Author Topic: NanoVNA Custom Software (Read 565875 times)

joeqsmith · « **Reply #2475 on:** October 25, 2024, 08:48:04 pm »

The new release of Solver will not be compatible with the earlier releases and requires a new Defaults file to be created. This is why the major revision number was incremented. Shown is a decoupage of the main changes involving the UI. Note that it supports several databased standard setups. Like the polynomial coefficients, the file locations for your databased standards are stored into the Defaults file. For those of you using a transfer relay with Solver, note it supports unique Touchstone files for each port. So if you are testing an insert able device, you would calibrate the two ports using your male and female standards. If you wanted to use the same sex, single set, just select the same Touchstone files for both ports.

I believe we have already discussed the other changes.

Solver's TDR/TDC math still differs from how CMT and METAS handle the reference plane, for now. Again, I am not apposed to changing Solver if I ever get it through my head why what they are doing makes sense. My skulls thick so it may not happen.

I have had people approach me about doing some sort of collaboration in the past. Typically I ask them about their experience with LabView, education level..... That normally ends the discussion. I assume they just wanted the source for their own personal gains. I am thinking about releasing a VI wrapper for the TDR function and let people post their own source code for it. If some bright young LabView guru came up with something that we could all agree on, I could just add it in. Maybe allow the user to select the method.

6.00, 10/25/2024
Update readouts when loading S2P files
Combine stability trigger to other cursor handler, all cursors now track
Change 2-Port Adv. Plots to SI units
Change TDR/TDT impulse curves to use the amplitude scale
Swap TDR/TDT samples/seconds graph axis selection
Expose all TDR/TDT filter settings to user
Add support for METAS time domain modes using an uneven step size
Change TDR FIR filter to Zero Phase IIR filter
Add 2PSwp to readout event
Add Reflection types of round trip and one way
Force the EXE location as a starting directory for Defaults and other files
Add full 2-port support for databased standards SOLT
Allow multiple sets of databased standards
Allow saving Databased standard's paths to the Defaults file

joeqsmith · « **Reply #2476 on:** October 25, 2024, 11:27:32 pm »

Attached is a VI with an array of sampled data for a Beatty standard. I would normally pass down my data into this array along with the nominal resistance. I have ran it though my normal processing and left the data displayed on the graph as the default to show basically what it should look like.

This was done using 2011. If you have a newer than 2020 LabView, save it to the older version (2020 or lower). Any functions are fair game to use assuming they are included as part of the professional package. In other words, don't use any functions that would require any third party add-ons.

joeqsmith · « **Reply #2477 on:** October 26, 2024, 03:11:43 am »

To get some idea of the warmup time for the new LiteVNa 3.2 revision, I let the VNA sit at room temperature overnight. I had attached a cable to connect port 1&2. I started Solver, then turned on the VNA. After connecting, I normalized the data to zero and recorded the change. It drifted a few dB in the first half hour. I then renormalized back to zero and let it run for another hour where it appears stable. Of course it would take a little longer to stabilize with cooler temperatures. Also notice that most of the high gain drift happens at the higher frequencies. You may be able to just turn it on a start collecting depending what frequency you are using it at and how accurate you are looking for.

joeqsmith · « **Reply #2478 on:** October 26, 2024, 05:10:15 pm »

Tosed the Lite3.2 into a sealed bag and into the fridge to cool it down for about an hour. Looking at the min/max, you can see even at a GHz, it only drifts about 1dB while warming up. Again, the higher in frequency, the more drift I see. Of course the temperature swing is much more than bringing it up from room temp and we see about a 12dB change at 8GHz.

Looking at the sweeps in 3D, starting at 0 (cold), it warms up fairly quickly and is quite stable after about a half hour.

joeqsmith · « **Reply #2479 on:** October 26, 2024, 05:37:14 pm »

For those attempting to write their own TDR, I created a new wrapper that contains the uncalibrated S11 data from the following post. The Touchstone file is also included and can be imported into METAS or CMT if you would like to view it. I also included the filter parameters I would normally pass down to the TDR routine. I added an X-Y graph as well that you can use for debugging. Or delete it.

For those not familiar with LabView, you can download a free copy of the community edition. It should work just fine with this VI. I just tried Solvers TDR with a student version of LabView 14 and it loaded fine as well. Point being, I think you could do all this using the base package (shouldn't require any of the advanced math). This means you could use Inverse Chirp Z, IFTT, or what ever you like.

Again, if someone creates a version we can agree on, I could easily include it into the next release.

https://www.eevblog.com/forum/rf-microwave/vna-for-cable-characterization/msg5684985/#msg5684985

joeqsmith · « **Reply #2480 on:** October 29, 2024, 12:56:26 pm »

While waiting to what the 18 people who DL'ed the wrapper come up with, next I wanted to repeat the T-Check using the original NanoVNA. After finding that transpose bug in Solver's coefficient calculations, the first step is to update the old software. Note it will jump from 2.x to 6.x., bringing in all workable changes that have been added to support the LiteVNA.

***
Updated screen capture

joeqsmith · « **Reply #2481 on:** October 30, 2024, 01:15:03 pm »

I pulled out the H4 to start testing the software. I had not used it since running the regression test with some new firmware that was found. Surprising, it turned on without being plugged in. However, it seems to have developed a major problem. Note S21 (blue). S11 is also bad. I thought maybe that new firmware had a problem that was missed when I tested it. So I tried the original FW. The battery was very low, so I let it charge. This had no effect. The hardware is dead. The H4 has been sitting in it's original box for all but maybe 40 hours of testing since I purchased it a few years ago.

That photo was taken with nothing attached to the H4. With both S21 showing such high coupling, and S11 with a load show very poor return loss, guessing something common failed.

joeqsmith · « **Reply #2482 on:** October 31, 2024, 12:08:45 am »

I inspected the H4 under the microscope and saw no visual signs of a fault. With S21 being so bad, I started out swapping port 2's mixer. I checked Digikey and they had no stock. So I sacrificed my original NanoVNA. The battery had expanded on that unit and the connectors need to be replaced anyway.

Notice that the two mixers are marked different. Photo of the screen is with port 1 terminated and port 2 floating. I had saved the original data I collected with the H4 when I purchased it and it seemed to perform better than this.

Next I connected it to the PC and calibrated it. METAS plotting the before and after repairs of the return loss and dynamic range. Possible one or both port1 mixers is out, or the replacement mixer is causing a problem.

As a side note, when I popped the cover the first thing I noticed was a mezzanine board floating in the air. They had attempted to hot glue it to the main board. They had used that white glue which I never have good luck with it adhering. I replaced it with some of the orange glue.

joeqsmith · « **Reply #2483 on:** October 31, 2024, 01:33:49 am »

Swapping the other two mixers has no measurable effect. Shown S21 sweeping from 50k to 1M with data normalized with open.

I thought maybe gluing the mezzanine board back down close to the main board may be causing a problem but lifting it had no effect. I had also installed the latest firmware, which was rolled back the what it was shipped with. Again no effect. Looking all each component from the connectors to the mixers, all appear fine.

joeqsmith · « **Reply #2484 on:** October 31, 2024, 02:03:35 am »

Comparing the H4 with my original NanoVNA that I robbed the mixers from.

joeqsmith · « **Reply #2485 on:** October 31, 2024, 02:12:02 pm »

If Dislord is still out there, I am curious if you have made access to the IFBW, number of data points, control of the synthesizers, for the H4 available from the remote (PC USB)?

This was a real pain with the software as you would have to use the menus on the H4 to set these up. Because I am doing a major update of my software to support these older VNAs, if you have these controls available and documented somewhere, I would like to include them.

joeqsmith · « **Reply #2486 on:** October 31, 2024, 02:24:25 pm »

I tried swapping one of what appears to be a good, original mixers out of the H4 back into port 2. This had no effect on S21. Looking at the the original mixer from my H4's port 2 input, it is marked completely different than the other two. I wonder if that was a coincidence or did they purposely use a better part for this mixer?

I then started to look at other areas of the PCB. The soldering isn't the best. Looking at C48, which is one of the large bulk caps, it appears on lead never wicked up. Applying a small bit of force to see if it had ever soldered, the part popped off. There are no signs of the common side tacking to the lead, the other appears to have been cold at best. No damage to the part when it popped off. I removed the old solder and reinstalled it.

joeqsmith · « **Reply #2487 on:** November 01, 2024, 06:18:08 pm »

After correcting the bad solder joint, I saw some improvements in the higher frequency but this is not the problem. I added a few other caps to the board in a last ditched effort. It did improve the overall noise levels but it was obvious there was still a problem.

Even with 100 averages, the S21 was about 60dB at 200MHz. I suspect if it was an ESD event, that the transient increased the leakage of the clock driver. They already push that part way beyond it's limits. So, I once again reached for my original NanoVNA and removed the driver which I then swapped into the H4.

They claim a dynamic range of:
70dB 50kHz-300MHz, Measured 80dB
50dB 300M-600MHz, Measured 75dB
40dB 600M-900MHz, Measured 64dB
---dB 900M-1.5GHz, Measured 63dB

This data was taken with the latest Hugen firmware and 1KHz IFBW (default). Comparing it with the original data I collected, we can see below 100MHz, the unit still does not perform as good. Beyond 100MHz, we can see the effect of adding the caps. I had added 4X4.7uf cer. One for each mixer and one for the ADC's reference. I also doubled capacitance of the 2Xtants.

The low frequency performance is better than the original NanoVNA than two of the mixers and clock driver were taken from.

joeqsmith · « **Reply #2488 on:** November 01, 2024, 08:56:19 pm »

Looking at the data I collected from 50k to 2GHz when the H4 was new, the low frequency noise peaks up to 80. Very similar to what I am seeing after my repairs. I am not sure why that one data set is so low. I'm sure I wouldn't have been averaging or lowing the IFBW. It almost looks like that first data point was a fluke.

The previous data was taken with the two ports open but cables attached. Terminating both cables lowers the floor even further up till about a GHz but had no effect on floor below 100MHz.

Next step, see if we can use the T-Check with the H4 to get anything meaningful beyond 50MHz.

joeqsmith · « **Reply #2489 on:** November 02, 2024, 12:25:04 am »

Comparing my original T-Check data for the NanoVNA sweeping 1-50MHz, with the now slightly modified H4 sweeping from 50kHz to 50MHz using the low cost T-Check. The H4 was calibrated using the same standards and METAS derived coefficients as before:

https://www.eevblog.com/forum/rf-microwave/nanovna-custom-software/msg5668129/#msg5668129

Was -1.3 to +3%, with the H4 -0 +1.5%.

joeqsmith · « **Reply #2490 on:** November 02, 2024, 12:34:43 am »

The Solver manual has a section on how to use a transfer relay along with the T-Check. At the time I had purchased the V2Plus4 and show it within +/-10% to about 200MHz. This was all done using the ideal models. Also, I had discovered a transpose error in my error corrections which would be in play at the time this data was taken. It's not very impressive results when you consider I was testing my Agilent to 9GHz and holding much tighter.

joeqsmith · « **Reply #2491 on:** November 02, 2024, 12:51:06 am »

But, the results with the V2Plus4 were much better than what I ever measured with the original NanoVNA. Shown sweeping from 1 to 150MHz. It stays within +1/-10 up to 130MHz. This was a lot better than what I was able to get with the first GaAs homemade transfer relay.

The H4 after repairs and modifications, using the new Solver software with the low cost T-Check will hold +/-5% to 900MHz. It will hold +/-10% to over 1.2GHz now. Sure, it's not near as good as the LiteVNA but this is a major improvement in the accuracy I have been able to achieve with these low cost VNAs.

The big take away from these last few weeks is that I have seen many people posting about the need for expensive cal kits with these low cost VNAs. The data just doesn't support it. Using these same standards supplied with the V2Plus4 (which measure the same as the ones supplied with my LiteVNA), with nothing more than a better load, could be used to calibrate my Agilent and measure the same T-Checks to 9GHz.

joeqsmith · « **Reply #2492 on:** November 02, 2024, 07:04:20 pm »

Stepping the attenuation from 0 to 80dB and sweeping from 50k to 2GHz with the H4. The IFBW was set to 4kHz. Results are decent to about 180MHz or so.

Also shown is a 11th order lowpass filter I had put together for the beginner section. This filter achieves 90dB rejection at 30MHz. The repaired/modified H4 with 100Hz IFBW is note quite able to measure it.

joeqsmith · « **Reply #2493 on:** November 07, 2024, 04:41:40 pm »

22 downloads of the TDR wrapper so far but I have not been contacted by anyone wanting to go over their design. If you know LabView and want to contribute, there is still plenty of time as I have no plans to release an update to Solver anytime soon.

https://www.eevblog.com/forum/rf-microwave/nanovna-custom-software/msg5690523/#msg5690523

The next software release will include both versions of Solver. One supporting the LiteVNA/64 and V2Plus/4. The other for the original NanoVNA and the H4.

***
To be clear, both programs would be 64-bits, using the same runtime engine and NIVISA, rather than requiring a separate 32-bit for the older software.
***

I have done a fair amount of testing with H4 version of Solver and while I unimpressed with having to use the VNA's display and touchscreen to set it up, it doesn't appear that we will see a new version of firmware that makes all of the settings available using the remote connection.

I have now shifted my focus toward the manual which is woefully outdated. There have been a lot of features added to Solver that the manual doesn't cover. Some of the information in the manual is no longer valid. The plan to to clean that up over the next few weeks.

When the manual was written it really only covered the old 32-bit version of the software. The new manual will cover both programs and discuss the differences between the various VNAs where applicable.

If there are areas you would like to see included, feel free to ask.

The current cost of a perpetual license of LabView is $13,682 USD. Sales tax at 6% brings it to $14,693.72. I had hoped that Emerson would reconsider raping their customer base but we are a few years into it and I don't see this changing. They do talk about negotiating the price which tells me they are trying to sort out how much they can charge to maximize profits. The plan at some point may be to create a kick start or something to try and fund it. I have not yet looked into the taxes and other details yet but am considering it. There would still be no support out side of being able to continue Solver's development.

https://www.newark.com/ni/776678-35/labview-software-professional/dp/15AJ5392

nealix · « **Reply #2494 on:** November 08, 2024, 04:50:43 pm »

Joe, or other experienced Solver64 user(s):

I just received a NanoVNA V2Plus4. I downloaded Joe's dropbox folder set for the software.

Question: In the sub-folder for NI_Runtime_VISA, he indicates to download:

LVRTE2011SP1f2std.zip
ni-labview-2020-runtime-engine_20.1.1_offline.iso
ni-visa_20.0.0_offline.iso

When I go to the N.I. Website, there are newer versions of the two packages.
Does Joe recommend staying with the specific versions mentioned, or can you
download the latest Labview Runtime and latest NI Visa packages?

Also, is it correct to assume that you only need ONE of these two runtime packages:
LVRTE2011SP1f2std.zip
ni-labview-2020-runtime-engine_20.1.1_offline.iso

Sincere Thanks for any guidance/assistance.

Neal

joeqsmith · « **Reply #2495 on:** November 08, 2024, 06:15:45 pm »

Hello Neal,

Your V2Plus4 uses the later protocol. This means downloading "ni-labview-2020-runtime-engine_20.1.1_offline.iso"
and "ni-visa_20.0.0_offline.iso".

You will also need the files contained in the Solver64 directory.

joeqsmith · « **Reply #2496 on:** November 10, 2024, 08:10:52 pm »

The directory NanoVNA now contains NanoVNA64.exe and associated files. The 32-bit version of the software revision 2.11 for the original NanoVNA and H4 has been rebuilt with the 64-bit toolset. This means only one version of the runtime engine and NIVISA are required to run both applications.

The revision was incremented from 2.11 to 2.12 and the program is marked to indicate that it is now 64-bits. The defaults file name was changed to defaults_h4.set rather than defaults.set. This should allow you to place both programs into the same directory.

No other updates or bug fixes were applied. The plan will be to eventually release a full blown upgrade to both programs, bringing them to the same revision, along with a much needed users manual.

***
The dropbox directory structure will also change from three to two directories. The Solver64 directory will contain both executable programs along with the manual. The NanoVNA directory will be removed.

joeqsmith · « **Reply #2497 on:** November 23, 2024, 04:18:58 pm »

I have been working on the new manual and am looking for people willing to review it. If you are interested in helping out, send me a PM and I will provide you a link to where you may download a copy of the draft once it is available.

joeqsmith · « **Reply #2498 on:** December 05, 2024, 02:37:40 am »

With the new release of Solver, I am planning to allow the user to set many of the colors and save them as your defaults.

The main graph is an oddball with its raster images. These were fixed to green. In the early days I was using black. You can now select black, green or white.

Shown would not be my color choice but it demonstrates how it can be changed.

RobNC · « **Reply #2499 on:** December 07, 2024, 12:20:04 am »

Quote from: joeqsmith on November 23, 2024, 04:18:58 pm

I have been working on the new manual and am looking for people willing to review it. If you are interested in helping out, send me a PM and I will provide you a link to where you may download a copy of the draft once it is available.

Can't send messages (I tried), but if I can help to review this, please let me know. Thanks!
I have the LiteVNA64 but playing catchup with all your posts and connectivity, firmware updates, etc. >25 yrs VNA experience but LiteVNA64 newbie.
Per what others have said, it would be helpful to have a maintained (I volunteer if requested) single place for instructions instead of "just Google it" methods. No need to make all newbies like me suffer, ha ha...


EEVblog Main Site	EEVblog on Youtube	EEVblog on Twitter	EEVblog on Facebook	EEVblog on Odysee

Author Topic: NanoVNA Custom Software (Read 565875 times)

Share me