开云体育

On 3/1/23 6:33 AM, Brian Beezley wrote:

On Wed, Mar 1, 2023 at 05:54 AM, <btomek@...> wrote:

I think both separators should be supported when reading the file.

The compiler may not support comma separators. For example, my compiler yields 1 when evaluating this string: 1,234. You can get around this by evaluating the number character by character, which is complicated if you allow scientific notation in all its variations, or you can replace commas with periods. The latter is easy, but it fails when the code processes a CSV file, which some VNAs generate and which most of my programs are designed to handle. It was easy to accommodate commas in the program that checks S-parameter magnitudes since it does not have to deal with CSV files. Incidentally, the updated program is here:

Getting down in the weeds here, but it's not so much the compiler as the parsing library it uses (which might be different at run time and compile time).

You don't have to do it character by character, you can look for either period or comma, split it there, then convert the two pieces separately. You also have to deal with the (optional) E.

But it's a pain, no matter how you do it. On the other hand, you only have to write the generalized routine once.
And for something like Python, I'll bet someone's already written it, but it might take longer to find it than to just write it.

On 3/1/23 5:54 AM, btomek@... wrote:

I changed the code to accommodate your files. One offending point for
OPEN_CHANGED.s1p occurs at 300 MHz. That may provide a clue::

@Brian - My s1p was generated in NanoVNA-App. Whether there is a period or a comma probably depends on operating system's regional settings.
According to wikipedia more countries use a comma instead of a period.
In NanoVNA-App, the lower left corner says what the decimal separator character is. In NanoVNA device, you can choose what the separator should be: Menu/Config/Expert settings/More/Separator
I think both separators should be supported when reading the file.

For SnP (Touchstone, S1P, S2P, etc.) files it's easy, because the values are space delimited.

What's *not* easy is that most of the libraries won't do it for you. You have to parse the data fields, split it at the delimiter, then convert the left and right side and combine them.



When writing, it should be as the standard specifies. If it specifies.

it does not. (If someone wants to get involved in standards making, this would be a nice thing to add. Just define the "decimal separator", either specify one, or say that it's localized and could be either, or whatever. The spec just says "you can use scientific notation".

is a great explanation


is the spec (linked from the Microwave101 site)

Going back to the test, I wanted to check whether the calibration data from the frequency 888750625 is responsible for the measurement result on 888750625 or maybe for something else, such as a bug on a neighboring result. However, it came out that every single calibration point affects 4 results, which was a surprise to me. The value of the result is not important, because the calibration data at the changed points is random, but such that it deviates significantly from what it was. Hence, the result will also be random, and may be > 1.

On 3/1/23 4:18 AM, Brian Beezley wrote:

On Wed, Mar 1, 2023 at 02:25 AM, <btomek@...> wrote:

Full s1p in attachment.

I'm curious why your Touchstone files use commas rather than periods in decimal numbers. The utilities and programs I write used to check for commas and replace them with periods, but to simplify the code I dropped that check after not encountering any files with commas for a long time. None of that software will work with your files. This includes a utility that checks for |Sij| > 1, which is invalid but occurs in several of your files.

Some software is localized, and when you put out a number, by default it uses comma as the "decimal point". In particular the various MS Visual whatevers can do this for dialogs. What's sort of pernicious is that by default, if you bought and installed Windows in the US, it's turned off, so even if you set a different region, it stays "." for decimal point; but if you bought and installed Windows somewhere else, it's turned on, and the decimal separator (and other stuff) is localize to the region setting.

I've also seen this for thousands separator (. in some EU countries, , in US). And, of course for dates (MM-DD-YY vs YY-MM-DD)

it's been a number of years (~5) since I did any significant coding specifically for the Windows platform, so it might be better now, but....

And BTW, packages like QT (on all platforms) have similar issues. They try to make localization easy, but you always wind up with a heterogenous combination.


(I've been caught by this before..)

On 3/1/23 2:25 AM, btomek@... wrote:

Roger, thank you for reproducing the experiment. We have clarity that it is a Bug after all.
I did an experiment in which I change the calibration data for a 4 specific frequencies by editing a copy of the .cal file in Notepad and I compared the s1p results for the correct cal file and the changed one.

Did you change the frequencies in the cal file, or the actual calibration data, keeping the frequencies the same?

There is probably some function that tries to smooth the calibration data, average it or filter it. Perhaps this function is used for interpolation if the calibration frequencies do not match the measurement frequencies.

In general, yes, there's likely some interpolation function.

I would recommend keeping the firmware version that it comes with until you
have used it for awhile and become very familiar with its operation. Newer
versions of firmware have added many useful features, but all versions have
been very solid in basic operation, including your intended use case of
analyzing antenna vswr performance. So there is no real need to upgrade
until you find a feature you want but don't already have.

And no need to worry about chipset performance if the correct firmware is
loaded.

On Thu, Mar 9, 2023, 4:04 AM John Knight via groups.io <john_knight1=
[email protected]> wrote:

Hi,
I am awaiting delivery of my first nanovna H4. It is going to be used to
examine various types of mobile VHF aerial sytems swr performance.
I have never used one and I do NOT know which version or chipset it is
using until it is delivered.

But while I am waiting for it to arrive I would like to try and understand
the following -
1. Which is the "best" stable version of code that I am likely to require.
2. I read there may be "suspect" chipsets, how do I find out if the unit I
purchased is a lemon or not?
3.Is there a professional looking case for the device available anywhere?

Thanks for looking.

Best regards - John

On 2/28/23 5:14 AM, Bob Ecclestone VK2ZRE wrote:

Hi Don,
Wow, what an insight that document is.
The comment on SMA connectors is very telling :
"These connectors do not work well in applications that require repeated connections; they wear out quickly. They work best as one-time only connectors, or in applications that require very few reconnections."

Well, they're typically rated at 500 mate/demate cycles, which is a lot, compared to some of those tiny "snap on" connector which are more like 5-10 cycles.

And that spec is more like a "what it's practical to test to in a reasonable amount of time, so it's in the MIL spec" - They will last a lot longer with moderate care. It's like the VSWR or loss spec - it's what's practical to test to.

The moral: Get yourself a set of SMA M-F Connector Savers.
Also, DO NOT turn the connector, ONLY TURN THE NUT.
The specified torque for brass SMA connectors is 5 in-lb / 56N-cm.

Very much so - Then you can replace your connector saver every once in a while when it gets dirty, or the center socket gets deformed, or the rim of the thread gets dinged.

One thing that's a bit tricky is that with the usual gold plated connector, you don't know if it's brass or steel under the plating.


from the Amphenol data sheet:
Mechanical
Items
Mating Cycles 500 Min
Coupling Mechanism Threaded
Interface Specification MIL-STD-348
Mating Torque (Stainless Steel Plug) 0.8 - 1.1 N-m (7 - 10 in-lbs)
Mating Torque (Brass Plug) 0.3 - 0.6 N-m (3 - 5 in-lbs)



MIL-STD-348 is here


But it's just the dimensions, no torque or performance specs.

Hi,
I am awaiting delivery of my first nanovna H4. It is going to be used to examine various types of mobile VHF aerial sytems swr performance.
I have never used one and I do NOT know which version or chipset it is using until it is delivered.

But while I am waiting for it to arrive I would like to try and understand the following -
1. Which is the "best" stable version of code that I am likely to require.
2. I read there may be "suspect" chipsets, how do I find out if the unit I purchased is a lemon or not?
3.Is there a professional looking case for the device available anywhere?

Thanks for looking.

Best regards - John

On Tue, Mar 7, 2023 at 04:19 PM, Ezequiel Reinaldi wrote:

Is there a way to measure the impedance of a balun (current type, connecting
mesh to mesh), but not as rejection in dB, but directly as impedance values
using the 2 ports?
As a guide, I have seen this website (),
where it exports the sp1 file and lifts it with Excel to convert to impedance,
but when exporting from the nano-vna, the curves are not displayed correctly,
and I cannot find the fault.

Try this link for discussions on this subject...

/g/nanovna-users/topic/97089104?p=Created%2C%2C%2C20%2C1%2C0%2C0

Roger

Terry, yes, this list has lots of traffic.
To Unsubscribe, use the Unsubscribe link at the very bottom of any one of
the emails you get from the group.

There is also a link to Mute a topic, i.e. the thread that email is part of.

Stan

Unsubscribe

Many Thanks!
That helps me a great deal.
--
Mike, W0IH

On Sun, Mar 5, 2023 at 03:23 PM, Roger Need wrote:

OneOfEleven developed the NanoVNA App for personal use and after moving on to
other projects kindly posted it as open source for the community to use and
modify. Owen Duffy has created a fork of the OneOfEleven App and corrected
some of the errors in the program especially on the info sidebar.
----------------
My suspicion is that something is wrong with the implementation of the cubic
spline interpolation in NanoVNA App. Perhaps DiSlord or Owen Duffy will
consider adding linear interpolation as an option and removing or fixing the
cubic spline interpolation.

Well that was quick! Owen Duffy has released a new version of the NanoVNA App using linear interpolation and it gives almost identical results to my stand-alone NanoVNA-H4. Thanks Owen for doing the update and sharing with the community. You can read about the update at the link below.


You can download the latest release OD13 from GitHub:


I took OD13 for a test drive today and compared the measured results against NanoVNA-H4 calibration and NanoVNA App calibration. Measurements for both were done with 8 time averages and trace smoothing set to 1. The same 3M coaxial cable as my other posts was used with calibration and testing done at the far end with the SMA pin receptacle test jig posted earlier.

Four screenshots are attached:
1. 50 ohm SMD cal load
2. 100 ohm SMD resistor
3. 1000 ohm SMD resistor with R-jX and R//jX plots
4. 47 pF SMD capacitor.

Roger

Dear DiSlord,

As the impedances of the NanoVNA are not precisely 50ohm (they can even be
something like 43ohm), I was wondering if this could be compensated for
within your software?
I think that doing the normal calibration is able to handle such a
compensation.

I have documented my experiences with this here:
www.archaeocosmology.org/eng/S21S11.htm
But Owen Duffy has similar ideas: .

Thanks for considering and of course, if questions, let me know.

All the best,

Victor

--
73 de PE1ATN or NL413

On Tue, Mar 7, 2023 at 06:49 PM, Steve Bowers wrote:

Roger, From where did windows get the update? I tried to pick the driver and
when it couldn’t I let windows go out and choose but it came back after what
seemed like an hour and responded no compatible drivers found.

All I did was plug the NanoVNA-H4 (in DFU mode) into the USB port of a Windows 11 computer that had not been configured for this device before. I did not do anything else. Windows brought up the new device icon on the task bar and after a minute I saw that it had installed the dfu driver all by itself. However it would not do this without an Internet connection so it was getting the info or driver from somewhere unknown to me.

My only suggestion is that you open Device Driver in Control panel and right click on the STM bootloader and uninstall it (also check the box to uninstall driver if it is there). Then unplug your NanoVNA and restart the computer. Then put the NanoVNA in DFU mode and plug it in. Then wait for a bit to see if it installs. If it doesn't leave it plugged in and do a Windows Update and see if it is in the optional update list.

Roger

Roger, From where did windows get the update? I tried to pick the driver and when it couldn’t I let windows go out and choose but it came back after what seemed like an hour and responded no compatible drivers found.

Thanks
Steve

On Sun, Mar 5, 2023 at 04:56 PM, Steve Bowers wrote:

I have Windows 11 pro so maybe that’s the problem.

Steve,

I just ran some tests on a newly installedWindows 11 and it automatically loaded the COM port and DFU drivers within about a minute. I had to have an Internet connection active in order to do this.

Here are two screenshots taken one minute apart. Plug in -H4 (after pressing down jog button and powering on) and it is in DFU mode. Windows first detects as STM32 bootloader and then downloads and automatically installs driver as "STM device in DFU mode"

Roger

On Tue, Mar 7, 2023 at 03:17 PM, Michael Foerster wrote:

I have a NanoVNA-SAA2 and I've found that there is a "Measure" parameter that
I can set to "L/C Match".

So, for my 80m dipole, it shows:
L/C match for source Z0=50.0(ohms)
Src shunt Series Load shunt
1.6uH 298pF
1.1nF 3.4uH

I'm assuming that the descriptions are for an "L" match, but I'm not sure what
the "Src shunt", "Series" and "Load shunt" circuits are.

Can anyone help me to understand this?

Hello,

Is there a way to measure the impedance of a balun (current type, connecting mesh to mesh), but not as rejection in dB, but directly as impedance values using the 2 ports?
As a guide, I have seen this website (), where it exports the sp1 file and lifts it with Excel to convert to impedance, but when exporting from the nano-vna, the curves are not displayed correctly, and I cannot find the fault.



Thanks, LU1FP

I have a NanoVNA-SAA2 and I've found that there is a "Measure" parameter that I can set to "L/C Match".

(The forum won't allow me to attach the picture.)

So, for my 80m dipole, it shows:
L/C match for source Z0=50.0(ohms)
Src shunt Series Load shunt
1.6uH 298pF
1.1nF 3.4uH

I'm assuming that the descriptions are for an "L" match, but I'm not sure what the "Src shunt", "Series" and "Load shunt" circuits are.

Can anyone help me to understand this?
--
Mike, W0IH

NanoVNA-APP (on PC) can also update firmware, without DFuse. And apparently without special driver.
(It's been a while since I used it, my memory details are fuzzy now.)
I also had plenty of troubles with DFuse Demo.

" I don't know if they still support DfuSeDemo" - -
It appeared to me a few months ago that they do not. I couldn't get their newer program to work with my (non-updated) Windows 10, either.
--
Doug, K8RFT