. I didn't record the exact height of
> the base-winding in all my tests, but it should have been in the
> vicinity of about 0.5m for this coil.

I ran these through the model, taking the walls at 1.5m and the
ceiling at 2.84m,

    measured   modeled
f1:  147.7      150.1  +1.6%
f3:  382.45     382.1  -0.1%
f5:  558.42     543.9  -2.6%
f7:  709.28     673.5  -5.0%

I don't expect much accuracy for f7 and beyond with the current
version of the model, which, for speed, has limited spatial
resolution.

To summarise, I think I'm justified in continuing to maintain that
the Wheeler/Medhurst gets the right answers by accident, but it would
take open-space measurements over a decent ground on Sk-long and the
inverted Sk-12cm to settle the matter. With Sk-long the two
predictions differ by some 10% and your instruments are more than
adequate to distinguish the two cases.

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