Active noise cancellation, Part 2: Implementation

Part 1 of this FAQ looked at the basic concept of active noise cancellation, how it might be implemented in principle using analog circuitry, and the harsh reality of actually providing the ANC function. Part 2 looks at modern ANC systems and how they function, as well as their practical issues.

Q: Is ANC using digital circuits and components simply an issue of digitizing the microphone signal, inverting it numerically, adding it to the original signal (also numerically), then converting back to analog to driver an earpiece (or, in the case of wide-area ANC, using the inverted signal to drive speakers?

A: Yes, but that’s only step 1. There are actually three techniques for ANC: 1) feed-forward cancellation, 2) feedback cancellation, and 3) hybrid cancellation. All require that the microphone signal be digitized to allow DSP-based algorithms to be executed on the microphone’s digitized signal, Figure 1.

If the noise signals that need to be canceled were simply played back after phase reversal, oscillation will occur at relatively high frequencies because of delays in phase reversal as the sound moves along the path from microphone to noise canceling circuit to driver unit to air and finally to audio output. To prevent this phenomenon known as “howling”, it is necessary to use a filter circuit to eliminate the high frequencies that trigger oscillation.

Q: What’s feed-forward cancellation?

A: In feed-forward ANC, a reference microphone is placed outside the headset to capture the noise signal, inverts it, and then adds it to the audio to provide cancellation; this is all done digitally, of course. Although this sounds simple, it is not, because the inversion must also account for the path traveled by the noise, then determine (calculate) the anti-noise signal before the correct noise goes from the microphone and reaches the eardrum. Do the math: with the speed of sound at 343 meter/sec (dry air, 20°C), that isn’t much time. While there would be more time if the microphone were placed further from the ear, the noise that the microphone picks up might then differ differently from the noise which is actually reaching the eardrum.

Q: What about feedback ANC?

A: In this approach, an “error” microphone is located inside of the headset or earbud, so it monitors the sound going to the user’s ear. The system then compares the sound in the user’s ear to the audio source, while a feedback algorithm identifies the noise and generates anti-noise to cancel it out. Despite the attractiveness of this approach, it is very hard to design an effective feedback system with filters that work across the many variations in ears and noise; also, the system can become unstable if the feedback is not “just right”— which will be the case under many circumstances.

Q: Then a hybrid system does what?

A: A hybrid feed-forward/feedback ANC solution combines the virtues of feedforward and feedback approaches, Figure 2. It has the external microphone of the feedforward system to monitor ambient noise, plus the internal microphone to monitor what the user hears beyond audio playback. With the right algorithms and filtering, this combination can be very effective.

Q: What does it take to provide an effective ANC system when using the hybrid approach?

First is the use of dynamically adjustable filtering, If the filer – even a digital one — has fixed coefficients, then it will be suboptimum in many cases, especially as the user moves or the type of noise varies (airplane, office, conversation, music, car: are all very different). Now, the latest generation of ANC systems use sophisticated, self-adjusting dynamically adaptive algorithms which change the filter parameters “on the fly” to provide the best performance.

Second, the most advanced ANC systems use multiple microphones. By using two, three, or even four, the ANC system cab better “hear” the noise and better adapt to canceling it, and can even deal with noise which changes source locations such as when someone nearby is walking and talking. Of course, a dynamic, smart ANC has a higher cost in components and software commitment but can significantly improve performance. Despite its additional cost, it is being used in conference room phones and even smartphones.

Q: If I need ANC, do I have to build my own system?

A: Absolutely not. In fact, IC vendors offer ultra-tiny ICs and chipsets which are specially designed for ANC headset and earbud applications, For example, Cirrus Logic, Sony, Analog Devices, Texas Instruments, Maxim, and ams offer different ANC ICs, including some which implement the hybrid approach.

Q: Who is using these latest ANC-based end products?

A: Travelers, pilots, industrial workers, and workers in noisy offices are among the users.

Q: What about ANC in open areas such as cars and aircraft cabins?

A: This is a difficult problem since it includes a physical environment with more unknowns and less control than provided by a closed environment of a headset or earbuds. Still, it is being tested in some cars using the car’s speakers, to generate a viable antinoise field to cut down inside power-train and outside road noise.

Note that a somewhat similar idea has already been used for canceling unwanted internal light reflections at lens/air, and lens/lens interfaces, using destructive interference — a sort of antinoise — created by precision thin-film coatings on the lens surfaces optical systems and cameras; this results in a “purplish” haze you can see if you look at the lens from just the right angle.

This FAQ has looked at implementation of a long-sought function of audio noise cancellation, first discussed almost a hundred years ago, long before the needed technology existed. Doing cancellation of sound was impossible until the availability of tiny microphones, digitized audio signals, and high-speed DSP algorithms, especially when coupled with adaptive filtering.

References

“Noise Control for Internal Combustion Engine Exhaust”
The David Clark Company, “How Does Hybrid Electronic Noise Cancellation Work?”
Sony Corp,  “The Science Of Sony’s Digital Noise Canceling”
Cirrus Logic, “Low Power Smart Codec with Active Noise Cancellation and Echo Cancellation”