HP app an57 3 figure measurement 2 5980 1916e schematic

10 Hints for
Making Successful
Noise Figure Measurements

Application Note 57-3

2

Table of Contents

Introduction 3

● HINT 1: Select the appropriate noise source 4

● HINT 2: Minimize extraneous signals 5

● HINT 3: Minimize mismatch uncertainties 6

● HINT 4: Use averaging to minimize display jitter 7

● HINT 5: Avoid non-linearities 8

● HINT 6: Account for mixer characteristics 9

● HINT 7: Use proper measurement correction 11

● HINT 8: Choose the optimal measurement

bandwidth 12

● HINT 9: Account for path losses 13

● HINT 10: Account for the temperature of

the measurement components 14
Appendix A: Checklist 15
Appendix B: Total uncertainty calculations 16
Appendix C: References 18
Appendix D: Abbreviations 18
Appendix E: Glossary and definitions 19
Key web resources 20

3

Introduction

To achieve accurate and repeatable results at RF or
microwave frequencies, measurement uncertainties and
barriers to measurement repeatability must be minimized.
The performance of a device can be obscured if errors
are allowed to accumulate. For the most accurate
measurement, it is important to understand the nature
of the error contributors and identify which of these can be
influenced or changed to improve the quality of the results.
This application note provides useful tips that
will assist in making accurate noise figure measurements.

The checklist in Appendix A is a helpful tool to verify
that all hints have been considered for a particular
measurement.

A detailed explanation of the uncertainties in noise
figure measurements is provided in Agilent Technologies’
Application Note 57-2, Noise Figure Measurement
Accuracy (see reference 5). For a general understanding
of noise figure and its variety of measurement techniques,
refer to Agilent Technologies Application Note 57-1,
Fundamentals of Noise Figure Measurement (see
reference 2).

4

Frequency range

Commercial noise sources cover frequencies up to
50 GHz with choices of co-axial or waveguide connectors.
The frequency range of the noise source must include the
input frequency range of the DUT, of course. If the DUT
is a mixer or frequency translation device, the output
frequency range of the DUT must also be addressed.
If one source does not include both frequency ranges,
a second source will be required. A second noise source
may also be necessary when measuring a non-frequency
translating device with low noise and high gain. Low
ENR is best for the measurement, however, high ENR
is necessary to calibrate the full dynamic range of the
instrument. In either case, a full-featured noise figure
analyzer can account for the different ENR tables
required for calibration and measurement.

Match

If possible, use a noise source with the lowest
change in output impedance between its ON and
OFF states. The noise source’s output impedance

changes between its ON and OFF states, which varies
the match between the noise source and the DUT. This
variation changes the gain and noise figure of the DUT,
especially for active devices like GaAs FET amplifiers.
To minimize this effect, 6 dB ENR noise sources are
commercially available that limit their changes in
reflection coefficient between ON and OFF states to
better than 0.01 at frequencies to 18 GHz.

Adapters

Use a noise source with the correct connector for the
DUT rather than use an adapter, particularly for devices

with gain. The ENR values for a noise source apply only at
its connector. An adapter adds losses to these ENR values.
The uncertainty of these losses increases the overall
uncertainty of the measurement. If an adapter must be
used, account for the adapter losses.

● HINT 1:

Select the appropriate noise source

ENR

The output of a noise source is defined in terms of its
frequency range and excess noise ratio (ENR). Nominal
ENR values of 15 dB and 6 dB are commonly available.
ENR values are calibrated at specific spot frequencies.
The uncertainties of these calibrations vary over the
frequency range of the noise source and add to the
uncertainty of the measurement. This uncertainty is
typically limited to approximately 0.1 dB using the
root-sum-of-squares method (RSS).

Use a 15 dB ENR noise source for:

• general-purpose applications to measure noise
figure up to 30 dB.

• user-calibrating the fullest dynamic range of an
instrument (before measuring high-gain devices)

Use a 6 dB ENR noise source when:

• measuring a device with gain that is especially
sensitive to changes in the source impedance

• the device under test (DUT) has a very low
noise figure

• the device’s noise figure does not exceed 15 dB

A low ENR noise source will minimize error due to
noise detector non-linearity. This error will be smaller
if the measurement is made over a smaller, and therefore
more linear, range of the instrument’s detector. A 6 dB
noise source uses a smaller detector range than a
15 dB source.

A low ENR noise source will require the instrument to
use the least internal attenuation to cover the dynamic
range of the measurement, unless the gain of the DUT
is very high. Using less attenuators will lower the noise
figure of the measurement instrument, which will lower
the uncertainty of the measurement.

If ENR values must be entered into the instrument
manually, double-check them to ensure that the table in
the instrument is correct.

5

Follow these guidelines:

(a) Ensure that mating connectors are clean and
not worn or damaged (see Reference 8 for further

information). If the measurement becomes unstable
when the cables and connectors are shaken lightly by
hand, try other cables or connectors.

(b) Use threaded connectors in the signal path
whenever possible. (For example BNC connectors are

very susceptible to stray signals.)

(c) Use double shielded cables (common flexible
braided coaxial cable is too porous to RF).

(d) Use shielded GPIB cables.

(e) Move the measurement setup to a screened room.

If a transmitter that has any frequency content within the
measurement bandwidth is nearby and any covers are off
of the DUT, move the measurement setup to a screened
room. Test for such signals with a spectrum analyzer with
a simple wire antenna on the input. Attenuate these stray
signals by 70 to 80 dB.

(f) Use shielding. This is especially important for
making measurements on an open PC breadboard. (See
Reference 5 for further information.)

(g) Use an analyzer with minimal electromagnetic
emissions. Devices may be susceptible to stray emissions

from some measurement instruments. Some modern
noise figure analyzers have electromagnetic emission
characteristics low enough to have negligible impact on
the measurement.

● HINT 2:

Minimize extraneous signals

A noise figure analyzer measures the noise power from
the noise source as affected by the DUT. It uses the power
ratio at two detected noise levels to measure the noise
figure of whatever is between the noise source and the
instrument’s detector. Any interference, airborne or
otherwise, is measured as noise power from the DUT
and can cause an error of any magnitude.

Figure 2-1 demonstrates the types of stray signals that
can get coupled into the signal path and affect the
measurement. Fluorescent lights, adjacent instruments,
computers, local TV and Radio stations, pocket pagers,
mobile phones and base stations are notorious for their
adverse effects on noise measurements. Random stray
signals can cause several tenths of a dB difference
between individual readings, and result in unstable
measurements (i.e. jitter that will not average to a
stable mean).

Figure 2-1

Interference

Noise

Source

Lights ( esp. fluorescent)

FM
TV
Computers

Double-shielded cables for IF (ordinary braid is

too porous)

Shielded GPIB Cables
Enclose all circuits

NF

Analyzer

RF comms basestation

Test connectors by shaking leads

DUT

6

Alternately, insert a well matched attenuator
(pad) between the noise source and the DUT to

attenuate multiple reflections. As an example, with a
10 dB attenuator, the re-reflections are attenuated by
20 dB. The advantage of an attenuator vs. an isolator is
broadband response. The disadvantage is that the noise
source’s ENR values will be reduced by the attenuator’s
insertion loss (10 dB in this example).

If the DUT has a high output reflection coefficient
(S22) and low or no insertion gain (S21), then, in some
cases, place a low noise pre-amp between

the DUT and the measurement instrument to
reduce the total measurement uncertainty. The

pre-amp should have an input reflection coefficient (S11)
as low as possible. The effective bandwidth of a pre-amp
with very low S11 is usually narrow. More than one may
be required for the entire frequency range of interest. To
identify an appropriate pre-amp, see Agilent Technologies
Application Note 57-2 (Reference 5).

● HINT 3:

Minimize mismatch uncertainties

Mismatch at connection planes will create multiple
reflections of the noise signal in the measurement and
calibration paths (as shown in Figure 3-1). Mismatch
uncertainties at these planes will combine vectorially
and will contribute to the total measurement uncertainty.

One method to reduce the mismatch uncertainty is to

place an isolator in the RF path between the noise
source and the DUT. This isolator can prevent multiple

re-reflections from reaching the DUT and can suppress
the build-up of error vectors. Isolators, however, operate
over restricted frequency ranges. Several may be needed
for the frequency range of interest. Isolators also add to
path losses. As a result compensation is required. Full­featured noise figure analyzers have a loss compensation
feature to account for the insertion losses of any isolators.

Figure 3-1

Measuring

System

Noise

Source

DUT

Calibration

Measurement

Mismatch

Uncertainty

ρ1ρ

2

ρ

4

ρ

3

ρ

= reflection coefficient at a reference plane