Agilent Technologies N9010A - Manual (Page 38)

38 Chapter 1
Agilent EXA Signal Analyzer
Amplitude Accuracy and Range
a. Absolute amplitude accuracy is the total of all amplitude measurement errors, and applies over the fol-
lowing subset of settings and conditions: 1 Hz
≤ RBW ≤ 1 MHz; Input signal −10 to −50 dBm (details
below); Input attenuation 10 dB; span < 5 MHz (nominal additional error for span
≥ 5 MHz is
0.02 dB); all settings auto-coupled except Swp Time Rules = Accuracy; combinations of low signal
level and wide RBW use VBW
≤ 30 kHz to reduce noise. When using FFT sweeps, the signal must be
at the center frequency.
This absolute amplitude accuracy specification includes the sum of the following individual specifica-
tions under the conditions listed above: Scale Fidelity, Reference Level Accuracy, Display Scale
Switching Uncertainty, Resolution Bandwidth Switching Uncertainty, 50 MHz Amplitude Reference
Accuracy, and the accuracy with which the instrument aligns its internal gains to the 50 MHz Ampli-
tude Reference.
The only difference between signals within the range ending at –50 dBm and those signals below that
level is the scale fidelity. Our specifications show the possibility of increased errors below –80 dBm at
the mixer, thus –70 dBm at the input. Therefore, one reasonably conservative approach to estimating
the Absolute Amplitude Uncertainty below –70 dBm at the mixer would be to add an additional
±0.10 dB (the difference between the above –80 dBm at the mixer scale fidelity at the lower level scale
fidelity) to the Absolute Amplitude Uncertainty.
b. Absolute Amplitude Accuracy for a wide range of signal and measurement settings, covers the 95th
percentile proportion with 95
% confidence. Here are the details of what is covered and how the compu-
tation is made:
The wide range of conditions of RBW, signal level, VBW, reference level and display scale are dis-
cussed in footnote
a . There are 44 quasi-random combinations used, tested at a 50 MHz signal fre-
quency. We compute the 95th percentile proportion with 95
% confidence for this set observed over a
statistically significant number of instruments. Also, the frequency response relative to the 50 MHz
response is characterized by varying the signal across a large number of quasi-random verification fre-
quencies that are chosen to not correspond with the frequency response adjustment frequencies. We
again compute the 95th percentile proportion with 95
% confidence for this set observed over a statisti-
cally significant number of instruments. We also compute the 95th percentile accuracy of tracing the
calibration of the 50 MHz absolute amplitude accuracy to a national standards organization. We also
compute the 95th percentile accuracy of tracing the calibration of the relative frequency response to a
national standards organization. We take the root-sum-square of these four independent Gaussian
parameters. To that rss we add the environmental effects of temperature variations across the 20 to
30
°C range. These computations and measurements are made with the mechanical attenuator only in
circuit, set to the reference state of 10 dB.
A similar process is used for computing the result when using the electronic attenuator under a wide
range of settings: all even settings from 4 through 24 dB inclusive, with the mechanical attenuator set
to 10 dB. Then the worst of the two computed 95th percentile results (they ere very close) is shown.
c. Same settings as footnote a , except that the signal level at the preamp input is
−40 to −80 dBm. Total
power at preamp (dBm) = total power at input (dBm) minus input attenuation (dB). This specification
applies for signal frequencies above 100 kHz.