Pipe Cap Filters for 2.695GHz

Back

This page describes my experiments with pipe cap filters for 2.695GHz.

These filters are renowned for their simplicity and can be readily constructed and aligned in an evening.

Filters using 22mm and 28mm have been built with the greatest success using the 28mm size caps.

How do they work?

Broadly, and missing out lots of the maths which I don't really understand, the cap acts like a self tuned L/C cavity that is fine tuned using the central screw. Coupling is by leaving a short probe into the cavity from the coax or from the pcb. The bandwidth (dictated by the Q) and the insertion loss are both dependant on the length of the probes. The shorter the probes, the narrower the bandwidth (usually a good thing), but the insertion loss increases (definitely a bad thing). So there is a compromise to be reached. The bandwidth of the 2.695GHz RA allocation is only 10MHz (from 2.69GHz to 2.70GHz) so a filter that demonstrates low insertion loss but a high attenuation within a few MHz to suppress out of band signals is highly desireable. An additional requirement of the filter is to provide good image rejection. The downconverter will have an IF of 151MHz, so assuming that the wanted frequency (2.695GHz) is higher than the local oscillator, then good rejection at 2.393GHz is essential.

(So far) I have built 7 pipe cap filters. Three using a 22mm cap and four with a 28mm cap. The 28mm cap filters have so far been the most replicable. I will detail the 28mm cap filters here.

Version 1

This was built onto a 'U' shaped saddle of 24swg brass sheet. The probes are constructed from SMA connectors and are 7.5mm long. Probes at 10mm have also been tried. The filter with 10mm length probes had very low insertion loss (1.1dB) but the bandwidth was very high with noticeable ripple between 2.6 and 2.8GHz.

With 7.5mm PTFE shrouded probes, the insertion loss was approx 1.2dB. The diagram shows the layout.

and here is a photo of the finished item

Version 2

A second version was made directly onto PTFE based pcb material of 0.8mm thick and Er 2.6 The probes were manufactured using standard veropins which co-incidentally were also 7.5mm long. The track width for this material is 2.3mm wide for a 50 ohm stripline.

The remaining measurements were the same as the original. However, it should be noted that there is no PTFE sleeve over the veropins, therefore the electrical length is shorter. Insertion loss was higher at 3.4dB, but the response of the filter was much sharper. See the graph below

Version 3

This is the same filter as Version 1, but with the probes reduced to 6mm long.

Version 4

I bought a second sample of a 28mm pipecap from a trade supplier. This had an internal height approximately 1.5mm shorter than the original (from the Plumb Center) although it was considerably cheaper. (1.25UKP v 2.75UKP). The filter shows very similar charactaristics to the original pcb filter, although the insertion loss is marginally higher.

Conclusion

The 2nd filter is far more useful and will be incorporated into my 2.695GHz down converter.

References

ARRL UHF/Microwave Projects Manual - "Cheap Microwave Filters from Copper Plumbing Caps" - Kent Britain WA5VJB

ARRL UHF/Microwave Projects Manual Vol 2 - "More Pipe Cap Mania" - John Sortor KB3XG

Last Updated 2005-10-09