I actually did a deep dive into this exact subject a few months back. The summary, according to patents, white papers, and various other research articles:
-It’s harder than it looks
The foundation of most automotive ANC systems is the adaptive filter with feed forward and feedback control. The principle being that you have a reference signal (just the noise) and a primary signal (what you measure in the car). The reference signal is contained within the primary signal. An adaptive filter generates the error between the two, in other words, it outputs the parts of the reference (noise) that are present in the primary (mic) input. That output is used to generate the cancellation signal.
While it seems simple in theory, the practical application is anything but. In this case the challenge is gathering quality reference signals. These are typically obtained via accelerometers on the body and engine RPM data. You need multiple reference inputs and multiple primary inputs. You can’t just mix them all together and be done either. Each input needs to be processed with every other input, and the output of those processes needs to be further filtered and correlated with each output.
What you end up with is a large array of sensors fed into a series of special filtering, filtering those output signals some more, and then sending the final output to the correct speakers.
Don’t expect headphone level isolation either. The coherence between noise source and interior noise rapidly falls apart above 90hz or so. It really works with steady state noises. And the reduction is no more 10dB at best. That’s not to say it isn’t any good, but it’s not the end all be all.
It’s most commonly used to alter the sound of engine drone and very low frequency rumbles caused by road impact. It is noise cancellation, but it’s better understood as a tool to shape the noise that gets through.
Also, the high output stereos common in today’s vehicles adds even more stress to the system (or really, its designer).
-It’s harder than it looks
- It only works below 100hz
- It’s used for a very specific purpose, not as a total replacement to sound absorption
- It requires special attention to tuning
The foundation of most automotive ANC systems is the adaptive filter with feed forward and feedback control. The principle being that you have a reference signal (just the noise) and a primary signal (what you measure in the car). The reference signal is contained within the primary signal. An adaptive filter generates the error between the two, in other words, it outputs the parts of the reference (noise) that are present in the primary (mic) input. That output is used to generate the cancellation signal.
While it seems simple in theory, the practical application is anything but. In this case the challenge is gathering quality reference signals. These are typically obtained via accelerometers on the body and engine RPM data. You need multiple reference inputs and multiple primary inputs. You can’t just mix them all together and be done either. Each input needs to be processed with every other input, and the output of those processes needs to be further filtered and correlated with each output.
What you end up with is a large array of sensors fed into a series of special filtering, filtering those output signals some more, and then sending the final output to the correct speakers.
Don’t expect headphone level isolation either. The coherence between noise source and interior noise rapidly falls apart above 90hz or so. It really works with steady state noises. And the reduction is no more 10dB at best. That’s not to say it isn’t any good, but it’s not the end all be all.
It’s most commonly used to alter the sound of engine drone and very low frequency rumbles caused by road impact. It is noise cancellation, but it’s better understood as a tool to shape the noise that gets through.
Also, the high output stereos common in today’s vehicles adds even more stress to the system (or really, its designer).