Measuring receiver bandwidth
Overview
The bandwidth of a receiver determines the total power that reaches the detector from a wideband source of noise or interference. The response of receivers is not ideal, and knowledge of the Equivalent Noise Bandwidth is important to measurement of wideband noise and interference.
Mathematically, it can be written as follows.
Avo is the reference audio voltage or the audio voltage at a reference frequency. In some sense it might seem natural to choose Avo to be the maximum of Av(f), but the purposes of this article is to find an equivalent noise bandwidth that allows determination of receiver Noise Figure from a standard sensitivity measurement or specification that establishes the S/N, (S+N)/N or SINAD ratio at 1kHz, AVo is taken to mean the audio voltage at 1kHz.
This article describes how to measure the bandwidth of an SSB receiver using a PC based audio spectrum analyser.
Procedure
The IF filter is the dominant determinant of the end to end frequency response of an SSB receiver. Audio shaping commonly employed can modify the response, typically applying a slope across the filter passband.
- Step 1: standardise the receiver filter settings (eg IF shift "normal", notch filters disabled, audio filters disabled, noise reduction / noise blankers disabled).
- Step2: connect the receiver audio output to the PC sound card and set the level for substantial audio without risk of clipping. It is very important that the audio level is high, but that there is NO clipping. Clipping will create distortion that results in an increase in levels outside the filter passband.
- Step 3: with wideband noise input to the receiver, the audio output is connected to a PC sound card and the audio analysed using a PC spectrum analyser package, in this case Spectrogram. I have used 30 seconds averaging for a smoother line on which to make the measurements. I have exported the spectrum log and analysed it using the sl2enb.pl Perl script to calculate the equivalent noise bandwidth (1875Hz) wrt to the gain at 1kHz.
Figure 1 shows the display of the receiver audio response. Not the substantial difference between the level within the pass-band and above the pass-band. Lack of audio level will result in too small a difference, and too much audio (clipping) will also result in too small a difference. Adjust audio level for maximum difference between the in-band and out-of-band level.
 |
Figure 2 shows the receiver frequency response plotted on a linear power axis, and the response of an equivalent ideal filter with the same gain as the actual filter at 1kHz (where sensitivity measurements are usually made). The equivalent filter has the same total noise power admitted (the area under the red line) as the total noise power admitted (the area under the blue line) by the actual filter.
 |
Resources
Perl and Python3 utility to read Spectrogram spectrum log to calculate Equivalent Noise Bandwidth
Example spectrum log from Spectrogram 11 for IC7400 with 2.4kHz SSB filter setting
Excel spreadsheet to perform ENB calcs from data extracted from Spectrogram 11 log.
Links
Audio spectrum analysis
Windows is not supplied with a good, high productivity, flexible, rich interpretive scripting language. Perl fills that gap very well.
Changes
| Version | Date | Description |
| 1.01 | 15/01/2005 | Initial. |
| 1.02 | ||
| 1.03 | ||
| 1.04 | ||
| 1.05 |