Info

Fig 8—Relative correlation of the most likely matches of different 5-digit numbers to the ZRO level-A data. The predicted start time is 227.0 s; correlations were computed every 0.04 s from 226.5 to 227.5 s.

(a) S ON spectrum

Frequency (Hz)

Frequency (Hz)

(b) Average ON spectrum

Frequency (Hz)

Frequency (Hz)

(c) Single OFF spectrum

Frequency (Hz)

Frequency (Hz)

(d) Average OFF spectrum

(d) Average OFF spectrum

Frequency (Hz)

Fig 9—A comparison of the receiver noise-power spectrum looking at OSCAR 13 with similar spectra looking at blank sky. All spectra have a frequency resolution of

8 Hz. Spectrum (a) is derived from 0.125 seconds of data and is dominated by random noise, (b) shows the result of average 852 such spectra, allowing for changes in Doppler frequency; the random noise has been reduced dramatically, (c) shows a single spectrum looking away from the satellite, with (d) showing the average of 868 such spectra. Any weak emission from OSCAR 13 would show up as small differences between the average ON (b) and OFF (d) spectra.

the duration of a CW dot-space combination. The 5-digit number showing the highest correlation amplitude was submitted to WA5ZIB as the ZRO level-A report. There was a small celebration at AA7FV when Andy MacAllister confirmed that this was the correct number!

Other Applications

Moonbounce

One obvious application of these techniques is to low-power moon-bounce (EME) communication. For this to be successful, conventions need to be agreed on the detailed format of transmissions: the precise speed, timing, and the synchronization sequence (eg, CQ CQ CQ ..., or repeated call signs). It is essential to use machine-sent CW or other modulation. Although the processing described here has been for CW, similar techniques can be applied to most forms of modulation. An additional 3-dB gain in S/N with CW could be realized by using frequency-shift keying (FSK) instead of on-off keying. With signals buried so far in the noise, active error-correction modes, such as AMTOR, bring little gain in sensitivity on their own, but could be combined with these DSP algorithms. The gain in sensitivity achievable with the processing presented here depends on the exact circumstances (CW speed, etc), but is at least 10 dB when compared to the human ear alone.

OSCAR 13 Leakage Radiation

DSP algorithms were used in August 1993 to search for possible leakage radiation from the 436-MHz exciter of OSCAR 13. The Mode-JL downlink transmitter of AO-13 failed in May 1993.7 8 The cause of failure is not yet known, but if any weak signal from the Mode-JL exciter could be detected, for example via leakage though the defective power amplifier, this might help the understanding of the problem. During August 1993, the S-band beacon and the L-band beacon (ie, the L-band exciter) were both switched on simultaneously for part of AO-13's orbit (see Note 7). An unsuccessful attempt was made to detect low-level leakage from the Mode-JL exciter during these periods.

Data were recorded with the antenna pointed towards AO-13 and the Doppler-corrected 70-cm beacon frequency centered within the 2-kHz receiver passband. More than 850 power spectra, with 8-Hz frequency sampling and covering -100 seconds of raw data, were generated using the FFT algorithm. An example spectrum is shown in Fig 9(a); it is dominated by random noise. Each spectrum was individually shifted in frequency to allow for the Doppler drift of-0.3 Hz/second. These

Doppler-corrected spectra were then averaged together giving a combined spectrum which showed the receiver noise passband with any possible additional energy from AO-13 superimposed; an example of the average spec-

Fig 10—A spectrum of the radiation at 436 MHz from AO-13, when the Mode-JL exciter was turned on and should have been sending RTTY, Maximum leakage radiation from AO-13 is less than 11-nW EIRP in any 8-Hz band.

trum is shown in Fig 9(b). The overall receiver passband shape is shown very clearly. A similar spectrum, generated from data taken with the antenna looking at blank sky, was also generated. Figs 9(c) and 9(d) show a single-component OFF-satellite spectrum and the corresponding average spectrum. In both averaged spectra, the random noise has been reduced by the square root of the number (-850) of spectra averaged together. Any weak emission from OSCAR 13 would show up as slight differences in the ON-satellite and OFF-satellite spectra. The two averaged spectra were subtracted and the spectrum difference was normalized to give uniform sensitivity across the 2-kHz passband, calibrated as a fraction of receiver system noise. This analysis removes frequency-dependent offsets and gain variations in the spectrum, which result mainly from ripples and slopes in the receiver IF passband. The final result is shown in Fig 10. There is a small residual dc offset of about -4% of the average system noise, which is consistent with a gain change of -0.2 dB between the observations on AO-13 and on blank sky. The intention is to search for a narrowband emission from AO-13, so this offset is unimportant. The raw data had been recorded while AO-13 should have been sending

Was this article helpful?

0 0

Post a comment