Using ceramic resonators in SSB filters

Ladder filters using low cost standard frequency quartz resonators are well known in amateur circles, but what about ceramic resonators? Actually, they can be quite useful in lowish (half MHz) SSB filters, despite their lower Q. The reason for this is that their lower inherent impedance allows top coupled parallel resonance operation to be quite practical, which makes playing with networks that much easier. Associated network capacitor values end up being very convenient, and 'suck it and see' filter development becomes quite fun. So here are two examples, each using three resonators, that operate at about 0.5 MHz. The shunt capacitors allow some movement of resonator frequency, and the series capacitors control the coupling between stages and therefor filter width. Oh, with this topology, it is easy to arrange for 50 ohm input and output matching impedances:

455 kHz 560 kHz

It's a while since I did these, so details, other than those shown above, have become a little hazy. The 455 kHz version was fitted into a small 80m monitoring receiver, and that certainly sounds OK.

The same idea was tried at 1 MHz, but I have lost the circuit values:

3 x 1MHz resonators 4 x 1MHz resonators 5 x 1MHz resonators

I must have tried higher frequency ceramic resonators, but cannot now find any graphs of those. They probably weren't sharp enough as sideband filters, but 3.58MHz or 3.68MHz units should make good image filters for 80m receiver front-ends or transmitters, though they will only cover a portion of the band. Likewise, 14.3MHz ones for 20m.

There definitely have to be other worthwhile applications for these for instance, SSB generation for an audio RF processor - or this 0 - 20 kHz carrier frequency USB generator, built for low frequency optical comms use.

speech processor


RF Speech processor

This is a little bit 'belt and braces', and the two SA602s and SA604 could probably be replaced with a single SA605, but it works really well, and keeps the to clip level with no overshoot at all, that I can see:

ssb speech processor

afprocessor
Although the output is flat with frequency, pre-emphasis is added to the mic audio before the first mixer, since this seems to enhance readability of my particular voice. Compression is about 10 dB with the values shown and the lapel mic that I use.


In the past, I have used back to back diodes to clip the IF signal, but using one half of the SA604 limiter sector worked first time and the clipping looks really predictable. There was only 2mV of signal coming out of the ceramic filter, so the higher gain (60dB) second limiter stage was used on this occasion. To be sure that harmonics due to limiting didn't also get demodulated in the second SA602 mixer, a LPF was added after the limiter, and I suspect that this is why the processor ouyput is so well defined. It really needs a video to show this at its best, but the following still will have to do:

processor output
 

single board
              processor
A single board version was then built (to replace the processor in the Weaver transceiver that has never sounded brilliant on transmit) and the same ICs used - the SA605s that were found and which would reduce the IC count were found to be in a very compact package with horribly close pin spacings).

The rather over-etched pcb opposite was just useable. This unit used 456 khz resonators, with coupling and shunt capacitances very similar to the first version. Having ordered 200 455 khz resonators from China, it will be interesting to apply a bit more logic in their use, since currently, I do not select resonators but use them randomly. With luck, the really cheap Chinese resonators (1.32 per 50) will have a much wider tolerance actual frequency, so that selection of frequency rather than the pulling down of frequency with capacitance can be tried. You can certainly see the Q fall as parallel capacitance is added - sometimes this requires that the higher frequency resonator (with no pulling capacitance) has to be damped down with shunt resistance to make it similar to the other two, in order to achieve a flat top.