Hi Alan,
I did some modelling and there are quite a few places where errors can creep in. I was a little surprised on the significance of Tx driver source impedance and balanced Rx output loading impedance have. It will take more analysis to understand what can be calibrated out.
The excitation driving source impedance has a resistor attenuator pad to improve the source impedance control but a compromise must be made on the tolerable drive signal loss. The capacitance of the synthesizer driver output and input of the SA612 effect high frequency end.
The sophistication of the calibration correction algorithm may be an area to examine as the correction algorithm for low Z measurements can be different from high Z measurement. How you might interpolate calibration points may depend on what is creating the errors, resistor matching, absolute resistor values, and/or stray reactance. At low freq end stray reactance effects are minimal but may have to begin to worry about spurious mixing products not being attenuated enough by LPF effecting ADC readings.
But this is a low cost 'hobbyist' unit. I would like to see a better synthesizer IC and mixer IC but that would run up cost. I have a NanoVNA, NavoVNA-F, and N1201SA unit. I like the N1201SA with its better AD4351 synthesizer chip and high frequency capability but it is limited on low freq end and doesn't seem to have open source code available. My N1201SA was built with AD4351 chip which could go down to 35 MHz but firmware stops it at 137.5 MHz as if it has the AD4350 chip. I would definitely go for the extra bucks of the NanoVNA-F over the original NanoVNA. The -F has 5v boosted supply for mixers where the original NanoVNA runs directly from 5v USB that drops to LiIon battery voltage when USB supply ls not connected. The -F also has the more powerful M3 processor and extra flash memory along with several other improvements including a real housing.
