Very accurate phase and frequency measurement using the nanoVNA-H4 HW

[erikka]

For those that want to do very accurate phase and frequency measurements, but are not willing or capable to buy a PhaseStation or Timepod a new FW has been made available for the nanoVNA-H4 HW that converts it into a mix between a DMTD and PhaseStation.
Accuracy of phase measurements is better then 1e-12 / (Tau in seconds) and the device support sending the phase measurements over USB to programs like TimeLab or writing to an internal SD card for later analysis.

During testing and comparing to other frequency and phase measurement devices, the tinyPFA outperformed conventional frequency counters and delivered phase measurements identical to a PhaseStation when used above the noise floor.

A wiki containing all the information for creating and using your own tinyPFA can be found here: https://www.tinydevices.org/wiki/
A support forum is available in case you have questions regarding the operation of the tinyPFA or want to share results
As the FW is fairly new you can expect some bugs.

The nanoVNA-H4 HW does not have to be modified so after use as a tinyPFA, it can always be converted back into a VNA.

[bingo600]

Thank you Erik   
And even with appraisal from "Bob" ...... Well done.

Now i need to get a H4 .... as i "Just have a " V2plus4

/Bingo

[rockwell]

Is there a difference between between the nanoVNA-H4 HW and the nanoVNA-H4 ?

[erikka]

No
The nanoVNA-H4 HW is the hardware of a regular, normal, unchanged nanoVNA-H4

[Mechatrommer]

is NanoVNA V2.2 included?

[erikka]

It is impossible to make this work on any of the V2, Lite or Libre VNA
And the H is too slow
And I will not make sw for the F

[syau]

Just ordered one from taobao, will test it out once received.

[thinkfat]

I've ordered one from Eleshop, it should arrive some time today. Maybe I'll fire up one of my Rubidiums and then let it run against my GPSDO

[thinkfat]

The nanoVNA was delivered a couple of hours ago and I took the time to update the device and do the noise floor test. I'm getting reasonable figures with a very ragtag setup I cobbled together in half an hour. The 0.1s ADEV is a tiny bit worse than what the graphic shows on tinydevices.org, but 1s, 10s, and 100s ADEVs are very reasonable and match or undercut the graphic.

In the same ragtag manner I'm currently running a quick test of my GPSDO against the OCXO timebase of my 53132A counter that has been switched off for months. It's reasonable to assume the OCXO will do a bit of retracing its previous lifecycle until it settles.

Overall, I'm quite happy with the results, especially knowing that I put essentially zero effort into it

[thinkfat]

So, here are the first results after letting the 53132A warm up for a few hours.

The measurement setup was relatively primitive, I just put the tinyDMTD on my lab bench, connected it to my laptop via the supplied USB cable and started logging in Timelab.
No special cables were used, I took some PTFE SMA cables, connected the timebase output of the 53132A to one channel and the GPSDO output to the other channel. Temperature was not controlled during the test and I was in the room the whole time.

I've attached a screenshot out of timelab after collecting 1 hour worth of data. I've also attached a screenshot from the GPSDO monitoring program, covering the same time interval.

The blue trace in timelab is the unedited phase data out of the tinyDMTD. The magenta trace is after removing the frequency offset against 10MHz and after correcting also the frequency drift - as mentioned, the 53132A was only recently switched on and it's reasonable to assume that it will drift for a while before settling.

Also, if you look at the first column of graphs from the GPSDO monitoring, the TIC ADEV is the phase error of the GPSDOs LO against the GNSS 1PPS output, after correcting quantization error ("sawtooth correction"). The actual GPSDO output cannot be worse than what is shown in that graph. Therefore it's reasonable to assume that the "uptick" in the blue Timelab curve is due to the 53132A timebase drifting against the GPSDO output. The magenta curve is likely closer to the actual GPSDO performance, as it corrects for said drift.

The timelab traces are all above the noise floor of the tinyDMTD.

[bingo600]

I have a H4 on the way from eleshop 

[edpalmer42]

I finally got around to buying a NanoVNA and flashing it to a TinyPFA.  I'm very impressed with this little gadget.  I'm still in the process of getting used to it and finding out what I can and cannot do with it, but I thought I'd post some of the ADEV graphs I've come up with.  Some of these create more questions than they answer and require more analysis so don't just accept them as absolute truth.

The last two graphs demonstrate the kind of questions I'm coming up with.  For some reason, everytime I try to include my HP 5065A Rb Standard in a graph, I get the strange results shown.  I'm wondering if there's a problem with my unit that has previously been below my visibility threshold.  More measurement and analysis required!

What would be useful is to see others repeat my tests to confirm or challenge my results.

Ed

[erikka]

Ed, a second oscillator at 0.1Hz offset leaking into one of the tinyPFA inputs could cause such an oscillation.
Make sure to use double shielded or semi rigid coax cables and switch of al oscillator not needed or imove them away from the measurement setup

[edpalmer42]

True, but if there was such an oscillator, it would show up in other measurements.  It didn't.  Besides, I just checked all my active oscillators and they are all within 0.005 Hz of their nominal value (either 5 or 10 MHz).

Edit:
I should also state that my frequency counter is a Fluke PM6881 referenced to the same 5065A referred to above.  When checking the frequency of my HP Z3801 and Trimble Thunderbolt GPSDOs, the reported error is less than 0.0005 Hz, i.e. < 5e-11 for the 10 MHz outputs of the GPSDOs.

Ed