From: Paul
Date: Sat, 09 Jun 2001 00:38:06 +0100
Subject: [TSSP] Progress on Cint
This afternoon I re-coded the Cint calculations to use a different
method, increasing the spatial resolution from 32x32 to 32x512.
This fixes a long standing defect, in that the interpolation required
for the 32x32 Cint had the side effect of 'flattening' the Cint curve,
(see pn2511, figure 2.3). The peaks were sagging slightly - reducing
the height of the short range peaks and spreading out the long range
tails. The result was about 10% too much Cint at long range, and about
the same amount too little at short range.
I had a suspicion this would be the case, but when I saw the excellent
set of results from Marc:
mm3: bare d=0.221m h/d=4.66 sr=0.93 b/h=0.35 turns=2989
measured modeled
f1 61.9kHz 63.3kHz +2.3%
f3 157.9kHz 152.5kHz -3.4%
f5 229.7kHz 212.9kHz -7.3%
f7 294.4kHz 264.8kHz -10.1%
f9 355.6kHz 314.2kHz -11.7%
supported by similar figures from Kurt's coils, it was confirmed. The
trend downwards in error at the higher modes is just what is expected
with this defect, which shows up most at large h/d.
With the new Cint code, the flat tops in figure 2.3 are now replaced by
sharp peaks, and the overall shape of the curve is more justifiably
correct. Consequently, the results are much better with the higher mode
frequencies than before. For example, Marc's #3 coil is now
mm3: bare d=0.221m h/d=4.66 sr=0.93 b/h=0.35 turns=2989
measured modeled
f1 61.9kHz 63.4kHz +2.5%
f3 157.9kHz 157.4kHz -0.3%
f5 229.7kHz 229.3kHz -0.2%
f7 294.4kHz 295.0kHz +0.2%
f9 355.6kHz 358.7kHz +0.9%
An astounding set of results there. Nice one Marc - it was well worth
the effort you put in counting those turns!
Some of the other test coils are
tfltr: bare d=0.261m h/d=2.92 sr=0.67 b/h=0.03 turns=1000
measured model old Cint model new Cint
f1 148.4kHz 145.7kHz -1.8% 146.5kHz -1.3%
f3 353.4kHz 341.0kHz -3.5% 353.4kHz +0.0%
f5 513.8kHz 485.4kHz -5.5% 522.7kHz +1.7%
sk20b49: bare d=0.205m h/d=3.26 sr=0.90 b/h=0.73 turns=943
f1 217.2kHz 206.3kHz -5.0% 206.4kHz -5.0%
f3 497.8kHz 475.6kHz -4.5% 488.8kHz -1.8%
f5 709.9kHz 668.5kHz -5.8% 716.8kHz +1.0%
sk16b55: bare d=0.161m h/d=8.71 sr=0.90 b/h=0.39 turns=1976
f1 161.4kHz 155.7kHz -3.6% 155.5kHz -3.7%
f3 386.4kHz 377.3kHz -2.4% 385.5kHz -0.2%
f5 562.0kHz 530.6kHz -5.6% 566.7kHz +0.8%
f7 710.3kHz 654.4kHz -7.9% 725.2kHz +2.1%
which also show excellent agreement on the higher modes.
Note the residual rising trend in error. This is the same trend, albeit
gentler, that we see on the small h/d and small radius coils. This could
be accounted for by the coil former material permittivity.
Note also the f1 errors. These are the ones most sensitive to the coil's
surroundings, and we should be able to reduce these by more careful
accounting of walls, ceiling, and ground plane.
I'll recompute the rest of the coils overnight.
Please, could all those who've sent in coil results please let me know
the material on which the secondary is wound, and its wall thickness.
I'll attempt to put in a first order correction for the dielectric.
Cheers All,
--
Paul Nicholson,
Manchester, UK.
--
Maintainer Paul Nicholson, paul@abelian.demon.co.uk.