Screws Mess With Filter Design

I was given a task of designing a band pass filter to install in front of a receiver at a frequency that was classified since it was for the defense department. Now the frequency was too low to use cavity rods since at a .25 wave, the rods would be too long and could not meet the size requirements.

The only filter design that would fit the size requirements and the low insertion specifications would be a helical resonator design. This design has the highest Q inductors since lumped coils have Q's too low to meet the narrow band and low insertion loss specifications that were 1 dB in the pass band.

Using Zverev's Handbook of Filter Design charts, it was decided that a five-section (pole) filter was required to meet the in-band and out-of-band-response. The metal enclosure to obtain the highest Q for the resonators was calculated and would fit the required size format.

The coupling between the resonators was difficult to calculate since the coil diameter and the number of turns was critical in this determination. So I built a two-stage helical filter and had machined-out slots on the top and bottom to adjust the resonators. The top slot ran in the center of longest dimension to accommodate the piston capacitors while the bottom slot ran as above to accommodate the brass screw to secure the coil that was wound on a threaded Delran rod. Since the enclosure was aluminum, a brass sheet had a similar slot machined out and installed on the bottom of the resonators so that the bare wire coil at the bottom could be soldered to make the ground. This brass plate was secured to the aluminum using flathead 4/40 brass screws.

BNC connectors were installed at the lower end of the box ends so that a bare wire could be soldered to that part of the coil, providing the best return loss. This two-stage filter was tested using a network analyzer, and the frequency and coupling adjustments were made and recorded prior to building the five-section filter.

The five-section filter was completed, but long brass screws were inserted down from the top between the piston capacitors and the resonators to provide for fine-tuning of the coupling. The filter was aligned using a dual trace of return loss (S11, S22) and pass band (S12) measurement. It met all the specifications, so the unit was installed in the front end of the receiver.

We ran tests using a signal generator with the proper modulation and every specification was met. Next we tried it with the proper external receiving antenna, and there was some interference occurring during the test. Since this head end RF narrow band filter ejected out-of-band spurious signals we knew it could not be the typical nf1 plus or minus nf2 order of signals causing the problem.

I remembered that when checking the stability of a broadband amplifier, you should sweep wide to see any re-entry of response that would cause oscillation or k<1 stability factor. The next day I took the filter out of the receiver system and did a wide scan of the response (S21) and discovered a response at three times the frequency with only a 2dB insertion loss. Going back to my transmission line book I found out that .25 wave lines invert, but they repeat three times the frequency.

Knowing this, I designed a low pass filter to reject this response, and installed one at the input and the output between the inside BNC connector pin and the tap. This rejected the extra response. Incidentally, this type of filter has more rejection at the low end than at the high end since the equivalent capacitance coupling does not provide a transmission at high frequencies, so there are four transmission zeros less at high stop band frequencies.

This entry was submitted by William J. Garner and edited by Rob Spiegel.

William J. Garner is an RF microwave engineering consultant with 47 years design experience. He has published papers in trade journals and holds seven patents.

Tell us your experience in solving a knotty engineering problem. Send stories to Rob Spiegel for Sherlock Ohms.

Related posts: