|
is disconnected and an internally generated calibration signal is applied to the Input Amplifier. The calibration signal is a highly accurate 15/85 duty cycle pulse train which provides a 10 kHz fundamental frequency compo- nent along with odd and even harmonic components spaced at 10 kHz intervals (Figure 3-13). The magnitude of the pulse is such that the fundamental frequency component produces full scale deflection when the instrument is properly calibrated. The amplitudes of the harmonic components are not meaningful. The calibration signal can be used for amplitude calibration or to verify the frequency accuracy of the instrument.
3-79. In the Amplitude Calibration Procedure (Paragraph 3-173), the front panel 10 kHz CAL potentiometer is adjusted so that the 10 kHz fundamental frequency compo- nent of the cal. signal produces full scale deflection. This calibrates all circuitry following the input attenuator to full scale accuracy of ± 1.5% at 10 kHz. 3-80. Bandwidth Setting. 3-81. Refer to Figure 3-14 for the following discussion. The 3581 uses a hetrodyne technique where the 0 Hz to 50kHz input signal is mixed with a 100kHz to 150kHz signal from a Voltage-Tuned Local Oscillator (VTO). To select a given frequency present at the input of the Mixer, the VTO frequency is tuned so that the difference between it and the frequency of interest is 100kHz. The 100kHz intermediate frequency (IF) is fed through the IF Filter, detected and applied to the meter. Signals outside the passband of the IF Filter are rejected. The BANDWIDTH setting determines the bandwidth of the IF Filter and thus, the selectivity of the instrument.
3-85. Equivalent Noise Bandwidth. When making noise measurements with the 3581, it is necessary to use the "equivalent noise bandwidth" rather than the 3 dB band- width indicated by the BANDWIDTH setting. In the 3581, the equivalent noise bandwidth is 12% wider than the absolute 3 dB bandwidth. Note that the absolute 3 dB bandwidth can be about 15% wider or narrower than the BANDWIDTH setting. For optimum accuracy, measure the absolute 3 dB bandwidth of your instrument and use that figure to calculate the equivalent noise bandwidth. 3-86. Bandwidth Selection. There are 4 things to consider when selecting a BANDWIDTH setting:
3-87. Resolution. Resolution is the ability of the analyzer to separate signals that are closely spaced in frequency. An important point here is that the response of the analyzer to a CW signal is an amplitude vs. frequency plot of the IF Filter (Figure 3-16). The width and shape of the filter skirts are, therefore, the major limitations of resolution. If two CW signals appear in the passband (± 3 dB points) simultan- eously, they cannot be separated (Figure 3-17). If two signals differing widely in amplitude are both inside the filter skirts, the response of the larger signal can hide or obscure that of the smaller signal (Figure 3-18). If the
-82. For operating purposes, the 3581 input channel can be pictured as a bandpass filter that can be manually tuned or swept over the 0 Hz to 50 kHz frequency range. The instrument responds only to signals passing through the filter and thereby sorts out the various frequency compo- nents present at the input. The BANDWIDTH setting determines the width of the filter skirts at the - 3 dB points above and below the tuned frequency:
3-83. IF Bandpass Characteristic. Many signal analyzers use active filters that have very steep skirts and a square-shaped bandpass characteristic that approaches the ideal "window filter". This type of filtering provides a high degree of selectivity, but because of its long transient response time, is not well suited for swept frequency applications. The 3581 IF Filter consists of 5 synchron- ously-tuned crystal filter stages. The bandpass characteristic of the synchronously-tuned filter (Figure 3-15) closely approximates a gaussian response. The gaussian filter provides good selectivity and, because of its relatively short transient response time, is considered optimum for sweep ing. 3-84. Shape Factor. The shape factor of the 3581 IF Filter is approximately 10:1 on the 3 Hz through 100 Hz band widths and 8:1 on the 300 Hz bandwidth. A shape factor of 10:1 means that the filter skirts are 10 times wider at the - 60 dB points than at the - 3 dB points. Similarly, a shape factor of 8:1 means that the skirts are 8 times wider at the - 60 dB points than at the - 3 dB points. On the 10 Hz bandwidth, for example, the - 3 dB points are 10 Hz apart and the -60dB points are 10 x 10 or 100 Hz apart. The filter is, in effect, centered on the tuned frequency, fo, and exhibits 3 dB of rejection to signals that are ± 5 Hz away from fo and 60 dB of rejection to signals that are ± 50 Hz away from fo. amplitude of the smaller signal is greater than that of the skirt produced by the larger signal, the peak of the smaller signal can be resolved (Figure 3-19). For optimum resolu tion, the bandwidth should be narrowed to the point where only one signal is inside the filter skirts at any given time. Generally, the width of the filter skirts at the - 80 dB point does not exceed 15 times the 3 dB bandwidth. Thus, optimum resolution can always be obtained when the frequency separation between signals is at least 15 times the BANDWIDTH setting.
|
