Digital microphones: from specification to finished product. Digital Microphone Blue Microphones Raspberry Studio (Red) Digital Microphones

In recent years, digital MEMS microphones have appeared on the electronics market. Their advantages include: high sensitivity, linearity of frequency response in the operating frequency band, repeatability of parameters and small overall dimensions. The use of a digital MEMS microphone also eliminates the problems associated with analog circuit noise and allows the microphone to be directly connected to the processor. We were interested in these advantages, and we tried to put them into practice.

At the time of the start of work, Second Laboratory LLC had several prototypes of ADMP421 microphones manufactured by Analog Devices. Then we got the SPM0405HD4H-WB digital MEMS microphones from Knowles Electronics. The results of work with the listed microphones became the basis for writing this article.

A digital microphone can be connected to an audio codec that has an appropriate interface [for example, 8–10]. But we were interested in the possibility of directly connecting a digital microphone to a microcontroller. This decision made it possible to abandon the use of an audio codec, which reduced the overall dimensions and further reduced the price of the product. For a preliminary assessment of the expected values of the parameters (required microcontroller performance, power consumption, sensitivity, dynamic range, THD, bandwidth), a small R&D was performed. Based on its results, the final decision was made on circuitry, software and the applied element base.

Connecting digital microphones to microcontrollers

The interface between a microcontroller and a digital microphone is simple, and information on its implementation is available in sufficient volume on the manufacturers' websites and is described in detail by other authors. As a rule, digital microphones have five leads, short description which are given in the table. The electrical and timing parameters of the microphone outputs are given in their specifications.

Table. Pin Descriptions for Digital Microphones

№	Name output	Short description
1	VDD	Microphone Power
2	GND	"Earth"
3	CLK	The input clock signal, synchronously with which the DATA line switches its states
4	DATA	During one half of the CLK cycle, this output is in a high impedance state, and during the second half serves as a conclusion for reading data from the output of the Σ-Δ modulator microphone
5	L/R_Sel	This pin is used to control switching the DATA line. If L/R_Sel connected to VDD, then some time after rising edge detection of CLK signal DATA pin goes high impedance, and after the arrival of the falling front CLK signal, the DATA pin is connected to the output Σ-Δ microphone modulator. If L/R_Sel connected to GND, the edges of the CLK signal, on which the DATA line is switched, change to opposite

To evaluate the required performance of the microcontroller, the ADSP-BF538 EZ KIT Lite development board from Analog Devices was used. Microphones could be connected to this board using SPI or SPORT interfaces. The first of these interfaces is more common, and therefore we used this interface in Slave mode. To generate the clock signal CLK, the hardware timer available in the microcontroller was used. To obtain output samples with a standard sample rate of 16 kHz at a decimation factor of 128, the required CLK clock frequency must be 2.048 MHz. As a clock source for the processor on the development board, a 12.288 MHz oscillator was used, which, when divided by 6, provided the required clock frequency for the digital microphone. To minimize the load on the processor, when receiving initial information from microphones, the DMA transfer mechanism was used.

During the simulation, it was calculated and experimentally verified that the processor must have a performance of about 8 MIPS to process data from a microphone. Evaluation of the required performance allowed us to conclude that it is possible to use a simpler microcontroller with less power consumption. Of the three alternatives (ARM, PIC, MSP430), the MSP430F5418 microcontroller manufactured by Texas Instruments was chosen, which has a minimum power consumption (165 μA / MIPS). In the future, to check the power consumption and work off software the debug board MSP-EXP430F5438 Experimenter Board of the same company was used.

On fig. Figure 1 shows simplified diagrams for connecting digital microphones to the debug boards used in prototyping, which allow you to completely simulate devices for reading, playing or storing data from microphones.

Rice. one. Diagram of connecting a digital microphone to the board: a) ADSP-BF538 EZ KIT Lite; b) MSP-EXP430F5438

The process of converting the input audio signal in the microphone

Rice. 2. Simplified model of a MEMS microphone

Each digital MEMS microphone can be simplified as the model shown in Fig. 2. The input sound vibrations are converted by a MEMS membrane into a weak electrical signal, which is then fed to the input of amplifier A. The pre-amplified signal then passes through an analog low-pass filter, which is necessary to protect against aliasing. The final element of signal processing in a microphone is a 4th order Σ-Δ modulator that converts the input analog signal into a one-bit digital stream. The data bit rate from the output of the Σ-Δ modulator is equal to the frequency of the input clock signal CLK and, as a rule, lies in the range from 1 to 4 MHz.

Measurement of parameters of digital microphones

The following equipment was used for measurements: a CENTER-325 sound level meter, a G3-118 low-frequency signal generator, a S6-11 non-linear distortion meter, a Dialog M-881HV headphone emitter and a PC.

Rice. 3. Frequency response of the ADMP421 microphone

In the time domain, the output of the Σ-Δ modulator is a jumble of 0s and 1s. However, if each high logic level of the microphone output is assigned a value of 1.0, and each low level is assigned a value of –1.0, and then Fourier transform, then we get a spectrogram of the output data from the microphone. On fig. Figures 3 and 4 show the responses of the ADMP421 and SPM0405HD4H-WB microphones to a 1 kHz, 94 dB SPL sine wave audio input. The measurements were carried out for three values of the CLK signal frequency - 512, 1024 and 2048 kHz. (To reduce the volume of the published article, the materials for the frequency of 1024 kHz are not given.) The spectrograms were built on a sample length of 128 × 1024 samples.

Rice. 4. Frequency response of the SPM0405HD4H-WB microphone

Judging by the spectrograms, the quantization noise is shifted outside the audio frequency range and does not affect the input audio signal. In this case, the quantization noise is shifted farther into the region of high frequencies, the higher the sampling frequency of the microphones. Approximately the cutoff frequency from which the increase in the noise level begins can be determined as F clk/one hundred. Although in the specifications for microphones, the operating frequency is normalized approximately in the range from 1 to 3 MHz, but, as the spectrograms show, microphones work normally at lower clock frequencies. This can be very useful when the need arises to reduce the number of calculations on the microcontroller, although, of course, this will also narrow the working audio bandwidth.

It can also be observed that both microphones have a constant component in the output signal (this effect has been eliminated in the latest microphone modifications). Moreover, the level of the constant component is comparable in level with the measured signal. In addition, the value of the DC component at least depends on the supply voltage. This property required the implementation of a recursive algorithm in the microcontroller that eliminates the constant offset.

If we compare microphones in terms of introduced noise levels, it is easy to see that the ADMP421 microphone has best attitude signal to noise ratio compared to the SPM0405HD4H-WB microphone by about 5-6 dB, as well as lower quantization noise.

If we compare the levels of nonlinear distortions, we will see that only second harmonics are present in the spectrograms of both microphones, while the second harmonic amplitude of the Knowles Electronics microphone is significantly lower than that of the Analog Devices microphone. This fact is of particular interest, since both companies standardize only the maximum THD and only for a certain sound pressure level. In reality, this data is not enough. For example, it is not possible to compare the actual THD values of different microphones. In addition, it is now common practice to normalize SOI according to the line input of recording devices, without taking into account the distortions introduced by microphones.

Therefore, in order to assess the nature of the dependence of SOI on the sound pressure level, an experiment was set up, which included the following steps:

Exposing the microphone input to a sinusoidal audio signal with a frequency of 1 kHz and recording one-bit data from the microphone output to flash memory (the sound pressure of the input signal varies from 87.5 to 115 dB SPL in 2.5 dB SPL steps).
Mathematical processing of one-bit data from a microphone using a digital low-pass filter in order to obtain a deterministic digital signal and cut off quantization noise.
Playback of processed digital data on a PC and measurement of the THD signal from the output of a PC sound card using a C6-11 non-linear distortion meter (non-linear distortions introduced by the sound card itself do not exceed 0.1%).
Registration of instrument readings C6-11 for each value of the sound pressure of the input audio signal.

Rice. five. Dependence of SOI microphones on the sound pressure level

The results of the experiment are shown in fig. 5. From the above graph, it follows that at a sound pressure level of less than 97 dB, the SPL of the ADMP421 and SPM0405HD4H-WB THD microphones does not exceed 1% and 0.3%, respectively. At higher sound pressure levels, the THD of the ADMP421 is significantly higher than that of the SPM0405HD4H-WB, and at pressures above 110 dB SPL, both microphones experience a sharp increase in the level of non-linear distortion. In general, it can be concluded that the Knowles Electronics microphone is suitable for use over a wider sound pressure range. It should also be noted that the THD values of microphones given in the documentation are normalized at the maximum sound pressure. The actual THD values at lower sound pressures are much lower and the microphones can be used for high quality audio recording.

However, the ADMP421 microphone has another advantage. This microphone model is practically insensitive to noise on the power bus, even if the latter reach values of 200-300 mV. On fig. 6 shows the case when artificially introduced impulse noise is present in the microphone power bus. Such a case is possible if the audio device operates in a pulsed consumption mode (for example, cyclic recording of data from a microphone to a flash memory when powered from a low-power source).

Rice. 6. Impulse interference in the power supply circuit of microphones

Rice. 7. Timing diagram of the signal from microphones when exposed to impulse noise in the power circuit

On fig. Figure 7 shows the output signal from the microphones, passed through a digital filter with the frequency response shown in fig. 9. No reference sound signal was used to record power interference during the recording process. In order to be able to estimate the noise amplitude from the microphone output, in the upper part of Fig. Figure 7 shows an 80 dB SPL sinusoidal audio signal recorded in the absence of power noise.

Rice. 8. Simplified circuit of digital signal converter Σ-Δ modulator

Rice. nine. Frequency response of software decimator implemented on ADSP-BF538F and MSP430F5438 processors

To eliminate the effect of noise on the power circuits, we had to use a smoothing RC filter.

Digital Microphone Output Processing

To isolate the audio band signal, the microphone data must be filtered and downsampled (typically 50 to 128 times the sample rate of the Σ-Δ modulator). The digital low-pass filter filters out external noise and the microphone's own noise outside the working band ( f >F clk /2M) to protect against aliasing, and also makes it possible to reduce the data repetition rate. On fig. 8 shows one of the possible options for processing a single-bit data stream from a microphone, implemented in software on a DSP or in hardware in audio codecs.

Shown in fig. 8, the sampling rate compression circuit (compressor) lowers the sampling rate due to the fact that from each M filtered signal samples w(mm) is discarded M-1 sample. The input and output of the converter shown in fig. 8 are related by the following expression:

In the software implementation of frequency converters, both FIR and IIR filters can be used as a digital low-pass filter. Developers should be very careful when choosing the type of filter, its length and bit depth, since the performance of the entire system as a whole directly depends on this. A correctly calculated and implemented decimator (frequency converter) in some cases will significantly reduce the cost of production and increase it. specifications. For reference, we note that during the development of the Soroka-1 and Soroka-2 recorders, software decimators that lower the frequency by 64 times (from 1.024 MHz to 16 kHz) were successfully implemented both on the high-performance ADSP-BF538F processor and and on the MSP430F5438 microcontroller with an operating clock frequency of 12.288 MHz. The amplitude-frequency characteristic of the digital low-pass filter, which is part of the implemented decimator, is shown in fig. 9. For complete information on practical matters digital filtering should refer to chapters 6-9 of the book.

As a second option, to convert data from the output of a digital microphone, you can use audio codecs adapted for this, which will significantly reduce the development time of the product. For example, Analog Devices suggests using the ADAU1361 and ADAU1761 codecs, which are equally suitable for the ADMP421 and SPM0405HD4H microphones.

Measuring the frequency response with the required accuracy for the operating frequency band turned out to be a rather difficult task due to the lack of an acoustic emitter in the laboratory with a linear amplitude response in terms of sound pressure. Estimates of the resulting frequency response show its linearity in the operating frequency band with an error of about ±4 dB. Therefore, when assessing the linearity of the frequency response, we considered it right to rely on the declared characteristics of manufacturers and the calculated characteristics of low-frequency filters with a ripple in the passband of less than 1 dB.

MEMS microphones open up new possibilities for audio developers. The process of creating digital audio devices becomes simple in terms of hardware implementation and complex in terms of writing programs for the microcontrollers used. We hope that the information on techniques and parameters given in this article will be of interest to many engineers.

DIGITAL MICROPHONE WITH
FAST AGC AND
SENSITIVITY ADJUSTMENT

MICROPHONE FOR VOICE

STELBERRY M-50 is a completely new solution for sound recording systems and the best voice microphone in its class. High-speed digital signal processing effectively emphasizes the speech range, significantly reducing unwanted sounds in the low and high frequencies.
The STELBERRY M-50 is equipped with a dual digital Automatic Gain Control system with a response time of less than one thousandth of a second.
An external control allows you to adjust the sensitivity of the digital microphone for any operating conditions.

IP MICROPHONE

The STELBERRY M-50 digital microphone is ideal for connecting to the line input of IP cameras, perfectly transmitting the acoustic picture environment.
This application actually makes it a full-fledged IP microphone.
Also, a definite plus this decision, is the ability to install a digital microphone anywhere, regardless of where the IP camera is installed.

Comparison Chart for STELBERRY M Series Omnidirectional Microphone Models

Characteristics and parameters of omnidirectional microphones
Fixed sensitivity value	✔	➖	✔	➖	➖	➖	➖	➖	➖	➖	➖
Adjustable sensitivity	➖	✔	➖	✔	✔	✔	✔	✔	✔	✔	✔
Sensitivity setting method	Resistor	Resistor	Resistor	Resistor	Resistor	Resistor	Resistor	Resistor	Resistor	Joystick	Joystick
AGC - automatic gain control	➖	➖	✔	✔	✔	✔	✔	✔	✔	✔	✔
Possibility to change AGC speed	➖	➖	➖	➖	➖	➖	➖	➖	➖	➖	✔
Ability to disable AGC	➖	➖	➖	➖	➖	✔	✔	✔	✔	✔	✔
Switchable low-impedance output for audio inputs of a number of IP cameras	➖	➖	➖	➖	➖	✔	✔	✔	✔	➖	➖
Maximum Bandwidth (Hz)	100...6100	100...7200	100...8300	100...9200	270...4000	80...16000	80...16000	270...4000	270...4000	80...16000	80...16000
Bandwidth adjustable	➖	➖	➖	➖	➖	➖	➖	➖	➖	✔	✔
Ability to cut a frequency selected from a set of frequencies	➖	➖	➖	➖	➖	➖	➖	➖	➖	➖	✔
Signal to noise ratio (dB)	48	48	48	48	48	63	63	63	63	67	67
Acoustic range (meters)	8	10	10	12	20	20	20	20	20	25	25
Sound processing	➖	➖	analog	analog	digital	analog	analog	digital	digital	digital	digital
Settings lock	➖	➖	➖	➖	➖	➖	➖	➖	➖	✔	✔
Output Level (V)	1	1	1	1	1	1	1	1	1	1	1
Maximum line length (meters)	300	300	300	300	300	300	300	300	300	300	300
Rated supply voltage (V)	12	12	12	12	12	12	12	12	12	12	12
Current consumption (mA)	3	3	8	8	25	8	8	25	25	25	25
Detachable cable connection with microphone	➖	➖	➖	➖	➖	✔	✔	✔	✔	✔	✔
Anti-vandal housing	➖	➖	➖	➖	➖	➖	✔	➖	✔	➖	➖

For reliable operation of the STELBERRY M-50 digital microphone, high-quality power supply with low ripple is required. The best solution is to use the STELBERRY MX-225 PoE splitter, which has an output voltage filtering system. Also, STELBERRY MX-225 has built-in protection against short circuit at the output or exceeding the maximum allowable current.

The STELBERRY MX-225 miniature pass-through PoE splitter is installed in the cut of the cable that connects the IP camera and the switch and can be glued to any surface or hidden inside the box through which the cable is laid. To power the STELBERRY M-50 digital microphone, the PoE splitter is equipped with self-locking connectors that ensure reliable contact.

FAST DIGITAL
SIGNAL PROCESSOR

A miniature digital signal processor (DSP) digitizes the audio signal from the sound capsule at a sampling rate of 44100 Hz and 16-bit sampling.
A distinctive feature of the processor is the presence of 2-speed AGC, providing lightning-fast automatic gain control, both at the input and at the output of the device.
6 digital filters processors process the signal in such a way that only the speech range remains at the line output.
A precise built-in preamplifier guarantees a high signal-to-noise ratio.

CONTROL PROCESSOR
DIGITAL MICROPHONE

The central control processor of the STELBERRY M-50 digital microphone provides microphone gain control and control of signal processing parameters.
The processor ensures that the microphone quickly enters the operating mode after power is applied, thanks to a high-speed exchange line with the signal processor.

WIND PROTECTION FOR DIGITAL MICROPHONE
STELBERRY M-50

For perfect sound transmission, the digital microphone is equipped with a wind protection filter.
By eliminating the wind component, the acoustic material filter cuts off unwanted sounds that occur when wind currents collide with a sensitive membrane, resulting in a crystal-clear sound.
The presence of wind protection allowed us to create an effective microphone for voice.

OPTIMIZING THE MICROPHONE UNDER THE SPEECH
RANGE

The bandwidth of the STELBERRY M-50 digital microphone is tuned to the frequency range of human speech and lies within 270...4000 Hz.
This bandwidth achieves excellent speech intelligibility, regardless of extraneous noise sources.
Signal processing is carried out by six digital high-speed filters, which guarantees a high slope of the amplitude-frequency response in the low and high frequencies.

DOUBLE AGC SYSTEM

The microphone is equipped with two digital high speed Automatic Gain Controls (AGC).
The first AGC adjusts the gain at the microphone input, immediately after the signal from the capsule is digitized, and the response time to changes in sound level is less than 1/1000 of a second.
This allows you to respond to any, even the most insignificant changes in the sound environment of the environment.
A second AGC processes the signal at the microphone output, reliably maintaining a stable output level. The response time of the output AGC system is also less than 1/1000 of a second.

COMPARISON OF DIGITAL AUTOMATIC GAIN CONTROL (AGC) WITH ANALOGUE AGC

Digital microphone Stelberry M-50 with adjustable gain, built on a specialized processor. The microphone operation process is an analog-to-digital conversion of the microphone capsule signal, subsequent digital filtering of the received signal, and inverse digital-to-analog conversion. The digital filters of the sensitive M-50 microphone are tuned to the range of human speech. Sound frequencies outside the frequency range of 270...4000 Hz are significantly attenuated by the microphone. The very fast AGC (automatic gain control) of the digital microphone makes it comfortable to use in a room with sudden changes in the volume of sound or human speech.

The M-50 digital microphone is well suited as a voice recording microphone for projects that focus on recording conversations. Ideal as an external high-sensitivity microphone for video cameras and audio recorders that are sensitive to the input signal level and do not have own funds sound filtering.

The sensitive Stelberry M-50 microphone is used as an external microphone for various surveillance cameras, including IP cameras, for audio monitoring of premises, as a highly sensitive microphone for voice recording in call recording systems and speech recognition systems.

The location of the digital microphone with AGC Stelberry M-50 in the room

When the M-50 microphone is placed in the corner of the room and the maximum sensitivity of the microphone is set, the comfortable listening area will correspond to a quarter circle area of 50m². With further distance from the microphone, the level of its output signal will gradually weaken up to the acoustic audibility limit of 20 meters.

Connecting STELBERRY M-50 Digital AGC Microphone to IP Camera

The M-50 digital microphone connects directly to the camcorder's audio line input. Connecting a microphone to the camera is done in this way. The yellow wire of the M-50 microphone, to the 3.5mm jack input connector of the camera, is connected to the end (center) and to the ring contact of the connector (Check in the camera manual.). If the camera or IP camera uses an RCA ("tulip") connector for audio input, then - to the center pin of the RCA connector. The black wire of the M-50 digital microphone is connected to the common (body) contact of the "Jack-3.5mm" connector (or to the ring external contact of the RCA connector), and to the "negative" common wire of the stabilized power supply. The red wire of the microphone is connected to the "positive" wire of the stabilized power supply.

Directional pattern of a digital microphone with AGC and gain control Stelberry M-50

The Stelberry M-50 digital speech frequency microphone is omnidirectional and has a circular pickup pattern with a slight attenuation of the microphone sensitivity from the sensitivity control. The radiation pattern is given for the microphone capsule used in the microphone, taking into account the influence of the microphone body.

Microphones

Description STELBERRY M-50

STELBERRY M-50 is a completely new solution for sound recording systems and the best voice microphone in its class. High-speed digital signal processing effectively emphasizes the speech range, significantly reducing unwanted sounds in the low and high frequencies. The microphone is equipped with a dual digital Automatic Gain Control system with a response speed of less than one thousandth of a second. An external control allows you to adjust the sensitivity of the digital microphone for any operating conditions.

IP microphone

The digital microphone is ideal for connecting to the line input of IP cameras, perfectly transmitting the acoustic picture of the environment. This application actually makes it a full-fledged IP microphone. Also, the undoubted advantage of this solution is the ability to install a digital microphone anywhere, regardless of where the IP camera is installed.

High speed digital signal processor

A miniature digital signal processor (DSP) digitizes the audio signal from the sound capsule at a sampling rate of 44100 Hz and 16-bit sampling. A distinctive feature of the processor is the presence of 2-speed AGC, providing lightning-fast automatic gain control, both at the input and at the output of the device. 6 digital filters of the processor process the signal in such a way that only the speech range remains at the line output. A precise built-in preamplifier guarantees a high signal-to-noise ratio.

Digital microphone control processor

The central control processor of the digital microphone provides microphone gain control and control of signal processing parameters. The processor ensures that the microphone quickly enters the operating mode after power is applied, thanks to a high-speed exchange line with the signal processor.

Wind protection for digital microphone

For perfect sound transmission, the digital microphone is equipped with a wind protection filter. By eliminating the wind component, the acoustic material filter cuts off unwanted sounds that occur when wind currents collide with a sensitive membrane, resulting in a crystal-clear sound. The presence of wind protection allowed us to create an effective microphone for voice.

Optimizing the microphone for the speech range

The bandwidth of the digital microphone is tuned to the frequency range of human speech and lies within 270...4000 Hz. This bandwidth achieves excellent speech intelligibility, regardless of extraneous noise sources. Signal processing is carried out by six digital high-speed filters, which guarantees a high slope of the amplitude-frequency response in the low and high frequencies.

Dual AGC system

The microphone is equipped with two digital high speed Automatic Gain Controls (AGC). The first AGC adjusts the gain at the microphone input, immediately after the signal from the capsule is digitized, and the response time to changes in sound level is less than 1/1000 of a second. This allows you to respond to any, even the most minor changes sound environment. A second AGC processes the signal at the microphone output, reliably maintaining a stable output level. The response time of the output AGC system is also less than 1/1000 of a second.

Convenient adjustment

A convenient location for adjusting the sensitivity makes it easy to adjust the microphone gain. A feature of a highly sensitive microphone is that the gain adjustment occurs before AGC processing begins. This makes it easy to achieve the desired sound quality. The bandwidth of the microphone has been chosen to pass voice frequencies while eliminating unwanted sounds from high-frequency sources.

Specifications STELBERRY M-50

Unit: 1 piece
Dimensions (mm): 10x10x52
Weight (kg): 0.01
Acoustic range: up to 20 meters
EMI protection: yes
Wind protection: Acoustic foam
bandwidth (after digital processing): 270...4000 Hz
Line length: up to 300 meters
Gain adjustment range: 350 times
Number of digital AGC: 2
"Angle of attack" input AGC: 0.7 ms
"Angle of attack" output AGC: 0.7 ms
Digital Low Pass Filtering: 2 1st order filters
Digital High Pass Filtering: 3 2nd order filters
Signal to noise ratio: 38 dB
Discretization: 16 bits
Sampling frequency: 44100 Hz
Housing: aluminum
Power: 7.5...16 Volts
Consumption: 20 mA
Dimensions: Ø10x52 mm
Weight: 10 grams

It might be useful to read: