NanoVNA Custom Software

[purpose]

Not a single thing?  Really?
Really.
Same exact procedure with the different versions of software gives me different results.

No fixture, turn on the vna, then start the software and select defaults. Hit link and 1.08 will automatically update the data points to 401. Even if I use the 1.10 defaults. 1.10 just remains at 101.

And I am not pulling your plonker.

Peter

Same firmware that I informed you of a week ago. You even gave me an attaboy for documenting the firmware version.

[purpose]

Just a quick note.
My NanoVNA H4 arrived, I then loaded the latest Dislord firmware (1.0.53 with 401 data points) into it.
Lo and behold it behaves itself very nicely with Mr. Smith's software, which eats up the extra data with ease.
3D charts take a little longer to display, but that would be expected.

Good to know.  Also smart idea to document the firmware. 

[purpose]

Just saying that something has changed between 1.08 and 1.10.

[joeqsmith]

Not a single thing?  Really?
Really.
Same exact procedure with the different versions of software gives me different results.

No fixture, turn on the vna, then start the software and select defaults. Hit link and 1.08 will automatically update the data points to 401. Even if I use the 1.10 defaults. 1.10 just remains at 101.

And I am not pulling your plonker.

Peter

Same firmware that I informed you of a week ago. You even gave me an attaboy for documenting the firmware version.

So you have gone back to the problematic firmware that I gave you an attaboy for telling us what not to use?    Last I knew:
https://www.eevblog.com/forum/rf-microwave/nanovna-custom-software/msg3525476/#msg3525476

Which is why all your earlier screen shots show 101 data points up until now.   At least we now know what has happened. 

Just saying that something has changed between 1.08 and 1.10.

Of course it did.  Did you think I roll the version for the fun of it?     


The way the Nano works, there is a command to request the frequencies that it uses.  In my early versions of the software, I would request the frequency, then the two data sets.  To speed things up, I would request only the first data set if I was using S11.   Later I only requested the frequencies when I changed one of the setting that would effect it.  Again, to increase the speed.   

When the V2+ came out, they were smart enough to do away with all that.  It's up to the software to calculate the table.   When I went to a common code base, I had left the frequency request in there.   The number of elements the Nano returns is the number of data points I used.   

Now we want to sweep / zoom faster.  Simple solution, ditch that frequency command.  The Nano has always used 101 data points, so lock it down and calculate the tables like how the V2+ works.  This is much faster as we don't have to wait for the Nano to send us that data.   

This works fine with the 1.08 firmware which at least defaults to 101 data points.  Of course, if you use firmware that has a different number of data points, its not going to magically use them.  It's not just the number of data points that are a problem,  the frequencies the software calculates would be wrong as well.   

They may have added similar commands as the V2+ that allows you to program in the data points.  You can imagine the nightmare this firmware mess can cause.  This is why the manual provides a link to the version I tested with. 

The software was tested with the following firmware.

https://github.com/ttrftech/NanoVNA/releases/tag/0.8.0

An attempt was made to test hugen79 1.0.45 NanoVNA-H_20210130.dfu which caused random errors in the data.  This was something that was found early on when I first started to look at the NanoVNA.  Sadly, a year has passed and the firmware is still unstable.   

The fact you have gone back to that version tells me there is some feature it has that is worth having bad data. 

[purpose]

So you have gone back to the problematic firmware that I gave you an attaboy for telling us what not to use?

No, this is my new Nanovna H4 with firmware that gives up to 401 data points.
It worked fine with 1.08. It was nice to have a well rounded Smith Chart.

Of course it did.  Did you think I roll the version for the fun of it?   

OK... I left myself open to that one.

Of course, if you use firmware that has a different number of data points, its not going to magically use them.  It's not just the number of data points that are a problem,  the frequencies the software calculates would be wrong as well.

I've certainly seen a few Hz difference from 101 to 401 testing crystals, but v1.08 does appear to be crunching the greater numbers. Whether accurate ... only you would know.

The fact you have gone back to that version tells me there is some feature it has that is worth having bad data. 


Different beastie altogether.

[joeqsmith]

There are a few things that could be done.   
1) Obviously, do nothing requires the least effort.  Just stay with the old software.     

2) Add the frequency command so the table is scanned upon connecting.   It would need to be done before the Span and Center  are set.   It would work for a while but someone would have the bright idea to remove the command and break the software again. 

3)The software for the V2+ is user settable.   I have no idea if your new firmware supports a command like that or not.   The problem with this approach is people are not going to be smart enough to know how many points they need to set it to.   

4)It's easy enough to stay with the old method, but it's too slow if you want to use the segments and like the first approach, may brake in the future.   

5)Another option would be to lock the software to a version of firmware. 

Maybe there are other options.  I get the feeling that the firmware will continue to cause problems.   I'm thinking option 3.  If they remove the command in the future, I don't care as I no longer use it.  It is just that closer to the V2+ software.   It could be saved in the defaults file for both flavors of software.   And again, lets not forget the reason I ditched that frequency command was for the speed.   

The other problem with option 3, people will think they can set the Points to what ever number they want.  They may think it has something to do with segmented sweeps.  Option 2 is looking better....   

[purpose]

There are a few things that could be done.   
1) Obviously, do nothing requires the least effort.  Just stay with the old software. 

That sounds right up my street.

2) Add the frequency command so the table is scanned upon connecting.   It would need to be done before the Span and Center  are set.   It would work for a while but someone would have the bright idea to remove the command and break the software again. 

I wouldn't have a clue where to begin.

3)The software for the V2+ is user settable.   I have no idea if your new firmware supports a command like that or not.   The problem with this approach is people are not going to be smart enough to know how many points they need to set it to.   

I would have to include myself in the not smart enough group.

4)It's easy enough to stay with the old method, but it's too slow if you want to use the segments and like the first approach, may brake in the future.

I spent the best part of the day re-reading this thread in search of radiolistener's elusive firmware. I think he has deleted most of his links, but think I found his github (https://github.com/qrp73/NanoVNA-Q), as the dates seem to be about right.
I'm willing try it out in my first little nano, in the hope of replicating your setup.

5)Another option would be to lock the software to a version of firmware.

Firmware takes moments to change. I wouldn't mind that in the slightest.

Maybe there are other options.  I get the feeling that the firmware will continue to cause problems.   I'm thinking option 3.  If they remove the command in the future, I don't care as I no longer use it.  It is just that closer to the V2+ software.   It could be saved in the defaults file for both flavors of software.   And again, lets not forget the reason I ditched that frequency command was for the speed.   

Speed is nice, accuracy even better, but I wouldn't know what to do with either.

Peter

[joeqsmith]

If we go with option 3 and save it to the defaults, it certainly makes the code cleaner.   It could be a pull down menu to force 101 or 401.   Then again, the firmware will then break it in the future.   So keeping it flexible would make more sense.   Use the default of 101, but just allow you you change it.   

Of course, in cases where people have a Nano fetish, they may want one default file for their entire collection.       There was a person who had bought five of them and I think they wrote they had damaged three. 

Yes, that looks like their code.  A shame that the people doing the large releases didn't follow suit.   That code has problems as well but I am not having to power cycle the Nano with it, no random jumps in the data, it's faster and the range change transitions are cleaner.   

If you want to try it with a programmable points setting like the V2+, let me know.  A few minutes and we could have it in there. 

****
Because the V2+ has a programmable number of data points, I store this with each calibration.  When you load a calibration, it sets up the VNA's defaults. 

I do the same thing with the original Nano.  I don't think this would be a problem as I doubt you would try and use the same calibration files from one version of the firmware to the next.   

 

[joeqsmith]

The manual and both programs have been updated.   You will need to manually set the number of data points and save your defaults for it to stick. 

[purpose]

Morning Mr. Smith,
In v1.11 I adjusted the points box to 401 (highlighting and inputting from the keyboard) and saved to defaults. Restarting shows 401, but as soon as I link, it jumps back to 101. Again putting 401 in the box while linked, as soon as I click on a parameter it again jumps back to 101.
I tried changing the numbers with the up/down icons and it made no difference.
I tried calibrating and saving, no joy.
101 points works exactly as advertised.

Am I missing something other than grey matter?

Peter

[joeqsmith]

Remember how I posted about hard coding the number of data points to 101?  Well, I never removed that initialization.    Of course, this causes no problems with the old firmware as I am expecting it to be 101.       Testing the software will continue to be a problem as people diverge from the original hardware / firmware.   

[purpose]

Remember how I posted about hard coding the number of data points to 101?
No. I must have been drunk.

Well, I never removed that initialization.
V1.08 would beg to differ.

Of course, this causes no problems with the old firmware as I am expecting it to be 101.
Of course.

Testing the software will continue to be a problem as people diverge from the original hardware / firmware.
I can live with 101. Less data to crunch, but I know it's there if I need it.

Peter

[joeqsmith]

Remember how I posted about hard coding the number of data points to 101?
No. I must have been drunk.

Well, I never removed that initialization.
V1.08 would beg to differ.

Of course, this causes no problems with the old firmware as I am expecting it to be 101.
Of course.

Testing the software will continue to be a problem as people diverge from the original hardware / firmware.
I can live with 101. Less data to crunch, but I know it's there if I need it.

Peter

Life didn't start or end at software v1.08.  I thought I had made it clear in this post the steps I had taken AFTER 1.08 to improve the speed but you didn't seem to follow. 
https://www.eevblog.com/forum/rf-microwave/nanovna-custom-software/msg3552000/#msg3552000

Simple solution, ditch that frequency command.  The Nano has always used 101 data points, so lock it down and calculate the tables like how the V2+ works.

In other words, 1.09 hard codes the initialization.  I thought it would be understood that this hard coded initialization is what my comment referred to.  Version 1.11 allows the user to change the points setting but the software would still initial it to 101 data points.  Again, because it may not be clear, version 1.11 is after 1.08.  Version 1.12, which is also after 1.08, removes this initialization.    Because you seemed to miss it, prior to 1.09 which includes 1.08, the software would initialize the number of point by sending the frequency command.  This command is no longer in use.   

If the firmware being used with software 1.09-1.11 has anything other than 101 data points, it's going to cause a problem.
 

[purpose]

I thought I had made it clear in this post the steps I had taken AFTER 1.08 to improve the speed but you didn't seem to follow.

I'm just miffed that you would take my precious points away. The naughty step is a lonely place.

because it may not be clear, version 1.11 is after 1.08

I'm not buying that. Surely some mistake on your part. Hic.

Version 1.12, which is also after 1.08

Now you're just being ridiculous and it's mean of you to confuse someone with dyscalculea.

Sniff

[purpose]

Well would you believe it. The one that comes after 1.11, you know the one, now works.

Mr. Smith... may you live for a thousand years.

Cheers
Peter

Spoke a little prematurely.

101 points behaves correctly, but 401 throws a minor wobbly.
Zoom goes though its motions with angular waveforms and shifts right at the 500Hz step.
Hitting peak center does not set it back.

What comes after 12?

[joeqsmith]

Sadly, once again I am given little data.  Maybe it's to the right 100Hz.  Maybe 10Hz.  Maybe 1.  Wit the technology available to us today where we can easily do screen captures,  its still a problem.   

Oh well, I don't have the hardware to test it anyway and with firmware that places random errors in the data sets, it may not be useful to even try and sort it out.  Your call.  You want me to look at it, provide some actual data. 

Don't tell me the car pulls to the right.  Show me a picture so I can see the front wheel is missing....   

Also, it would have been helpful to know if you could replicate the problem with the old hardware and the version of firmware I tested with.

*****************
Also, it would have been helpful to know if it 1.08 zooms correctly when using the new 401 data point Nano.   

*****************
Also, you may want to make sure that it's not some sort of resolution problem with your new firmware/hardware.    You know the original firmware  firmware was 101 data points.  You know the software zooms into a span of 500Hz.   So you can calculate the step size.   You can also calculate the step size for your new configuration.   You could then test the step size to see if its acceptable.   

*****************

101 points behaves correctly, but 401 throws a minor wobbly.

I am not even sure what this comment means.   Are you running the new configuration with the 401 points but then setting the software to 101 and suggesting that behaves correctly?  Or are you running the original hardware with the firmware I tested with and that behaves correctly?   

[purpose]

Good Morning Mr. Smith,

Sadly, once again I am given little data.
You scientists and your bloody data.

Here we go then.
Kindly assume that these experiments are all carried out by a monkey under duress.
All will be with your software version 1.12 and are with the standard procedure for crystal measurement, up to and including zooming the waveform.
Start up defaults 10M centre, 2M span, 1M start-300M stop.

The first slide please, Mrs. Coldheart.

Old hardware with eddy 0.8.0 firmware.
No problems whatsoever, Exactly as I expect it to perform.


Old hardware with radiolistener's 0.4.4 firmware.
First zoom attempt missed the mark by 500k and did not recover.
Yes the crystal was plugged in and no I wasn't distracted by something shiny.


Second zoom attempt grabbed it nicely.


Now on to the fun stuff.
New H4 hardware with Dislord's 1.0.53 firmware, with 51-101-201-301-401 data point choices.

101 data points set.
First attempt was much the same as the first with radiolistener's. I forgot to take a screenshot (because I was distracted by something shiny) and the second attempt behaved as would be expected.


401 data points set.
The waveform 'throwing a wobbly'. It's not consistent, but disforms half a dozen times throughout the zoom process.


This is where we end up after it kicks to the right at the 500Hz step.
Cursor centre set. Cur B-Cur A shows -29.


Also, it would have been helpful to know if it 1.08 zooms correctly when using the new 401 data point Nano.
Indeed it does.

Also, you may want to make sure that it's not some sort of resolution problem with your new firmware/hardware.    You know the original firmware  firmware was 101 data points.  You know the software zooms into a span of 500Hz.   So you can calculate the step size.   You can also calculate the step size for your new configuration.   You could then test the step size to see if its acceptable.
I'm sorry, but anything containing the word 'calculate' frightens the life out of me. Simple arithmetic is within my ability, but I'm still in the dark. Could you rephrase your last paragraph for a child.

Are you running the new configuration with the 401 points but then setting the software to 101
No. I set the vna to whatever points I'm using, then change the software to match before linking.
1.08 would detect whatever the vna was set to and change automatically.

Or are you running the original hardware with the firmware I tested with and that behaves correctly?
I refer you to my previous answers.

Peter

[purpose]

Just to add to the last screenshot.
Pressing peak centre has no effect at pulling the waveform back.
On this test CurB-CurA is showing -50.

[joeqsmith]

Old hardware with radiolistener's 0.4.4 firmware.
First zoom attempt missed the mark by 500k and did not recover.

I've ran into this as well, even with the newer firmware.   I suspect its a result of my trying to speed the zoom process up.   

401 data points set.
The waveform 'throwing a wobbly'. It's not consistent, but disforms half a dozen times throughout the zoom process.

If you manually zoom in using the software (setting the span and center), do you get the same effect?  Can you replicate it without the PC attached? (I am guessing you could)

No. I set the vna to whatever points I'm using, then change the software to match before linking.
1.08 would detect whatever the vna was set to and change automatically.

There's a new bit of info I did not have before.  I had assumed it was fixed at 401.  Of course I've covered the difference of how 1.08 worked compared with the newer versions and the reasons for the change and won't repeat it a third time.  Yes, it's different.  If your too old for a trophy, would you like a gold star? 

Also, it would have been helpful to know if it 1.08 zooms correctly when using the new 401 data point Nano.   

Sadly I am left to guess if this actually zooms in correctly or not.   All I know from your previous statement is that it picks up the number of points automatically.       

The fact the software can't tell your Nano to move the peak to the center, seems to point to something with the firmware.  Again, you should be able replicate this without the PC. 

[purpose]

If you manually zoom in using the software (setting the span and center), do you get the same effect?
Exactly the same. Warped waveform shapes and stuck 50Hz right. Unable to centre.


Can you replicate it without the PC attached? (I am guessing you could)
I can't.
If I turn off the software and re start the vna, then use the span to zoom in increments, the final 500Hz step is solid.

If you're too old for a trophy, would you like a gold star? 
Yes please, Sir.

[purpose]

Sadly I am left to guess if 1.08 actually zooms in correctly or not.
Answered, but yes it does.

[joeqsmith]

If you manually zoom in using the software (setting the span and center), do you get the same effect?
Exactly the same. Warped waveform shapes and stuck 50Hz right. Unable to centre.

Can you replicate it without the PC attached? (I am guessing you could)
I can't.
If I turn off the software and re start the vna, then use the span to zoom in increments, the final 500Hz step is solid.

If you're too old for a trophy, would you like a gold star? 
Yes please, Sir.

You mention the centering problem and you read that I asked you to try and replicate it without the PC.   I am not sure if you actually started the Nano with no PC, set the span to 500Hz, set the center for 9.9974MHz as your graph shows, saw the same peak location, and assuming you did, entered a center of 9.997446MHz and saw the peak center.  Maybe it moved.  Maybe it didn't.  Maybe you didn't try it.  I don't have a crystal ball.  If you want a gold star, you're going to have to step up your game.   Don't try and out guess what I am asking by entering  9.9974 and thinking that is good enough.   You need to use the same settings that the software is using.   

Jot down the center and spans where you have noted the odd shapes and enter them in manually as well.   

Sadly I am left to guess if 1.08 actually zooms in correctly or not.
Answered, but yes it does.

So we are clear, 1.08 will zoom in correctly with 401 data points?   Or are you saying that 1.08 will zoom correctly with the old firmware and 101 data points?   

[purpose]

You mention the centering problem and you read that I asked you to try and replicate it without the PC.
And because you asked so nicely, I did just that.

I am not sure if you actually started the Nano with no PC
I did and also shut down the nano and restarted.

set the span to 500Hz, set the center for 9.9974MHz as your graph shows, saw the same peak location, and assuming you did, entered a center of 9.997446MHz and saw the peak center.
(nano only)As the software was off and I wasn't looking at any pictures, I would have to put on a peak marker and centre to that frequency.

If you want a gold star, you're going to have to step up your game.
I'm thinking of settling for the bronze.

You need to use the same settings that the software is using.
400k-60k-16k-3.2k-640-500. Done.

Jot down the center and spans where you have noted the odd shapes and enter them in manually as well.
Software 1.12
400k-hit peak centre, no worries
60k-hit peak centre, no worries
16k-waveform deformation-hit peak centre fine
3.2k-hit peak centre-not moving, but could be because it really is centered
640- mild deformation, hit peak centre jumped 50Hz left of centre
500-stays at 50Hz left of centre-hit peak centre jumped to 30Hz right of centre

I need a Whisky... Oh look, there's a bottle of Grant's.

[purpose]

So we are clear, 1.08 will zoom in correctly with 401 data points?

YES

Or are you saying that 1.08 will zoom correctly with the old firmware and 101 data points?   

Probably also yes.

[joeqsmith]

set the span to 500Hz, set the center for 9.9974MHz as your graph shows, saw the same peak location, and assuming you did, entered a center of 9.997446MHz and saw the peak center.
(nano only)As the software was off and I wasn't looking at any pictures, I would have to put on a peak marker and centre to that frequency.

You can't type in the center frequency from the menu?   

Jot down the center and spans where you have noted the odd shapes and enter them in manually as well.
Software 1.12
400k-hit peak centre, no worries
60k-hit peak centre, no worries
16k-waveform deformation-hit peak centre fine
3.2k-hit peak centre-not moving, but could be because it really is centered
640- mild deformation, hit peak centre jumped 50Hz left of centre
500-stays at 50Hz left of centre-hit peak centre jumped to 30Hz right of centre

So you're plugging these numbers in without the PC using just the Nano and seeing deformation?   Or are you using my software to manually enter them? 
 
The fact it seems to work with 1.08 and how you are seeing these odd spaced jumps when zooming,  I am leaning towards some sort of rounding problem with the Nano's firmware.   Again, 1.08 is asking for the table and working from that rather than telling the Nano what to do.   If there is a problem with the resolution 1.08 would mask it.   

Sort out how to enter the set points from the menu, then maybe you can help answer this.