# Accuracy of frequency measurement of spectrum analyzer N9020a

Question asked by guoguo on May 20, 2013
Latest reply on Jul 9, 2013 by tabbott
I am very sorry for disturbing you. I am doing research length measurement based on frequency measurement by using Agilent spectrum analyzer N9020. The accuracy of length measurement depends on frequency measurement. Rigorous error analysis and accuracy assessment of frequency measurement of spectrum analyzer is very very important for our research and paper.
The oscillation of frequency is about 20GHz, and the oscillation frequency may vary 1MHz with the variation of length. I am very concerned with the accuracy and the relative accuracy (ppm) of measurement of the frequency variation. My questions are listed below:
*1. The repeatability of frequency measurement*
If the oscillation frequency f0 is stable and the frequency is measured for several times, do the readout frequencies have the same value, or do they distributed in the range of f0±Δf ~readout~, where Δf ~readout~ is the readout accuracy, and can be stated :
*2. The accuracy of relative frequency measurement*
As is mentioned in AN-150, the accuracy of relative frequency measurement Δf ~relatice~, defined as the accuracy of how far apart spectral components are, depends on the span accuracy, which means the uncertainty in the indicated separation of any two components on the display. For N9020,  Δf ~relative~ =0.35%×Span.
(1) Difference between measuring respectively and Marker Delta function
If f0 and f1 exist on the same display, (f1-f0) can be get by measuring f0 and f1 respectively, or by the Marker Delta function. Then, is the accuracy equal?
(2) Relative frequency measurement on the same signal component
The oscillation frequency of signal varies from f0 to f1 during measurement. Could the function of Marker Delta be used to measure (f1-f0)? What is the relative accuracy of (f1-f0)?
(3) Accuracy of relative frequency measurement between two components not on the same display
Δf ~relatice~ shows the uncertainty in the indicated separation on the display. But if the components f0 and f1 are not on the same display, and I tune the central frequency by Δf ~central~ in order to make f1 on the display, what’s the relative accuracy between them? Should I increase Span in order to improve relative accuracy?
*3. Accuracy of relative frequency measurement by the counter*
Marker counter accuracy Δf ~counter~ can be stated as:
Δf ~counter~ =±[(Marker frq×freq ref error)+counter resolution]
Delta counter accuracyΔf ~Deltac~ can be stated as:
Δf ~Deltac~=±[(delta frequency×freq ref error)+counter resolution×1.5]
(1) Minimum signal-to-noise ratio for counter
As is mentioned in AN-150, a minimum signal-to-noise ratio is required to eliminate noise as a factor in the count. However, I have not found the value of minimum signal-to-noise ratio.
(2) Counter mode with delta marker
When use counter mode with delta marker, should the two frequency components be on the same display. If tune the central frequency, what influence will be had on the relative accuracy(f1-f0) ?
(3) Relative frequency measurement on the same signal component
The oscillation frequency of signal varies from f0 to f1 during measurement. Could the function of Marker Delta be used to measure (f1-f0) with frequency counter mode.
*4. Any other reference material?*
Is there any Application Note that discusses accuracy of frequency measurement in detail? Just like AN1316 (Optimizing spectrum analyzer amplitude accuracy) for amplitude accuracy.