So i have these LTW455IT ceramic bandpass filters in my parts box, 455kHz Bandwidth +/-2kHz at 6dB with 6db insertion loss. So i thought i would plot out its response to see how it looks, and to be honest it is very close to spec, but a little bit high in freq, centre should be 455 + or – 1khz, mine showed a centre of 457khz, 2khz higher, now i did not impedance match to 2K ohms, but i probably should have. (BLUE Graph Line)
So i thought to myself, if one is good, two must be better and then plotted out the response (RED GRAPH) The added capacitance pulled the freq down closer to 455khz. Each filter is 6th order so this essentially made a 12th order filter, I kind of expected the skirts to steepen, but they did not, I do not know why, but i would like to find out.
Measured with the signal generator on the input and scope on the output with the grounds connected together, isolated from the filter grounds. I did connect the filter grounds together but was getting strange readings, i think you have to use dc blocking caps on the output if you use the inbuilt ground pins.
Anyway, it was a little bit of fun, that did not amount to much other than one of these filters would make a nice narrow SSB filter for a tranceiver that has a 3khz or larger front end bandpass.
I did some tests on these filters quite some time ago. Also cascaded. The idea was to try to improve the bandwidth enough to maybe do SSB since mechanical filters are rather large and difficult to find. Works well with two filters (tighter skirts and narrower bandwidth) but results deteriorate pretty quickly with more.
The termination impedance is rather important and would account for a lot of the shift you are seeing. That’s partly why the cascaded arrangement resulted in a downward shift as the termination between the filters would have improved collared to a direct 50 ohms termination. These filters I think from memory require 2k ohms without any reactance. I did the measurements on a HP8753D with appropriate series resistors on each port to achieve the right termination. You can de-embed this by taking the S-parameters and working back through an RF simulator.
Hope this helps.
Hi Bryan, thanks for the information, its been a while since i have played with these. I did have a lot of fun though and learned a lot, I might even go back and revisit them again sometime soon, after i buy a VNA so i can Bode Plot filters easily. Oh a bit of a surprise to have you drop into my humble blog. I was looking at your website just the other day as someone I know was after information on fox hunt sniffers and your site came up in the google search. Keep up the good work.
Rob.