Last week's update included how I'd become dissatisfied with the bandpass and Low Noise Amplifier (LNA) which is embedded within my design. I had other problems with the board so it was time for a reset on the filter design circuit. What I've come up with is applicable to my aviation radio project. It is also applicable (in general principles) to ham radio receivers in the 2M band.
All of this design work is for SCARP, which is my code name for my project: the Software Controlled Aviation Receiver Arduino Project, AKA SCARP! (or should it be SCARAP?)
As for engineering, I've decided to open up the design for this small portion of the project. I had been using a Butterworth filter, but I've switched to an Elliptic filter. If those don't mean much to you, this may not be your post. Oh well.
Here's what I had been using, it is the Butterworth bandpass filter:
I had tested it, and I thought it was working OK. Somewhere in my recent PCBa iterations, I discovered it was not working well at all. Back to the drawing board. After some research, and using a web-based filter design resource, I came up with the following improved Elliptic filter for my redesign:
This filter had the benefit of being more easily prototyped, because the capacitive values are less affected by PCB and prototyping stray capacitance. I ordered all the parts from Mouser (which has recently provided me with better service than Digi-Key) and earlier today I hand-assembled the RF bandpass filter prototype.
The ideal response, shown above, compares well with the real world response on my spectrum analyzer:
In this photo, I'm sweeping a test frequency from 50 to 250 mhz. You can see that everything below 108 mhz is dropping rapidly, and also everything above 138 mhz drops off. (This is looking at the blue line.) I'm running a signal with a strength of 0 dbm out of the analyzer, then attentuating (reducing) that strength by 20 db. The filter is next in the path, stopping everything outside of 108 -- 138 mhz, and then the Low Noise Amplifier is next, increasing the signal strength back by 20 dm. The 'net-net' in band is 0 db, and out of band, the background is down by 40 decibels.
The LNA I used is the Qorvo QPL9547TR7. It works great. It has the additional characteristic of rolling off response the lower the frequency below 100mhz. As a result, it also stops some FM and a lot of AM interference. This is demonstrated by looking at the entire response spectrum from 0 to 1.2 ghz:
I think this is geeky cool. The bandpass is so freaking obvious. The rolloff down to AM frequencies is nice and sharp. I don't understand the peak around 500 mhz, but it doesn't make any difference for this design.
The acid test was hooking this up to my existing design and listen to it. The following video is 30 seconds of multi-channel listening using the new filter design and the new LSA. I am scheduled to have this dropped into the design two board spins from now.
Comments