Active noise cancelling (ANC) headsets are an attractive proposition to consumers, since they offer a superior listening experience in conditions that are normally hostile to audio reproduction, such as trains, airplanes and busy urban areas. In fact, while the idea of silencing ambient noise is a simple one, its practical implementation is complex.

Nowadays the trend for music headsets goes to big over-the-ear headsets, whereas five years ago in-ear systems were the most popular. Consumer preferences have obviously changed in the last couple of years and besides the original use case (listening to music), headsets have turned into a fashion statement.

Developing an ANC headset requires considerable know-how, especially if you want to combine a piece of art with modern ANC technology. No matter in what kind of headset ANC is implemented (in-ear, on-ear or over-the-ear headsets) there are basically three different concepts to tackle the ambient noise.

ANC ConceptsThe most common is the feed-forward topology (see Figure 1), in which a micro-phone exposed to the exterior senses ambient noise, and the ANC circuit generates an anti-noise signal that the speakers reproduce (together with the user’s audio playback signal). This headset type consists of four blocks: the speakers, battery, ANC circuit, and ANC microphones for the left and right channels.

Figure 1: ANC Feed Forward Block Diagram

The feed-forward topology is typically used in communication headsets like Blue-tooth headsets because of its wide ANC bandwidth. Such systems can cancel noise up to 3kHz with properly designed acoustics.

The higher frequencies in particular help to improve speech intelligibility if you are making phone calls since this is the typical frequency range of human voice. Another important advantage of a feed-forward ANC system is that there is absolutely no influence on the audio signal path.

Figure 2: Feed Forward Block Diagram

Figure 2 shows the signal flow of a standard feed-forward ANC headset. The “Dff” block represents the feed-forward delay caused by the speaker due to the conversion from electrical impulses to air pressure waves. The noise signal that is picked up by the microphone is treated with a gain and phase compensation filter G(w) and mixed together with the sound signal. The music playback path is completely independent from the noise reduction path.

Another interesting mode that is commonly used in ANC headsets is the monitor mode. In this mode the ANC microphone is in turn being used to actively amplify the ambient noise. In this special mode the gain and phase compensation filter G(w) is bypassed and the microphone is connected directly to the speaker amplifier. This helps to overcome the passive attenuation of a headset when having a conversation with your neighbor or flight attended in a plane without removing the headset.

Typically this mode can be activated by pressing a push button on the headset. A disadvantage that comes with feed-forward systems is that they are susceptible to wind noise if the electronics and the acoustics are not properly designed.

Feedback Concept A solution to overcome the wind noise issue is a different ANC topology. The second topology in the ANC industry is the feedback topology. This topology makes use of the same hardware blocks like we have it in feed-forward applications. The only difference is the location of the microphones, which are inside the ear capsule. This makes the headset insusceptible to wind noise.

Another advantage that comes with feedback systems is the automatic compensation of little leakages. With feed-forward systems it is important to have a good sealing of the headset to the ear, otherwise you get only very limited ANC performance. This is a typical problem and very often the reason why feed-forward systems (especially in-ear systems) do not deliver the same ANC performance across various test persons.

Although this behavior can be compensated with a proper acoustic design, feedback systems usually do not show this behavior. Up to a certain degree they can compensate sealing tolerances of the headset.

A major difference between the two ANC topologies is the noise reduction characteristics. Feedback systems usually have better low frequency performance (<100Hz) but do not reach the bandwidth of feed-forward systems. Typically feed-back systems can work up to 1kHz and have a more flat ANC distribution with lower peak values. In turn, feed-forward systems show superior peak performances (typically up to 25dB) with a cone-shaped characteristic.

Figure 3 shows the typical block diagram of a feedback headset. The music signal is directly fed to the speaker. The ANC microphone and the speaker are located in-side the ear capsule and build a closed environment, which leads to a general drawback of the feedback topology.

Since the microphone is located inside the ear capsule it picks up the noise signal and the music signal. Thus, the ANC microphone cannot distinguish between music- and noise-signal. This signal is then fed back and treated with a gain and phase compensation filter G(w). Because both signals, noise and music, are used in the feedback path, the headset tries to cancel music and noise. This phenomenon is well-known as low-frequency loss in ANC feedback systems.

Since the feedback ANC system shows typically good low-frequency performance, the low-frequency content of the music is reduced by the ANC performance level in the frequency range where the ANC is active (20Hz – 1kHz). A simple trick to over-come this problem is shown in Figure 2. One way to compensate for the low-frequency losses is the subtraction of the music signal from the microphone signal. Very often a gain compensation filter M(w) is used to further improve this compensation circuit. The output of the subtraction circuit is the pure noise signal that is used for noise reduction.

This solution looks quite simple on paper but is rather difficult to implement because we have an acoustic signal path (speaker to microphone) involved. A simpler approach is an EQ circuit on the music input. With a simple bass-boost function the low frequency losses can be compensated for, with the possible drawback being headroom issues with the amplifiers.

Hybrid Topology While the feed-forward and feedback topologies are commonly used in headsets, the new “Hybrid” technology has rarely been used up to now. Figure 4 shows the block diagram, which is nothing more than a combination of the feed-forward and feedback topology.

This technology combines the advantages of both systems into a single unique technology. It allows you to achieve best ANC performance levels (>30dB) and wid-est bandwidth. One system compensates for the lacks of the other system and vice versa. These systems typically show superior ANC performance from 20Hz up to 3kHz, which is not possible with a standalone feed-forward or feedback system.

Although the system is the most expensive one (two ECM microphones per channel) and requires a lot of mass production expertise, it is a solution that allows you to be silhouetted against ordinary ANC headsets. Therefore it is now a surprise that the trend for new high-end ANC headsets goes definitely toward “Hybrid”.

ANC Production Challenges A problem that all ANC solutions have in common is production trimming. Variations in the microphone have a very marked impact on headset performance.

The tolerances of a typical electret microphone are around +/-3dB. In order to com-pensate for those tolerances a fairly complex tuning procedure is necessary. Today, this is normally accomplished by adjusting two potentiometers on the PCB. This is a manual process – the adjustment is made by an operator using a screwdriver. In or-der to optimize performance, the operator must adjust the screw while viewing the output of the test rig until the optimum gain setting is found.

This process is both time-consuming and error-prone, because the operator must concentrate on the test results while making the adjustment. In addition, the headset assembly cannot be finished until the headset is trimmed. If the ANC circuit is integrated into an over-the-ear capsule the back of the headset must be removed during trimming. This creates the risk that the acoustic behavior of the headset will be different after re-assembling the capsule, and thus generating a new, uncompensated variation in ANC performance.

New Production Technology If this manual process is slow, unreliable and expensive, how can it be automated? The answer is by implementing a circuit with a dedicated digital trimming capability – a feature found in a new generation of ANC ICs such as the AS3410/AS3430 feedback/feed-forward ANC solution from ams. In an automated trimming system, the audio jack becomes the trimming input of the ANC headset.

The 3.5mm audio jack is the only connection to the outside world, and therefore the only route into the internal hardware besides the battery terminal. Used as a trim-ming input, the jack can configure digital potentiometers in the ANC circuitry. The benefit is clear by comparison with today’s headset production method: it eliminates the need for careful manual adjustment with a screwdriver, and the operator handling the headsets does not need any knowledge of the way that trimming works, since a fully-automated trimming process can be installed.

An automated trimming system such as that enabled by ams’ AS3410/30 thus elimi-nates human error, improves production yield and generates more reliable, higher-performance outputs. The prospect is that ANC headsets can be manufactured more cost-effectively, and therefore are within reach of the mass consumer market, rather than being, as now, a high-end item for wealthy consumers.

Any ideas which consumer products use the AS3410 / 30? I'd like to get a decent performing set of non-expensive noise cancellation headphones.

Hi, for an on-ear product I’d recommend the AKG K490 and in ear type the Audio Technica ATH-ANC23. The audio technica is one of the best in-ear headsets right now on the market and the price is more than fair. You get for ANC only a playtime more than 130h out of a single AAA battery! Hope this helps! If’ve got further questions do not hesitate to ask!

Hi Horst, how do the digital mics fit into the scheme of this. I expect the AS3410/30 is analog only. Could a similar design be accomplished through a uC (with audio focus) using digital mics? I am currently designing audio systems using processors, and need a similar solution.

Hi, you are right, the AS3410/30 has a digital control interface but the signal processing is analog. We, ams, started also a research project making use of a DSP to do a noise cancelling headset and it is basically also possible to do it. The big drawback with digital solutions is the latency issue/propagation delay that limits your bandwidth of the ANC system. This latency caused mainly by the DAC and the DAC filters cause a phase shift at the higher frequency range that affects/limits the ANC performance. To overcome or limit the propagation delay you can of course increase the sample rate of the system but this influences also the power consumption. If you compare the total system power consumption of an analog solution it is approximately 1/3 compared to a digital solution typically with less ANC performance and more hiss noise. Therefore I would say, in terms of ANC, analog makes sense. By the way, another problem we see with digital mems microphones is the frequency response of the microphones. Usually these mics have already a high pass filter integrated which is contra productive to do ANC because it cuts of the low frequency signals you want to cancel. Kind Regards Horst

Thank you, Horst. So would I be correct in thinking that it would be best to use the analog ANC as a post source then pass through the DSP system for required effects? I will down load the data sheet and consider the possibilities. It would be nice if we could utilize this system to reduce our own noise impact on the environment. Especially on marine life. Bryan

Hi Brian, yes you are basically right but usually you do it the other way around. First you do the signal processing of the audio signal with any kind of audio effects and want to do. The output of your DSP is then connected to the ANC chipset that incorporates also a speaker driver. If you do the signal processing at the end of the signal chain you influence the gain and phase response of the ANC signal path each time you change the audio filter in you DSP, thus ANC is not functional any more and requires re characterization of the headset to match the gain and phase response of your headset for good ANC performance. If you plan to go wireless I would recommend the AS3421/22 that has already two fully differential line input that can be directly connected to a wireless chipset e.g. Bluetooth without a differential to single ended conversion. Kind Regards Horst

A typical MEMS microphone does not have a high-pass filter integrated in the digital output processing. The high pass characteristics of many MEMS microphones is a result of the transducer design and is not caused by an additional analog or digital filter. Many of the early MEMS mic designs had a low-frequency corner at about 100 Hz, but now there are many analog and digital mics available that have much lower corner frequencies.

Hi Jerad, I never ever said that there is a analog or digital filter integrated. We are also in the MEMs business and we got microphones to test that operate also at lower frequencies but MEMS microphones with a good frequency response below 100Hz are not very common yet and I am not sure I they give you a price advantage over ECM microphones, do they? Therefore we do not see a point switching to analog MEMs microphones unless you have physical size constraints in a design.

Hi Horst, I wanna ask some question about AMS ANC chips. I want to develop ANC system for industrial environment. Is it possible to use AMS chips with external amplifier and some sort of flat panel speakers for this purpose. What is your recommendations. Mali Dalin. Regards.

Hi Mali, basically it is possible but our products are optimized for portable systems like wired and wireless headsets. Of course you can use it with an additional amplifier but I would recommend going another direction since system power consumption is not really an issue. There are companies out there in the field that have specialized on industrial noise canceling applications that would probably be a better fit than our portable chips. Kind Regards Horst

Hi Horst! I was hoping you could provide insight if this is possible for an ‘off the ear’ solution. There’s a noisy garbage truck that comes by once a week and wakes me up very early. Is it possible to use AMS IC’s with a small amp and speaker? I envision placing a left and right channel on either side of my head maybe hovering a few feet above me. If it is possible, might there be a solution to cancel frequencies 20-20k Hz? Thanks!

Hi, well there are many applications and basically it is possible to chancel noise is you put speakers in a headrest but the performance is rather low and bandwidth is also very narrow. In an ideal world it is possible to chancel noise up to 20kHz but unfortunately the speed of sound makes this difficult. I am planning a second article how to design an ANC headset that will also explain why we get the bandwidth limitation in ANC headsets. Kind Regards Horst

Hi Horst, How can I use this in earmuffs to make it active noise cancellation earmuff? Which kind of speaker and microophone can be used with this AS 3430? Thanks

Hi, earmuffs are usually always big over the ear “headsets”, therefore I would recommend a 16-32Ohm speaker and a standard ECM microphone with good SNR next to the speaker inside the cavity of the earmuff. Next step would then be the open loop characterization of the headset to determine the necessary ANC filter, start filter development and test it out with the AS3430 eval board. Do you want to develop a product or is it just for private use? Kind Regards Horst

Thanks for the reply. Yes I want to develop a product.

then I would recommend that you get in contact with one of our sales offices. We can also get you app notes and provide you with the necessary support needed.

“Something worth adding about the feedback method is that (due to something called bode’s inequality) you also get regions where noise gets boosted. One below the frequency range where the loop works and one above. The subsonic hump places extra strain on the power handling capability of the driver.”

You must Sign in or Register to post a comment.