On Wed, 9 Oct 2019 at 06:40, Daniel Marks <profdc9@...> wrote:
I built my own VNA similar to the NanoVNA, and I can tell you of my
experiences of measuring small and large impedances with it.
That¡¯s interesting.
You can see the design, which is completely open source hardware and
software, at
Great
With my design, I have measured impedances as low as 1 ohm and as high as
10000 ohms at HF.
Impressive
The VNA is most stable when there is a warm-up period of about 15 minutes
and it is calibrated after that period.
I am not sure about the NanoVNA, but the calibration method of the VNA I
designed works with test fixtures. For accurate measurements of extreme
impedances, a good test fixture is necessary. I made one out of a ZIF
socket that seems to work pretty well for HF. The calibration procedure
corrects for the open, load, short, and thru at the measurement plane of
the test fixture. One can calibrate in other ways, such as at the end of
coaxial cables, if one would like to remove the effects of these cables and
measure the impedance at the input of an antenna. I used the DG8SAQ guide
to come up the method I used to calibrate the VNA.
There are some differences between the NanoVNA and the VNA I built. The
first is that rather than using a resistive bridge, it uses a RF choke to
sense the load current as the EU1KY analyzer does.
That¡¯s a fundamentally different design. Keysight call that the RF-IV
meter. I assume that it is more difficult to design these well, but
Keysight sell instruments up to 3 GHz using this method. As far as I am
aware, the instruments using this technique are only used for reflection
measurements (S11) and not transmission (S21).
What¡¯s the frequency range of your instrument?
I have here an old HP 4291B impedance/material analyzer, which covers 1 MHz
to 1.8 GHz. That uses RF-IV. Calibration of that is a 3-stage process, as
calibration is performed at two planes.
1) Short, open and 50 ohm load calibration to an APC7 connector. Optionally
calibrate with a low loss capacitor too, in order to get an accurate 90
degrees phase reference.
2) Add a fixture which connects the DUT to the APC7 connector.
3) Perform a short, open and *optionally* a load at the fixture
I never understood why the NanoVNA applies two in-phase signals to both
inputs of the mixer. Mixers do a highly imperfect job at subtracting
common-mode, and so if an opposite-phase signal is not provided, the mixer
can be used as an unbalanced mixer by applying the signal to one input.
The second issue is that there is no common clock between the
processor/audio codec and the clock generator. Therefore there is
significant phase jitter/phase noise that can limit demodulation accuracy.
I think this limits the effective number of bits to far less than 16 that
can be captured from the IF signal. I actually triggered the acquisition
off of the IF signal zero-crossings which more or less eliminates that
noise, and makes processing of the signal much easier as well.
I have examined the source code to the NanoVNA and I don't quite
understand how it calibrates.
Because you know that for a short load, voltage is supposed to zero, for an
open load current is zero, and then you apply a known resistive load, you
can solve for the linear transformation between the empirical
current/voltage and true current/voltage, and based on this you can figure
out the impedance and/or reflection coefficient assuming any port impedance
you like.
The ¡°short¡± is more properly called an ¡°offset short¡±, and the open an
¡°offset open¡±. So it¡¯s not quite as simple as that.
Anyways, that's my take on it given my experience designing and building a
VNA.
Thank you for sharing. Can you please confirm if you can make transmission
measurements.
--
Dr. David Kirkby,
Kirkby Microwave Ltd,
drkirkby@...
Telephone 01621-680100./ +44 1621 680100
Registered in England & Wales, company number 08914892.
Registered office:
Stokes Hall Lodge, Burnham Rd, Althorne, Chelmsford, Essex, CM3 6DT, United
Kingdom
