From: Paul
Date: Thu, 07 Jun 2001 13:47:53 +0100
Subject: Re: [TSSP] Interesting article on Medhurst, Wheeler, modeling,etc.
Hi Finn,
Thanks for the info. Up to now the smallest h/d examined in detail is
h/d=1.36, so therefore you are taking us into new territory.
Finn Hammer wrote:
> The wire diametre is 0.25mm, and with 20m insulation, it ends up
> occupying 0.29mm per turn. The coil is closewound.
Gosh, that makes a high spacing ratio, 0.29*500/146 = 0.99!!
You must have a superb coil winding machine! Have you really
got exactly 500 turns?
The measured inductance, 29.4mH, compares with 28.9mH from
Nagaoka, +1.7%, quite a bit on the high side - we need to resolve that.
Can you take a DC resistance reading? Do you know what frequency the
Iso-tech runs at when measuring the inductance? If not, perhaps you can
observe it on the scope?
> The generator is 50 ohms. impedance.
Fine.
> I had no groundplane under the coil, but have PCB sheets 1 by 1 metre
> square, which I can use. Is that big enough?.
Do you have four of them, to make a 2m x 2m square? In view of comments
below, this may not be a big issue at the moment, so don't worry too
much about this right now.
> Is it ok to just pick up the oscillations from an antenna, or
> should I rather measure the voltage drop across a resistor in series
> with the coil at the base?
Either. It's important to make sure the field pickup is not perturbing
the coil. With plenty of signal from the generator and a sensitive
scope input, you can put the antenna a long way away, with little risk
of upset. Confirm this by taking set of readings, then bring the
antenna in significantly closer and take another set. They should change
by less than say, 0.2%, if the pickup is far enough away.
If the generator provides insufficient power for a distant pickup
reading, then the base current is a good alternative. With a resistance
in series with the base feed, just tune for a dip in the voltage
downstream of the resistor.
With careful measurements, you may notice that the frequencies obtained
by tuning for maximum antenna pickup are not quite the same as the ones
measured by looking for maximum base current.
> Nordmende SRG 389 (2mHz)
> Grundig TG 11 (300kHz)
> Fluke 87
> Iso-tech 9053 handheld LCR meter
> Gould 10000 (20mHz)
> Tektronix 485 (350 mHz)
OK.
> By creating lissajous figures on the scope, it is possible to
> determine the relationship btwn 2 frequences.
That will work fine, providing the Nordmende doesn't drift too fast.
We need the frequencies to about 0.1% accuracy, which should be no
problem for this method.
> Pls. allow me into the weekend to finish the measurements, I need some
> coax and hardware, and time to take the whole setup out into the
> backyard.
Well, no rush. These things take their time, and it might take quite a
while to fully understand the behaviour of this coil. We need first to
resolve the discrepancy in the DC inductance.
Here are some preliminary comparison results for your coil,
fh1: bare d=0.160m h/d=0.91 sr=0.86 b/h=6.85 turns=500
measured modeled error = (modeled-measured)/measured * 100%
f1 340.0kHz 361.8kHz +6.4%
f3 725.0kHz 878.5kHz +21.2%
f5 1100.0kHz 1355.9kHz +23.3%
f7 1450.0kHz 1838.2kHz +26.8%
By comparison, the lowest h/d tested so far (one of my big CW coils)
gives:
pn1: bare d=0.590m h/d=1.36 sr=0.91 b/h=0.05 turns=356
f1 150.7kHz 151.5kHz +0.5%
f3 360.0kHz 357.8kHz -0.6%
f5 543.0kHz 538.6kHz -0.8%
You can see, we have some way to go. Can you let me know how thick the
coil former material is? Do you have available any polyethylene film,
suitable for wrapping several layers around the outside of the coil?
Incidentally, if I switch off Cint in the model, and re-run the pn1
coil, we get:
pn1: bare d=0.590m h/d=1.36 sr=0.91 b/h=0.05 turns=356
f1 150.7kHz 172.8kHz +14.7%
f3 360.0kHz 617.5kHz +71.5%
f5 543.0kHz 1232.8kHz +127.0%
from which we can see that Cint is contributing around 30% of the total
capacitance at f1, and considerably more at the higher modes. As h/d
falls lower still, we expect an even bigger proportion of Cint.
Can you see why I believe the long-range Cint theory? There are no
fiddle factors in the code - it is just Coulombs law and I have no
choice about it, so to get the pn1 readings to be all better than 1% as
they are, would otherwise require a miracle of coincidence!
Why are your initial readings so far out? Not your measurements - that
is almost certain. Rest assured I have a number of other readings from
similar coils which have much the same sort of error pattern. For
example, two of Malcolm Watts' h/d=1.0 coils:
mwa1-1hd0: bare d=0.168m h/d=1.00 sr=0.91 b/h=3.94 turns=272
f1 600.0kHz 639.8kHz +6.6%
mwa2-1hd0: bare d=0.168m h/d=1.00 sr=0.49 b/h=3.94 turns=272
f1 605.0kHz 641.8kHz +6.1%
also exhibit the same offset at f1 (higher modes not measured on these
coils, so diagnosis is limited).
I can account for around 5% to 8% of the f7 error because the short-
range Cint calculation is not quite right. Shows up more at lower h/d.
We see the errors appear asymptotic to around 30% as the mode number
goes up, which is a sure sign that Cint is being under-estimated by the
model, for some reason. (You can reach f9 on the Nordmende - see if
it's getting on for 1764 kHz, (1764 + 30% = the predicted 2294 kHz).
Cint almost completely dominates the coil's capacitance at these high
modes - so they are important to look at.
Why does my coil, at only a little larger h/d give a good account,
compared with yours? I'm beginning to suspect that this is due to the
coil former material - the model does not account for its permittivity.
In my coil, 1.5mm thickness of polyethylene is much less, in proportion
to coil length, than is likely for your coil. Lines of E-field
contributing to short-range Cint have to pass inside the material, and
we are not allowing for this. I think this may increase short-range
Cint by up to 50%, which would account for all our problems at small
radius, high elevation, and small h/d. All coils tested with diameter
less than around 20cm are exhibiting this Cint shortfall.
In view of the initial errors from your coil, probably no need to run
it outside - if the external C was badly wrong, we would see a different
pattern of errors altogether - so stay indoors, at least for now!
But can you let me know the distance to the nearest wall, and to the
ceiling?
Anyway, Finn, I'll hope that you write back to say that your coil former
has a really thick wall, and that you can wind another using a
paper-thin cylinder of the same size! Of course, that may be not be
possible, so the plan (b) would be to wrap the outside of the coil in
plenty of plastic, which should - if my theory is correct - make the
higher mode frequencies fall even lower than they are now. That in
itself would give a good indication that the former material cannot be
ignored when computing Cint.
Cheers,
--
Paul Nicholson,
Manchester, UK.
--
Maintainer Paul Nicholson, paul@abelian.demon.co.uk.