The 8MHz 2nd IF Filter module requires no alignment. It contains four 8Mhz ceramic resonator filters, these came from a bag of miscellaneous components from my junk box. A BC109 transistor amplifier followed by AD603 variable gain amplifier compensate for losses in the filters and provide additional sensitivity to the spectrum analyser.
It's debateable how much of this circuit is really necessary. Some further gain and further IF filtering is certainly required as seen from the temporary analyser configuration shown in the 2nd mixer section. At first I used 2 filters and the BC109 transistor amplifier only. I built the AD603 experimentally and found that at maximum gain the sensitivity of the analyser does improve somewhat, but the noise floor is also substantially amplified.
Since the addition of the 2 further ceramic filters did seem to improve the shape of the frequency peaks (more sharply defined) and because extra sensitivity can be useful, I decided to leave the AD603 in the circuit. The gain of the AD603 can be adjusted from 10dB to 50dB in this circuit, via varying the voltage differential between pins 1 and 2 from zero to +1V. Supplying this gain control from a front panel potentiometer permits use of minimum IF gain most of the time and a nice "thin" noise floor, but if extra sensitivity is required to detect a very weak signal the IF gain can be cranked up.
I also tried cascading two of the BC109-transistor-plus-2-filters circuits but found the results unsatisfactory. The picture to the left shows the unused transistor module before removing the 2 filters for use in the AD603 circuit.
The circuit of the 2nd IF Filter at 8Mhz is shown below. Other IF frequencies would present no problem provided all the ceramic resonators are the same and the 2nd local oscillator frequency is adjusted accordingly. I believe ceramic resonators usually have input/output impedences of approximately 500 ohms so approximately correct termination is provided using 470 ohm resistors. If desired, the two rightermost ceramic filters and AD603 can be left out of the circuit. What we have here are the wild wanderings of the mind of a madman with too many identical ceramic filters to hand and most probably too little knowledge of how to use them properly. But the result works.
Again, note the extensive use of inductor and capacitor filtering of all DC supply and control signals entering the shielded enclosure, vergeing perhaps on the paranoia. A firm believer in the theory that you can never have too much decoupling and shielding.
Shown here is an analyser trace from 0 - 70MHz of the signal generator set to 30MHz and using input attenuation of 26dB. On the left, the IF gain is set to minimum (10dB), and on the right it is maximum (50dB). The improvement in sensitivity is only about 7 or 8 dB. Though the noise floor appears as a continuous band in the photograph (right) it is in fact a continuously fluctuating strip of tiny peaks constantly appearing and dissappearing.
With the 2nd filter in line between the 2nd mixer and logarithmic amplifier modules, the spectrum analyser will now show good selectivity (sharp peaks) and greater sensitivity than the first module hook-up made on completion of the 2nd mixer.
Again, note the extensive use of inductor and capacitor filtering of all DC supply and control signals entering the shielded enclosure, vergeing perhaps on the paranoia. A firm believer in the theory that you can never have too much decoupling and shielding.
