NanoVNA Custom Software

[joeqsmith]

Step one, turn off the data while changing the frequency.
Not entirely sure what you mean by this.

Just tried it as normal. More deformed waves and ending up 30Hz to the right of centre.

Need some sleep, but expect you'll have 14 by the time I wake up.

Toodle pip.

Explain what is meant by "normal"?   

Did you try both 101 and 401?  Does 101 still work? 

Change the Refresh time from 200 to 600 and the PgmTimeout from 1200 to 1800 and let me know if it has any effect.   There may be more than one problem.  That centering sure sounds like a resolution problem but if you say it works fine with the menus,  maybe not.   No point in wasting too much more time with it.   

You should plan on staying with 1.08 for now.   It will take some time for the VNA to show up.  I will still need the Github link for the exact same firmware image you are using. 

[purpose]

Morning,

Explain what is meant by "normal"? 
Because you didn't explain what "Step one, turn off the data while changing the frequency", I didn't do it, so ran the normal test.

Did you try both 101 and 401?  Does 101 still work? 
Last night only 401, but this morning tried 101 and 101 works well.

Change the Refresh time from 200 to 600 and the PgmTimeout from 1200 to 1800 and let me know if it has any effect.
None, or at least none that I noticed. Waveform still ends up to the right of centre.

No point in wasting too much more time with it.
Agreed.

One other thing that I forgot to mention is that all versions after 1.08 (in the SetupDiagnostics page info box), only displays 'the quick brown fox', whereas previous versions have all the nano info.
Might be something.

Peter

[joeqsmith]

Morning,

Morning.

Explain what is meant by "normal"? 
Because you didn't explain what "Step one, turn off the data while changing the frequency", I didn't do it, so ran the normal test.

You may have noticed that this was the same comment in the readme file for that release.  It's not instructions for you to follow, its documenting what has been changed.  More so for my own benefit.   The manual now contains a Software version section that I will maintain, at least internally, to help keep track of the development.

>>>Just tried it as normal.
>>Explain what is meant by "normal"? 
>Because you didn't explain what "Step one, turn off the data while changing the frequency", I didn't do it, so ran the normal test.
 
Ok, it's clear with your circular explanation of what normal is, you're not sure, which is normal.

Did you try both 101 and 401?  Does 101 still work? 
Last night only 401, but this morning tried 101 and 101 works well.

Good info.

Change the Refresh time from 200 to 600 and the PgmTimeout from 1200 to 1800 and let me know if it has any effect.
None, or at least none that I noticed. Waveform still ends up to the right of centre.

More good info. 

No point in wasting too much more time with it.
Agreed.

One other thing that I forgot to mention is that all versions after 1.08 (in the SetupDiagnostics page info box), only displays 'the quick brown fox', whereas previous versions have all the nano info.
Might be something.

Certainly a possible clue.  Some firmware does not report any info.  If you run the terminal menu and select the info command does it send anything back?   

No matter.  Once the hardware arrives and you provide a link for the exact firmware you are using,  I can take it from there.     

In the mean time, I suggest you get a pull test jig finished so we can start to validate some of your results. 

[purpose]

Ok, it's clear with your circular explanation of what normal is, you're not sure, which is normal.
Nothing is as it seems.

If you run the terminal menu and select the info command does it send anything back? 
It does. Fills the info box with what I would expect to see.

Tried running another crystal test, but same right of centre result.

In the mean time, I suggest you get a pull test jig finished so we can start to validate some of your results.
I'll get on it. Cough.

Peter

[purpose]

Afternoon,
I think I may have inadvertently made this...




You're going to tell me to start reading some lengthy paper now, aren't you.

[joeqsmith]

Looks good.   Do you know how to use it?  If not....

Because the person who first asked me about making this measurement had brought up the Dishal software, this is what I used.    I used version 2.0.5.2.  This has support built-in for that pull test.   There is a decent help file for it. 

In my case, I used my old HP5328A counter.  I have a GPS that runs 24/7 for the time reference.   

Running the 1.08 software and working firmware (if you can find one) you should be able to compare Lm, Cm  and Fs.   For Cp, I just used my RLC meters to compare results.   

Make sure the parts are stable before you measure them.   I noticed that the  0.8.0 firmware would still glitch which is why I went back to what I know works.   

[purpose]

Evening,

Do you know how to use it?
Of course not, but finally got some numbers out of it. Missed a couple of connections.
I don't have a counter with a reference in, so will have to make do with what I have.

Browsing the 'crystal test set' pdf you linked mentions this in regard to the G3UUR circuit...
Although this circuit is useful for general
purposes, it is terribly inaccurate for those
performing demanding design tasks such as
narrow-band crystal filters.

That, plus a probably inaccurate counter might just be a waste of your time and mine.

Talk me out of it.

Peter

[joeqsmith]

Evening,

Do you know how to use it?
Of course not, but finally got some numbers out of it. Missed a couple of connections.
I don't have a counter with a reference in, so will have to make do with what I have.

Browsing the 'crystal test set' pdf you linked mentions this in regard to the G3UUR circuit...
Although this circuit is useful for general
purposes, it is terribly inaccurate for those
performing demanding design tasks such as
narrow-band crystal filters.

That, plus a probably inaccurate counter might just be a waste of your time and mine.

Talk me out of it.

Peter

Not sure why you feel that's my job.  If you have things you would rather be doing, I am perfectly fine with that. 

[purpose]

Not sure why you feel that's my job.
You threw me into this quagmire, at least throw me a rope.

[joeqsmith]

Not sure why you feel that's my job.
You threw me into this quagmire, at least throw me a rope.

No, you make your own choices.  While I have some of my own time invested in your project (my choice) I have no problem stopping the efforts if that is what you would like to do. 

[purpose]

Good grief, Mr. Smith, ... I merely suggested that my setup was possibly inadequate.
I'm not going to let you abandon the last few weeks of interaction because of a misunderstanding.

Have a smoke if you've got em... pour yourself a whisky and I'll pop back in after I've committed my mother to the flames.

Peter

[joeqsmith]

Good grief, Mr. Smith, ... I merely suggested that my setup was possibly inadequate.
I'm not going to let you abandon the last few weeks of interaction because of a misunderstanding.

Have a smoke if you've got em... pour yourself a whisky and I'll pop back in after I've committed my mother to the flames.

Hopefully you understand that this is my choice to make.

I didn't think it was going to be such a big deal for you to run the test and post your findings.

[purpose]

Hopefully you understand that this is my choice to make
Sure, but leaving a destitute orphan to navigate the wilderness is cruel.
Besides, you've just ordered a new gizmo and you have to at least tell me I'm a bellend for even ordering it.

Peter

[purpose]

Well, accuracy aside, I did some mathematics and came up with...
Cm 9.0616fF. Software 1.08 came up with 9.6216fF.
Lm 27.98mH. Software 1.08 came up with 26.33mH.

Would this be considered a validation?

[joeqsmith]

Hopefully you understand that this is my choice to make
Sure, but leaving a destitute orphan to navigate the wilderness is cruel.
Besides, you've just ordered a new gizmo and you have to at least tell me I'm a bellend for even ordering it.

I see people buy all sorts of things for various reasons.  What you spend your money on is really your choice.   I have no use for this VNA outside of seeing why my software isn't working with what ever flavor of the day firmware you are trying to use.   I had no plans to support anything beyond the original hardware and firmware called out in the manual.

Well, accuracy aside, I did some mathematics and came up with...
Cm 9.0616fF. Software 1.08 came up with 9.6216fF.
Lm 27.98mH. Software 1.08 came up with 26.33mH.

Would this be considered a validation?

I have no way to know what you did unless you provide the details?  What did you use for CSW?  How did you derive it?   Guess?  Saw something on the internet?  Don't think it matters?  Guessing gets old.   

It does seem like what ever you are doing is causing a fair amount of error.   

[purpose]

I have no way to know what you did unless you provide the details?

Here we go...
I measured the crystal with S1 and S2 closed (crystal to ground) and got a number.
I then measured with S1 open (through Cs) and got another number.
Punched these numbers into some formulas and got some other numbers.
Mathematics is not my thing, so could easily have made a mistake.
The formulas were this lot and are not my numbers...


My numbers were...
fgnd  9997742
fopen 9998439
Cs 62pF
Co 2.99pF
Delta f 697

What did you use for CSW?  How did you derive it?
Cs was 62pF. W? Not a clue. Measured at 10M on the vna.

It does seem like what ever you are doing is causing a fair amount of error. 
Again, mathematics is not my thing.

[joeqsmith]

I thought the software made all the calculations for you.  You just had to plug in the values you read from your equipment. 
There was a pretty good help file for it and I would have assumed you read it. 

What did you use for CSW?  How did you derive it?
Cs was 62pF. W? Not a clue. Measured at 10M on the vna.

How did you connect it?  If you use your RLC meter and replicate my steps, then what do you get?

[purpose]

I thought the software made all the calculations for you.
Now that you mention it... cough.


Pretty much bang on to what the vna reports.
Software 1.08 came up with 9.6216fF
Software 1.08 came up with 26.33mH

[joeqsmith]

I could believe those numbers.  My problem was  I never had a way to prove the absolute accuracy, which is where that standard crystal would have been nice to have. 

[purpose]

Same crystal, different day.
1.08 software, 401 data points, 650 sample crystal sweep.


Input software reported Co of 2.49pF into dishal calc.

[joeqsmith]

Looks like the VNA will arrive sometime next month.  In the mean time, I have removed the newer versions of the software and have left 1.08 based on your claims of it being more compatible with other hardware/firmware.  You may not be the only person trying to run unsupported combinations of firmware and hardware.  While the newer software is much faster running segmented sweeps, I have only seen one person ask about it.  With them now using the open source software, I consider it a very low priority.   Reading the groups, you get a feel for how unstable the software/firmware/hardware combos are.   

I would imagine if you ran some sort of repeatability study, you may get a better feel for how stable the readings are.   That seems like a deep rabbit hole.  Much more effort than just collecting some data off your new fixture which was already a stretch.   You could just play with that software to try and see how these errors would effect the filter you are trying to design. 

[purpose]

I would imagine if you ran some sort of repeatability study, you may get a better feel for how stable the readings are.
OK. Exact same everything, say 10 times and jot down any differences. I can do that.
Probably prudent to restart software and hardware between each pass.
 
Peter

[joeqsmith]

Achieving the exact same conditions will be impossible. 

I doubt restarting the PC, software, or VNA would do much beyond adding the time required to run it.  The VNA will drift and would rather be just turned on and running. 

Depending what you want to learn, you could try different firmware.  Glitch free stable versus the current versions.   Leave the cables attached and the crystal plugged in, not touching anything.  Maybe even put a towel over it to minimize the effects of air currents.    Or maybe you want to try swapping the same crystal in for each test.  Or maybe even reinsert the cables for every test.   You could try with them torqued vs hand tightened.  At the frequencies you are running, finger tight should be fine and I doubt you would see a difference unless you were being very sloppy.   

You could play with the software to determine what sort of sample size you need.  Time to blow the dust off that stats book. 

[purpose]

Time to blow the dust off that stats book.
A bit short of breath these days, but I'll get as much done as I can.

[joeqsmith]

Another option, now that the data is within reason between your pull fixture, RLC meter and Nano, is try to measure your bag of parts and build up the filter you are after.