Clock Radio
Active Audio Filtering
I love the engrish you find on some electronics sites.
- High gain—when there are many the transistor so many gains.
- Easy—just add two resistors you so a good audio amplifier.
- Inexpensive—Now IC is low price than a transistor in the same working.
The electronics advice is usually decent. But boy is it hard to understand.
I'm learning about active audio filtering. Basically, I need to make a mono mix for the single speaker in the clock radio. The I2S chip is stereo, so I need to mix that down.
The simplest circuit for that is just putting a couple of resistors connecting the inputs.
(L)---[ R1 ]--\
(out)
(R)---[ R2 ]--/
If R1 and R2 are equal, you get a 50/50 resistor divider. Any difference between the L and R channels is averaged into the output.
But there are a couple of problems here. This design depends on a couple of things being true:
- your inputs aren't being used to drive anything else
- the output is going into a high impedance amplifier.
The first one isn't true by design (The I2S output also goes to a line out jack), and the second one is only kind of true.
I connected this circuit up and while it did technically work, it attenuated the signal a lot. And I think that's because the I2S output doesn't have a lot of drive strength. I think maybe the passive filtering in the OK BOOMER did not work well with the divider.
But the major implication for point 1 is the passive mixer will pass some current between the L and R channels. This is what they call crosstalk. So how do you fix that? You add some buffers. An op amp has high input impedance and the ability to drive the output from its own power supply. It can turn a weak signal into a strong one.
But most importantly, high input impedance means the input signal doesn't get disturbed. It can go on its merry way and we can do whatever we want with it on the other side of the buffer.
Buffers are also used in digital electronics to strengthen digital signals. You can think of them like amplifiers with a gain of 1.
So I buffered each of the L and R inputs and fed that into the resistor divider. And that worked a little better but it still wasn't ideal. Still attenuating and I could definitely see some distortion on the scope. The passive filter network was probably still interfering.
So I put another buffer after the resistor divider to isolate that from the next stage. And that worked great. :)
The least distorted sound I've found is to keep the buffering after mixing, and feed it directly into the input cap on the amp. And use a big enough input cap that you don't attenuate the bass (somewhere between 200nF and 1uF).
Not totally sure why I can't feed directly into the amp but whatever! This is easy enough to implement.