YU1LM HF/6M BPF designs - anyone built any ?

[VK6ZFG]

Has anyone built and tested the YU1LM bandpass filters as per http://yu1lm.qrpradio.com/HF-6m%20BP%20YU1LM%20ver3.pdf  ?

I haven't run the Ansoft Designer spice-like software on them but I have built a couple of them up (10 & 14MHz) using the values as per Table 1 to find that they don't provide the expected (/published) results. My measurement process is manual as I don't have a tracking generator but am using a microwattmeter (displays in dBm) plus a sign gen at +6dBm to validate the actual response values. 

I have had to tamper with the capacitance values a bit to get (1) the pass loss low at 10 and 14; and (2) provide the desired centre frequency characteristic.  Untouched, they are off-frequency and the shape of the response is far slower to fall off than in the 'BP for 10MHz' and 'BP for 14MHz' graphs in the article.

Mine are each built on a small PCB about 40mm long x 12mm wide using SMD components, the boards then plug onto a "test" baseboard so that their results could be checked.

Fortunately I have a degree of flexibility in the actual filter design as the PCB layout is two rows of a series of rectangular pads that will allow a lot of different configurations. I can experiment a bit to try to improve the outcome.

The destination of these BPFs is my Raspberry PI WSPR beacon project (Pi Zero plus small PA ~ 200mW + 4 x BPFs switched by SMD relays) and I am currently targetting good 10, 14, 18 and 50 MHz designs. I need to clean up the PI's output spectrum hence the need for the BPFs.

Hi Doug

I have not built any of those but many other filters.  

What I have found that you can take advantage of the series resonant frequency of capacitor.  This makes it possible to achieve a higher order filter without really using any more components.  This can be done by making the shunt capacitors series resonant on a frequency that you want.  This can be achieved by careful lead length selection or  coiling the leads to get more inductance or adding a bit more wire length.  A tracking generator makes this readily achievable.

Component self resonance do not show up in computer plots.

73s

Igor