开云体育

Hi Roger

You are right, i tested a 4.63 K Ohm serial resitor S21 gives me 5.35 Khom +15.6% error but regular static value across the 30Mhz span , S11 shunt gives me a very irregular values across the 30Mhz span , thats mean if we can calibrate better S21 resistor function it will give better accuracy measurements for hight values > 1K Ohm , I dont Know if there is a trick for this possible S21 compensation ?
see illustrations below
73's Nizar .

Stabilize the S22 port with a 6 dB or greater attenuator for starters and
90% of the problems.

In addition any VNA has trouble measuring high impedances.

For measuring CMRR of my common mode chokes (transmission line
transformers), I use a spectrum analyzer with a tracking generator - a
transmission mode measurement.

Dave - W?LEV

On Sun, May 18, 2025 at 3:55?PM Roger Need via groups.io <sailtamarack=
[email protected]> wrote:

Nizar,

The NanoVNA H4 has difficulty making accurate S21 measurements for several
reasons that have been posted many times in this group. I suggest you try
measuring a 5K and a 10 K resistor using your test jig and S21 and see what
you get for results.

Roger

--

*Dave - W?LEV*


--
Dave - W?LEV

Nizar,

The NanoVNA H4 has difficulty making accurate S21 measurements for several reasons that have been posted many times in this group. I suggest you try measuring a 5K and a 10 K resistor using your test jig and S21 and see what you get for results.

Roger

Thanks, Nizar. There is almost no difference between the two methods for Z (|Z|), a bit more for R and I. This suggests that the residual shunt capacitance in your test fixture is negligible. The S21 method is noted as Ser, which stands for series-through (Snt for the shunt-through method comes next). I include R because some builders judge choke performance by it rather than |Z|. The I plot shows where the choke changes from being inductive to capacitive.

Brian

Hi Brian

Here S1p and S2P attached

73's Nizar

Hi thanks

Sent from Outlook for Android<>
________________________________
From: [email protected] <[email protected]> on behalf of Paul Hardcastle via groups.io <paul.a65dr@...>
Sent: Sunday, May 18, 2025 8:03:31 AM
To: [email protected] <[email protected]>
Subject: Re: [nanovna-users] mint 22.1 git install error

Hi Rod,

Confession - I'm not a software guy.

It does look like an issue with PIP/python rather than nanovna-saver.
Google tells me it is likely the libssl-dev was not installed when
python was built.

If you are happy with the version installed from the repos then stick
with that. I normally prefer to stick with a slightly older package
than disappear down a rabbit hole in an area I am lacking skill.

Cheers,
Paul



On Sat, 2025-05-17 at 23:46 -0700, Rod via groups.io wrote:

Hi Paul,

when installing nanovna-saver 0.7.3 from github I get the error shown
here:

a
Thank you again for helping me. I love the software.

Cheers, Rod

Nizar, please post the .s2p file. I'd like to see whether the Y21 method offers any improvement over the S21 method for your test.

Brian

Hi Roger

Thanks for all this papers and sharts, here my simple direct measurements comparaison of this morning for the same balun 1/1 commun mode impedances done by two differents methods :

1) S11 Shunt IZI responses ( commun mode connections).

2) S21 IZ_Seriall response ( commun mode on serial connection).

see illustrations attached below , it gives a very different impedances result , that's Why the S11 IZI is not usable here due to high values of impedance measurements where S21 Serial methode is still a very correct methode .

Conclusion : S21 Serial impedance or LogMag is the only correct methode to mesure the commun mode rejection of baluns Z> 1KOhm ( or any S21 derived nanoVNA graphes ) .

73's Nizar

Hi Rod,

Confession - I'm not a software guy.

It does look like an issue with PIP/python rather than nanovna-saver.
Google tells me it is likely the libssl-dev was not installed when
python was built.

If you are happy with the version installed from the repos then stick
with that. I normally prefer to stick with a slightly older package
than disappear?down a rabbit hole in an area I am lacking skill.

Cheers,
Paul

Hi Paul,

when installing nanovna-saver 0.7.3 from github I get the error shown here:

a
Thank you again for helping me. I love the software.

Cheers,? Rod

Hi Paul,

Thank you, installing via apt JustWorks! I am using it an it's great!

I had been attempting an install from? github. Later this weekend
I will send feedback to you with screenshots on how I could not get it
to work.

Cheers, Rod

Hi Rod,

Did you have trouble installing it, or you install it and it doesn't
run?

I have mint mate (i.e. not cinnamon). I can't recall if it is 22 or
22.1. ?All I did was 'sudo apt install nanovna-saver' and it installed
from the standard repos. For me that did 'just work'. No issues running
it after either.

I didn't have to worry about python versions - the software manager and
repos looked after that for me.

On my previous system (a 2019 vintage ubuntu mate) I did have to add in
a compatible python version, but not on my current system.

Did you try installing from standard repos? If that fails please tell
us how, and any error messages you get.

Best regards,
Paul Hardcastle
G7SLP (A65DR/KW4CM etc)

All:

In my recently emailed out method of measuring inductors and capacitors
using the Smith Chart presentation on the NANOVNAs, I seriously also
considered treating accuracy and determination of SRF. At the time of
writing, I chose not to include these subjects to keep it on a simple basis
for those visiting this for the first time or those not familiar with the
Smith Chart or the VNA in general. When all the comments are in, I may add
those items as well, as, if it were aimed at those familiar with the VNA
and Smith Chart, I certainly would have included them.

Thanks for the inputs (for the future).

Dave - W?LEV

On Sat, May 17, 2025 at 5:07?PM Roger Need via groups.io <sailtamarack=
[email protected]> wrote:

On Sat, May 17, 2025 at 09:56 AM, Roger Need wrote:

? Components (resistive and reactive) with an impedance range of
approximately 1 to3K can be measured with reasonable accuracy using the

1 port

S11 shunt method

Below are plots of a 3K SMD resistor and a 1 ohm resistor. Also a 10 pF
capacitor which shows good results over a wide reactance range. Note that
you need a good test jig and this becomes critical when measuring higher
frequencies.

Roger

--

*Dave - W?LEV*


--
Dave - W?LEV

On Sat, May 17, 2025 at 09:56 AM, Roger Need wrote:

? Components (resistive and reactive) with an impedance range of
approximately 1 to3K can be measured with reasonable accuracy using the 1 port
S11 shunt method

Below are plots of a 3K SMD resistor and a 1 ohm resistor. Also a 10 pF capacitor which shows good results over a wide reactance range. Note that you need a good test jig and this becomes critical when measuring higher frequencies.

Roger

Comments by Jean-Roger, Nizar and others concerning accuracy of S11 shunt measurements inspired me to write this post.

The accuracy of any measurement is important and an excellent paper on this subject was
written by Brian Walker of Copper Mountain Technologies (a VNA manufacturer).
"Make Accurate Impedance Measurements Using a VNA" available from


This paper has been discussed on groups.io before and has been subject to misinterpretation
by myself and others. So I contacted the author and he was kind enough to answer my
questions and send me his spreadsheet for error analysis. The paper goes into considerable depth on how to calculate the impedance measurement and an error equation is derived to calculate ΔS11(max).

The author then goes on to derive ΔS11 for a Copper Mountain Technologies S5065 VNA which can measure from 9 kHz to 6.5 GHz with a reflection accuracy (S11) specified to be ±0.4 dB for measurements from ?15 to 0 dB. The curves and charts in his article are all based on THIS specification. They are NOT applicable to the NanoVNA which has better performance when used over a much narrower frequency range.

I wrote a technical note that provides more detail on his paper and how it can be applied to the NanoVNA. I hope you find it informative. It can be downloaded from my Box account at the link below"


A summary of the report is as follows:
? The NanoVNA is capable of making accurate measurements of component parameters provided that quality calibration loads and a proper test jig are used.
? Components (resistive and reactive) with an impedance range of approximately 1 to3K can be measured with reasonable accuracy using the 1 port S11 shunt method
? The uncertainty of measuring the S11 reflection coefficient will determine the error to be expected when measuring components.

Roger

Hello,

Using Mint22.1

All attempts to install NVNAsaver have failed B/C ssl problems, including the instructions at

/g/nanovna-users/files/NanoVNA%20PC%20Software/NanoVNA-Saver/nvna-s-pve-rev-d.pdf

I have spent two days on this, and I would greatly appreciate any help (that JustWorks!)
installing NanoVnaSaver on Mint 22.1

Thanks,

Rod

"Because in case one is measuring a reactive component at a frequency
range of interest, which coincidentally is around the self resonant
frequency."

On the silver mica new unit the leads are about 1.5 inches and using both leads, ~ 50 nH. The 130 pF in Dave's pdf might
have a SRF at 62 MHz (1/root(LC)). So as a rule of thumb a measurement at 1/10th or near
the 40 meter band is appropriate.

Hi Nizar,

Be careful with this NanoVNA accuracy sheet. It remains indicative, and as you say, accuracy remains dependent from hardware bridge and calibration load. The initial purpose of this picture here is to show that depending circumstances, it is sometimes relevant to choose another measurement method, than the common s11 one.

In the context of this topic (s11 measurement), you can't totally rely on these relative error measurement values. But for example if you get an X=1000 ohms value or higher, you are sure not to be in a confident frequency range, for capacitor or inductance measurement. Here, we have to consider this graph (configuration 1), as an helper here, no more.

73 - Jean-Roger

Hi Jean

Your NanoVNA accuracy Shart is Very Nice : imposed naturaly by the S11 NanoVNA hardware 50 Ohm bridge , indeed S11 accuracy is better around 50 Ohm bridge resistor : S11 accurate prety good 3% from 10 Ohm to 200 Ohm , Calibration Load should be known as accurate as possible.

73's Nizar

Thank you Dave. It is a quite interesting overview.

I just suggest to add an advice about having also an eye of the jX (reactance) value. The idea is to check that this value combined with R (Z module) remains in a relevant accuracy field of measurement for the NanoVNA. See captures below.

1: 10 pF capacitor @ 145 MHz : R and C values
2: 10 pF capacitor @ 145 MHz : R and X values
3: Indicative accuracy for s11 measurement (configuration1), suggested Z module limits for a maximum relative error measurement of 3%, between 10 ohms and 200 ohms. The previous capacitor result of the measurement respects this criteria.

73 - Jean-Roger