In tx test mode, Linrad shows four different spectra at the same time. The spectrum in white is the conventional spectrum always present in Linrad. It is the average power spectrum at the bandwidth specified by the fft1 parameters as usual. The spectrum in red is the average spectrum in a flat bandwidth of 2.4 kHz. The red spectrum shows what one would experience when using a normal transceiver in SSB bandwidth and measuring RMS power. The green spectrum is the peak power spectrum in SSB bandwidth. It shows the maximum peak amplitude that has occurred since the latest reset. (Press Z for reset) The yellow spectrum is the averaged peak power. As a first step the peak powers are determined in the time span determined by the duration of a fft1 transform for each 2.4 kHz wide frequency segment. These peak powers are then integrated by means of a leaky integrator (equivalent to a RC filter with a time constant of 1 second).

The green peak spectrum will show splatter peaks and all kinds of emissions that may be taking place at irregular intervals or just once, when the PTT was pressed. It is like the peak hold function on any spectrum analyzer - but it is a fft analyzer so all frequencies are seen simultaneously.

Figure 1 shows a pure crystal oscillator. This is a demonstration of filter responses, spurs and dynamic range. The hardware is WSE RX144 -> RX70 -> RX10700 -> RX2500 -> Delta44. The red curve shows a dynamic range of about 111 dB. This corresponds to 145dBc/Hc when the 2.4 kHz bandwidth is taken into account. The signal is at about 144.012 MHz, the spur at 144.088 is the mirror image spur. This is a typical level for a calibrated system, just a little bit better than 70 dB. The spur at 144.048 is the zero frequency noise. In a direct conversion receiver, the center of the passband is at frequency = zero and there 1/f noise degrades performance. The peak to average ratio is a bit larger on this spur compared to the 10 dB that is typical for white noise. Note that the peak to average ratio is 0dB for a pure carrier.

Figure 2 shows an IC-706MKIIG emitting a constant carrier. The phase noise is at -118 dBc/Hz at a separation of 20 kHz. The spurs are small and there are no spurious emissions at the sides of the signal, the peak power is 10 dB or less above the average power. The yellow curve is about 2 dB below the green one. The separation between the green, yellow and white curves is the same at all frequencies and reflects the gaussian distribution of white noise.

Fig. 3 shows the spectrum of the same IC-706MKIIG when hand keyed. There are no visible keying clicks. This figure differs from figure 2 only in that the peak to average power is 6 dB reflecting the 25 % duty of the keying at the time when the image was captured. Note that this is valid only at close frequency separations. At large frequency separations, the noise is the same regardless of whether the key is up or down and therefore both figures 2 and 3 have the noise floor at the same level at large frequency separations.

Fig 4 shows the spectrum in SSB mode. The splatter goes into the noise floor at about 30 kHz frequency separation. With a "DX voice" into the microphone, the peak to average power is about 10dB. At a frequency separation of 10 kHz, the peak power of the splatter is about 15 dB above the peak power of the noise floor due to the sideband noise (see fig. 2) This is 52 dB below the peak power of the signal itself.

Figure 4 shows that this IC706MKIIG is well behaved on SSB on 144 MHz. The separation between the green and the yellow curves is larger than 2 dB at separations below about 40 kHz. This is how far the infrequent splatter peaks reach. The separation between the red and yellow curves is above normal only at frequency separations below 20 kHz. This is the frequency range in which the splatter is constantly present.