How A2B technology and digital microphones can achieve outstanding performance in emerging automotive applications

Future vehicle architectures will increasingly rely on high-performance acoustic detection technologies such as microphones and accelerometers. A fully digital approach that includes sensors, interconnects, and processors can bring important performance and system cost advantages. ADI is working with Harman International to provide cost-effective solutions to create value and differentiation for end customers.

Authors: Ken Waurin, Dietmar Ruwisch and Yu Du

Introduction

This article is about Automotive Audio Bus® (A2B®, Car audio bus) technology article introduces the latest developments in digital microphones and connection technology. These innovations are prompting the rapid adoption of transformative applications that support the new generation of automotive infotainment systems.

Market and application overview

In the automotive cockpit electronics market, as automakers strive to differentiate their vehicles from their competitors, an increasingly obvious trend is that audio, voice, and acoustic-related applications are expanding rapidly. In addition, as ordinary consumers become more aware of technology, their expectations for driving experience and the level of personal interaction with the vehicle are also greatly increasing. Home theater-quality audio systems have become a common configuration for all price-priced vehicles, and now there are complex voice hands-free (HF) and in-car communication (ICC) systems. In addition, the active noise reduction and road noise reduction (ANC/RNC) systems traditionally deployed only in high-end vehicles have now entered the mainstream market that is affordable for ordinary people. Looking to the future, audio or acoustic technology will become a key component of the engine control unit (ECU) of L4/L5 autonomous vehicles, because the ECU needs to detect whether there is an emergency vehicle.

What all these traditional and emerging applications have in common is their reliance on high-performance acoustic detection technologies, such as microphones and accelerometers. Almost all emerging applications require multiple acoustic sensors (such as microphones or microphone arrays) to achieve high-quality system-level performance, so a simple but cost-effective interconnection technology is needed to ensure that the total system cost is minimized. For a long time, the lack of microphone-optimized interconnection technology has been a major pain point for car manufacturers. Each microphone needs to be connected directly to the processing unit using expensive and bulky shielded analog cables. These increased costs—mainly actual wiring, followed by weight gain and reduced fuel efficiency—have hindered the widespread adoption of these applications in many cases, or at least restricted them to the high-end market. The latest developments in digital microphones and connection technologies are expected to promote the rapid adoption of transformative applications in the new generation of vehicle infotainment systems. A2B technology will be promising.

Implementation and limitations of traditional analog microphones

Most countries and regions prohibit the use of handheld phones while driving and support Bluetooth®The hands-free device has become standard equipment in almost all vehicles. There are a variety of hands-free solutions on the market-from simple stand-alone units containing speakers and microphones to advanced solutions that are fully integrated in the vehicle’s infotainment system. Until recently, most hands-free systems were implemented in a very similar way. Such systems contain only one (a few have two) microphones, and the related microphone technology is an electret condenser microphone (ECM) type 50 years ago. The voice quality of the transmitted audio is often unsatisfactory, especially for simple independent units, where the distance between the microphone and the speaker’s mouth may be quite large. If the microphone is installed as close to the mouth as possible (for example, in the roof panel of a vehicle), the communication quality can be improved. However, in this case, if you want to support the driver and passengers equally, then both seats in the front row need to have microphones.

A typical automotive ECM microphone is a device that integrates an ECM unit and a small amplifier circuit in a single housing. The amplifier provides an analog signal whose voltage level allows the signal to be transmitted through several meters of wire, which is also a requirement for typical automotive applications. Without amplification, the original ECM signal is too low for such a long wire, and the signal-to-noise ratio (SNR) will decrease too much due to electromagnetic interference on the wire. Even if the signal is amplified, a shielded cable is required-usually a two-wire cable, which supplies power to the microphone device through a bias voltage (8V). Considering this wiring requirement, due to weight and system cost constraints, it is obvious that the number of ECM devices used in mainstream vehicles is very limited.

One of the few advantages of ECM is its built-in acoustic directivity, which is usually adjusted to a supercardioid polar pattern (MEMS microphones can also be made directed, but usually require more complex acoustic designs). Usually a backward attenuation of 10 dB or more can be achieved. “Backward” refers to the direction toward the windshield, from which only noise (ie, no desired signal, such as the speaker’s voice) will be generated. Higher sensitivity in the incoming direction of the desired signal is very conducive to improving SNR. However, the directional ECM unit will introduce unnecessary side effects, such as high-pass characteristics-sensitivity will be reduced at lower frequencies. The 3 dB cutoff frequency of this high-pass response is usually in the range of 300 Hz to 350 Hz. In the early days of HF technology, this high-pass characteristic was an advantage, because engine noise mainly existed at lower frequencies, and the engine sound itself was attenuated by the microphone. However, since the advent of broadband or HD calls, this high-pass feature has become a problem. In broadband calls, the effective bandwidth increases from 300 Hz to 3400 Hz to 100 Hz to 7000 Hz. The microphone’s own high-pass filtering characteristics make it necessary to amplify signals from 100 Hz to 300 Hz in the post-processing unit, and if the microphone itself can provide better audio bandwidth, it is not necessary to amplify signals in this range. Another disadvantage of ECM technology is that the sensitivity and frequency response of different devices vary greatly. The manufacturing tolerances of ECM are relatively large, which may not be a problem for single microphone applications. However, if multiple microphone signals are deployed in a microphone array application with a small pitch, strict matching between the microphones is essential to achieve high-quality array performance. In this case, ECM is difficult to use. In addition, from the perspective of physical size, traditional ECM units are generally not suitable for small microphone arrays.

Microphone arrays have a wide range of applicability, including in cars, because they provide similar (and often superior) directional performance compared to traditional ECMs. The spatial information about the sound impact direction can be extracted from the microphone signal using two or more suitable microphones grouped in the array. This type of algorithm is often referred to as beamforming (BF). The term “beamforming” is derived from the analogy with phased array antenna technology, using a simple pure linear filter and summation algorithm to focus the radio “beam” emitted by the antenna array in a certain direction. Although there is no such beam in a microphone array, the term beamforming is also very common in the field of microphone signal processing. Compared with simple linear beamforming processing, it covers a wider range of linear and non-linear algorithms and supports higher implementations. Performance and greater flexibility.

In addition to beamforming processing, the original microphone signal almost always requires post-processing, because each HF microphone captures both the desired voice signal and the interference in the environment (if the cockpit). Wind noise, road noise, and engine noise can reduce SNR, and the signal played through the loudspeaker-usually called the loudspeaker echo-is also an unwanted signal source. In order to reduce this interference and improve voice quality, complex digital signal processing techniques are needed, often referred to as echo cancellation and noise reduction (AEC/NR). AEC eliminates the speaker sound from the microphone, otherwise it will be transmitted as an echo of the human voice speaking at the other end of the line. NR improves the SNR of the transmitted signal while reducing the constant driving noise. Although the International Telecommunication Union (ITU) has issued detailed specifications (such as ITU-T P.1100 and P.1110) to define many performance details of the HF system, when talking in a moving vehicle, if the AEC/NR processing fails to meet the standard , People may not be satisfied with their subjective impression of communication quality. Together with the aforementioned BF algorithm, the combination of AEC/NR/BF enables a wide range of new applications, all of which are related to a certain degree of digital audio signal processing. In order to support these applications, a new generation of microphone technology that eliminates the shortcomings of traditional ECM is required.

Technical and performance advantages of digital MEMS microphones

Micro-Electro-Mechanical System (MEMS) technology has quickly become a new industry standard for microphones because it offers many advantages over traditional ECM. First, MEMS makes acoustic sensors much smaller than existing ECM units. In addition, a digital microphone obtained by integrating a MEMS sensor and an analog-to-digital converter (ADC) in a single IC can provide signals that can be processed immediately by AEC/NR/BF.

Analog interface MEMS microphones also exist, but they have many of the same shortcomings as analog ECM, and if the traditional two-wire analog interface is used to work, even more complicated amplifier circuits are required than ECM. Only by adopting all-digital interface technology can the inherent interference and SNR problems of analog lines be significantly reduced. In addition, from a production point of view, MEMS is also selected, because the production specification deviation of MEMS microphones is much smaller than that of ECM units, which is very important for the BF algorithm. Finally, the manufacturing process of the MEMS IC microphone is greatly simplified, because the automated installation technology can be used, and the overall production cost can be reduced. From an application point of view, the smaller size is the biggest advantage, and because the sound entrance is very small, the MEMS microphone array can actually be made invisible. The entrance and sound channel of the sensor require special care in design and production quality. If the acoustic seal is not strong, the noise from the internal structure may reach the sensor, and the leakage between the two sensors may reduce the performance of the BF algorithm. Unlike typical ECM units that can be designed and manufactured to be omnidirectional or directional, MEMS microphone elements are almost always manufactured to be omnidirectional (that is, there is no inherent directionality for sound reception). Therefore, the MEMS microphone is an omnidirectional sound pressure sensor that is faithful to the phase, and provides an ideal signal for the advanced BF algorithm. The attenuation direction and beam width can be configured by the user through software.

Generally speaking, it is very important to organize all signal processing modules in an integrated algorithm suite. If the functional modules are implemented in isolation from each other, processing delays will increase unnecessarily and overall system performance will decrease. For example, the BF algorithm should always be implemented together with AEC, preferably by the same provider. If the BF algorithm introduces any non-linear effects in the signal, AEC will definitely produce unsatisfactory results. The ideal result of digital signal processing is best achieved through an integrated algorithm package that receives undegraded microphone signals.

The following is a detailed comparison of the standard linear BF and ADI proprietary algorithms, so that everyone can fully understand the performance potential of the advanced BF algorithm. The curve in Figure 1 shows the polarity characteristics and frequency response of three different BF algorithms in the in-beam and out-of-beam directions. A standard linear supercardioid algorithm based on a dual microphone array is used as a reference (black curve). The reference curve shows the maximum attenuation in the typical zero-angle direction (that is, the maximum out-of-beam attenuation), and the “backlobe” at 180°, where the out-of-beam attenuation is low. The resulting backlobe is the result of a trade-off with the beam width in the linear algorithm. The heart-shaped beam (not shown) has the maximum attenuation at exactly 180°. However, its acceptance area is wider than the super-cardioid configuration. Beams with less obvious backlobes and higher out-of-beam attenuation can be realized by nonlinear algorithms. The red curve shows this type of ADI proprietary dual-microphone algorithm (microphone spacing: 20 mm).

How A2B technology and digital microphones can achieve outstanding performance in emerging automotive applications
Figure 1. Polarity attenuation characteristics of different BF algorithms.

There are two omnidirectional microphones in the array, so the beam shape always has rotational symmetry. In other words, the attenuation of X° in the polarity diagram is the same as the attenuation of 360°-x°. This assumes that the 0° to 180° line of the polarity diagram is equivalent to the imaginary line connecting two microphones. The three-dimensional beam shape can be imagined by rotating the two-dimensional polar curve around the microphone axis. An asymmetric beam shape or narrower beam without rotational symmetry requires at least three microphones to be arranged in a triangle. For example, in a typical overhead console installation, a dual microphone array can attenuate sound from the windshield. However, when so oriented, the dual microphone array cannot distinguish the driver from the passenger. Rotating the array 90° can distinguish the driver from the passenger, but the noise generated by the windshield and the sound in the cockpit will be indistinguishable. Only by using three or more omnidirectional microphones arranged in an array can the windshield noise be attenuated and the driver and passengers can be distinguished. The green curve in Figure 1 shows the exemplary polarity characteristics of the corresponding ADI proprietary three-microphone algorithm, where the microphones are arranged in an equilateral triangle with a spacing of 20 mm.

The polarity map is calculated using band-limited white noise reaching the microphone array from different angles. The audio bandwidth is limited to 100 Hz to 7000 Hz, which is the wideband (or high-definition voice) bandwidth of advanced cellular telephone networks. Figure 2 compares the frequency response curves of different algorithm types. In the direction of the beam, the frequency response of all algorithms is flat within the desired audio bandwidth, which is in line with expectations. The out-of-beam frequency response is calculated for the half-space outside the beam (90° to 270°), and it is confirmed that the out-of-beam attenuation is high in a wide frequency range.

How A2B technology and digital microphones can achieve outstanding performance in emerging automotive applications
Figure 2. In-beam (dashed line) and out-of-beam (thick line) frequency responses of different BF algorithms.

The relationship between array microphone spacing and audio bandwidth and sampling rate is worthy of further discussion. Broadband HD voice uses a sampling rate of 16 kHz, which is a good choice for voice transmission. Compared with the 8 kHz sampling rate used in earlier narrowband systems, the current 16 kHz wideband sampling rate is very different in terms of voice quality and voice intelligibility. Due to the promotion of speech recognition providers, the demand for higher sampling rates (such as 24 kHz or 32 kHz) continues to grow. Voice band applications may require sampling rates up to 48 kHz, which is usually the main system audio sampling rate. The underlying motivation is to avoid sampling rate conversion internally. However, the additional computing resources required to support these high sampling rates are not commensurate with the actual effects they produce, so 16 kHz or 24 kHz is now widely accepted as the recommended sampling rate for most voice band applications.

For beamforming applications, high sampling rates are problematic because spatial aliasing occurs where the frequency is equal to the speed of sound divided by twice the microphone spacing. Beamforming cannot be performed at this aliasing frequency, so spatial aliasing is undesirable. If you limit the microphone spacing to 21 mm or less, you can avoid spatial aliasing in a wideband system (16 kHz sampling rate). If the sampling rate is higher, the spacing needs to be smaller to avoid spatial aliasing. However, too small a distance between the microphones will not work, because microphone tolerance, especially the inherent (non-acoustic) noise of the microphone sensor will become a problem. If the spacing is small, the interference (such as inherent noise) and sensitivity deviations between the microphones of an array may overwhelm the signal difference between the microphones, causing the signal difference to become insignificant. In practice, the microphone spacing should not be less than 10 mm.

A2B Technical overview

A2The B technology is specifically designed to simplify the connectivity challenges of emerging automotive microphones and sensor-intensive applications. From an implementation perspective, A2B is a serial topology with a single master device and multiple sub-nodes (up to 10). The third-generation A currently in full mass production2There are five members in the B transceiver family, all offering automotive, industrial and consumer electronics temperature ranges. The full-featured AD2428W and four derivative devices with reduced functions and lower cost-AD2429W, AD2427W, AD2426W and AD2420W-constitute ADI’s latest pin compatible enhanced A2B transceiver series.

The functions of AD2427W and AD2426W have been reduced (only for sub-nodes), mainly for microphone connection applications such as hands-free, ANC/RNC or ICC. AD2429W and AD2420W are entry-level A2B derivative devices have significant cost advantages over full-featured devices, and are particularly suitable for cost-sensitive applications such as automotive eCall and multi-element microphone arrays. Table 1 compares various third-generation A2The characteristics of the B transceiver.

Table 1.A2B transceiver feature comparison

characteristic

AD2420/ AD2420W

AD2426/ AD2426W

AD2427/ AD2427W

AD2428/ AD2428W

AD2429/ AD2429W

Support master node

no

no

no

Yes

Yes

Number of discoverable child nodes

Up to 10

Up to 2

TRX function module

A only

A only

A + B

A + B

B only

I2S/TDM support

no

no

no

Yes

Yes

PDM microphone input

2 microphones

4 microphones

4 microphones

4 microphones

4 microphones

Maximum cable length between nodes

5 meters

15 meters

15 meters

15 meters

5 meters

The AD242x series supports the connection of a single master device and up to 10 sub-nodes through a daisy chain. The total bus distance can be up to 40 meters, and the distance between nodes can be up to 15 meters. Compared to the existing ring/parallel topology, A2B’s daisy chain topology is an important advantage, which is beneficial to the integrity and robustness of the overall system. If A2One connection of the B daisy chain is affected, and the entire network will not collapse. Only the nodes downstream of the failed connection will be affected. A2B’s embedded diagnosis can determine the cause of the failure, send out an interrupt signal, and initiate corrective actions.

Compared with the existing digital bus architecture, A2B’s master device-slave node topology itself is more efficient. After starting the simple bus initialization process, the bus can operate normally without the intervention of more processors. A2An additional advantage brought by B’s unique architecture is that the system delay is completely deterministic (less than 50 µs), and the delay is the same as that of the audio node in A2The position on the B bus is irrelevant. This feature is extremely important for voice and audio applications such as ANC/RNC and ICC, where audio samples from multiple remote sensors must be processed in a consistent manner.

All A2B transceivers can transmit audio, control, clock and power signals on an unshielded twisted pair cable. This can reduce the total cost of the system for the following reasons.

• Compared with the traditional implementation, the number of physical cables is reduced.

• The actual cable used can be a lower cost and lighter weight unshielded twisted pair, rather than a more expensive shielded cable.

• Most importantly, for specific application scenarios, A2B technology can provide bus power supply capability and transmit current not more than 300 mA to A2B Audio node on the daisy chain. With this bus power supply capability, there is no need to use a local power supply on the audio ECU, thereby further reducing system costs.

A2The total bus bandwidth of 50 Mbps provided by B technology can support up to 51 upstream and downstream audio channels using standard audio sampling rates (44.1 kHz, 48 kHz, etc.) and bit widths (16, 24 bits). This provides considerable flexibility and connectivity for a wide range of audio I/O devices. Maintaining an all-digital audio signal chain between audio ECUs can ensure high-quality audio quality without any degradation in audio performance due to ADC/DAC conversion.

Open circuit, wire short circuit, wire reverse connection, wire short circuit to power or ground. From the perspective of system integrity, this function is very important, because in the event of a fault such as an open circuit, a short circuit of a wire, or a reverse connection of the wire, the A at the upstream of the fault point2Node B can still work normally. The diagnostic function also provides the ability to efficiently isolate system-level faults, which is critical from the perspective of the car dealer/installer.

The recently announced fourth-generation A2The B transceiver AD243x is a development on the basis of existing technology, which has improved the key functional parameters (the number of nodes has been increased to 17, and the bus power supply has been increased to 50 W). At the same time, an additional SPI control channel (10 Mbps) has been added, which is an intelligent A2The remote programming of Node B provides an efficient software over-the-air update (SOTA) capability. The new features of the AD243x series make it very suitable for new applications, such as microphone nodes equipped with LEDs in ultra-advanced microphone architectures.

A2B Microphones and sensors in the automotive industry

From a single voice microphone to a multi-element BF microphone array for HF communication, from ANC to RNC, from ICC to alarm sound detection, microphones are increasingly used in the automotive industry. In accordance with technology and market trends, almost every new car on the road today is equipped with at least one microphone module for HF communication. High-end and luxury cars may have six or more microphone modules, which are necessary to realize the full potential of BF, AEC, ANC, RNC, ICC, etc., digital MEMS microphones have obvious advantages in these applications.

More and more microphones present a major challenge to vehicle infotainment system engineers-how to simplify the wiring harness and make it lightest. For traditional analog systems, this is not a simple task. An analog microphone needs at least a pair of double shielded wires (ground and signal/power), pins, and connector cavity for interconnection. The amount of wires is always twice the number of microphone modules in the system. At the same time, the length of wire required to connect each microphone module will cause the total weight of the wire harness to increase faster. A simple way to alleviate this problem is to share microphone signals between multiple applications, thereby reducing the number of microphones used in the system. For example, the same microphone signal can be used for HF communication, and can also be used as the Error input in the ANC system. However, different applications may require different microphone characteristics. In the aforementioned examples, HF microphone signals are often more desirable to have a rising frequency response shape (ie, sensitivity decreases with decreasing frequency) to eliminate low-frequency noise content in the cockpit. This is a useful and very effective technique that can improve the intelligibility of the voice delivered by the voice microphone. On the contrary, ANC microphones need a sufficiently high sensitivity level at low frequencies, because the main purpose of the ANC algorithm is to reduce low-frequency noise. Therefore, in order for two applications in an analog system to share the same microphone, the signals from the microphones need to be fed to different circuits for proper frequency filtering. In this case, one or more ground loops may be formed, which may cause serious noise problems.

As a digital bus with daisy chain connection capability, A2Together with digital MEMS microphones, B technology provides a multi-microphone signal interconnection and/or sharing solution, which is very suitable to meet the needs of the rapidly expanding audio, voice, noise cancellation and other acoustic applications in vehicles. Consider a fictitious but demonstrative situation: an automotive application requires a HF microphone module, an ANC microphone module and a simple array microphone module composed of two microphone elements for BF, all three modules are integrated in the ceiling light Around the module. Figures 3a and 3b show how to use traditional analog systems and digital A, respectively2B system to achieve this design.

How A2B technology and digital microphones can achieve outstanding performance in emerging automotive applications
Figure 3. (a) Analog system design (shielded wire) using analog microphone components.
(b) Digital system design using digital microphone components (A2B technology and UTP line).

Since the analog system cannot easily support microphone sharing, each application module (HF, ANC, and BF) requires a dedicated microphone and a separate wiring harness to connect the corresponding functional circuit. This results in the need for four separate microphone elements and three sets of wiring harnesses (a total of seven wires plus shielding). On the other hand, the number A2System B can easily support shared signals, so the number of microphone components can be reduced from four to two. In this specific example, a single microphone module composed of two broadband omnidirectional microphone elements can be used to provide two acoustic signal channels to meet the needs of all application modules. Once the signals of these two channels reach the central processing unit (such as a host or independent power amplifier) ​​through a simple UTP line, they can be shared and digitally processed to support HF, ANC and BF applications.

Although the example shown in Figure 3 may not represent the actual situation, it clearly shows A2The advantages of B technology over traditional analog technology. A2Digital audio bus systems such as B technology solve the challenges of automakers, allowing them to propose new audio and acoustic-related concepts to enhance the user experience, and support the faster introduction of these concepts to the market.

In fact, A2The commercialization of B technology has made many applications in the automotive market possible, including both new applications and previously difficult applications. For example, Harman International, a leading provider of car audio solutions, has developed a series of digital microphones and sensor modules, which use A2B system to empower various automotive applications. Figure 4 shows some common car A2B microphones and sensors and how they are used in cars. These sensors include: Single A2B microphone, multi-element microphone array for ANC and voice communication, A for RNC2B accelerometer, externally mounted bumper A2B microphone, and roof A for emergency alert detection and acoustic environment monitoring2B microphone array. In these A2With the empowerment of B microphones and accelerometers, more and more application solutions that require multi-sensor input are being developed to further enhance the user experience in the automotive industry.

Summarize

Future vehicle architectures will increasingly rely on high-performance acoustic detection technologies such as microphones and accelerometers. A fully digital approach that includes sensors, interconnects, and processors can bring important performance and system cost advantages. ADI is working with Harman International to provide cost-effective solutions to create value and differentiation for end customers.

How A2B technology and digital microphones can achieve outstanding performance in emerging automotive applications
Figure 4. Common A2B microphone and sensor.

author

How A2B technology and digital microphones can achieve outstanding performance in emerging automotive applications

Ken Waurin

Ken Waurin is the strategic marketing manager of Analog Devices, responsible for automotive infotainment systems. His main focus is advanced audio, active noise cancellation and in-car connectivity.

How A2B technology and digital microphones can achieve outstanding performance in emerging automotive applications

Dietmar Ruwisch

Dietmar Ruwisch is a senior audio technology expert at Analog Devices. He studied physics in Mst, Germany, and received his PhD in 1998 with a thesis on artificial neural networks. Since then, his focus has been on audio signal processing, and he holds a number of patents in this field. He is committed to improving the quality of audio communication-between humans and between humans and machines-and the corresponding processing of microphones and microphone array signals.

How A2B technology and digital microphones can achieve outstanding performance in emerging automotive applications

Yu Du

Yu Du is a senior chief acoustic engineer at Harman International Industries. He holds a bachelor’s and master’s degree in automotive engineering from Tsinghua University (Beijing, China), and a PhD in mechanical engineering from Virginia Tech (Blacksburg, Virginia, USA). He has more than 20 years of research and development experience, involving multiple fields of acoustics, including structural acoustics, active and passive vibration and noise control, MEMS sensor design and simulation, hearing science, and acoustic signal processing. His current work at Harman focuses on the development of advanced microphone and sensor technology for automotive applications. Dr. Du is a member of the American Acoustic Society (ASA), Audio Engineering Society (AES) and American Society of Mechanical Engineers (ASME). He currently serves on the AES Car Audio Technical Committee.

The Links:   LM215WF3-SLC5 G649DX1R010

Author: Yoyokuo