Hello,

my sincere thanks to David Blackham for providing the VNA uncertainty calculator. It is a great tool that we use a lot.

In most applications we were especially interested in the uncertainty related to power level, i.e. changing the measurement

AND calibration power level to perform the uncertainty analysis of our ENA.

Now we needed to adapt only the calibration power level and I do not understand the calculation results.

Changing the calibration power from 0 to -10dBm and to -20dBm I see a global minimum in the S21 uncertainty curve

which is correlated with the calibration power - this is fine. However, I thought that lowering the calibration power level

will introduce a corresponuding amount of noise (i.e. lowering the power from 0 to -10dBm increases the noise floor or

dynamic range by 10dB). However, the results show that there is almost no influence on the measurement uncertainty.

I need to lower the power levels to something like -60dBm to become a significant additional error related to the cal-power-level.

I assume that your calculations are correct, but could you explain why this has to be like this?

I have looked through all sorts of publication, including the online manuals of the instruments/web as it has been mentioned that the formulae

are "all there". However, the formulae include the effective system data, only and do not seem to include power levels. Is there any publication

that includes this influence as well?

Cheers,

mleibfritz

my sincere thanks to David Blackham for providing the VNA uncertainty calculator. It is a great tool that we use a lot.

In most applications we were especially interested in the uncertainty related to power level, i.e. changing the measurement

AND calibration power level to perform the uncertainty analysis of our ENA.

Now we needed to adapt only the calibration power level and I do not understand the calculation results.

Changing the calibration power from 0 to -10dBm and to -20dBm I see a global minimum in the S21 uncertainty curve

which is correlated with the calibration power - this is fine. However, I thought that lowering the calibration power level

will introduce a corresponuding amount of noise (i.e. lowering the power from 0 to -10dBm increases the noise floor or

dynamic range by 10dB). However, the results show that there is almost no influence on the measurement uncertainty.

I need to lower the power levels to something like -60dBm to become a significant additional error related to the cal-power-level.

I assume that your calculations are correct, but could you explain why this has to be like this?

I have looked through all sorts of publication, including the online manuals of the instruments/web as it has been mentioned that the formulae

are "all there". However, the formulae include the effective system data, only and do not seem to include power levels. Is there any publication

that includes this influence as well?

Cheers,

mleibfritz

soure power = -60 dBm,

S21 = 1 (0dB)

Noise floor = -120 dBm (depends on instrument and freq range and IF BW).

so convert -60 dBm to mW, then convert -120 dBm to mW then add and subtract and convert back to dBm and look at the result.

Simple calcualtions: if noise floor is -18 dB it will cause 1dB error, if noise floor is -38 dB it will cause 0.1 dB error, if noise floor is -58 dB it will cause 0.01 dB error, so in my example, the noise floor contribution is less than 0.01 dB. pretty small on the uncertainty curve.