TLV3231: I need a timing based audio rejector.

Hi Shawn,

Interesting application. I'm not entirely sure if a comparator is the right device. What signal do you expect the comparators to output? What does your system do with those signals? It would be helpful if you could provide some sort of high-level block diagram of the system you are trying to create.

Hi Shawn,

Like Ho mentioned, a comparator may not be the correct device. Comparators are typically used to compare a signal to a reference signal, but you want to compare the signals with each other and block out them out when they’re not close to each other. which leads to more of an AND logic gate device?

Maybe is more related to a differential amplifier and common mode rejection? In the case of common mode rejection any signal with the same phase is rejected because its assumed to be noise. This would be how XLR works to eliminate noise. Well, in this circuit I want to keep the noise so to speak and reject everything else. In the case of my circuit, the 3 mic signals being in phase indicates they're all the same distance from the sound source which is the only sound I want to accept through the circuit (theoretically). It doesn't have to be precise so to speak, just a nice roll off for anything that is out of phase. Now, if I could turn a dial and adjust the amount of suppression of out of phase signal, and adjust the precision of the phase alignment, that would be cool. Basically so that with the turn of a dial I could widen or narrow the area in front of the mics where sound would be picked up from. That would be like adjusting the phase acceptance from <= .01 milliseconds offset to <= 1 milliseconds of offset. It would also allow arranging the mics in a larger or smaller pattern or using different mics. 

Hi Shawn,

My thinking is that since you don't really want the differential but more-so the common mode signal, here's an idea that may work for you.

The idea would be to observe the differential between mics A, B, and C with audio op amps and use a comparator to compare them with some reference voltage to have some adjustable threshold. 

Once the differential voltage exceeds the threshold on any comparator, it sends an active low digital signal to shut down the analog signal paths using the MUX.  

The third comparison logically isn't necessary but probably important for matching, if not you could do 2 channels for amp and comp. 

Is this in the ballpark of feasible solution? 

These aren't precise parts for the application but an idea:

4-channel Audio Op Amp: OPA1604

4-channel Comparator: TLV1844(preview)

3-channel multiplexer: TMUX4053

Let me know your thoughts.

Best,

Michael

At first glance this looks like it could work, but it does not work. The issue is that I don't want to select from the inputs. What I want to do is perform an analog signal analysis, very similar to common mode rejection, but call it "uncommon" mode rejection. If that is a thing. The reason for the microphone array is to create a spatial arrangement in 3D space that measures time delays of same sounds as they arrive at different mics. Then, by looking for these delays using a chip, we can then roll off the out of phase sound (time delayed) and retain the in phase sound (time aligned) using an adjustable in time (maximum delay allowed) and amplitude (sound reduction in decibels), 2nd or 3rd order filter. Essentially an adjustable 3 channel common mode rejection filter in reverse in that it's uncommon mode instead of common mode. With CMRR basically the signals are intentionally sent out of phase by 180 degrees, thus,when received it is known that the true signal is 180 degrees out of phase to itself, using this, any other signal is rejected. In theory, it can reject itself if there is a problem, and that is kind of where this design comes into play. Its almost like a broken CMRR in that it should be adjustable to allow a range from 0 degrees to say 180 degrees, or more or less. If that makes sense. So, maybe its harder than CMRR because of the time scales but basically similar to that.

Hi Shawn,

Thank you I believe I understand your "uncommon" mode rejection. My idea here was to find the differences between mics, then at a chosen reference threshold the comparators would 'filter' whether one of the mics is out of phase or larger sound on one mic by shut the system down which gives it somewhat of an uncommon mode rejection, but this won't work if you want to pass the in signals that are in phase.

Let me talk this over with my team and see what we come up with, and I will get back to you on Monday.

Best,

Michael

Hi Shawn, 

This goes much further than the scope of comparators so I'm moving this ticket to the audio team for better assistance.

Best,

Michael

Hi Shawn,

I'm one of the audio applications engineers at TI. What you're describing is a beamforming algorithm. We've evaluated something very much like this before and our portfolio can't support it due to processing limitations. We have a solution in development but it will be at least a year before it sees the market.

Right now audio solutions do beamforming via DSP rather than a single chip like a codec. The logic of this algorithm makes it very difficult if not impossible to do purely in analog, partially due to the "un-common mode" rejection you were discussing earlier.

My recommendation is to do this in the digital domain as it will likely save you cost and time.

Best regards,
Jeff McPherson