Obtaining better accuracy with VNA

Hello,
As a hobby I have been designing a UHF receiver. When designing a UHF BJT amplifier I have been measuring the S parameters with the VNA, creating a S parameter simulation model, importing this into a simulator and then designing the matching circuits.

I wish to obtain better accuracy with my measurements as my amps typically achieve 6db less than what the simulator or datasheets claim I should be able to achieve.

Will I achieve better accuracy by creating a CAL kit that reflects the PCB my circuit is mounted on and calibrating using this? Or otherwise should I use the SOLT CAL kit I have and simply adjust the phase offset and electrical delay to a value that is appropriate for the connectors on my PCB?

Does anyone else have any other recommendations?

I have recently purchased some expensive cables in the hope that they may help. I am considering measuring the performance of my library of resistors, caps,inductors and using these in my simulation models.

FYI my design frequency is 450.325MHz.

Thank you

I would prefer in-circuit calibration with semi-precise standards (e.g. good chip reistors) over port extension.

I don't believe that measuring S-parameters of passive components will bring you much advantage. Usually the effect of PCB pads and traces in different circuit topologies will be larger than the parasitics of the naked component. At least the test setup should be well considered. It's more important to model the basic circuit wiring correctly. For small SMD designs, you can still use a lumped element approximation as LC ladder.

cool Thanks

Are generic thin/thick/metal film resistors suitable for reflection measurement calibration? I've found differing results between using cheap (but precise) 0603 resistors vs fancy "RF" resistors like these ones. But I can't really verify which actually gives the better calibration.

There will be surely differences. My suggestion is primarly intended for the sub-GHz frequency range. According to my experience, small chip resistors are pretty accurate in this range. You can also refer to manufacturer application notes about RF behaviour of chip resistors that tell the same.

For many VNA applications, a small s11 error (e.g. < 0.02) of a calibration resistor won't matter that much, for example when making impedance measurement of electrical small antennas that underly huge accidental errors by nature. I found that the said "semi-precise" in circuit-calibration is always outperforming an exact calibration with an unsure piece of transmission line in front.