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

Re: Variation of secondary Q



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

Terry wrote:
> I tried my Q = Vtop/Vin thing and I don't like it.  It does give
> a reading and all but there is a lot to go wrong.

Yes, it gives users of the data umpteen excuses to query the
accuracy by interpreting variations of Q as artifacts of the 
experiment.  The trace ringdown method has that nice unequivocal
feel to it. The results from it should be pretty well unarguable.

> Instead of using the signal generator that only does a tiny 20
> volt P-P, why not just charge the sucker up to 1000 volts and
> let an IGBT short it to ground.

Yes, if you want to, it will mean you can move the field pickup
further away.  As you'll see below, the data you supplied in
TEK00001.CSV for the single shot is more than adequate.

See
 http://www.abelian.demon.co.uk/tmp/tek00001.raw.gif

for the single shot waveform as captured.

The FT (fourier transform) is in 
 http://www.abelian.demon.co.uk/tmp/tek00001.ft.gif

The Fres is at 82544 Hz +/- 10 Hz.  You can also see a little
3/4 wave ringing at around 311kHz.

The carrier at 227.4 kHz +/- 75 Hz is nothing to do with the coil,
have you a strong broadcast station at that frequency nearby?

We then crop all FT components outside the range Fres +/- 20% and
do an inverse FT to get back a cleaned up TD (time domain) waveform

 http://www.abelian.demon.co.uk/tmp/tek00001.postft.gif 

As you can see, the DC offset and low frequency RC pedestal have
been stripped away, and most of the noise has gone too.  The noise
reduction is seen more clearly when we look closely at the tail end
of the ringdown,

 http://www.abelian.demon.co.uk/tmp/tek00001.cmp.gif 

The green trace is the original raw data (along with its original DC
offset) and the red trace is the post-FT cleaned up version. Note
how there is no phase error introduced by this 'digital' filtering.

At this step in the processing, we crop the TD to miss out the
preamble, the initial impulse, and anything beyond say the 5%
amplitude of the ringdown.  We end up using the portion of the 
ringdown highlighted in green in 

 http://www.abelian.demon.co.uk/tmp/tek00001.croptd.gif 

The next step is to match a synthesised waveform of the form

  A * cos( 2*pi*Fres*t) * exp( -r*t)

to the cleaned up signal.  We already know the complex number A
from the FT, so we only have a to do a least squares match over the
the exponent r, and from this, Q = pi * Fres/r.  For this sample
I get roughly 219.6 +/- 2.  In order to get the full accuracy on Q
we'll need to subject the above synthesised waveform to the same FT
band limiting as the trace signal is processed through, so that
they have compatible envelopes (the band limiting means the decay
is no longer quite exponential).

Was the single shot TEK00001.CSV a top-volts (direct probe) or an
E-field pickup, a base current pickup, or a B field pickup?  Try to
avoid any probes near or on the coil. 

> I have been looking at the "NEW!!" Tektronix TDS3GV plug in to
> enable RS-232.

Might be worth trying the soundcard capture first, as a cheap and
cheerful option.  If that could be made to work (quite likely I 
think) then this trace analysis method could be used by many more
people.  With your sig gen set to 75kHz, a diode and an RC or two
should give you a noisy 10kHz ringdown waveform at the soundcard
input.

I'll tidy up the code and send it to you.  You'll have to compile
and run it on the Sun - I haven't a Green-haired Turtle in Taiwan's
chance of getting this to run on windoze (yuck, spit).
--
Paul Nicholson
--