How to find "best" SMA terminator with nanovna?

[flolic]

So, I got nanovna V2 and calibration kit that comes with it. But I also have two more SMA 50 ohm terminators of the unknown specs/history.
Obviously, if I calibrate nanovna with any of those, other ones show different results. How to determine which one has best characteristics?

[eb4fbz]

You will never know. The calkit is your reference.

[0culus]

If it's worse than your cal kit's load, you'll be able to characterize it. If it's better, you'll need a better cal kit (and possibly also a different VNA with more dynamic range than the nanovna)

[TheSteve]

As others have said you'll need a much better reference load - which will cost more then your nanovna.

[flolic]

Ok, that make sense.
 I don't see a point to invest any serious money into a better calibration kit, so let's hope supplied terminator is good one 

[cdev]

I'd like to try to make one, (within a usual hobbyist cost constraint), in your opinions, what kind of SMD resistor is best to use?

what kind of resistors  should I use, "that dont have inductance"? 

I have a nanovna2 so it goes up it seems to 4.4 GHz, as far as I can tell.

As others have said you'll need a much better reference load - which will cost more then your nanovna.

[joeqsmith]

I'd like to try to make one, (within a usual hobbyist cost constraint), in your opinions, what kind of SMD resistor is best to use?

what kind of resistors  should I use, "that dont have inductance"? 

I have a nanovna2 so it goes up it seems to 4.4 GHz, as far as I can tell.

As others have said you'll need a much better reference load - which will cost more then your nanovna.

I've used homemade standards for GHzish with reasonable hobby results for several years.   1.5GHz is pushing it but I did run some experiments a bit above 2GHz but confidence was low in the results.   
     
Assuming you really plan to work in above 2GHz,  I again suggest you check out Mario's website.  Link is in the attached post.   
https://www.eevblog.com/forum/projects/20db-rf-attenuator-seeking-feedback-to-improve/msg2987982/#msg2987982

Shown is my new poor mans standards and low grade connectors.  Included is a Beatty and T-check.  The T-check results were poor.    Also shown is the TRL standards which also gave poor results.   The above thread shows some of the data collected for these standards.

[cdev]

Are the long ones to serve as a known length/delay?

[Roger Need]

You can build some SMA calibration loads like this fellow did...

https://www.qsl.net/in3otd/electronics/VNA_calkit/SMA_female.html

Roger

[joeqsmith]

Are the long ones to serve as a known length/delay?

The larger diameter, longer coaxial parts are the attempt at a Beatty and a section of air line.  These are made from some hobby brass. 

The smaller diameter coaxial parts are the attempt at a TRL set.  I think I was attempting to get down to 100MHz with it.  Yes, it requires physically long lines.

All these bits sort of do sortasomethingish...      The T-Check for example, uses a couple of resistors and like trying to make a home made load, it yields poor results at higher frequencies. 

[radiolistener]

So, I got nanovna V2 and calibration kit that comes with it. But I also have two more SMA 50 ohm terminators of the unknown specs/history.
Obviously, if I calibrate nanovna with any of those, other ones show different results. How to determine which one has best characteristics?

you can measure it's resistance with multimeter. Close to 50.00 Ohm is better.
You can calculate SWR in the following way:

SWR = max(50, R) / min(50, R)

For example, if you're measured 49.98 Ohm, it means that this dummy load has SWR = 50 / 49.89 = 1.0022 at DC.

This cannot guarantee that it's impedance will be the same 50 Ohm at high frequency. But at least you can choice better one for low frequency in such way...

If dummy load has very flat response from 10 kHz to 500-1000 MHz, then there is a reason to believe that your measurement with multimeter for DC will be very close to the value for high frequency.

[Wallace Gasiewicz]

I have never used a nanovna. I use an old HP VNA. You are in a tight spot. If you have several 50 ohm loads which one to believe?
All VNAs that I have seen require good calibration. The newer inexpensive ones like the nano require a lot of calibration (lots of correction done by the computing power). The older expensive equipment is much closer to "accurate" without any calibration. In stuff like nanovna, there is a lot of adjusting of the result you see that is done by comparing to calibration. I think this means the "standard" you use for calibration is even more important.
You said you have several loads. I would trust the most respected brand. Huber and  Suhmer for example. However some loads I have used that are very expensive have been disappointing. Pasturnak is one disappointment.
Maybe someone else who has used better calibration standards on their nanovna would be better able to tell you which cal standard is acceptable. Maybe someone has evaluated the cal standards that come with the nano on a known good VNA.
You can buy good 50 ohm SMA loads for less than $10 that have been verified on very good equipment.
Maybe ask on the for sale forum.
Wally