Over the years I’d seen a number of projects that were using one or more pipe-cap filters. They looked as if they should be fairly simple to make so I thought I should try one to help reduce the unwanted local oscillator (LO) spur at 2256MHz (2400MHz minus 144MHz). A quick search of the Internet resulted in a number of good write-ups covering both the technical side as well as the construction of these filters.
For 2400MHz I needed a 28mm end-feed pipe-cap plus a couple lengths of RG402 semi-rigid coax, two SMA connectors, a M4 brass screw, two M4 nuts and a small piece of good double-sided PCB. The following photo shows the parts (albeit with only one nut!):
Construction of these filters is well documented on many sites so I will not repeat this information other than add one small tip: I found that that by drilling and tapping a M4 hole in the top of the pipe-cap I could use a steel screw to hold the brass nut in place during soldering. The steel screw did its job and could be simply unscrewed after soldering.
I cut the probes to a length of 8mm and spaced them 16mm apart. The PTFE dielectric of the RG402 semi-rigid cable was not removed. The photo below shows how the probes are fitted. This was taken prior to them being soldered and trimmed to length.
Once the filter construction was complete the next step was to to tune and establish the filter characteristics. I would normally do this using either a VNA or Spectrum Analyzer + Tracking Generator. Unfortunately neither of my units work as standard above 1.5GHz so I needed an alternative method. I used my ADF5351 Synthesizer and AD8317 Power Meter plus an Excel spreadsheet. This is the test setup:
Using these two devices it was very easy to tune the filter for minimum insertion loss (IL) at 2400MHz. Then with a simple press of the 10MHz step button (many times!) it was easy to step through from 2200MHz to 2600MHz and record the power meter reading in a spreadsheet and plot filter response curve. When I constructed the power meter I also included a socket that allows me to connect an analog volt meter to show the AD8317 Vout voltage. It’s far easier to tune for a minimum or maximum reading using an analog meter rather than the standard LCD display.
What was also a very interesting exercise was to compare the filter characteristics when using a steel screw in place of the brass tuning screw.
Using the brass tuning screw is clearly the better choice giving much steeper skirts and 25dB rejection at the LO frequency (2256MHz). Also the IL was approximately 1dB compared to 2.4dB when using the steel screw.
For more technical information on pipe-cap filters see: Pipe-Cap Filters Revisited by Paul Wade, W1GHZ