Ambient noise
Ambient noise usually takes the form of random noise with uniformly distributed power density spectrum, ie the power per Hz of bandwidth is approximately uniform over a wide frequency range.
Ambient noise is often expressed as an ambient noise figure Fam or ambient noise temperature Tam, see ITU-R P.372.
Captured power
There is a direct relationship between ambient noise level and power captured by an antenna system in a given bandwidth.
SDRs and SAs are often calibrated in absolute power units, commonly dBm. SAs often have useful filters to slow the response and help to find the average of a varying signal (usually termed Video Bandwidth VBW).
Any features in an SDR to reduce or cancel noise will probably compromise its use as a measuring instrument.
Bandwidth
SAs as measuring instruments usually have calibrated filter bandwidths (usually termed Resolution Bandwidth RBW). SDRs often have selectable filter bandwidths, but they are often nominal bandwidths rather than Equivalent Noise Bandwidth (ENB), some work may be required to find the actual ENB.
SAs often have useful filters to slow the response and help to find the average of a varying signal (usually termed Video Bandwidth VBW).
Total, internal and external noise power
Both SDRs and SAs have internal noise, and their display is usually calibrated to display the equivalent total power at the input terminals, ie internal and external power, so to arrive at the external power, the internal power must be deducted. That said, if the internal noise is relatively low, it has little influence on the result.
The graph above shows the error in reading the level of two combined noise sources as the level of the higher one. The error is very small when the measurement is more than 20dB above the noise floor, below that the result to be calculated for best accuracy. If for example the noise power displayed on the SDR or SA with 50Ω termination on its input is xdBm, and the antenna noise reads x+10dBm, the true antenna power is x+9.5dBm.
Example
Above is a sweep around 7MHz when connected to the station antenna. Noise floor with a 50Ω termination is -108dBm, so the trace is more than 20dB above the noise floor and we can assume that the external noise is approximately the value shown.
The challenge is to choose a combination of RBW and VBW to find the true noise floor between signals. In this case, let’s focus just below 7MHz where this is less activity, and we see at the marker frequency (6.9MHz) a broad flat area in the trace, power is -83dBm.
Let’s use a handy online calculator to do the calcs.
Above is the calculator input form.
Above is the result table, and it contains a lot of result values, most of which are not relevant to the question asked. The one figure of interest is the Field Strength Noise Figure (yellow highlight) of 47.2dB, this is the ambient noise figure Fa as given in ITU-R P.372.
So. what does ITU-R P.372 predict for this precinct, a rural residential neighborhood that usually measures a bit lower than ITU-R P.372 Rural (Curve C).
The predicted median noise Fam is 43.96dB, measured is a few dB above that. Expect quite a variation in measured noise, experience is here that it mostly measures lower than Rural.
Above is a survey from quite some years ago, but it shows the variation in measurements around the neighborhood. The measurement reported in this article is using a different type of antenna, different time, different place (but near to location 3) closer to a building (more LED lighting and other switched mode power supplies in use now), and is higher than the historical survey. This graph is scaled in dBµV/m QP in 9kHz bandwidth (a common practice), and location 3 median was around -1dBµV/m QP, the measurement reported in this article is about 7dBµV/m QP (pink highlight in the results table), so 8dB higher than the historical median on this single measurement.
Quasi Peak (QP)
QP is commonly used to express the magnitude of noise in the context of emissions and interference potential. QP derives from a very old CCIR recommendation describing the response of a noise measurement instrument. The recommendation included a specified meter response to short tone bursts. The QP detector has a slowed rise time, and slower decay time, so it under captures isolated very fast impulses, but accumulates high repetition impulses. The QP response reads higher than an average power detector, and lower than a peak power detector. My experience is that QP is about 4-5dB higher than average power on white noise, and typically around 6-7dB higher on off air noise with little impulse noise, but high impulse noise can result in a much higher QP reading. For more information, see (ITU-R. 1986).
The concept behind its original use for interfering emissions is that it better captured the impact of impulse noise on AM broadcast signals than an average power detector.
It turns out that common communications receiver’s S meter has a sort of QP response, a quite finite rise time and slow decay. Icom showed us with the IC-7000 what happens if the AGC has an attack time that is too fast, occasional impulses would hold the receiver gain suppressed. The saving grace in that transceiver was that the noise blanker was very good and compensated the problem of the short AGC attack time (but caused some distortion in the process).
References / links
- ITU-R. Aug 2022. Recommendation ITU-R P.372-16 (08/2022) Radio noise.
- ITU-R. 1986. Recommendation ITU-R BS.468-4 (1986) Measurement of audio-frequency noise voltage level in sound broadcasting.