开云体育

Hi Gene,
I'm always open to suggestions for improvements, including code - either by
email (mihtjel@gmailcom) or as "pull requests" to the GitHub repository. :-)

Thanks!
--
Rune / 5Q5R

Yes,

Look in /g/nanovna-users/files/NanoVNA%20PC%20Software/TAPR%20VNA
Don't complain about the manual, it is being adapted but not done yet

The screen protector in mine looked real ugly, so I got rid of it quick. It also gave me a chance to check out the shields and general workmanship. I didn't see any issues with scratching the screen.

Hi Rune,

If it's ok with you, I can push the changes to the repo and you can decide if you want to merge. I also have another machine running High Sierra that I can test to see if it has the same issue.

73
Gene KJ4M

Hi Dave,

Yes I agree that the 20:1 VSWR impedance ratio for my 1K hi Z test is higher than for the 5 Ohm low R which is half that at 10:1 VSWR. I selected the values of my test loads based on the range of impedance I would like to have the ability to measure with a reasonable degree of accuracy. I never made up any loads for less than 5 Ohms as with typical antenna matching tasks one would not often run across the need to measure a load less than 5 Ohms. I guess there would be exceptions for example a short mobile whip where the radiation resistance is less than 5 Ohms. Of course the input Z of solid state RF power amplifiers might be another example where there would be need to measure very low Zs. I agree that if one were to try and measure a very low resistance load it would, as you point out, be also difficult. Series inductance and I suppose skin effect R could become significant factors.

I do think the need to measure high impedance loads is relatively more common for antenna building which I think is the most common application for the nanoVNA. For example a coaxial stub might be needed to match an end fed antenna. The use of 450 Ohm balanced line for feeding non resonant multiband wire antennas with VSWR exceeding 3:1 is not that uncommon. Most antenna feedpoints present a series impedance bucket curve that approaches 50 Ohms at resonance and off resonance the Z typically increases dramatically due to reactance. Thus 50 Ohms is more or less the minimum encountered in that case. My impression is that there is more of a real world need to have the ability to measure very hi Z loads than for measuring very low Zs. There are probably methods to measure hi Z loads using impedance step down techniques but much more convenient to be able to make the measurement directly.

My original intent was to raise awareness of the hi Z measurement difficulty at higher frequencies. I think if in the future parallel equivalent impedance data is provided by software/firmware that will be very helpful for those making hi Z measurements. In my case it would have given me a much better perspective. I would have become immediately aware that the real R value measurement was close and the problem was just due to parasitic shunt capacitance. But by just looking at the serial Z data I gained the false impression that the real R measurement accuracy had fallen off around 100 MHz when in fact with my nanoVNA it remains quite good up to 450 MHz - what a wonderful little instrument!

Best Regards,
Tom
VA7TA

On Tue, Sep 3, 2019 at 06:33 AM, <erik@...> wrote:

On Tue, Sep 3, 2019 at 06:26 AM, <hugen@...> wrote:

Hello erik, I am constructing a USB-VNA that can use your software. It
includes a USB-HUB, USB Audio Codec and an Arduino-compatible MCU. I think
this is a good solution.

hugen

Please have a look at
It has all the functionality you need.
I will update over the coming weeks.
I'm willing to help you test the SW on your new HW

Is Tapr-VNA software for NanoVNA available now?

I'm suggesting something very basic. Accumulate enough data to make a meaningful statement of the mean and variance of S11 and S21 using the parts supplied with the unit by doing uncalibrated SOLT measurements with two open measurements. The bare SMA connector and the supplied open. Both ports for 5 cases, and if possible, a 6th case with both ports terminated with a 50 ohm load. No unknowns. Just poor quality "knowns".

Many of the people who buy these do not, and will not, have access to anything better. An Amphenol cal kit from SDR-Kits is roughly 1/2 the cost of the nanoVNA. The Rosenberger cal kit is 2x the price of a nanoVNA.

This is hobby science. The purpose is to provide a statistical estimate of measurement uncertainty. Fundamentally calibration consists of assuming that any deviation of the measured values from the theoretical values is device and fixture error. In this case we also have significant errors in the calibration standards.

Once we have sufficient, i.e. several thousand, measurements, one can make pretty good estimates of the quality of measurements a buyer can expect. Aside from the practical utility, I think it would be an interesting study in the quality of low budget Chinese electronics manufacturing.

To collect the data I'll set up an email account for people to send their results to and write a program which will collect the data. I already have software which will compute the mean, standard deviation and the probability density function at each frequency. Then on a regular basis I'll rerun the statistical codes, plot and post the results.

It will take me a few days to develop the data collection routines as I'll need programs for multiple platforms. If a python program proves portable to Windows, Linux, Solaris and MacOS I'll use that, The TDR program uses that. So users have strong incentive to install python. And I figure it's about time I learned it.

At the moment the question is: 101 samples from 70 kHz to 900 MHz? 901 samples from 1 MHz to 900 MHz? or something different?

Have Fun!
Reg

I've certainly not seen it before - I think it might well be a "Dark Mode"
thing changing default font look. I haven't done anything in particular to
force the backgrounds to be white, but it is the default when not running
any theme, I believe.

If anyone who has a Mac, access to these themes, and knows how to code
Python & PyQT5, comes across a fix - do let me know ;-) Otherwise, I'll try
to see what I can find out on my own.

I may write to you later, Gene, if I think I find a fix - I hope you will
be up for testing it.

--
Rune / 5Q5R

Thanks Ron for the testing effort! The update has been pushed to GitHub,
and will be included in the 0.0.9 executable in a few days. :-)

--
Rune / 5Q5R

I just realized something else by making these changes, that improves things. My laptop has a native 3840x2160 screen, plus I have the 1680x1050 monitor. I dragged it from my "4K" display to the lower screen and made it full-screen and it sized up PERFECTLY. Prior to making these changes, it did not fit the screen properly... I had to scroll up and down but now it fills the full screen.

On Tue, Sep 10, 2019 at 02:30 PM, tinhead wrote:

I did a little testing with help provided by Rune and essentially made those changes. He first suggested the line "QtWidgets.QApplication.setAttribute(QtCore.Qt.AA_EnableHighDpiScaling, True)" to be inserted before "app = QtWidgets.QApplication([])". This resulted in an error that was resolved by changing line 19 to read: "from PyQt5 import QtWidgets, QtCore"; importing QtCore in addition to QtWidgets. I don't think the other line for pixmaps is required.

Great job Rune and thank you for this excellent software!

I am running on Mac OSX and the QLineEdit boxes are white on white text. I can highlight the contents and see the text. I set the text color to black in NanoVNASaver.py and that seems to work but I'm not sure if it's only happening to me or why it's happening. It may have something to do with Mojave Dark Mode. Anyone else seeing this?

Thanks again,
Gene KJ4M

On Tue, 10 Sep 2019 at 21:09, <erik@...> wrote:

We should go back to the original question. Resistor linearity of the
nanovna. So everything should be kept equal and only the resistor value
should change. That includes calibration with the same setup of SMD
resistors of zero' very high and 50 ohm.
In the nanovna here are other stray capacitance but as they areconstant
all arecalibrated out

But it was originally quoted

“Load values below 50 Ohms were found to be roughly within 10% across the
entire 900 MHz sweep.“

then extremely high values of resistors were tried. To me, the implication
is that the instrument is struggling more at high impedance than low. That
maybe so due to the effect parasitic elementals would have. But using a
maximum of 10:1 VSWR for low values of resistance and a maximum of 40:1 for
high values of resistance, would be expected to give poor accuracy for high
values of resistance.

Dave

--
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

On Tue, Sep 10, 2019 at 12:04 PM, <norbert.kohns@...> wrote:

Hi George,
that's a very good Idea! If I have some time tomorrow I will redo the test by
using the same setup on both analyzers. One thing that we have to consider,
the VNWA has much better calibration features than the Nanovna.

We shall see!

73, Norbert, DG1KPN

Hi Norbert,

That is a great idea. I'm looking forward to the results!

You probably are aware of this, but I can't help mentioning that the H-probe must be in exactly the same position for both tests. Its proximity to objects may influence the measurement. Also I underestimated the nanoVNA step size. For your study the range is 220 MHz which yields about 2.18 MHz per step. So, being off by one step can be significant when comparing two units.

Good luck with the test.

We should go back to the original question. Resistor linearity of the nanovna. So everything should be kept equal and only the resistor value should change. That includes calibration with the same setup of SMD resistors of zero' very high and 50 ohm.
In the nanovna here are other stray capacitance but as they areconstant all arecalibrated out

On Tue, 10 Sep 2019 at 18:39, Reginald Beardsley via Groups.Io <pulaskite=
[email protected]> wrote:

I'm proposing collecting raw, unaltered data and using a statistical
analysis of that to determine what the calibration routines should do. I
am well aware that this has been done in various forms. But I always feel
more comfortable if I go back to first principles when solving a problem.
There are very often approximations made which are not described well and
may be significant in certain cases.

A few things you might want to be aware of Reg. Perhaps I have
misunderstood what

*1) Mating similar connectors *
There are several connector types which will mate, although they are not
identical

1) SMA, 3.5 mm and 2.92 mm (K connector)
2) 2.4 mm & 1.85 mm

When different types are mated, there’s a discontinuity which has been
analysed and can be corrected for. There’s a paper that does this for

1) 3.5 mm and 2.92 mm (K connector)
2) 2.4 mm & 1.85 mm

The author(s) looked at the correction for SMA, but concluded that the
variability on the semi-precision SMA connector was too large to make it
worthwhile.


*2) Polynomial fits*

The usual polynomial is certainly not based on any accurate physical model.
However, for almost any high-end commercial calibration kit below about 50
GHz, that is what you will be given. You *might* be given measured data
too.

Given the accuracy of the polynomial fit method *far* exceeds that
warranted for an SMA connector, I can’t help feeling that coming up with a
physical model based an an SMA connector is of no practical use whatsoever.

Perhaps I have misunderstood your intentions.

Have Fun!
Reg

Dave

--

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

Hi Erik,

I find your design for an open reference interesting and I am thinking of making a set of similar references - it is a great idea for nulling out the shunt C effect! For the use of my set of references I was thinking of building an OPEN reference with 4, 10M 0805 resistors that would closely match the shunt C of my set of load references. But that would be only good for testing the functionality of the VNA as the loads would not provide a good absolute value pure resistive reference. It would be much better if I had a way to null out the shunt C in the first place which your design does.

But I am puzzled about how to establish a starting point calibration for the nanoVNA prior to optimizing the position of the resistor. How does one know that one isn't compensating for a skew in the calibration? For example by adding a small C to skew the OPEN reference I can use the calibration to compensate for the shunt parasitic C in my load but of course that doesn't provide me with a good set of pure R references. I don't know how to establish a starting point reference without the availability of an expensive standard OPEN reference - I wonder if there is a magical method to do that?

--
Tom
VA7TA

Hi George,
that's a very good Idea! If I have some time tomorrow I will redo the test by using the same setup on both analyzers. One thing that we have to consider, the VNWA has much better calibration features than the Nanovna.

We shall see!

73, Norbert, DG1KPN

Thank you very much for the generous offer! I do enjoy a beer, but I
wouldn't want *anyone* to feel like they had to donate anything in order to
use my software! I enjoy making it, and I'm in no risk of going bankrupt
or, worse, running out of beer because of doing so ;-)

If you really do want to, I shan't stop you though - mihtjel@... is
my PayPal address :-)

Again, thank you very much for even offering it! It really means a lot to
get all these kind words from the entire community.

--
Rune / 5Q5R