This article describes the necessary steps to design an active noise reduction feedforward headset based on the AS3415 active noise reduction chip.
Before we officially start making an active noise reduction headset, we need special audio equipment.
- First of all, we need an audio measurement system for measuring frequency response and phase response.
- In addition to the audio measurement system, the human ear simulation device is also an important part. The ear simulation device can be used to simulate the response of the human ear when measuring the characteristics of the earphone. These artificial ears integrate highly accurate microphones that can measure the sound that people actually hear when wearing headphones.
- In addition, we need a speaker. It is used to measure the passive attenuation characteristics of headphones. This is part of the filter design. The speakers should be two-way speakers. We’d better choose coaxial two-way speakers. This can ensure that the transmission distance of the high frequency and low frequency from the speaker to the earphone is equal.
- Finally, the AS3415 evaluation board is needed, which contains all the necessary connectors and preamplifiers to make the performance test process as smooth as possible.
Why Do We Need to Perform a Performance Test on the Headset?
The acoustic performance of each headset is different. Because headphones use different components, such as speakers with different impedances and transfer coefficients. The elastic cushions and front and rear acoustic cavities of each headset are also different.
If we want to make active noise reduction headphones, it is important to understand the characteristics of the headphones. In this way, good noise reduction performance can be obtained. Active noise reduction feedforward headphones use ECM microphones to capture noise outside the headphones. The electronic circuit will generate an anti-noise anti-signal, which is then played out through the loudspeaker. Theoretically, the ANC loop is a simple inverter circuit, but this is not the case. Because the different components of the headset will affect the frequency response and phase response. Simple inversion cannot make ANC meet the performance requirements. In order to understand the performance of the headset in terms of gain and phase, the performance test of the ANC headset is very important.
Calculate the Ideal Value of the Active Noise Reduction Filter
After the performance test of the headset, we can use the measurement results to calculate the ideal value of the ANC filter. The required filter amplitude is calculated as follows:
Figure 1
The phase of the required filter is calculated as follows:
Figure 2
The calculation results can be easily obtained through Excel spreadsheets. An example of a filter is shown in Figure 5. The frequency response and phase response of the example show that it is difficult to make an ideal noise reduction signal using only a full-bandwidth inverting amplifier.
Development of the Filter
The key to a good ANC headset is the design of the filter. The goal of filter design is to match the gain and phase response as much as possible. The better the match at a specific frequency, the better the ANC performance. Because it is analog signal processing, the simulation of the filter is usually completed by spice simulation tools. Figure 3 is a spice simulation circuit, which embodies the signal path of the ANC microphone filter.
Figure 3: A Simulation Example of Spice Filter
The purpose of the ANC filter design engineer is to match the gain and phase response in the filter simulation circuit of Figure 3 with the calculated ideal curve. The typical filters that people use today are first-order low-pass filters, notch filters, high-shelf and low-shelf filters. Designers must understand different topologies and calculation methods for cutoff frequency, passband, and stopband. This is certainly not a simple task, especially when they are not used to using spice simulation tools and analog filter development tools.
Figure 4: AS3415 Feedforward Filtering Simulation Tool
To solve this problem, the AS3415 evaluation software integrates a feedforward filter simulation tool, as shown in Figure 4.
Design engineers can use this tool to design ideal ANC filters. This tool provides a set of predefined filter architectures instead of modifying part values and filter structure. Based on filter structures simulated for many different customers, these predefined filter structures can cover 90% of the ANC acoustic requirements. Figure 5 shows the simulation results of the tool. The green curve represents the ideal ANC gain and phase response, and the blue curve shows the simulation results of the ANC filter made with the tool in Figure 4.
When designing a filter, one thing is very important, that is, which frequency bands we need to pay attention to. The operation of ANC headphones has a specific frequency range. This is not due to the limitations of the AS3415 itself, but is related to the speed of sound propagation and the acoustic characteristics of the headset. If we only focus on the gain response in the ideal filter curve, it is easy to design a filter that conforms to this curve. But the problem is that the phase must also be matched in the design of the ANC filter. Since the phase at higher frequencies is rotated by almost 180 degrees, the designed filter is likely to match the frequency response, but not the phase. Depending on different earphones and their phase response, we can usually achieve filter matching below 1.5kHz. The higher frequency part needs to be attenuated as much as possible. If these unmatched high-frequency parts are not attenuated, noise may be introduced. We attenuate the noise in the low frequency part, but if the phase mismatch of the high frequency, it will cause the noise to be amplified.
In order to avoid this phenomenon, we will try to reduce the gain in the unmatched area. The green transparent area in Figure 5 represents the smallest mismatch of gain and phase that we can usually achieve. The red area is the part we want to attenuate as much as possible. A good balance must be reached between high frequency attenuation and phase response. If it is attenuated too much at high frequencies, it will affect the phase response of low frequencies and may lose the effect of ANC.
Figure 5: Simulation Results
Filter Inspection and ANC Test
When we get a satisfactory filtering curve, the AS3415 filter simulation tool also provides a list output function. Since this tool matches the evaluation board of AS3415, the items listed in the list can be soldered on the evaluation board. In this way, the performance of the ANC with this filter can be tested. The test contains two contents: one is the passive attenuation test when the headset is worn on the artificial head, and the other is the frequency response test when the AS3415 chip is turned on and the feedforward noise reduction function is configured. ANC performance calculation is as follows:
These calculations can be done through an Excel table and generate a graph of the ANC noise reduction performance in the audio range. This graph of ANC noise reduction performance is very important and common in the design and production of noise reduction headphones. AS3415 development tools and application notes and samples related to the development of ANC noise reduction headphones are now online.
For information on Easybom and other electronic component, please visit:electronic parts