Re: More Coupling...

["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>

Bart,

Your new measurements have certainly altered the picture. I've plotted
these along with acmi predictions on a graph,

  http://www.abelian.demon.co.uk/tmp/bart-k1712.gif

A number of things are apparent. Overall the measurements are now less
than predicted, eg 

 s-p   measured      predicted    error

 +2"     0.125        0.141       +12.8%
  0"     0.157        0.178       +13.4%
 -2"     0.183        0.215       +17.5%

with the worst error occuring when the secondary is at its lowest 
position. This discrepancy is still quite a bit higher than I would
have expected from acmi's predictions. We need to try to account for
both the reduced coupling and also the downward trend to the readings
as coupling is increased.

I have some more questions for you!

You mentioned 1uF + 18k ohms in series with the secondary - can I ask
what these are for? What is the resistance of your DMM input during
the secondary voltage measurement? If it's of order 200K ohms or less
on the range in use, then an allowance for the voltage drop across the
18K would bring the secondary voltage up by about the required amount.

The DMM may suffer a 0.2V or 0.4V shortfall in its AC readings due to 
forward drop across internal diodes - are you able to compare the DMM
with say a 'scope reading to establish whether this is the case? If we
were entitled to correct for this, the additional secondary voltage
would bring the overall level of the predicted and measured readings
into good agreement. (I expect that with an expensive professional
instrument like the Fluke 77 this will not be the case, but we have
to ask!).

Were you able to continuously monitor the primary current during the
measurement run? I'm wondering whether the current changed, perhaps
due to the resistance warming up. If you began with the low-K
readings, and if the primary current crept down due to resistance
increasing with temperature, then the later high-K readings would show
a shortfall in secondary voltage - perhaps enough to cause the 5%
downward trend in K.

Other than the above suggestions, the possibility remains that some
inductively coupled conductor loop is present, which would disturb the
induced voltage on the secondary. This would show up as a reduction of
the measured secondary self inductance by a few percent. Can you
measure the secondary inductance in-situ, with the primary open
circuit, to see if it remains at the nominal 87.6 mH in the current
test setup?

I've been looking at the acmi code to see if I can account for the
discrepancy. The trouble is that acmi calculates the self inductances
by summing all the mutual inductances between the turns in a winding,
and the result is good to a percent or so. The same summing routine is
used for the mutual between two windings, but the code doesn't 'know'
the turns in question are in different windings, so its hard to see
how it can be as much as 17% out. It might be argued that the mutual
involves on average a longer range of coupling than the self
inductances, but then if the coupling was over-estimated at longer
ranges, the trend would be for reduced discrepancy when the coils
overlap more - the opposite to what we see.
 
Bart, this is the first time a careful check of acmi has been made for
the case of direct primary to tesla secondary coupling. I've only ever
used it for transformers with overlapping symmetric windings (which is
why the bug found by Antonio went unnoticed for so long). Thanks to
your help I feel we have a chance of validating a potentialy useful
tool for coil designers.

Cheers,
--
Paul Nicholson,
Manchester, UK.
--