SM 5 BSZ - Linux pc radio, the second time function (selective limiter).
(Dec 30 2000)

Why all this effort ?

Man made noise is often in the form of pulse trains. In case the pulse repetition frequency is high pulses will overlap due to the pulse stretching caused by the filters in use.

For efficient noise blanking SSB bandwidth is not enough, more bandwidth is required. In a conventional analog noise blanker more bandwidth is used, but one has to make a compromise because with too much bandwidth for the noise blanker strong signals from neighbouring channels will come into the blanker pass band. Such signals will not only make it difficult to detect the noise pulses, they also degrade blanker performance since it will be meaningful only to remove pulses that are stronger than the unwanted signals in the pass band. An attempt to remove pulses that destroy S/N for the desired signal will fail when the "keying clicks" from the signals in adjacent channels caused by gating them together with the pulse become stronger than the original pulse.

Check The MSDOS package for some additional information on blankers and selective limiters.

The selective limiter at 7MHz

While the effect of the selective limiter is pretty obvious on the VHF bands where only a few strong signals are occasionally within the passband it is less evident what performance to expect on the crowded 40m band.

Figure 1 is a screen dump showing the sum spectrum from two different antennas separated by about 1 wavelength, the complex time functions from the two antennas and the total power vs time. It is clear from figure 1 that pulses down to the size shown and well below that can be removed without problems.

Fig 1. The white track is the summed power vs time for both receiver channels. Yellow and magenta are the real and imaginary parts of the signal from one antenna, the red and blue are real and imaginary parts from the other antenna. The upper track of each colour is generated by the frequencies with white points in the power spectrum while the lower track of each colour is generated by the frequencies with red points. The lower tracks have reduced gain to fit in the 16 bit representation used for the second time function.
Figure 2 is produced the same way as figure 1 (a few minutes later) with the difference that the splitting into two parts is disabled. Obviously pulses of the size seen in figure 1 are no longer visible.

For typical powerline noise and many other sorts of pulse type interference where pulses come in groups so they can not be resolved at narrow bandwidth, the selective limiter and noise blanker approach is a very efficient method to improve receiver performance.

Fig 2. Same as figure 1 but all spectrum points are made white so all signals contribute to the upper track of each colour.