"Thus the receiver is . . . a non-linear device at the threshold level."
This article originally appeared in The Milliwatt, Vol. 1 No. 4, August 1970. Ade Weiss, WØRSP, who edited The Milliwatt has kindly granted me permission to republish it here. It has been reformatted and lightly edited.
When QRPp [in 1970, "QRPp" referred to 5-W or less input, as QRP-ARCI still defined "QRP" as 100-W or less. -- KR1S] signals are received, often the signal-to-noise ratio is low. Under these conditions, a receiver, which is linear according to conventional reasoning, can for all practical purposes become a non-linear device, and the noise can suppress the signal. A simplified explanation is as follows. We do not observe the signal by itself. We first observe the noise, then we observe the signal-plus-noise and look for a difference, and this difference is less than the output which the signal would produce in the abscence of noise.
For example, suppose that the noise produces a detector output of one volt, and further, suppose that the signal, in the absence of noise, would also produce an output of one volt. Then, because there is a random phase relationship between signal and noise, the signal-plus-noise output corresponds to the sum of their powers and not to the sum of their voltages. And therefore, the actual detector output due to signal-plus-noise is the square root of two volts (1.41) rather than two volts. The difference between the signal-plus-noise output and the noise output is then only 0.41 volt (which we recognize as readable signal), and the signal has been in effect 'suppressed' in the ratio of 1:0.41 or about 4 dB by the presence of the noise.
As the signal is increased, however, the effective signal output increases
a. by the direct increase in signal power and
b. by a reduction of the suppression ratio.
Thus the receiver is in effect a non-linear device at the threshold level. It can be demonstrated mathematically that near the signal-to-noise threshold an otherwise linear receiver becomes a square-law device.
This suppression is much more pronounced with wide-band FM receivers, a fact that is well known. However, it does take place to a lesser degree with other types of receivers. Its existence is not merely a matter of mathematical juggling, but can be demonstrated in the laboratory. For further information, the reader is referred to "Microwave Receivers," in vol. 23 of the M.I.T. Radiation Laboratory Series, ed. S.N. Van Voorhis (McGraw-Hill: New York, 1948), Sec. 7.9.
.... Because receivers operate in a non-linear manner near their thresholds, it is important to eliminate every bit of loss in the transmitting system.
Yardley Beers, WØJF