Crystal Filters

The following screenshots are based on my 10 MHz quartz crystal parameters (I measured fs – resonance frequency series, fp – resonance frequency parallel, BW 3dB – bandwidth of fs at 3dB, cp – parallel capacitance).

Entering the above details in AADE software, it calculated the parameters marked with pink colour in the screenshot below.

I designed the filter using AADE software a Cebyshev filter with 4 crystals (bandpass 2.5kHz @ 3dB, 0.1dB ripple) with impedance matching network for 50 ohm and then I simulated the same circuit in Spice.

The Spice simulation showed simular results, which is expectable. There are some differences in the power loss and frequency. I am not really sure what are the reasons behind these differences.

One factor for the difference may be due to the fact that LTSpice does not show the loss between input and output, but the loss before the 50 ohm left resistor. If we measure the difference between the blue line (input) and green line (output), there are about -15.6 dB, not too far from AADE analysis, of -14.25dB

If anybody can explain in more detail what causes this power loss difference, please let me know.

Now, more interesting for me is to see how this shape is modified when there is no impedance matching at the input and output of the filter.

First simulation using LTSpice is to replace the 50 ohm impedance of the signal generator and the load and the matching networks simply with the design impedance of 174 ohm.

The simulation is very similar with the screenshot above (please note I changed the frequency limits for a better view of the attenuation range).

Now, let’s see when the same filter sees on each side an impedance of only 50 ohm.

We can see the filter shape is not so symmetrical at the top and there is also an extra loss of 5dB. I was expecting worse than that…

Finally, here we have the simulation of the BitX circuit with the filter properly matched, according to the first diagram from AADE.

We can see that for Transmitter Mode, there is a total gain of 7dB. Please note the capacitances going into and out of the filters are set to 100nF (0.1uF).

Also, the signal marked V(rxsignal) in the graph below is actually the ouput signal in Transmitter Mode!

In Receiver Mode, the overall gain is higher (14.2dB), because of having two transistors in the output circuit.

The signal marked V(txsignal) is in fact the output signal in Receiver Mode!


You can find the asc file (schematics for LTSpice) here

Any comments, suggestions, ideas, critiques about the above text
should be addressed on the BITX discussion group at yahoo or contact me.