Design of Handheld Communication Products Based on Miniature Array Microphone Technology
Spatial constraints in handheld product structures are increasing. Speakers are getting smaller while required sound levels are rising. Suppressing various noises and echoes (linear and nonlinear echoes during video hands-free calls) to achieve clear voice communication in noisy environments is a challenging problem.
The FM2010 chip, introduced by ForteMedia Inc. (USA), is a low-power, low-cost single-chip solution based on patented Small Array Microphone (SAM) technology. It employs spatial filtering, directional sound pickup over far/near distances, acoustic noise suppression, and acoustic echo cancellation. This article introduces the design essentials of SAM technology in handheld communication products, the main features of the FM2010 chip, and its typical application in GSM phones.
Design Essentials of Miniature Array Microphone Technology
SAM technology can utilize two microphones, a Uni-MIC (primary mic) and an Omni-MIC (reference mic), to form the miniature array. They can be placed back-to-back or side-by-side. Leveraging the physical differences between the two mic types and processing by the FM2010 chip, a conical pickup beam is formed, acting as a spatial filter to suppress non-stationary noise. The microphone characteristics, structural design, and FM2010 parameter adjustments determine the direction, angle of the pickup beam, and the effectiveness of non-stationary noise suppression.
1 Microphone Selection
4mm Uni and Omni microphones are recommended. Uni-MIC sensitivity: -40dB ±3dB; Frequency response: less than 8.5dB dip at 300Hz, less than 3.5dB peak at 3.4kHz; Polar pattern: Cardioid, sensitivity difference between 0° and 90° >4dB, difference between 0° and 180° >10dB. Omni-MIC sensitivity: -40dB ±1.5dB; Frequency response flat from 300Hz to 3.4kHz. Recommended models: Uni-MIC B4015UL403 and Omni-MIC B4015AL-398 from Yilida (IEA), Weifang, Shandong.
2 Structural Design
Key structural design issues involve maintaining Uni-MIC characteristics and the conical pickup beam direction. If hands-free functionality is included, additional considerations for speaker/microphone vibration damping and microphone air-tightness are needed. The direction of the conical beam determines the direction of non-stationary noise suppression. Therefore, the useful signal must be within the beam; otherwise, it risks being suppressed as noise. Product form factor design must fully consider the Uni-MIC's orientation. The design should ensure that after installation into the mic sleeve and the entire housing, the Uni-MIC maintains a sensitivity difference >6dB between 0° and 180°, with sensitivity and frequency characteristics largely unchanged. Microphone vibration damping reduces nonlinear echoes; air-tightness reduces linear echoes, improving the signal-to-echo ratio.
3 Signal Comparison: Raw vs. FM2010 Processed
Signals picked up by the miniature array microphone compared to signals output after FM2010 processing are shown in Figure 4. The sound source is 0.3m from the array, intensity 83dB (SPL). Test signals are: Uni-MIC 0° and 180°, Omni-MIC 0° and 180°, Line Out (Lout) - signal after FM2010 processing for 0° and 180° pickup. Figure 4 shows that signals of the same magnitude can differ by 20dB in output level depending on whether they are inside (0°) or outside (180°) the pickup beam. Meaning, as long as non-stationary noise is outside the beam, it will be suppressed by 20dB relative to the useful signal. The effective range of the SAM conical beam is 2m. The beam angle depends on the Uni-MIC's polar pattern and FM2010 parameter adjustments.
The signals from the Uni-MIC and Omni-MIC are amplified by programmable gain amplifiers (PGA), converted to digital (ADC), and high-pass filtered before being sent to the voice processor (for linear echo cancellation, nonlinear echo cancellation, VAD detection, noise suppression, and mic volume setting). The processed voice data is converted back to analog (DAC) and output single-ended from the Line Out (LOUT) to the MICIP/MIC1N inputs of the TWL3014/16 analog baseband processor. After uplink processing, the signal is sent to the digital baseband processor OMAP733/750. The received GSM signal is demodulated, channel decoded, and decrypted by the OMAP733/750 to obtain the digital audio signal. This is sent to the TWL3014/16 analog baseband processor. After downlink processing, the signal is output to the receiver (HSO) and headset (EARP/EARN). One path goes to the receiver, another to an external power amplifier driving the hands-free speaker. Both signals are also fed into the FM2010's Line Inputs (ADC -> HPF -> Voice data) as echo reference signals for the voice processor.
OMAP733/750 Control Flow for FM2010: Control is achieved via the SHI interface, PWD, RESET, and ANA_IRQ pins. After system power-on: First set PWD high and ANA_IRQ low. After reset, send parameters (clock source, clock frequency, DSP operating speed) to the FM2010, then put the chip into power-saving mode. As shown in Figure 8, the call mode is determined by incoming/outgoing calls or recording operations, waking the FM2010. After reset, based on Handset/Hands-free mode, send corresponding parameters. For Handset Noise Cancel mode, send parameters: number of mics, mic gain, mic volume, echo cancellation parameters, VAD parameters. For Hands-free Conference mode, send: number of mics, mic gain, mic volume, mic inversion, echo cancellation parameters (covering scenarios like low/normal/loud remote speech). Hands-free Personal mode tuning is similar to Handset mode, but echo cancellation parameters require more adjustment. After the call ends, turn off the FM2010 CODEC and set it to power-saving mode. As shown in Figure 9, GSM Phone Audio Test Mode is mainly for testing purposes. In test mode, the FM2010 works in bypass mode; the internal DSP performs no processing. The Uni-MIC input signal is amplified by the PGA and directly output from the LOUT amplifier. Mic amplifier and LOUT amplifier gain parameters can be adjusted online via the SHI interface.
The FM2010 chip, introduced by ForteMedia Inc. (USA), is a low-power, low-cost single-chip solution based on patented Small Array Microphone (SAM) technology. It employs spatial filtering, directional sound pickup over far/near distances, acoustic noise suppression, and acoustic echo cancellation. This article introduces the design essentials of SAM technology in handheld communication products, the main features of the FM2010 chip, and its typical application in GSM phones.
Design Essentials of Miniature Array Microphone Technology
SAM technology can utilize two microphones, a Uni-MIC (primary mic) and an Omni-MIC (reference mic), to form the miniature array. They can be placed back-to-back or side-by-side. Leveraging the physical differences between the two mic types and processing by the FM2010 chip, a conical pickup beam is formed, acting as a spatial filter to suppress non-stationary noise. The microphone characteristics, structural design, and FM2010 parameter adjustments determine the direction, angle of the pickup beam, and the effectiveness of non-stationary noise suppression.
1 Microphone Selection
4mm Uni and Omni microphones are recommended. Uni-MIC sensitivity: -40dB ±3dB; Frequency response: less than 8.5dB dip at 300Hz, less than 3.5dB peak at 3.4kHz; Polar pattern: Cardioid, sensitivity difference between 0° and 90° >4dB, difference between 0° and 180° >10dB. Omni-MIC sensitivity: -40dB ±1.5dB; Frequency response flat from 300Hz to 3.4kHz. Recommended models: Uni-MIC B4015UL403 and Omni-MIC B4015AL-398 from Yilida (IEA), Weifang, Shandong.
2 Structural Design
Key structural design issues involve maintaining Uni-MIC characteristics and the conical pickup beam direction. If hands-free functionality is included, additional considerations for speaker/microphone vibration damping and microphone air-tightness are needed. The direction of the conical beam determines the direction of non-stationary noise suppression. Therefore, the useful signal must be within the beam; otherwise, it risks being suppressed as noise. Product form factor design must fully consider the Uni-MIC's orientation. The design should ensure that after installation into the mic sleeve and the entire housing, the Uni-MIC maintains a sensitivity difference >6dB between 0° and 180°, with sensitivity and frequency characteristics largely unchanged. Microphone vibration damping reduces nonlinear echoes; air-tightness reduces linear echoes, improving the signal-to-echo ratio.
3 Signal Comparison: Raw vs. FM2010 Processed
Signals picked up by the miniature array microphone compared to signals output after FM2010 processing are shown in Figure 4. The sound source is 0.3m from the array, intensity 83dB (SPL). Test signals are: Uni-MIC 0° and 180°, Omni-MIC 0° and 180°, Line Out (Lout) - signal after FM2010 processing for 0° and 180° pickup. Figure 4 shows that signals of the same magnitude can differ by 20dB in output level depending on whether they are inside (0°) or outside (180°) the pickup beam. Meaning, as long as non-stationary noise is outside the beam, it will be suppressed by 20dB relative to the useful signal. The effective range of the SAM conical beam is 2m. The beam angle depends on the Uni-MIC's polar pattern and FM2010 parameter adjustments.
The signals from the Uni-MIC and Omni-MIC are amplified by programmable gain amplifiers (PGA), converted to digital (ADC), and high-pass filtered before being sent to the voice processor (for linear echo cancellation, nonlinear echo cancellation, VAD detection, noise suppression, and mic volume setting). The processed voice data is converted back to analog (DAC) and output single-ended from the Line Out (LOUT) to the MICIP/MIC1N inputs of the TWL3014/16 analog baseband processor. After uplink processing, the signal is sent to the digital baseband processor OMAP733/750. The received GSM signal is demodulated, channel decoded, and decrypted by the OMAP733/750 to obtain the digital audio signal. This is sent to the TWL3014/16 analog baseband processor. After downlink processing, the signal is output to the receiver (HSO) and headset (EARP/EARN). One path goes to the receiver, another to an external power amplifier driving the hands-free speaker. Both signals are also fed into the FM2010's Line Inputs (ADC -> HPF -> Voice data) as echo reference signals for the voice processor.
OMAP733/750 Control Flow for FM2010: Control is achieved via the SHI interface, PWD, RESET, and ANA_IRQ pins. After system power-on: First set PWD high and ANA_IRQ low. After reset, send parameters (clock source, clock frequency, DSP operating speed) to the FM2010, then put the chip into power-saving mode. As shown in Figure 8, the call mode is determined by incoming/outgoing calls or recording operations, waking the FM2010. After reset, based on Handset/Hands-free mode, send corresponding parameters. For Handset Noise Cancel mode, send parameters: number of mics, mic gain, mic volume, echo cancellation parameters, VAD parameters. For Hands-free Conference mode, send: number of mics, mic gain, mic volume, mic inversion, echo cancellation parameters (covering scenarios like low/normal/loud remote speech). Hands-free Personal mode tuning is similar to Handset mode, but echo cancellation parameters require more adjustment. After the call ends, turn off the FM2010 CODEC and set it to power-saving mode. As shown in Figure 9, GSM Phone Audio Test Mode is mainly for testing purposes. In test mode, the FM2010 works in bypass mode; the internal DSP performs no processing. The Uni-MIC input signal is amplified by the PGA and directly output from the LOUT amplifier. Mic amplifier and LOUT amplifier gain parameters can be adjusted online via the SHI interface.