NanoVNA Custom Software

[joeqsmith]

Okay, I am up and running and getting useful results it seems.  Thanks for your assistance.

I am a little puzzled by segmented sweeps.  With so few points in the built in frequency sweeps on the original NanoVNA, it seems that it would be useful to allow segmentation of sweeps for the "Main" tab -- SWR. Smith chart and so forth.  But when I tried that, it looks like I get a sequence of charts, updating once for each segment, rather than a single chart with all the segments on it.  I will experiment a bit more.

Thanks again.  This is a really great project and I very much appreciate your efforts on it.

Did you uninstall the wrong runtime first, or is it still install like I asked?   

There is a section on segmented sweeps which I assume you read and it is not behaving how the manual shows.  I came up with that mode for looking at narrow band parts to create models for SPICE and normally don't have a use it.     It's very possible I broke it when I combined all the software.   Then again, you may enjoy randomly selecting keys without reading the manual and then ask why things don't work.  Let me just give it a quick test....

***
To give us something to look at, here's a 12MHz crystal filter I put together using 101 data points.   Sure looks like we could use more data....     
***
So far it seems what I would expect.  With basically steps of 1Hz, it's collecting VERY slowly.  No reason to bitch as it's only $50.  Running a segmented sweep like this with my old HP8754A would require a lot of manual setup.   The Nano is far more advanced.

I must have put a limit on the number of data points.  So I dialed it back and changed the range.   Its no problem to increase the buffer size but I suspect  people wouldn't want to wait.   

It seems like it works as I would expect it to but that may not be how you expect it to work.  One thing I am seeing is how the data seems to jump.  It's possible this it a result of running newer firmware than what I had originally used.   I would need to roll back.     
     

[gardner]

Did you uninstall the wrong runtime first, or is it still install like I asked?   

I didn't mess with the LabVIEW runtime at all.  I assume I now have both 2020SP1 and 2011.  When I look in the NI package manager I see NI-VISA 5.40.491.52 which wasn't there before.  I am guessing that this is the runtime I actually needed.

One issue I probably have is that my NanoVNA firmware is very old and crappy.  I will update it, likely with ttrftech/NanoVNA/releases/tag/0.8.0, once I have the update tools set up.  Is this a reasonable firmware to choose?  Is there a recommended firmware for an old original NanoVNA?

Thanks again.

[joeqsmith]

The manual talks about my testing with one version and how other versions may cause random problems and a lot of wasted time.    That firmware seems to be anything but stable.   I'll reload the old image and see if it corrects this little snag.   I tested the segmented mode with the V2+ and it wasn't a problem.  Still, it could be my software.   

[joeqsmith]

I rolled back to the old firmware and tried running it without the PC.  The problem remains.  So it's been there at least a year.   Not a big deal for me.  Doing a quick search, I don't see anyone reporting it as a problem as well.   

It appears that the segmented sweep times could be improved by a factor of about 2X (much in how the Zoom was improved).  It would still be very slow.   There also is a compare problem which should be corrected.   I'm not seeing any other problems with it.   

[gardner]

One issue I probably have is that my NanoVNA firmware is very old and crappy.  I will update it, likely with ttrftech/NanoVNA/releases/tag/0.8.0

I now have edy555's 0.8.0 firmware on my NanoVNA.  This is the version mentioned in the r15 manual.

I've gone through the segmented sweep again and I think it still is wonky.  I am wanting to do a SWR across 420 to 480 Mhz in 5 Mhz segments and when I set that up and initiate a sweep, I can see the SWR graph display each segment in succession.  This is not what I expected, but appears to be the design.  Within each segment, the portion of the SWR curve looks reasonable.  When all the segments are complete, the SWR chart finally shows the X axis covering the whole range, but the SWR curve is completely wrong, not resembling the individual segment curves at all.

I'll put some screen shots.

[joeqsmith]

I noticed that I had forgot to enable port2 for polar transmission with the old Nano.   So there's a few things that need attention.  I've also had that firmware crash to where I had to power cycle the Nano to get it to recover.   This is happened twice since I started testing with it. 

Pictures would help. 
****
I suggest you also provide as much detail as possible about the setup and procedure  you are using.   
Did you cal it first?
If so, what standards did you use?
What range did you cal over?
...
Every detail you provide will help me reproduce it.   

[gardner]

I did calibrate it via the two port cal procedure.  The calibration standards are total rubbish -- my objective is to characterise twisted pair and bits of bent wire for use with those cheesy 434MHz OOK radios.  The setup is a female dupont connector soldered onto an SMA and the reference standards are a short and 50R metal film resistor soldered on a bit of 2.5mm header.


The procedure for segmented sweeps is not perfectly laid out in the manual.  Steps 2,3,4 below are set out, but not the rest.  Here is what I did.
(1) toggle sweeps off
(2) enter span 10Mhz, press enter
(3) enter start 420Mhz, press enter
(4) enter top, 470Mhz, press enter
(5) press Segment.  At this point the start and stop values begin to change, cycling through the required segments.  I wait for them to stop changing.
(6) I press sweep and the start and stop values once again change, cycling through the required segments.

The displayed SWR/return loss chart updates for each segment and within the segment appear plausible.  When the segment sweep is complete, the graph is drawn with the complete (or almost complete) X range specified.   Most times the final graph is a total mess.


Once I got the whole thing, sometimes I get all but the last segment like here.

[joeqsmith]

(1) toggle sweeps off
(2) enter span 10Mhz, press enter
(3) enter start 420Mhz, press enter
(4) enter top, 470Mhz, press enter
(5) press Segment.  At this point the start and stop values begin to change, cycling through the required segments.  I wait for them to stop changing.
(6) I press sweep and the start and stop values once again change, cycling through the required segments.

1) Why do you turn the sweeps off?  Shouldn't matter, just curious.

6) Why are you running a normal sweep after the segmented sweep? I doubt the software will set things up right but you should notice that.  Still that first segmented sweep should have been fine.

I did calibrate it via the two port cal procedure.  The calibration standards are total rubbish -- my objective is to characterise twisted pair and bits of bent wire for use with those cheesy 434MHz OOK radios.

Did you calibrate it for the 420-470 range, 420-480 range, some other range?  Want me to guess?
 
****
The partial is most likely the compare bug I mentioned.  The scrambled may be a problem with the communications.   

Showing my antenna simulator. 

[joeqsmith]

What I find interesting is the feel for needing higher resolution when looking at SWR.  As I mentioned, I came up with the idea of sweeping in segments after trying to collect S-parameters for a crystal to make a model for SPICE.   There you are working with a very narrow band.  An antenna, not so much.  Seems like it's just going to take longer to collect the data with little or any gains.    That said, the V2Plus4 I bought can scan over a 1000 points without segmentation and the performance in this frequency range is much better than the old hardware.   You may want to consider getting one.

Here's my antenna simulator at the standard 101 data points,  using a 2MHz Span for 3001 data points and a Span of 500KHz for 12002 data points.   I've heard that it's important to get that 3 places beyond the decimal as well...   

If you really feel you need this feature, let me correct the problems I noted and try to improve the sweep times.

[gardner]

Did you calibrate it for the 420-470 range, 420-480 range, some other range?  Want me to guess?

The calibration was over 384MHz-484MHz -- 434+/-50MHz

I did the sweep (6) because after the (5) segment, the graph never seemed to update.  I stopped the sweep at the start because it seemed like the right thing to do.  The manual doesn't say it should be sweeping or not prior to steps (2) (3) (4).  I took a guess.  Since the sweep seems to turn off after the segmented sweep and the graph gets updated/displays the final results, it seemed reasonable to guess that it should be off to start.

For grins I set up https://github.com/NanoVNA-Saver/nanovna-saver.  Their approach to segmented sweeps seems more what I would have assumed.  It is easy to configure and works alright with my unit.  It can't do as much as your LabVIEW software, but what it does do, I find easier to figure out and more stable with an old original NanoVNA.

[joeqsmith]

It's now available if you want to try getting your 12000 points of antenna data.    You should find it to be a bit faster.  Running from 1 to 300MHz with a 50MHz span required 1min 40 seconds with 1.08.   This new version requires about 17 seconds.   Other communications are faster as well and you may notice some commands are processed much faster.   

I have corrected the polar transmission mode.  Not sure how I missed that.   When using segments, the software always sweeps to or past the stop frequency.  It then truncates the data.   Greater than or equal bug.... 

Because it seems you want to run multiple segmented sweeps back to back, the software now shows you a popup telling you to press the Sweep to reset the VNAs ranges and enable sweeps.   It just saves you having to do this manually.  For me, it was no big deal as I was only collecting a single sweep. 

Not being able to replicate your problem with the firmware/hardware I have, I am not sure if this new version will have any effect on what you are seeing. 

[gardner]

What I find interesting is the feel for needing higher resolution when looking at SWR.

Well I don't, really.  Where I want more points is when I want to sweep from 100MHz to 500MHz and have the smith chart or the return loss chart look like something.  With too few points, the curve shapes wind up with large aliasing artefacts and aren't useful.  Once the bent wire, make-shift balun or section of twisted pair is in the ball park and I'm looking for tuning around the last 5MHz, the small number of points is fine.

With the nanovna-saver calibrations may be performed with a segmented sweep and have 1000 or more points to cover a large range of frequencies with enough points to keep artefacts somewhat at bay.  It also lets me do single port calibration, which is what I want at this stage of the game.

[joeqsmith]

What I find interesting is the feel for needing higher resolution when looking at SWR.

Well I don't, really.  Where I want more points is when I want to sweep from 100MHz to 500MHz and have the smith chart or the return loss chart look like something.  With too few points, the curve shapes wind up with large aliasing artefacts and aren't useful.  Once the bent wire, make-shift balun or section of twisted pair is in the ball park and I'm looking for tuning around the last 5MHz, the small number of points is fine.

With the nanovna-saver calibrations may be performed with a segmented sweep and have 1000 or more points to cover a large range of frequencies with enough points to keep artefacts somewhat at bay.  It also lets me do single port calibration, which is what I want at this stage of the game.

Not having ever tried it, I still have little doubt that the open source software would be much better suited for the radio hobbyist and general user.  It was written for the public which is why I continue to suggest people use it.   

I can understand the time required to measure the thru causes a loss in productivity over just performing the SOL.  Of course, one could always press the OK button without the thru inserted to save those few seconds.   

Here we have the simulated antenna being swept from 200MHz to 900MHz with 101 data points.   The one thing to notice that there isn't enough data to see the range crossover noise but the area I would expect we are interested in is clearly visible.   

******
Oh now I see the need. Those extra 34900 points made a huge difference.      

 

[joeqsmith]

Had to power cycle the old Nano again after the new firmware hung again.   The noise at the transition was much worse than the old firmware.  There have been some random glitches as well.  Not to mention, I can speed things back up with it.     The plan is to go back to the old firmware from RadioListener.   It's archived somewhere in this thread if anyone is interested. 

[joeqsmith]

I am a little puzzled by segmented sweeps.  With so few points in the built in frequency sweeps on the original NanoVNA, it seems that it would be useful to allow segmentation of sweeps for the "Main" tab -- SWR. Smith chart and so forth.  But when I tried that, it looks like I get a sequence of charts, updating once for each segment, rather than a single chart with all the segments on it.

After reading this comment a second time, I wonder if there was an expectation that when using segmented sweeps that you get a concatenated view of the data while it is being captured.  So the software would display all of the data captured up to the point of each segment.   Easy enough to add but a pure guess on my part if adding it would be useful to anyone.   
 
***
I tried it and personally like it, so it's in there. 

[gardner]

I wonder if there was an expectation that when using segmented sweeps that you get a concatenated view of the data while it is being captured.

Yes, that was how I imagined it would work, before I first tried it.  As I mentioned before, the appearance of the subcharts for different segments was not really how I pictured it working, but seemed to be according to design.

As it happens, NanoVNA-saver shows the same chart continuously during the sweep.  You can see the effect directly on the chart as additional sweeps are made and averaged, and as additional segments are swept in succession.  It's pretty well how I imagined it would work.

[purpose]

Mr. Smith,
Quick note to say that v1.10 will not work at 401 data points. v1.08 worked well and displayed the correct points value.
Switching the vna to 101 resolved the issue.

Peter

[joeqsmith]

That new version my be more to your liking then if you wanted to try it.  Still, I wouldn't spend too more time with it.     

I had not given much thought how the segmented sweep should work when I wrote it over a year ago.   I just knew I needed more data to run that particular demo.   
https://youtu.be/v01lSgWo904?list=PLZSS2ajxhiQD2gftdurGQoyGpUM_HobNI&t=989

It's not a big deal to change the code.  The problem is defining what it should do.  For this particular software, that hasn't been a problem as I wrote it for my own use.    Now that it is in the wild, I am open to making changes that would make it more useful to the general user.   

Recently someone posted that Labview is crap and the serial ports don't work.  Now that's some top notch feedback.   

[joeqsmith]

Mr. Smith,
Quick note to say that v1.10 will not work at 401 data points. v1.08 worked well and displayed the correct points value.
Switching the vna to 101 resolved the issue.

Peter

    Yeah, this is what happens when I do a quick and simple change. 

It's like my previous post.   What are the exact settings you are using? Span, Start, Stop, Function?
When you say it does not work, what does it do?   

Should be simple enough to fix if you can explain how to replicate it.   If you can't, then I am left to guess.. 

[purpose]

Yeah.... I like shiny things.

Standard defaults..
10M centre
2M span
Start 1M
Stop 300M

Crystal fixture as usual. Bridge, ref, chn ref and throw in the crystal.
Waveform not visible, just noise.
Points box showing 101, but with v1.08 it would show 401.
The auto sensing of data points seems kaput.

Peter

[purpose]

What I see


What I expected to see

[joeqsmith]

Yeah.... I like shiny things.

Standard defaults..
10M centre
2M span
Start 1M
Stop 300M

Crystal fixture as usual. Bridge, ref, chn ref and throw in the crystal.
Waveform not visible, just noise.
Points box showing 101, but with v1.08 it would show 401.
The auto sensing of data points seems kaput.

Peter

Because you did not say, I am going to guess you are using the transmission function then.   

I am sure I am missing something.   

Are you using segmented mode?  What steps did you perform with the old versus new software?   You would need to made sure you detail this as the manual suggests, the order is important.   

****
Ok transmission then.

Forget the new software for now.  In 1.08, what exact steps did you take to get the 401 displayed?   

[purpose]

Forget the new software for now.  In 1.08, what exact steps did you take to get the 401 displayed?  

Not a single thing. It auto detected it and worked flawlessly from the start.

[joeqsmith]

Forget the new software for now.  In 1.08, what exact steps did you take to get the 401 displayed?  

Not a single thing. It auto detected it and worked flawlessly from the start.

Not a single thing?  Really?  Its not magic and certainly requires the human to do things.     Guessing your steps will make it much more difficult.       

You turned on the Nano with the fixture inserted and a crystal install.  You started the software, selected your defaults file. You then selected link.  Next you selected sweep.   You are using normalize which I doubt magically was enabled so, you may have removed the crystal, installed the short, stored the reference then selected the CHn-Ref.  Next you reinserted the crystal.   

Of course I don't see what you have included because I am only guessing at what you have done.   

[joeqsmith]

Let me guess, you are playing with some new version of firmware which supports 401 data points and I was supposed to just know this.