“Noise Figures of Radio Receivers”*
HT. Fans
Dwight 0. ori? Dr. Friis’ article is a valuable con-
tribution: ards the standardization of techniques in.
receiver measurement and classification. His definitions
of available power and of gain in available power will be
especially useful, and should eventually become stand-
ardized terms in the lexicon. One wishes that his treat-
ment of earlier definitions in this field had been accorded
the same care.
Following reference to a paper of mine which defined
and formulated certain concepts relative to the rating
of noise in receivers,? he continues in reference to his
own text. “In this paper a more rigorous definition of the
standard of absolute sensitivity, the so-called noise fig-
yn of
might happily be overlooked; yet I
ure, of a radio receiver is suggested.” The asse
greater
* Paoc. LR, vol. 32 pp, 419-422; July, 1944.
1 RCA Laboratories, Princeton, New Jersey.
3D.0, North, “The absolute sensitivity of radio receivers,” RCA
‘vol. 6, pp. 332-343; January, 1942,
~ antenna, thus: F,
wave passing
erat the yet os general usage, partly through want of in-
= CRAG
of vetbal entanglement, 4 for ‘noise factor” versus
“noise figure,” there is cial plea from this quar-
ter beyond the cogent observation that, while the
term “noise factor” has found common usage in this
country, in Britain its use appears to be exclusive.
The formulation of absolute sensitivity ran as fol-
lows, in practical units.
240m? ART OAS [r+2z- ]
» DN2,¢)
where E is field strength, 0 is wavelength, & is Boltz-
mann’s constant, To is room temperature in degrees
Kelvin, Ts is a fictitious temperature assigned to local
space in recognition of the power of local noise fields,
DX(Q, ) is the antenna’s space-directivity function, Fis
the noise factor under review, and Af is the over-all noise
bandwidth? The quantity in square brackets 1 termed
~ modified through the use efi pel ine diana deeey
=F+(T./Ts)—1. Neither the “oper-
"ating noise factor” nor the “absolute sensitivity” has
Fora definitive description of T. and D'(0, ¢) see footnote refer
cence 2.That is, the improvement in performance
represented by a reduction in noise factor of, say, several
decibels cannot be evaluated at all without a knowledge of
T.. An appreciation of this fact is extremely important.
So, also, are further studies to catalogue Ts at all wave-
lengths, at all seasons, and in all localities of technical
interest.
A recent communication from the Radiophysics
Laboratory, Sydney, Australia, emphasizes the need for
revision of the definition of noise factor in a particular
circumstance. Noise bandwidth Af was defined as the
over-all bandwidth in my work; Dr. Friis’ definition
agrees. However, a superheterodyne receiver may pos-
sess a considerable response, between antenna and con-
verter, at image frequencies. Other things being equal,
and according to present definitions (if not usage), its
noise factor might, in consequence of an augmented Af,
be quoted several decibels lower than the noise factor of
a similar superheterodyne without image response! Be-
fore we decide to abandon this problem, let us agree to
redefine Af as the noise bandwidth in the useful signal
channel only. In this event the receiver with image re-
sponse yields a somewhat greater noise factor than one
without image response—as it should, in view of the
extra thermal noise of the antenna in the image chan-
nel; and, provided the local noise fields possess a uniform
power distribution over the entire pass band, the revised
‘expression for operating noise factor becomes
Pop = F + W(To/To — 1)
where —_has(entire noise bandwidth) +Af.
The formula for absolute sensitivity is revised only
through Fey.
Te is of great importance to understand this revision
and the need for it, since a faithful adherence to the
earlier rules for a determination of noise factor, when
exercised in the measurement of a receiver with equal
response in signal and image channels, would lead to a
quotation some 3 decibels below its deserts.
Another limitation upon the concept of noise factor,
this time constitutional, restricts its employ to that
class of receivers which exhibits no measurably signifi-
cant source of noise beyond the detector. For, the
nal-to-noise ratio before a detector is transformed within
the detector in a complex fashion which may depend
upon the magnitude of both the signal and the noise.
Fortunately the limitation is, on the whole, an academic
one, since it is generally demonstrable that detection at
low level is bad practice from a noise standpoint, and
such designs can usually be avoided.
Of course, even with high-level detection, the signal-
tomnoise ratio ps after detection (by any reasonable
definition) can be equal to the signal-to-noise ratio p,
before detection only when the latter is large, and de-
creases more rapidly as p, approaches unity, finally as-
“Although unable to agree with their view that no need for re-
ion exists, 1 am indebted to Drs. J. L. Pawsey and Ruby Payne:
‘Scott of the Radiophysics Laboratory for pointing out this Gificulty,
suming the form, p:% pi, as p70. An understanding of
this behavior is often a prerequisite to a comparison of
alternative systems for a specific service,* and to studies
of service quality or coverage. Even s0, the subject
would seem to fall outside the province of absolute sen-
sitivity, whose scope is conceived to embrace p; alone.
The relation between p; and service quality can be
treated in no general terms,
For purposes of analysis, particularly in the event of
nonuniform distribution of noise over the pass band,
one often wishes to refer to a quantity associated with
each differential element of bandwidth df, whose
weighted average over the entire band is the noise fac-
tor. The weight factor is, of course, the gain in available
power, or its equivalent, as employed in the definition of
noise bandwidth. In accord with a proposal by E. J.
Schremp# we have found the term, “single-frequency
noise factor,” a definitive title for this quantity. Never-
theless, it lacks the brevity it merits; perhaps something
better will eventually be suggested.
In order fully to convey a useful magnitude in the
quotation of a noise factor, one must convey also the
value of room temperature T» upon which it is based.
‘The latter information is ordinarily suppressed. 1 in
cated a preference* for 300 degrees Kelvin; Dr. Friis
suggests 290. Of course neither of us is suggesting that
the tester-work in a thermostatted atmosphere of 81% 3004|
degrees Fahrenheit or a comparatively chilly 63 degrees} > 290K
nor are we even suggesting that the dummy antenna be
maintained during test at a standard temperature. The
requirement is merely that its temperature be ascer-
tained and the computations be corrected to a standard
temperature before a noise factor is quoted. It should be
recognized that no such, procedure need ordinarily be
followed unless one’s quotation carries a claim to accu-
racy of 0.1 decibel or better, except perhaps in unusual
circumstances which subject the dummy antenna to a
humanly uncomfortable climate.
XH. T. Friis? It is stimulating to read the discussion |
by Dr. North, who has contributed so much on receiver
rating by his paper.? |
‘The use of the phrase “more rigorous” in the first
sentence of the second paragraph of the paper under dis-
cussion was perhaps unfortunate since I did not wish to
imply that Dr. North’s definitions were at all inaccu-
rate, but merely that my suggested definitions were be-
lieved to be sufficiently precise to apply to four-terminal
networks in general, including low-gain amplifiers and
converters, and cases where a mismatch existed.
As far as the nomenclature is concerned, I am per-
fectly willing to leave the choice of terms up to the
engineers who use the definitions and to those who will
finally write the standards on the subject. The term
‘For example, M. G. Crosby, “The service range of frequency
mnedtiong REA Ren sl" py MOAT Yaneaye S10
"Radiation Laboratory, Maseachusetts Institute of Technology,
oe, Mase
" Bell Telephone Laboratories, Inc., Red Bank, New Jersey.‘noise figure," as is known, is widely used in this coun.
tryand has appeared in at least one British publication.
T used the words “absolute sensitivity” in their ab-
ct sense. As defined by Dr. North or by.
I most certainly agree that the noise figure or factor is
not a measure of the absolute sensitivity of a complete
radio receiving installation. The importance of the local
noise field in this case has long been recognized,” but so
far we have not found it necessary or convenient to set
aside any particular phrase to designate this absolute
sensitivity. In radio-circuit design we use the noise-
figure definition given by formula (5) in my paper for
the four-terminal network made up of both the trans-
mitting antenna and the receiving antenna, and we as-
sign, as Dr. North has suggested, for the temperature of
"'D.K. C, MacDonald, “A note on two definitions of noise gure
in radio receivers," Phil. Mog, vol. 35, pp. 386-395; June, 1944.
# See papers by R. K. Potter, A. Reber, and K. G. Jansky in the
PRoceEDINGS OF TRETLICE. on atmospheric and other radio noise,
the receiving-antenna impedance, a value that takes
care of the external noise sources. This way of rating a
complete radio circuit is briefly outlined in my paper at
the end of the section entitled “Measurement of the
Noise Figure.”
‘As Dr. North points out, it makes little difference
whether the noise figure be defined for a temperature of
290 degrees Kelvin or 300 degrees Kelvin. I chose the
value 290 degrees merely because it makes the value of
KT a little easier to handle in computations.
1 am glad to see that Dr. North calls attention to
the effect of image response on the noise figure, For
reasons of clarity I made my paper as brief as possible
and a great many details have therefore still to be set-
tled, He has also pointed out that the concept of noise
figure cannot be applied indiscriminately to the final or
audio detector of a receiver. Fortunately, as he has
stated, the limitations are such that they will have little
effect upon the usefulness of the noise figure as a stand-
ard.