At the IARU Region 1 Conference in Hungary 1978 the need for a harmonised standard for the "S-meter scale" was expressed and a proposal was accepted for publication in society journals. The essential recommendation was 1 S-point is 6 dB . At the Brighton Conference in 1981 the recommendation was formally adopted as a standard for amateur radio equipment manufacturers. This recommendation is titled
"IARU Region 1 Technical Recommendation R.1". Sadly, I can no longer find the original document on the web. If anyone has that document I'd love to have a copy.
The current
IARU Region 1 VHF Handbook states in Section 4.1.1:
IARU Region 1 Technical Recommendation S-METER READINGS
BRIGHTON 1981, TORREMOLINOS 1990
Standardisation of S-meter readings
• One S-unit corresponds to a signal level difference of 6 dB,
• On the bands below 30 MHz a meter deviation of S-9 corresponds to an available power of -73
dBm from a continuous wave signal generator connected to the receiver input terminals,
• On the bands above 30 MHz this available power shall be -93 dBm,
• The metering system shall be based on quasi-peak detection with an attack time of 10 msec +/- 2
msec and a decay time constant of at least 500 msec. The current
IARU Region 1 HF Manager's Handbook states in Section 4.1.3:
S-METER STANDARDS
In order to make a uniform reporting system on the amateur bands possible, taking into account the widespread
use of the 'subjective' S-system, taking into account the large deviations between the characteristics of S-meters on
current amateur equipment, the IARU Region 1 recommends the use of the "S"-system for signal strength
reporting on the amateur bands, based on the following standards:
(a) One S-point corresponds to a level difference of 6dB.
(b) On the bands below 30 MHz a meter deviation of S-9 correspond to an available power of a CW signal
generator connected to the receiver input terminals, of -73dBm.
(c) On the bands above 30 MHz this power shall be -93dBm.Note that although these handbooks are labeled "Region 1" they are the only handbooks and all Region 2 and 3 references point back to them, i.e. there are not separate handbooks for each IARU region.
In no case do they specify the measurement bandwidth, probably because the standard is intended for the measurement of signal power and not noise power.
The good news is that when making signal level reports and the signal is at least 6dB (1 S-unit) above the noise floor the contribution of the noise power will make little difference to the S-meter reading for any typical amateur radio measurement bandwidth, i.e. a bandwidth below 3KHz. For instance, assuming a 2.7KHz measurement bandwidth and a corresponding S-meter noise power measurement of -120dBm, and a signal power measurement of -114dBm, the total power (signal + noise) is -113dBm, only a 1dB difference.
Thus measurement bandwidth is not significant when performing a level cal. on our radios. Indeed, most of us will be using signal generators with signal levels substantially above the noise floor. I usually try to calibrate with an input signal level close to -73dBm (S9). In such a case the difference between signal alone and signal+noise is much less than a thousandth of a dB.
In contrast to signal power measurements, where it is only important that the measurement bandwidth be equal to or larger than the bandwidth of the entire signal, when making noise power measurements the measurement bandwidth is critically important
if the results are to be compared to other, similar measurements. Thus in industry it is quite common for noise power measurements to be referenced to 1Hz so that the total noise power in any bandwidth can be calculated and receiver performance more easily compared. For example, given the case of a -120dBm S-meter reading with a 2.7KHz measurement bandwidth, this equates to a noise level of -154dBm/Hz (-120 - 10log(2700)).