I've been using my nanoVNA v2 plus4 since mid-November, mainly to check the common-mode impedance of various HF antenna tuner-balun arrangements on the HF ham bands (S11 port to input of tuner, S21 port to output from balun), and also did some checks with various resistances (33 ohms to 10K ohms) between S11 and S21 ports.  That's been with the same calibration file for 1-30 MHZ via the vna-qt software.  The loss figures were consistent over the several weeks and also very close to the results with another 2-port vna (an MFJ-225).

Yesterday I ran another test, looked over the loss curve for 1-30 MHz, recorded the results, and then just left the nanoVNA/vna-qt running, still showing the loss curve on the computer screen.  An hour and a half later, I got back to the project and noticed that theS21 loss curve looked somewhat different.  (I'd touched nothing in the meantime.)  I recorded the new figures.  So, for example, I had earlier measured a 28.8 dB S21 loss at the bottom of the 80 meter band (3.5 MHz).  Now it was 31.9 dB.   A similar shift occurred across the HF spectrum.  I then rechecked two other tuners and got a comparable shift from previous figures I'd recorded.  For known resistances, I also got a similar shift (for example, from 3.0 to 4.5 dB loss with 50 ohms between the vna ports, 19.5 to 23.6 dB with 1K, 38.5 to 42.8 dB with 10K).

I've now done another recalibration of the nanoVNA, reloaded the firmware (same version 201013 as came with the instrument), used a new set of cables between the vna and my test jig, scratched my head, etc.  I keep getting the same puzzling "new" figures.  (When I cross check with the MFJ-225, it still gives me the old loss figures, very close to the nanoVNA's "old" figures.)

It's almost as if some random component or setting change occurred within the nanoVNA.

What else can I do to track down the source of the problem?   I been "playing" with ham-band antenna tuners for years, but I'm a real novice with it comes to this new instrument.

Charlie Lofgren, W6JJZ

Here's another symptom of the problem with my nanoVNA V2Plus4.  I did a single-port (SOL) calibration for 1-30 MHz and checked the S11 resistances for several 1/4 and 1/2 watt carbon composition resistors.  I've used these on other VNA instruments and have found them to measure close to their nominal values (+/- a percent or so).

With the nanoVNA, the resistors with values 50 ohms and under measure OK.  Those above 50 ohms measure significantly over their nominal values *and* different sweeps give somewhat different results for the same resistor, for example:

100 ohms measures 110-112 ohms
150 ,  175-180 ohms
180, 218-225 ohms
390, 604-645 ohms
470, 858-948 ohms

I'm guessing there's a malfunction in the hardware, but that's only a guess and I don't have a clue what it might be.  Any help?????????

Charlie Lofgren, W6JJZ

Quite a strange failure mode.
Try calibrating with a different load and then measure?

On Thu, 17 Dec 2020 at 23:37, Charlie Lofgren <clofgren@cmc.edu> wrote:

> Here's another symptom of the problem with my nanoVNA V2Plus4. I did a
> single-port (SOL) calibration for 1-30 MHz and checked the S11 resistances
> for several 1/4 and 1/2 watt carbon composition resistors. I've used these
> on other VNA instruments and have found them to measure close to their
> nominal values (+/- a percent or so).
>
> With the nanoVNA, the resistors with values 50 ohms and under measure OK.
> Those above 50 ohms measure significantly over their nominal values *and*
> different sweeps give somewhat different results for the same resistor, for
> example:
>
> 100 ohms measures 110-112 ohms
> 150 , 175-180 ohms
> 180, 218-225 ohms
> 390, 604-645 ohms
> 470, 858-948 ohms
>
> I'm guessing there's a malfunction in the hardware, but that's only a
> guess and I don't have a clue what it might be. Any help?????????
>
> Charlie Lofgren, W6JJZ
>
>
>

Thanks, Dragan,

I did another single port (SOL) calibration with vna-qt, and got the correct S11 impedance figures for the various test resistors.  I'm not sure what happened the first time.

But that still leaves the problem I described in my first email--the S21 loss figures that differ from the ones I'd obtained over several prior weeks, with no change in the calibration.  As I mentioned those earlier figures agree with the ones I obtained with another 2-port vna.  I'll do some additional testing, this time with known loss pads.  (I have them for 3 thru 30 dB, and supposedly good at HF frequencies and some into the UHF range.)   I'll report back when I get the results.

Charlie, W6JJZ

Problem solved.  Actually, there turned out to be no problem with the "new" loss figures.  I pulled out my RF attenuators--an HP 355c certified for dc thru 1000MHz for 0-12 dB loss and a homebrew unit for 0-71 dB-- and checked the readings from the nanoVNA V2 Plus4 against the various attenuator settings.  They matched dead-on with the HP unit and within a fraction of a dB with the homebrew unit.

The V2Plus is a great instrument!

Now I can confidently redo my earlier measurements.

Charlie, W6JJZ