CRYSTAL LADDER FILTER SIMULATION WITH SPICE
For the purposes of this exercise, I’ll assume that the crystal parameters have been measured, and are as follows:
Frequency 4 MHz
Lm (motional inductance) 0.1971 H
Cm (motional capacitance) 8.03554e-15 F
Rm (motional resistance) 48 ohms
Co (static capacitance) 3.5 pF
The filter configuration is as follows:
Calculated capacitor values for this filter are:
C3 = 94 pF
C2 = C4 = 60 pF
C1 = C5 = 14 pF
The terminating resistance is 1200 ohms.
The Pulsonix package I use for PCB design came with Pulsonix SPICE, which I’ll be using here. I created the following schematic for the SPICE analysis:
Each crystal is represented by its equivalent circuit, with its parameters. The five shunt capacitors have been added, and the input resistance (1200R) is in series with the signal source. A 1200R load is placed across the output. The 1G resistors across the shunt capacitors are required by SPICE; it doesn’t like ‘floating’ nodes, and such a high resistance won’t affect the simulation.
Running SPICE gave the following plot:
It appears to be quite a useful filter, with a reasonable shape factor, and a 3 dB bandwidth of about 500 Hz.
Lets see what happens if we change the terminating resistances to 330R:
As you can see, a real mess has been made of the response. The terminating impedance of this type of filter is quite critical.
I've extracted the netlist for this filter from Pulsonix:
R7 $4 $0 1200With a few minor changes you should be able to use it with any flavour of SPICE.
Here are some free SPICE packages:
SPICE OPUS http://fides.fe.uni-lj.si/spice/SIMetrix is the SPICE used with the Pulsonix package.