search

NanoVNA-Saver / nanovna-saver

A tool for reading, displaying and saving data from the NanoVNA
GNU General Public License v3.0
884 stars 272 forks source link

Please add R+jX and |Z| graphs calculated from S21 with G3TXQ method #165

Closed yo2naa closed 3 years ago

yo2naa commented 4 years ago

Is your feature request related to a problem? Please describe. Measuring high impedance with NanoVNA is not reliable.

Describe the solution you'd like A better method of measuring high impedance to insert the device to be measured between CH0 and CH1, and calculate the impedance using S21 magnitude and phase. This method is described by G3TXQ here http://www.karinya.net/g3txq/chokes/#measurement The explanation and formulas are here http://www.karinya.net/g3txq/chokes/s21.pdf image

I've measured a 9.8 k small THD resistor using S11 and also the G3TXQ method, the results are much better using G3TXQ method: image

Describe alternatives you've considered Manually copy S21 data into excel for various frequencies and do the graph in excel

Thank you.

satfan52 commented 4 years ago

Good idea! What about measuring very low impedances with accurracy? Can the same technique be used with as much accurracy ?

cozycactus commented 4 years ago

Good suggestion!

yo2naa commented 4 years ago

Kind reminder, thanks!

akinad3 commented 3 years ago

this would be great to measure common-mode chokes!

no3m commented 3 years ago

+1

rjordans commented 3 years ago

To measure very low impedances you'd probably want to have a shunt-through measurement. You want to have the DUT cause a significant variation in the measured S21 with variation of the component values. W0QE has a nice video about the different approaches and how they impact the accuracy: https://www.youtube.com/watch?v=1UbEz73FGCU

rjordans commented 3 years ago

I had some time to play with this and implemented both the R+jX and |Z| graphs. Some quick tests with a few SMD resistors, capacitors, and a ferrite bead I had on my desk seem to confirm the correct operation.

no3m commented 3 years ago

I had some time to play with this and implemented both the R+jX and |Z| graphs. Some quick tests with a few SMD resistors, capacitors, and a ferrite bead I had on my desk seem to confirm the correct operation.

Getting the following errors when sweep segments are > 1:

S21 R+jX series plot enabled:

Traceback (most recent call last): File "/home/eric/Public/Downloads/amateur/software/nanovna-saver-rjordans/nanovna-saver/NanoVNASaver/Charts/Frequency.py", line 416, in paintEvent self.drawValues(qp) File "/home/eric/Public/Downloads/amateur/software/nanovna-saver-rjordans/nanovna-saver/NanoVNASaver/Charts/RI.py", line 206, in drawValues max_real = max(8, math.ceil(max_real)) OverflowError: cannot convert float infinity to integer Aborted (core dumped)

S21 |Z| series plot enabled:

Traceback (most recent call last): File "/home/eric/Public/Downloads/amateur/software/nanovna-saver-rjordans/nanovna-saver/NanoVNASaver/Charts/Frequency.py", line 416, in paintEvent self.drawValues(qp) File "/home/eric/Public/Downloads/amateur/software/nanovna-saver-rjordans/nanovna-saver/NanoVNASaver/Charts/MagnitudeZ.py", line 107, in drawValues maxValue = 10*math.ceil(maxValue/10) OverflowError: cannot convert float infinity to integer Aborted (core dumped)

rjordans commented 3 years ago

Hmm, I did test with multiple segments, that does work for me. But looking at the error I think this may be something different.

My guess is that your calibration is better than mine and you have nothing connected between the two ports. That would result in S21 going to 0 and the series calculation divides by S21. This gives an infinite impedance which apparently is causing trouble in the plotting code when it tries to get the automatic vertical scale correctly.

I'll have a look and see how to correct for this, maybe some large fixed upper bound no reasonable impedance values will work though that feels a bit hacky.

no3m commented 3 years ago

The DUT during these failures was a Murata 5000 series (4.7mH) surface mount CM choke. Respective ends of the CM choke tied together and in series with Ch0 and Ch1, ground pass-thru.

rjordans commented 3 years ago

Ok, I've added some code to catch the infinite values and not use them for the plotting. Hopefully this fixes your issue? It looks like my initial guess was indeed correct and that the plotting code didn't handle this case properly yet.

no3m commented 3 years ago

latest changes seem to work fine.

yo2naa commented 3 years ago

Many thanks for adding this nice feature, really appreciate the effort. I'm trying to run the v0.3.9 pre NanoVNASaver.x64 (nanovna-saver.exe) on my Windows 10 Pro 64 but I'm getting some errors:

File "c:\hostedtoolcache\windows\python\3.8.7\x64\lib\site-packages\PyInstaller\loader\pyimod03_importers.py", line 493, in exec_module File "NanoVNASaver\NanoVNASaver.py", line 28, in File "c:\hostedtoolcache\windows\python\3.8.7\x64\lib\site-packages\PyInstaller\loader\pyimod03_importers.py", line 493, in exec_module File "NanoVNASaver\Windows__init.py", line 2, in File "c:\hostedtoolcache\windows\python\3.8.7\x64\lib\site-packages\PyInstaller\loader\pyimod03_importers.py", line 493, in exec_module File "NanoVNASaver\Windows\AnalysisWindow.py", line 23, in File "c:\hostedtoolcache\windows\python\3.8.7\x64\lib\site-packages\PyInstaller\loader\pyimod03_importers.py", line 493, in exec_module File "NanoVNASaver\Analysis__init.py", line 1, in File "c:\hostedtoolcache\windows\python\3.8.7\x64\lib\site-packages\PyInstaller\loader\pyimod03_importers.py", line 493, in exec_module File "NanoVNASaver\Analysis\Analysis.py", line 22, in File "c:\hostedtoolcache\windows\python\3.8.7\x64\lib\site-packages\PyInstaller\loader\pyimod03_importers.py", line 493, in exec_module File "scipy\signal__init__.py", line 302, in File "c:\hostedtoolcache\windows\python\3.8.7\x64\lib\site-packages\PyInstaller\loader\pyimod03_importers.py", line 493, in exec_module File "scipy\signal\filter_design.py", line 16, in File "c:\hostedtoolcache\windows\python\3.8.7\x64\lib\site-packages\PyInstaller\loader\pyimod03_importers.py", line 493, in exec_module File "scipy\optimize\init__.py", line 421, in File "c:\hostedtoolcache\windows\python\3.8.7\x64\lib\site-packages\PyInstaller\loader\pyimod03_importers.py", line 493, in exec_module File "scipy\optimize_shgo.py", line 9, in File "c:\hostedtoolcache\windows\python\3.8.7\x64\lib\site-packages\PyInstaller\loader\pyimod03_importers.py", line 493, in exec_module File "scipy\spatial\init.py", line 107, in File "c:\hostedtoolcache\windows\python\3.8.7\x64\lib\site-packages\PyInstaller\loader\pyimod03_importers.py", line 493, in exec_module File "scipy\spatial\transform\init__.py", line 19, in File "rotation.pyx", line 5, in init scipy.spatial.transform.rotation ModuleNotFoundError: No module named 'scipy.spatial.transform._rotation_groups'

rjordans commented 3 years ago

@yo2naa That looks like an independent problem from this feature. Looking at the error messages I believe that there is some component of the scipy library of Python missing in the install.

r2axz commented 3 years ago

I wrote this Python script to calculate complex impedances in s11 shunt, s21 series, and s21 shunt through modes. Works OK with NanoVNA as well as with any other VNA that can export Touchstone files. You can probably get the formulas there or use it for testing purposes. I use the script mainly for HF baluns / chokes measurements, but the math to calculate the impedance is universal.

yo2naa commented 3 years ago

@yo2naa That looks like an independent problem from this feature. Looking at the error messages I believe that there is some component of the scipy library of Python missing in the install.

Many thanks Roel for all the good work. A friend of mine managed to build the exe on my PC, and now I can use the new features. Would you be so kind to add them to the markers too? image Thank you so much!

akinad3 commented 3 years ago

@rjordans Opened a PR for adding the markers. Initial tests with @yo2naa shows some weirdness on the graphs as even if R is close to 0, the y axis show -2000 or so value. https://github.com/NanoVNA-Saver/nanovna-saver/pull/384

rjordans commented 3 years ago

@akinad3 thanks for adding the markers!

Regarding the weirdness you see, is this also in the graph view or just in the marker values? And what would be an easy way for me to reproduce that so that I can see if it's an easy fix somewhere?

yo2naa commented 3 years ago

There is only a minor issue with R scale low limit for the S21 R+jX series graph: image I think is is caused by negative R from the calculations, due to S21 measurement errors. For example, S21 series R for 69.4 kHz is -0.272 Ohm. Of course this is wrong, but not due to software, but because of wrong S21 measurements. However, the lower limit on the graph is still odd, -3623 Ohm. image

Here is the s2p file:

7 turn RG142 toroid choke S21.zip

The markers are OK, they indicate correct values. When the resistance is negative, only a minus sign is displayed :-) image

rjordans commented 3 years ago

Ah, yes, this indeed looks like it comes from inaccuracies in the initial measurement. I'm unsure though how to handle this nicely but at least the lower bound of the axis could be closer to the measured value. Looking at your graphs, the plotted label looks a lot like it's related to the data on the other axis, this may simply be a case of the wrong value being printed at that location.

rjordans commented 3 years ago

Ok, I finally had a bit of time to look into the axis labeling. It seems that this happens for any real/imaginary plot and was a simple typo to fix!

rjordans commented 3 years ago

Getting the negative real part printed correctly in the markers also looks like an easy fix. I've added a comment to the pull request on how to get that fixed

yo2naa commented 3 years ago

Thank you all, working fine now!