NanoVNA Custom Software

[realfran]

Yes, the Nano-QT does not use the unknown thru only the SOL and the SOLT.
The xaVNA software seam to use the unknown thru possibly on the library their some information for it,  I'm NOT able to read or write any software.

[joeqsmith]

Dislord,

I had no problems with the install.  I started by testing the power level settings with both synthesizers in swept and CW modes (4 tests total).  This appears to now work as I would expect. 

Next I tried loading a very bad cal into memory 0 and rebooted the VNA.  Then connected using the PC.  I now get the raw data as I would expect. 

Next I tried sweeping my home made band pass filter.  The Lite was not calibrated for this test.  The data seems in the ball park as I would expect but again, I can't verify the results.    Just for fun I checked and it seems to lock it at 9.3GHz.  I normalized with the supplied cables and ran 7.8GHz HP filter.    Yellow is the normalized thru, purple the HP filter and red with the ports terminated.   Just a quick sanity test but again, it all seems expected.   

I did try to collect some data down at 9kHz.  Appears the same. 

Looks good.  Again, thanks!

[joeqsmith]

Without having the old kickstart hardware, it would be difficult for me to say if it works or not then.   Of course, anymore I would need you do define "works" as that particular word carries little weight with me anymore.   The software guy will tell me

my code built without errors.  See, it works!

  I would want to see the performance gains before I would suggested it worked.

[joeqsmith]

I went over the Lite VNA with a friend today.  They are well versed in the world of RF.   After showing them the specs and hardware , I spent some time going over my software.   They wanted to know how to measure group delay with it.  So I took 15 minutes and added it.  Then explained the power of LabView.   

Shown looking at the 12MHz lowpass I made for one of the demos. 

[joeqsmith]

Midwest Microwave 10 & 20dB attenuators using leakage error.   I have not tried it running from batteries or that linear supply I made to try and tame the V2Plus4.   

[joeqsmith]

Using my home made 70mm Beatty standard, 101 data points, SOL with no average (software or firmware).   

[realfran]

I hope to get the V.3 software with the news feature to test the Lite-VNA whit it,  I  don't get any substantial difference with the T/R relay on the HP test set with two relays getting good isolation considering the Lite-VNA spec.
The screenshot uses the leakage correction for testing one 50homs load.
Not averaging

[joeqsmith]

Good to see another person playing with the Lite.   The only features added to 3 are:

*Automatic collection of antenna radiation pattern
*Average sweeps
*Group delay
*Increased timeouts to support the Lite
*Allows setting the Lite's power levels
*Allow larger data sets (more points) to support the Lite in unsegmented sweeps
*Allows setting the Lite's internal average (or what ever Dislord decides to call it in the future)
*Maybe one day, new error models

Most of these changes would have very limited use.  For example, measuring radiation patterns requires a fairly expensive motor and would need the user to develop the gear drive.   Like smoothing, averaging is not very useful IMO and can mask problems.  The timeouts are only a problem when using the Lite at very low frequencies.   I was surprised that not one person had asked me about adding group delay but it also tells me that most users would have no use for it.   Error correction hasn't shown to have any use yet.   The documentation also is lacking but many people seem unable to read anyway so this may be of little concern.  Want some entertainment and drama, read the last issue posted on Git.  I had no idea that the NanoVNA had anything to do with the spread of the virus until recently.   

In a nutshell, there's no benefit to the antenna analyzer group. 

Dislord,
What are the default settings for your firmware's average and power levels?  Will these be user settable and made sticky so they are saved as part of the calibration?   If so, when using USB, will these settings reflect what the user has set of be some other default?  Is there a way to query the Lite (using USB) to know what they are set to?   I wonder if I need to consider initializing the Lite to some known defaults when connecting to it.  If so, maybe store them in the defaults file to allow my software to make them sticky.   Curious of your plans. 

I suspect there may be a small problem with your average feature.  I believe I have seen it not change in some cases but I will need to test it.   Is it similar to how the power setting was locked out in CW, where maybe you can't just change it in some modes?   If so, can you provide these requirements so I can automate the change.    If you are not aware of a problem, I will try and sort the cause and get back with you. 

[realfran]

Hello, I'm glad you added this feature my plan is to learn to use one VNA, I haven't the money to buy one professional one, on the market there is vintage but the working one hare expensive. The portability for me is not interesting, what is important is to use it with PC all possible VNA features.
The calibration system if there is a possibility to improve in the future.
The philosophy of the mobile phone is to put everything on.
Testing Minicircuit VAT-15+ attenuator very close to specs.
Touchstone file for VAT-15+
 

[DiSlord]

nanoVNA also have Group delay trace

Lite use inernal AVG setting for send to CPUm but you can always send need value before sweep start. You can read current avg settings by read 0x40 register (or set this valuse if write in it)
Read - need send 0x10 0x40 and read answer byte

Also for power levels, send on sweep start, or read current from register
0x40 - (byte) average
0x41 - (byte) si power
0x42 - (byte) adf power

PS soon i hope increase S11 (and others) dynamic range, i add AGC for all measures (AGC auto gain change, allow select best gain for every measure data). Before AGC have limited functional only in S21 measures (for > 140M)

[joeqsmith]

... value before sweep start.

I think that this may be tied to what I saw.  I would have been trying to change it after or during.   I'll double check it.

I'll go ahead and add the power and average to my defaults file and auto set them when connecting to the VNA.   It will be interesting to see how these effect the dynamic range.   Are you posting this data in that beta test area?  I may need to join up.   

[realfran]

(quote from joeqsmith)" Most of these changes would have very limited use.  For example, measuring radiation patterns requires a fairly expensive motor and would need the user to develop the gear drive.   Like smoothing, averaging is not very useful IMO and can mask problems.  The timeouts are only a problem when using the Lite at very low frequencies.   I was surprised that not one person had asked me about adding group delay but it also tells me that most users would have no use for it.   Error correction hasn't shown to have any use yet.   The documentation also is lacking but many people seem unable to read anyway so this may be of little concern.  Want some entertainment and drama, read the last issue posted on Git.  I had no idea that the NanoVNA had anything to do with the spread of the virus until recently."

Online, there is a lack of teaching to use the VNA with your video people learning a different way to use VNA not only "antenna analyzer" if not have request from people is for the lack "knowledge" (this is my case)  I thank you for your effort to teach me many think to use the VNA with your video.
Question: Is possible doing a T-Check on a wide range of frequencies from 10Mhz to 6.4 GHz?

[joeqsmith]

Question: Is possible doing a T-Check on a wide range of frequencies from 10Mhz to 6.4 GHz?

I suspect what you are asking is if under the best conditions, could the LiteVNA achieve acceptable results over this range.  I don't know.

On a more practical side, I think I made a demo showing a home made transfer relay with the original NanoVNA working up to 50MHz.   I don't know if I ever demo'ed the OEM relay.  Still, it would have been with home made FR4 standards which are not going to be near good enough. 

I attempted to make a better T-check standard and ran it on my vintage 4 x receiver PNA.  Good cables, properly torqued, home made standards I had characterize against some Agilent standards using Mario's scripts and SOLR correction.   About as good as I could do.   The performance was poor.   

That paper I showed during the demo from RS had an example at 4GHz or so.   But that standard was nothing home made.   I guess you would need to try it and report back. 

[joeqsmith]

... value before sweep start.

I think that this may be tied to what I saw.  I would have been trying to change it after or during.   I'll double check it.

I'll go ahead and add the power and average to my defaults file and auto set them when connecting to the VNA.   It will be interesting to see how these effect the dynamic range.   Are you posting this data in that beta test area?  I may need to join up.

There was a problem with my software.  I can change it in all modes as I would expect.

[joeqsmith]

Testing Minicircuit VAT-15+ attenuator very close to specs.

Your  measurement appears worse than I would have expected.   I tried using a more stable set of cables and again torqued everything.  Shown with the the leakage term enabled (I gain about 10dB with it) and Midwest Microwave attenuators at 20,30 and 40dB.   I would expect your VAT device to be much flatter. 


I tried the same settings using a very low cost 20dB attenuator like the ones listed below.   Its within 1dB over the range.  I would expect you to have similar results.   
https://www.amazon.com/Nooelec-SMA-Attenuator-Kit-Line/dp/B07YYMT65T

[joeqsmith]

Question: Is possible doing a T-Check on a wide range of frequencies from 10Mhz to 6.4 GHz?

I suspect what you are asking is if under the best conditions, could the LiteVNA achieve acceptable results over this range.  I don't know.

On a more practical side, I think I made a demo showing a home made transfer relay with the original NanoVNA working up to 50MHz.   I don't know if I ever demo'ed the OEM relay.  Still, it would have been with home made FR4 standards which are not going to be near good enough. 

I attempted to make a better T-check standard and ran it on my vintage 4 x receiver PNA.  Good cables, properly torqued, home made standards I had characterize against some Agilent standards using Mario's scripts and SOLR correction.   About as good as I could do.   The performance was poor.   

That paper I showed during the demo from RS had an example at 4GHz or so.   But that standard was nothing home made.   I guess you would need to try it and report back.

From the RS paper:

This parameter should be exactly 1, i.e. 100%, in the ideal case. Any deviation from 100% indicates inaccuracies of the network analyzer. Deviations of up to ±10% are considered as minor (green range). Deviations between 10% and 15% are still acceptable (yellow range) and those more than 15% should not occur in a good vector network analyzer after careful system error calibration (red alert).

V2Plus4 shown with my home made PCB standards and transfer relay.  Full 2-port cal was performed and sweeping from 1MHz to 1GHz.   The Transco part far exceeds my home made transfer relays.  At a GHz, its well within the +/10%.   I've used the FR4 standards to about 1.5GHz but that's about their limit. 

The home made standards I made for the PNA are coaxial and give fair performance to several GHz but I am using an OEM load that I had sorted.   The T-Check is standard resistors and it's just not good enough.  At 2GHz, its within 15% but not impressive at all.   

[tungsten2k]

I have not tried it running from batteries or that linear supply I made to try and tame the V2Plus4.

External linear might help but ferrite loop on stock LiteVNA USBC cable didn't seem to make a big improvment for me.



-=dave

[joeqsmith]

I'm not too surprised.  Using the linear supply with the V2+4 had no effect.  As soon as I run it off external power, the noise goes up.   Someone had recommended going right to the battery pack.  This would of course work.  Or, in my case, I just turn off the liner supply (basically cut the USB cables power pin only).   The Lite was not nearly as bad as the V2+4.  Easiest way to check for possible gains it just try it with and without external power using the firmware.   

Every time I try to use the firmware I waste time trying to get what I want on the screen.  Most of it, I find very easy.  For the Lite, I turn off everything but S21. Turning off the traces requires double taps.  You can't select what you want to display with each trace. Viewing S21 requires the second trace.  There's no autoscale and that reference position seems to be in some alien units.  Assuming you set the stop freq to 6.3GHz, you would think you want -50dB or maybe 50dB ref.   Of course, that drives the trace off screen.  Maybe 0 resets it.  Nope.   Set the reference to 9.  Once you have it all set up the way you want, save the configuration.   Surely it saves everything to NVRAM.  Power cycle the unit.  Oops.  All your work was lost.  Start over...   

I obviously never took the time to learn to use it as I just don't run it this way.

Anyway, once you have it setup. Try plugging in the unit to the PC. Don't connect with the software.  Just watch the LCD and see if it changes.  The Lite I have, using 10dB/div shows no difference.    If I do this with the V2Plus4, the noise goes up a fair amount.   OWO claimed the cables but really, it seems they screwed the pooch on the USB interface was all.   

[tungsten2k]

Yes, the whole VNA base FW it all started from is a bit of a UX disaster, but then so are many commercial VNAs (it’s just that you’ve grown used to them over the years ). DiSlord had made some improvments over the V2P4 atrocity but there’s only so many shades of lipstick that look good on a pig. And that the LiteVNA manual exists only in Chinese makes even learning how this “settings stored in 50 different procedural logic traps that aren’t documented in English” even more frustrating. But unlike you, I no longer have access to equipment that I can occasionally use for this purpose, so learning the parts I need to make it work for me I’m more motivated to endure. The portability of the device is also unparalleled (which, again, is moot for you, so I understand your dismissal of it because  a lot of these idiosyncrasies).

Regardless, we persevere

Btw- as you’ve undoubtedly noticed, there’s 2 main settings groups under the hood: the “config” and the cal data stored in cal slot 0. Unfortunately it’s still not fully known to me all of the various ways data is saved (and more importantly jettisoned) from the config nvram but I’ve learned the few parts that seem safe to touch in order to keep the “config” from changing over a power cycle. However, even now I occasionally stumble into doing something only to find the config changed (and as you’ve noticed, only after frustratingly finding out only because of getting impossible results after a lengthy setup process, then having to start fully from scratch [insert table flip here]). Yes, it’s all a UX hot mess nuclear disaster.

For most of these community based “designed by engineer” projects I just rely on the benefit of reading the code when something doesn’t make sense, but that’s not an option here. Regardless the fact that it is available for such a low cost is it’s only saving grace that makes hitting ourselves upside the head with a hammer getting it to at least work for the few things we need it for an overall net positive.

-=dave

[joeqsmith]

With the early firmware, less options = easier to setup but it was clear with fragments on the display and other oddities it was going to need a better UI.  Of course, I was never able to get the software that was supplied to work well enough to be used.   The solution seemed obvious.   With the software no longer an issue and limiting the the requirements for the firmware,  it  gets down to the hardware which has it's own limitations (square wave drive....).    I've shown in the right hands they can throw up some pretty decent numbers.    I've never had a problem saving/using the internal calibration, until the Lite was changed to send corrected data.   The latest firmware seems to work well.   I've put a few hours on the Lite now and not had any major problems.

All of them I have looked at have been under $200.  At that price point I'm not expecting much.  I continue to be impressed with what has been achieved.

[Alextsu]

Yes, the whole VNA base FW it all started from is a bit of a UX disaster, but then so are many commercial VNAs (it’s just that you’ve grown used to them over the years ). DiSlord had made some improvments over the V2P4 atrocity but there’s only so many shades of lipstick that look good on a pig. And that the LiteVNA manual exists only in Chinese makes even learning how this “settings stored in 50 different procedural logic traps that aren’t  an overall net positive.

-=dave

English version of Lite,s manual here,:
https://static.eleshop.nl/mage/media/downloads/LiteVNA_User-Guide.pdf

[joeqsmith]

Yes, the whole VNA base FW it all started from is a bit of a UX disaster, but then so are many commercial VNAs (it’s just that you’ve grown used to them over the years ). DiSlord had made some improvments over the V2P4 atrocity but there’s only so many shades of lipstick that look good on a pig. And that the LiteVNA manual exists only in Chinese makes even learning how this “settings stored in 50 different procedural logic traps that aren’t  an overall net positive.

-=dave

English version of Lite,s manual here,:
https://static.eleshop.nl/mage/media/downloads/LiteVNA_User-Guide.pdf

4. Calibration status
Displays the saved slot number of the calibration being used and the error correction applied.
C0 C1...: Each indicates that the corresponding calibration data is loaded.
5. Reference position
Indicates the reference position of the corresponding trace. You can change the position with:
DISPLAY →SCALE →REFERENCE POSITION.
6. Marker status
Displays the selected marker points and the frequency of the marker p...

Notice, not even a comment about the alien units used.

7 Device settings
The CONFIG menu contains general settings for the device:
Saving device settings
Select CONFIG →SAVE to save general instrument settings. General device settings are data
that includes the following information:
• Touch screen calibration information
• Brightness
The CONFIG →SAVE command does not apply to calibration setting 

Maybe it saves the trace settings, maybe it doesn't.  This manual leaves everything up to your imagination but I'm not Willy Wonka or playing with everlasting gobstoppers. 

***
Thinking about the poor results with the VAT attenuator and T-check.   Don't underestimate the importance the cables play.  You can't magically fix a bad cable with calibration and the higher up you go the more critical they become.  Same for proper torque.   

I played a bit with the cables supplied with the V2Plus4.  They were fairly stable and flexible.  I used them during my review of it.   I didn't try the ones that came with my Lite.   The ones that came with my original Nano I never even bothered with.   I have used this type of coax for the majority of what I have shown.     

[gf]

... the alien units used.

IIRC, in the NanoVNA V2 firmware there was at least one parameter whose unit were screen divisions.
Don't remember which one, but could have been the reference level (eventually it is just a vertical offset for the trace).

EDIT: I.e. at say 10dB/div, three units means that the origin is shifted by of 30dB then (or at 20 Ohm/div, three units mean 60 Ohm).
I also don't remember whether the origin was at the top or bottom of the screen when the offset is 0.

[joeqsmith]

IIRC, in the NanoVNA V2 firmware there was at least one parameter whose unit were screen divisions.
Don't remember which one, but could have been the reference level (eventually it is just a vertical offset for the trace).

Even yourself seem unsure and it sounds like you use it.  That's not good.  Once it goes off screen you can guess how many alien units to try to get it to display something again.  There's no undo, autoscale or even a full range button.   In my example, I selected a different trace function, then re-selected S21 log to start over and guess some more.

[DiSlord]

Reference = N ean shift trace 0 position to to N*Scale from bottom screen, yes not good

you can change reference / scale position of current selected trace by tap on left side (see attachement)
Scale factor change from middle of screen