From: Paul
Date: Wed, 17 Jan 2001 09:40:25 +0000
Subject: Re: [TSSP] Top voltage testing
Terrell W. Fritz wrote:
> It appears that the variation in frequency is due solely to the
> capacitance of the probe
Yep,
> while the variation in Q is due solely to the resistance of the
> probe.
Not quite. Your 'built-in' coil loss goes down with frequency,
offsetting to some extent the drop in Q due to the resistive part of
the probe load.
> Thus, it is pretty easy to figure out the effect of the probe
> loading.
>From your latest bare-coil results, I find 15.6pF of extra lumped
top cap is required to reproduce the freq shift:
| Measure Model
|Bare coil 147.745 146.500 -0.8% with no fiddles.
|With probe 105.05 105.05 0.0% with extra 15.6 pF ctop.
If the model wasn't already on the low side with the bare coil
prediction, I reckon the probe cap would come out real close to
the stated 16.6pF.
> What is odd is that the bandwidth Q and the voltage rise Q are
> always the same in the simple lumped element models.
When the current is non-uniform the voltage gain is no longer equal to
Q, as expected with an idealised lumped coil. The voltage gain is
given by
Vgain = Q * 1/(w * sqrt(Lee * Cee))
It's the factor 1/(w * sqrt(Lee * Cee)) that makes the difference.
For a TC with average or large h/d ratio, we have
Lee < Les,
and
Cee < Ces,
and for all coils, w = 1/sqrt( Les * Ces)
which means that the factor 1/(w * sqrt(Lee * Cee)) is a little larger
than unity for medium to large h/d. For small h/d the factor can be
smaller than unity, ie gain lower than Q. Adding topload pulls the
gain towards unity from either side. This stuff was in the section I
deleted from pn2511 - there was a sentence I couldn't prove so I've
deferred it to the next installment of 'theory notes'.
> Bandwidth Q Vtop = 74.055
> Bandwidth Q Iin = 78.36
> For the last Q measurement I used the current input multiplied
> by 0.7071 to find the Q but got a different answer than using the
> top voltage.
> Do you think using the input current as a measure of Q is valid
> and should it be equivalent to the top voltage Q?
They should be the same. The transimpedance ratio Vtop/Ibase varies
very slowly - I checked this with the model and the two methods
gave the same Q to within 0.5%. I think we need to take a closer
look at the base drive arrangement in order to understand this
discrepancy. We might also need to look at the harmonic content of
your signal generator output at some point too.
> here is a 0.995 surface mount resistor in series with the
> generator output, so the driving impedance is around 1 to 2 ohms.
I guess that's the resistor across which you sample the current. Is
that done with the scope probe? Which end takes the scope ground?
Something must be floating here. Is there a shunt resistor to the
groundplane upstream of the series resistor? Perhaps you could tell us
a bit more about what's connected to where.
> YT5060 through ~1 meter 30ga. wire
> Fo = 95.309 kHz
Well it seems the thin wire does nothing but add capacitance.
I thought using the wire might take a bite out of the extra C apparent
in the toroided case - presumably C between the toroid and probe body.
In the bare-coil case you don't seem to have any extra 'probe body' C
to begin with. In view of this I'll focus on the directly connected
105.05 kHz results.
If I fiddle the loss factor to obtain a match with your measured
Q=69.8, we get
Measured Modeled
Vgain 77.2 76.0 -1.6%
Zft 49.95 Kohm 44.7 Kohm -10.5%
Rin 647 ohms 588 ohms -9.1%
We should concentrate on the major Zft discrepancy to begin with, as
this is virtually independant of Q, ie it's a function of the react-
ances only. The weak area is the input current measurement. To fix
this discrepancy we need to register a higher current, ie a higher
drop across the 0.995 ohm series resistor, which leads me back to
wondering where the scope probe ground connects to - we may be loosing
some V from that reading. Getting to the bottom of this Zft discrep-
ancy may make your two methods of Q reading converge, since Zft is
heavily involved with the difference.
Cheers,
--
Paul Nicholson,
Manchester, UK.
--
Maintainer Paul Nicholson, paul@abelian.demon.co.uk.