Quote:
Originally Posted by bojan
After some time, I am back to the project :-)
Filter is tuned on the frequency, but I am still not happy with response - it is not flat (ripple is +/-1.5dB, and the attenuation is 3.5dB.. too much for placing this filter in front of the LNA.. so it will have to go after.
Getting there..
|
Hi Bojan,
Saw your note on the problems you are having. I've built a few of these now and have some observations. (Mine are 1275MHz centre frequency). If you use the Butterworth routines, then the bandwidth tends to be fine. The tapping points I have have found to be a bit too low. The rods seem to me to be slightly too short, as the tuning screws (I use 6mm screws for 6mm rods) are a long way into the case, when the design does not recognise the need for screws. The two characteristics that strongly drive passband attenuation are bandwidth and case depth.
By the time I add in 0.2dB for connector losses (not included in the design tool), then the passband attenuation is not far off.
I have discovered "Dishals tuning method", this is great for these filters, but you do need to be able to measure reflected phase. All you do is terminate the filter with a 50ohm load, wind in all the screws to short the resonators. Then using a VNA or similar, back off the first screw and watch for the phase flipping from -ve to +ve and lock off the screw at 0deg at your centre frequency. Move to the second screw and back that off until the phase is at 90degrees. This is an inverter, so the input impedance should be 0 + J50. If all is well you move onto the 3rd screw and back that off until it is at 180 degrees, then the input impedance will be 50 + J0 and the resonators are all bang on the desired resonant frequency. What is likely to happen, is that the input impedance will be off, but you will have a good guide on how much adjustment is necessary.
The issue I have, is on my current filter, I cannot get the last rod adjusted to resonance, once I've sorted what causes that, then I'll have good grip of these things.