Here is my take on this business:
Accurate measurements of L's and C's is as much "art" as science and
that's the reason there are so much discussion about results.
The "art" part comes into play when the instrument used for the
measurement at hand is not entirely appropriate for that measurement.
Since we are hams and working on a hams budget, we seldom have the
"right" instrument at hand.
Capacitors are generally easier to measure than inductors. For our use,
they are necessarily high Q devices. If the test leads are compensated
for, C's will pretty much read the same value when tested at any
frequency up to near the first series resonance. Chips caps with no
leads will measure the same from audio frequencies up to microwave
frequencies. Leaded caps like disk ceramic and silver mica show errors
as low as the upper HF region depending on their actual capacitance. The
big problem with caps in amps is the leads used to connect them to the
rest of the circuit. If you put 4 inches of wire (which is actually a
series inductance of about 80 nanohenries) on a 100 pF doorknob, it is
NOT going to look like 100 pF anymore. The difference between what it is
and what it looks like will depend on the frequency, little or no
difference at 1 kHz, a LOT at 30 MHz.
Inductors are far more difficult to measure accurately than capacitors
mostly because they have MUCH lower Q. Air core inductors will show less
variation over frequency than cored inductors. What variation does occur
is mostly due to capacitance between turns, capacitance across the whole
coil and proximity to metallic objects.
Cored inductors, no matter if it's ferrite, powered iron, soft iron,
brass or aluminum, (yes, all those are valid core types!) will vary
considerably over frequency due to the core material. It is imperative
that cored inductors be measured at the frequency of prime interest no
matter what instrument is used.
Here is an example: I designed a 1.8 MHz high pass filter for a friend.
He built it and meticulously set the C's and L's with an AADE meter. He
sent it to me to evaluate. When I looked at it on the network analyzer,
it had high insertion loss, skewed bandpass, and "lumps" in the
response. Since my friend used ceramic chips caps, they were within a
few Pf of what they were supposed to be. However the inductors were more
than 15% high in value. These were in the 2 uH range wound on 1/2 inch
#2 powered iron cores. When I "fixed" them using my MFJ-259B set at 1.8
MHz, the filter was as perfect as you could ask for, the plotted
response laid almost exactly over the computer predicted response!
A good high Q inductor does not suffer nearly as much from these
frequency variations. So, when you measure a typical PI tank coil on a
instrument that works at a low frequency (like the AADE or the B&K) the
result is close enough to actual that you can't tell the difference in
the final application. The only thing you need to worry about is how it
is installed, I.E. lead length and proximity to metallic objects. (FWIW,
I have done a lot of measurements on inductors installed in chassis. In
general, you MUST keep the sides of the coils at least one coil diameter
away from anything and the ends of the coil at least TWO diameters away
from anything. This includes the chassis as well as the tank capacitors
and the band switch!!! Certainly it can be made to work with everything
squeezed together like a cell phone mother board, but it is NOT going to
be optimum.)
Given the above criteria, various instruments will almost ALWAYS show
different results. This is where the "art" comes in...
Simple ( spelled C H E A P) instruments like the AADE and the B&K make
all the measurements at a frequency much lower than we are normally
interested in. This is not a problem as long as cored inductors are not
used and the final application does not require extreme accuracy. Most
HF amps fall into this class. I like both the AADE and the B&K
instruments, but don't own either. Wish I did!!
Simple (again spelled C H E A P) RF analyzers, like the MFJ-259B and the
Autek RF-1 for example, use a return loss bridge in the detector
circuit. One of the characteristics of return loss bridges (as well as
directional couplers of any type) is that they only work well at the
characteristic impedance they were designed for. In the case of the
MFJ-259B accuracy suffers greatly at impedance's more than 100 ohms or
less than 25 ohms. This means that inductors whose REACTANCE at the test
frequency is out of that range will not measure accurately. (another
problem with the MFJ is that ANY DC applied even once to the input will
cause damage to the detector diodes and effect the measurement accuracy
for ALL measurements. Static discharge to the center pin will do it. The
instrument will appear to work but will produce erroneous results.
Always check the calibration before doing any important measurements!)
Assuming the MFJ calibration is correct and the inductor is in the
"good" range, the resulting measurement will be accurate enough for most
any HF work.
Complex (spelled E X P E N S I V E) instruments such a vector network
analyzers suffer from some of the same problems, I.E. they mostly use a
return loss bridge or directional coupler in the detector. However, the
typical network analyzer uses a better bridge circuit and usually
employs OSL (open, short, load) calibration to extend the "good range"
well beyond what you can achieve with the "cheap" analyzers. For example
the AIM-4170 is good. I have a N2PK VNA and it is reportedly better. I
know that it does quite well from around 10 ohms to around 250 ohms
reactance with the normal return loss bridge. More about the N2PK VNA in
a bit. All of the "ham" VNA's will produce accurate measurements, both
inductance and Q, when used within their limitations.
Very complex (spelled H E A R T - A T T A C K) instruments like the
HP/Agilent network analyzers that are marketed as component test sets,
(sorry don't remember the model numbers offhand) use a system known as
RF-IV for the detectors. They actually measure the RF voltage and the RF
current at the test ports and use that to calculate the complex
impedance. Since this system is capable of accurate phase and amplitude
measurements over a VERY wide range of frequencies, these instruments
are the best there is for evaluating components, especially high Q
inductors, over a wide frequency range. The N2PK VNA has a RF-IV option.
I have the RF-IV bridge but have not yet put it together so have no
first hand knowledge, but it is reported to show accuracies similar to
the Agilent instrument (within the N2PK's normal frequency range).
For control purposes, I have a test setup that uses a HP signal
generator, a HP spectrum analyzer, a couple of shielded loops and a
polystyrene test stand. With that and a known very high Q capacitor to
resonante the inductor under test, I can very precisely determine the
inductance at the test frequency and the Q (at least up to Q=600). The
only art involved in this setup is the requirement to keep the shielded
loops at least two coil diameters away for the ends of the coil under
test. The results are accurate over the entire frequency range of the
test equipment, regard less of the Q of the coil or the core material.
It is not possible to measure toroids with his setup.
Hope this helps understand some of the mystery of measurements.
73, Larry
Larry - W7IUV
DN07dg - central WA
http://w7iuv.com
_______________________________________________
Amps mailing list
Amps@contesting.com
http://lists.contesting.com/mailman/listinfo/amps
|