I've been having a lot of fun with the $20 Bill rig, it works great, but, as with anything, there's always room for improvement.
When I designed the thing, I was going for simplicity, so the audio bandpass filtering was very rudimentary, basically consisting of the coupling capacitors to limit the lows and a couple of shunt capacitors to roll-off the highs. I like a wide receiver, but after a few days, even I thought it was a bit too wide.
My first few efforts used discreet components, but these either proved marginally effective or used more parts than the rest of the rig, so, alas, I broke down and resorted to the magic black boxes. But, rather than use a modern DSP subsystem, I went somewhat old-school and used an IC that's been around pretty much forever: The LM324. The '324 is a quad op-amp that's very common and dirt cheap; yeah, there are better parts, but the ol' '724 is good enough for this application.
There are excellent active filter tutorials elsewhere on the web, so I'll just describe the circuit at a very high level:
Each stage is comprised of a high-pass filter (3.3K in series with .01uF) on it's input, and a high-cut filter (68K paralleled by .01uF) in it's feedback loop. By cascading eight stages, I was able to get very decent stop-band attenuation; four stages probably would have been perfectly adequate, but more is always better, right?
Here's what LTSpice says the passband should look like:
When I designed the thing, I was going for simplicity, so the audio bandpass filtering was very rudimentary, basically consisting of the coupling capacitors to limit the lows and a couple of shunt capacitors to roll-off the highs. I like a wide receiver, but after a few days, even I thought it was a bit too wide.
My first few efforts used discreet components, but these either proved marginally effective or used more parts than the rest of the rig, so, alas, I broke down and resorted to the magic black boxes. But, rather than use a modern DSP subsystem, I went somewhat old-school and used an IC that's been around pretty much forever: The LM324. The '324 is a quad op-amp that's very common and dirt cheap; yeah, there are better parts, but the ol' '724 is good enough for this application.
Each stage is comprised of a high-pass filter (3.3K in series with .01uF) on it's input, and a high-cut filter (68K paralleled by .01uF) in it's feedback loop. By cascading eight stages, I was able to get very decent stop-band attenuation; four stages probably would have been perfectly adequate, but more is always better, right?
Here's what LTSpice says the passband should look like:
Built in the physical realm, it's a bit tighter than that, likely due to the loose tolerances of the parts I used, which is to be expected when the resistors and capacitors cost less than a penny each when bought in bulk. But, it's good enough!
With any filter, the real test is in listening to it: What's it sound like? Actually, very nice - it reminds me a lot of the 3KC filter in my SX-101A, very pleasant when listening to SSB signals and more than adequate for casual CW. My only gripe is that it's a bit noisy, but that may be due to where I have it tacked into the $20 Bill - I inserted it after the volume control, and a better place for it may be at the input to the power amplifier stage. Maybe I'll try that tomorrow, or maybe I'll do something else.
I'll close with this teaser: There's another hardware defined transceiver on my drawing board (screen?) and it'll have a very unique architecture - something I haven't seen anyone do before, and there may be a very good reason for that (like, it doesn't work!)
No comments:
Post a Comment