Noise Figure inferences of spectrum analysis of weak signals

This article reports some observations using two common PC based audio spectrum analysis tools to analyse the strength of weak signals.

In all of these measurements, signal level was below the AGC threshold, so the receiver gain was fixed at maximum. If signals cause AGC gain compression, the analysis below does not necessarily apply.

Stage 1 - SSG

In Stage 1 the receiver is connected to a standard signal generator. The signal generator was adjusted for 10dB SINAD at the receiver output using a Motorola 1012 SINAD meter. The Effective Noise Bandwidth of the receiver has been measured at 1450Hz (Measuring receiver bandwidth). Table 1 shows the calculated Noise Figure.

Table 1: Sensitivity measurement with SSG and SINAD meter.
Quantity	Value	Comment
SSG power	-124.5dBm
SINAD	10dB
ENB	1450Hz
NF	8.3dB
S/N	9.5dB

The Spectrogram analysis indicates signal level of -30dB, noise in 2.7Hz bandwidth adjacent to the signal of -62dB, so a S/N of 32dB.

Table 2: Analysis of Spectrogram display.
Quantity	Value	Comment
SSG power	-124dBm
S/N	32dB
ENB	2.7Hz
NF	13.2dB

The Noise Figure for the narrowband measurement around the carrier is quite different to the Noise Figure using the SINAD meter. The reason for this is the influence of the receiver audio response on the total audio noise power. Although the input noise power is broadband, and essentially flat across the receiver channel, the effect of the filtering and audio shaping is to reduce the contribution of higher audio frequencies to the total noise output power, having the effect of reducing the total noise output power for an improvement in SINAD and NF, albeit somewhat deceptively.

Calculation of SINAD from an export of the spectrum log from Spectrogram accounts for the amplitude / frequency response of the noise output, and reconciles with the measured SINAD.

Another factor that may contribute is the FFT windowing used in the software, but since neither package describes its windowing, nothing more can be said.

The Spectrogram analysis indicates signal level of -22.7dB, noise in 2.7Hz bandwidth adjacent to the signal of -57.0dB, so a S/N of 34.3dB.

Table 3: Analysis of Spectran.
Quantity	Value	Comment
SSG power	-124dBm
S/N	34.3dB
ENB	2.7Hz
NF	10.9dB

Again, the indicated Noise Figure is quite an amount better than measured with the SINAD meter. See the comments re Spectrogram, the same issues apply.

Stage 2 - Antenna

In Stage 2, the receiver is connected to an antenna pointing at the beacon at Nimmitabel on 144.413MHz.

The Spectrogram analysis indicates signal level of -26dB, noise in 2.7Hz bandwidth adjacent to the signal of -53dB.

Table 4: Analysis of Spectrogram.
Quantity	Value	Comment
S/N	34.3dB
ENB	2.7Hz
NF	10.9dB

The Spectran analysis indicates signal level of -22.6dB, noise in 2.7Hz bandwidth adjacent to the signal of -56.9dB.

Table 1: Analysis of Spectran.
Quantity	Value	Comment
S/N	34.3dB
ENB	2.7Hz
NF	10.9dB

Determining ambient noise level

The ambient (or external) noise level can be determined by measuring the change in noise between a known noise source and the antenna. The noise level measured in the first stage is from a known source, it is the noise in a resistor which can be calculated.

The change in noise from stage 1 to stage two can be assessed from the spectrum displays, additionally, it can be measured with a true RMS voltmeter.

Table 1:
Quantity	Voltmeter	Spectrogram	Spectran
Tcold (K)	290	290	290
Noise on SSG (dB)	-16.2	-62	-57.0
Noise on Antenna (dB)	-7.9	-53	-48.6
Y (dB)	8.3	9.0	8.4
Ta ( K)	14,510	17,430	14,900
Fa (dB)	17.0	17.8	17.1

Links

Changes

Version	Date	Description
1.01	15/10/2006	Initial.
1.02
1.03
1.04
1.05