Hello

I'm using N5230A to masure ~5pF capacitance (grounded)

I measure S11 and convert it to capacitance.

The question is

how to calculate possible uncertainty of such measurement based on PNA spec?

Thanks in advance

I'm using N5230A to masure ~5pF capacitance (grounded)

I measure S11 and convert it to capacitance.

The question is

how to calculate possible uncertainty of such measurement based on PNA spec?

Thanks in advance

The VNA uncertainty calculator gives S11 phase accuracy ~1.2deg at 10MHz -2 G band.

Is it worst case?

Is this value the same for 10MHz and 2GHz?

I'm pretty sure it is worst case.

It could be better at one frequency than another, but I would guess it would be no worse than the value from the calculator. Probably 2GHz would be better than 10 MHz.

Alsmost always the error is greater at the higher frequency. The error is due to uncertainty in the length of the open or the short, and this uncertainty in length produces greater phase error at higher frequencies.

But sometimes, other effects (such as coupler rolloff) add to the uncertainty, so we pick the bands based on the value being within a 2:1 or so ratio over the band. In fact, the error in phase, might even look a bit sinusoidal, as we use a polynomial fit for the data and some of the fitting is less good at low frequencies due to loss affecting the impedance.

The uncertainty calculator gives specified performance, which I think is 2 signma; this means the standard deviation of the phase error is about .6 degress. In practice, the results you get will be about 2 times better than this, so expect about .3 degrees, .03 dB errors, if you are quite careful (it's because it's very unlikely that you will get both a worst case NA and a worst case cal kit, and even if you did, it's unlikely that they will both be worst case in the same direction at the same frequency.

Ok

When i measure fringe of 83059 male adapter (using VNA5230A kit85052B)

i see 71 (70 69..) fF on the Smith all time with variouse calibrations

this is 0.05deg(=arg(1-j*w*c*50)/(1+j*w*c*50)) at 1GHz

Where is 0.3deg uncertainty?

Phase uncertainty seems to be less realy.

Last explanation gives me some insight.

Here's what @Discussion Forums

So some 2 sigma, mostly worst case.