There are curves for dynamic accuracy in the The Technical Specifications for the E836xB NAs, among other things. These curves can also be calculated using Agilent's na_calculator.

The Technical Specifications contains a link to a description of equations used to generate the uncertainty curves in the specs., however, there are no equations there for dynamic accuracy.

My question is this: can the mathematical form of the DA curve be described in terms of instrument specs., or parameters?

The Technical Specifications contains a link to a description of equations used to generate the uncertainty curves in the specs., however, there are no equations there for dynamic accuracy.

My question is this: can the mathematical form of the DA curve be described in terms of instrument specs., or parameters?

1) Compression of the receiver: The top end of the curve is set by the compression of the receiver, and this is a measured value. In the 8510, a model was used because they could not easily measure the compression; the model was based on a diode sampler, and the compression measured at low levels was extrapolated to high levels using the model.

2) non-linear effects in the ADC: these are hard to model and combine many physical effects including hysteresis in the filter inductors, dielectric memory in the fitler capacitors (this is the anti=aliasing filter in the PNA or DNA, the 8510 was a DC ADC), dithering effects, spurious signals, lots of stuff.

3) Non-linearity in the power sensor used to measure the PNA linearity: turns out this is the largest effect in the middle region, and is on the order of 0.004 dB/10 dB if I remember correctly, Drift of the power sensor is also an issues as the 10 dB range is "stitched" over several measurements and drift will affect it.

In the newest PNA-X, the 67 GHz model N5247A, we have created a new method for measuring the dynamic accuracy with greatly improved capability, on the order of 0.001 dB/50 dB change, and have dramatically improved the specs; at the low end it is more than 100 times better, and is I believe the worlds best receiver linearity, regardless of cost or method of measurement. I've attacted a portion of the data sheet. A