Specifications

Chapter 1 1-9
Overview
Characteristics
a. The Special Option H26 preamp can reduce the total NF measurement uncertainty substan-
tially above 3 GHz because it will reduce the effective noise figure of the measurement system,
and thus it will reduce the sensitivity of the total NF uncertainty to the Instrument Gain
Uncertainty. But if the signal levels into the preamp are large enough, the preamp may experi-
ence some compression. The compression differences between the noise-source-on and
noise-source-off states causes an error that must be added to Instrument Noise Figure Uncer-
tainty for use in the Noise Figure Uncertainty Calculator. Such signal levels are quite likely for
the case where the DUT has some combination of high gain, high noise figure and wide band-
width. Here’s an example: The measurement will be made at 18 GHz. The typical preamp gain
is 25 dB and the noise figure is 7 dB. We will assume the DUT has 20 dB gain, a 10 dB NF, and
a passband from 5 to 30 GHz. We will use a noise source with 17 dB ENR. When the noise
source is on, the DUT output can be computed by starting with kTB (–174 dBm/Hz) and adding
10×log(30 GHz – 5 GHz) or 104 dB, giving –70 dBm for the thermal noise. Add to this the ENR
of the noise source (17 dB) combined with the NF of the DUT (10 dB) to give an equivalent input
ENR of 18 dB, thus –52 dBm input noise power. Add the gain of the DUT (20 dB) to find the
DUT output power to be –32 dBm. The noise figure of the H26 preamp may be neglected. The
H26 preamplifier gain of 25 dB adds, giving a preamplifier output power of –7 dBm. The typical
1 dB compression point of this amplifier at its output is +19 dBm.
Therefore, the output noise is 26 dB below the 1 dB compression point. This amplifier will have
negligible compression. As a rule of thumb, the compression of a noise signal is under 0.1 dB if
the average noise power is kept 7 dB below the 1 dB CW compression point. The compression in
decibels will usually double for every 3 dB increase in noise power. Use cases with higher gain
DUTs could be compressed, leading to additional errors.
b. The band 0 to band 1 crossing should be avoided. In addition to the wear-out mechanisms
(see
Caution on page 1-8) involved in measurements that overlap the 3 GHz band crossing,
there will also be performance degradations. There will be thermal instabilities in such mea-
surements that will add nominally 0.2 dB Instrument Uncertainty. The uncertainty of some NF
or Gain measurements are greatly multiplied from the Instrument Uncertainty. See the Uncer-
tainty Calculator included with the Noise Figure Measurement for details.
c. In this frequency range, the preselector is well-controlled and there should be no need for spe-
cial measurement techniques.
d. In this frequency range, the preselector usually requires no special measurement techniques
in a lab environment. But if the temperature changes by a few degrees, or the analyzer fre-
quency is swept or changed across many gigahertz, there is a small risk that the preselector will
not be centered well enough for good measurements.
e. In this frequency range, the preselector behavior is not warranted. There is a modest risk
that the preselector will not be centered well enough for good measurements. This risk may be
reduced but not eliminated by using the analyzer at room temperature, limiting the span swept
to a few gigahertz, and not changing the operating frequency range for many minutes.