Band Pass Filters
If we refer back to part 1 of this project and look at the block diagram that outlines our project, the first block to evaluate is the band pass filter. Its purpose as the name suggests is to pass a band of frequencies while rejecting all others. In this case, i want the filter to pass the 40m band and everything else. A good question to ask at this point are what frequencies we want to pass, and what frequencies require special attention in rejection. Being that the overall design of this receiver is as a CW receiver to be mated with an existing CW transmitter, passing a 100khz range from 7.000Mhz through to 7.100Mhz would cover the part of the band in VK where most of the CW activity is. So, we now have our first design criterion, 100khz band width, 7.050 center frequency.
What about rejection, well the greater the attenuation the better of course, but there are 2 frequencies that special attention need to be given to. The first one is the IF frequency, in our case the IF is 12Mhz, so we need good attenuation there and the other is the mirror image of the mixing products. Our desired frequency is 7.000Mhz and our LO is 5.000Mhz, so 7 + 5 = 12, but also you have the image of that as well, so 7 – 5 = 2Mhz. Another good question is how greater attenuation is needed at those 2 points? Well, to be honest I am not all that certain at this point, but for argument sake lets through a number out there and see how we go, -50Dbm. So if we have a 0Dbm 7.000 mhz signal in the filter, both 2 and 12Mhz will be -50Dbm down or greater.
Now where did I get the -50Dbm number from, well its the legal requirement of transmitters to have its harmonic content -50Dbm down on the fundamental so if it is good enough for that, it seems like a good enough number to start with to keep the crap out for the specifications of a band pass filter. I might be wrong, I really dont know. We will see as the project progresses. One last thing to consider is insertion loss. All filter designs will have some loss in their design, there is nothing you can do about that its the nature of LC filters. And remembering that insertion loss is attenuation of our desired signals, the less we lose at the beginning, the less we need to make up for later on in the gain stages. -3db is a halving of our desired signal, so lets shoot for less than 2Dbm. It sounds like an achievable value without being overly critical on component tolerances.
Desired Filter Parameters:
Center Frequency: 7.050Mhz
LO and Image Attenuation: -50Dbm or greater
Insertion Loss: less than 2dbm
First up is a design i have used a couple of times in different projects. The topology is one we see often in internet designs. I have never simulated it in LT Spice before now, so i never knew how good or bad it was. I have tweaked the values a little, but the filter does fit all the design parameters. Bandwidth is 100khz, IF is -49Dbm down and the Image is greater than 100Dbm down.
Next up is the PA3AKE filter for which the design can be easily found and plenty of people have copied and used. You can buy parts kits for them and board even. It has an extra inductor over the previous design, but its levels of attenuation is far superior. Bandwidth of the design is for the full 40m band, I can narrow up the bandpass easily enough b changing some of the cap values. Both the IF and the Image are both over -70Dbm down. So that is a substantial improvement.
I include this filter as a bit of a novelty, you see similar on many direct conversion designs using NE602 mixers, hence the 1500 ohm termination, I have used it a few times in different designs myself, so it is about time to see how it actually performs. Well it was safe to assume this would never meet the design criterion, but it is better than expected.
Mostly the filters you find in schematics on the net kind of fit in with the above 3 designs. The only difference is the starting value taken for the inductors. The BITx uses the same topology as the PA2AKE filter above but starts with 6uH inductors and standard value capacitors for almost the same level of attenuation, as does the Universal RX project by Dave G4AON, who started with 13.8uH inductors.
There is also software like ELSIE and AADE which will spit out values for all sorts of filter designs. I have been playing with both and to be honest i cannot get a better looking filter than the better two above. And in the end I will probably use the same values used in the BITx, we all have 100, 470 and 1000pf caps in our parts bins and 6uH inductors are not all that burdensome to wind and it meets all design specs by some.
ADD BITx 40 simulation here:
In the next part we will build and test the bandpass filter to see if the simulation and the actualization meet somewhere close to each other.
2 thoughts on “CW Superhet Receiver Project PART-2”
A couple of things come to mind. I’m pushed for time today so please excuse my brevity.
Image rejection – 50dB may not be enough. An image signal, say S9+20, will be clearly audible with only 50db of rejection (S9+20-50=s9-30=s5 if 1 s point is 6db)
Your first filter, the 2nd order capacitor coupled bandpass, will have a high insertion loss. I’m unsure where you allowed for the Q of the inductors. Consider the loaded Q of your entire filter is 7050/100=70. It is really hard to build inductors with a Q greater than 70 so you’re already struggling to meet a 3db insertion loss.
I use something similar to the PA3AKE filter. See here http://vk6tt.blogspot.com/2016/07/direct-conversion-receivers-rf-bandpass.html which explains how I ended up with this topology.
I have tried both AADE and Elsie, and for really exotic stripline filters for 23cm I even have a student edition of Ansoft Designer which I loathe because of the complexity. I nearly always reach for the Iowa Hills filter software nowadays. But each to their own. But why go to all the trouble of putting a filter in LTspice, which I also use, when all of the filter software will allow you to check insertion loss and attenuation curves with so little trouble?
I admire you for your ambition regarding LTspice. I rarely use it myself so it will be interesting to hear your thoughts on the software once you have used it for 6 months.
thanks for the comments Richard, I have downloaded the Iowa Hill programs, will try them tonight and see how i go. I never did like ELSIE all that much. And noted with inductor Q also, will take that into account also.