AO-40 K-band Waveguide Filters for Dummies
(by a Dummy)
Created 1 April 2002
Updated 3 April 2002
Whilst not the be-all and end-all of waveguide filters, here the objective is to provide sufficient image rejection of the unwanted IF image of a 24GHz receive system. Without it, you'll suffer an extra 3dB of noise figure. The DB6NT solution I use has a problem in that at 24GHz, a 144MHz IF is rather small, so a sharp filter is what's needed.
In lower frequency operations, reducing unwanted IF images by 50, 60 or more dB is essential as the bands are so busy. At 24GHz operating AO-40 we're pointing a very narrow antenna out into space, so it's more important to reduce the noise as the chances of a real signal being on the unwanted image frequency is inconceivably small.
After some discussion with Charles G3WDG, considering the DB6NT configuration I use with a 2dB NF preamp, a 10dB notch in the unwanted image adds about 0.5dB to the system noise figure. I achieved about 13dB rejection.
So far I've built four filters. Three filters are to the specification described in this page and here, and one to the specification in the RSGB Microwave Handbook Vol. 3. I couldn't get the RSGB Microwave Handbook version to tune, and one of my other ones behaved like an attenuator too.
A potential improvement would be to use copper rather than brass screws.
How long did this take? Three weeks to find and order the waveguide & flanges, three hours to build, two hours to setup the tests & tweak.
Thanks to Simon GM4PLM for pointing me to http://www.eisch-electronic.com where I purchased the waveguide and flanges.
And special thanks to Charles G3WDG for putting up with my questions.
See Original Filter info from this link
First scribe the cuts & holes. Note that for the slots, scribes either side of the hacksaw blade's cut are made to aid cutting perpendicular to the waveguide sides.
A light punch for drilling - be careful not to bend the soft copper waveguide. The hammer I used is way over the top.
Bench Drill & Cutting fluid
Cut iris slots with a mitre guide
Iris's made from 0.45mm copper sheet (rwwww.com part #680-965) wrapped in copper foil (the outer of some Westflex-103 coax) to fill out so that the iris's are a tight fit. Although folded over, the copper foil will have a tiny amount of solder paste sneaked under the folded flap.
Yuck - solder paste. Looks a lot worse than it was, honestly. The trick here is to avoid solder getting into the waveguide, so use the minimum amount. This picture looks like I used loads, but in reality there was only a tiny amount. I used stainless steel screws to steady the brass nuts when soldering.
Like a bench drill, no kitchen (ahem, workshop) should be without this single hob.
Finished filter - proof that beauty really is in the eye of the beholder. Possible improvement on rejection may be found by using copper rather than brass screws.
Calibrating & testing the filter
The first job: peak the filter
Second job: confirm image rejection
Second test - filter response - unwanted 23760MHz image is 13dB down from the 24048MHz image
Mail Howard, G6LVB