Mechanical Filter Testing

Testing a mechanical, crystal, or ceramic filter requires only a stable signal generator for the correct frequency range for the filter being tested, a sensitive AF or RF voltmeter for that freq range, and a matching network to match the test equipment to the filter being tested.

Test Block Diagram

The signal generator should be adjustable in fine steps and have a high resolution dial or display. An added frequency counter may be necessary to properly measure the exact signal generator frequency. All the filter data sheets I've seen specify a max drive level of 2 V or more, so a 1 V RMS output level will be safe for most filters.

The impedance matching network will usually consist of only a resistor and a capacitor.

Most audio frequency voltmeters will have suitable sensitivity and frequency response. The HP 400C, for example, is flat up to 10 MHz.

Link to large image My filter test setup consists of an HP 3590A Wave Analyzer. This is more convenient than a standard AC voltmeter since it automatically switches ranges in 10 db steps when over or under range. For stability and small step sizes, I feed the external local oscillator input from a Videospectra 1010 Synthesized Function Generator. An external freq counter provides easy reading of the output freq rather than translate from the offset LO freq.

Preamp Schematic. To get better sensitivity to deep skirts and to easily match either high or low Z filters, I use the preamp shown in the schematic. R1 and R2 are changed as required to match the filter impedance, and the variable capacitors C1 and C2 are adjusted for best response. C1 and C2 can then be measured to determine the filter's preferred C loading values.

Link to large image The input and output sections of the preamp are mounted in separate shielded boxes with adjustment knobs for the variable capacitors. Clip leads are OK to see the nose area of the filter curve, but leakage from the high level signal at the filter input to the filter output will make the filter look worse than it really is. For accurate measurement of the 60 db and lower portion of the filter response, a shielded enclosure would be necessary. It's necessary to also ground the filter case or ground terminal.

Link to large image Filter curves can be plotted nicely by typing the numbers into Microsoft Excel and letting it draw the chart.


E-mail to John

Copyright © 2003 John Kolb Last revised Apr 30, 2003