MEMS microphone electrical implementation

AN558
MEMS microphone electrical implementation
About this document
Scope and purpose
This document covers important information about analog and digital microphone electrical interfaces and
signal requirements. It also imparts information about important parameters like Acoustic Overload Point
(AOP), Signal to Noise Ratio (SNR), dynamic range, sensitivity and bandwidth.
Intended audience
Infineon XENSIVTM MEMS microphone customers.
Table of contents
About this document . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1
Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .2
2
2.1
2.2
2.3
Electrical implementation of analog microphones . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .3
Analog microphone interfaces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
Device microphone signal path requirements - dynamic range . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
Device microphone signal path requirements - bandwidth . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .6
3
3.1
3.2
3.3
3.4
Electrical implementation of digital microphones . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
PDM interface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
Signal connections . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
Digital clock and bandwidth of a PDM interface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
L/R channel multiplexing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
3.5
3.6
3.7
3.8
3.9
3.10
Digital timing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
Digital I/O levels . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
Device microphone signal path requirements - dynamic range . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
Device microphone signal path requirements - bandwidth . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
Current consumption . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
Multi-mode PDM microphones . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .14
4
4.1
Electromagnetic compatibility (EMC) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
RF disturbance mitigation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
5
Power supply disturbances . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
6
Electrostatic discharges (ESD) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .17
7
Wind . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
8
Further information . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
Disclaimer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
Application Note
www.infineon.com
Please read the Important Notice and Warnings at the end of this document
V1.0
2018-01-16
MEMS microphone electrical implementation
1 Abstract
1
Abstract
Electrical implementation is a critical microphone success factor, alongside acoustical and mechanical
implementation. Microphones need a high-performance output signal path and a working environment that
enables getting the best possible signal quality out of the system.
Optimizing signal quality includes, for example:
•
A high-performance signal path that is able to carry the microphone output signal without degradation in
noise level, frequency response, phase performance, dynamic range, or other key parameters
• A clean and steady power source
• A clean, steady, low-impedance ground
• Well protected, clean signal lines
- Protection from external disturbances, such as conducted and radiated RF signals
- Separation from other data traces to avoid crosstalk
• Isolation from other noisy electrical systems in the device
• A reasonable electromagnetic radiation environment
•
A high-quality clock signal (for digital microphones)
In this application note, we go through key information and guidelines for the electrical implementation of
MEMS microphones into systems such as smartphones, smart speakers, IoT devices and laptops. Some of the
information is divided into two sections based on the output format of the microphone, analog or digital.
Analog signals are significantly more susceptible to disturbances than digital signals. Therefore, the tasks of
implementing analog and digital microphones into a device differ significantly. Typically, it takes more time and
effort to achieve a high performance level with analog microphones than with digital microphones. The design
of the circuitry, board layout and wiring are much more critical tasks in the analog domain than they are in
digital. Achieving a satisfactory result with analog microphones requires experience, careful design work,
prototyping and often several design iterations. With digital microphones, the rules are more straightforward.
Due to the easier implementation, the popularity of digital interface microphones has risen, especially in bigger
devices and devices in which the electromagnetic environment is especially hostile. For example, laptop
microphone traces tend to be long and they run next to disturbance sources so digital microphones are used
practically exclusively in that product category.
Application Note
2
V1.0
2018-01-16
MEMS microphone electrical implementation
2 Electrical implementation of analog microphones
2
Electrical implementation of analog microphones
2.1
Analog microphone interfaces
There are two commonly used interfaces for analog MEMS microphones: single-ended and differential. Singleended has dominated the market until recent years but differential has recently gained more popularity due to
increased performance requirements (for example, dynamic range) and the need for better signal robustness
against disturbances.
The single ended interface is a simple and low cost solution. The
microphone requires only three pins: VDD, signal and ground. Singleended interfaces do not offer significant protection against disturbances,
such as radio frequency interference, cross talk from other signal lines, or
differences in ground voltages between the sending and receiving ends.
A differential interface transmits the microphone output signal divided
into two complementary signal lines with opposite polarities. The effective
signal Vout at the receiving end is the difference between the two signals.
Therefore, differential interface offers a 6dB higher dynamic range than Figure 1
single-ended (assuming same interface rail voltage and same sensitivity for
both MEMS sensors). In other words, a differential interface doubles the
dynamic range of the interface. The key benefit in a differential interface,
what comes to interference, is that disturbances in the two complementary
signal lines are (ideally) the same and they get canceled from the signal
when received with a differential amplifier. The electromagnetic field
around an ideal differential line is zero, so a well-designed differential
interface also reduces disturbances emitted by the microphone output
signals to other traces. Good disturbance rejection performance requires
that the two signal traces are well matched (balanced impedance, matched
trace layout).
Figure 2
Microphone
in
single
ended
interface
Microphone
differential
interface
in
Figure 3 shows two examples of analog microphone interfaces. The single-ended interface on the left is an
interface to an audio codec. The differential interface is an interface with a differential amplifier. Both circuits
include a 0.1µF capacitor (a higher capacitance can be used, up to 10µF) to filter disturbances from the supply
voltage line. Additional smaller value capacitors (50 - 100pF) may be needed on the power supply line to filter
out RF disturbances. Typically, the power supply filtering capacitors should be placed physically as close to the
microphone as possible, the smallest value one closest to the microphone. There are also series capacitors
(>1µF) in the output lines to filter possible DC component out of the signal.
Figure 3
Single ended (left) and differential (right) analog microphone interface examples
See more details on microphone signal protection in section ‘RF disturbance mitigation’.
Note:
A differential interface should be chosen instead of a single-ended when high disturbance immunity
or high dynamic range is needed.
Application Note
3
V1.0
2018-01-16
MEMS microphone electrical implementation
2 Electrical implementation of analog microphones
2.2
Device microphone signal path requirements - dynamic range
A key success factor in high performance analog microphone interfaces is making sure that the dynamic range
of the signal path does not limit the performance of the microphone. The maximum signal swing of an analog
microphone output can be calculated using the sensitivity (output voltage with a 94dBSPL (1Pa) rms acoustic
reference signal input) and the acoustic overload point (AOP) of the microphone. The AOP of a microphone can
be specified either as a peak value or as an rms value. For output voltages, we concentrate on peak or peak-topeak values when calculating maximum signals levels.
The rms sensitivity of an analog microphone can be calculated with the equation below, where SensitivitymV/
Pa(rms) is the output voltage of the microphone with a 94dBSPLrms reference sound input.
The rms output voltage measured with the 94dBSPLrms sound input can be derived from the equation above:
If the AOP sound pressure level is specified for the microphone as a peak value (AOPpeak), the peak output
voltage of the microphone at the AOP can be obtained by calculating the difference (in decibels) between
AOPdBSPL(peak) and 94dBSPLrms and adding it to Voutput(rms)@94dBSPL(rms).
If the AOPSPL is given as an rms value (AOPrms), the peak AOP output voltage can be calculated by adding +3dB
to Voutput(peak)@AOP(peak) (see Figure 4).
Figure 4
Difference between AOP (peak) and AOP(rms) on the microphone sensitivity
The peak-to-peak signal voltage swing can be calculated by doubling the peak value. For example, for a
microphone with Sensitivity -38dBV and AOPpeak 130dBSPL, the maximum peak-to-peak output voltage is:
Application Note
4
V1.0
2018-01-16
MEMS microphone electrical implementation
2 Electrical implementation of analog microphones
This peak-to-peak voltage swing might barely be compatible with a single-ended interface that has a 1.8V rail
voltage and maximum signal swing of approximately 1.6V (from 0.1V to 1.7V; the practical maximum and
minimum signal values are typically about 100mV from the theoretical rail voltages). However, if the sensitivity
tolerances for the microphones are ±1dB, the sensitivities of the outlier microphones could be -37dBV. The
resulting Voutput(p-p)@AOP(peak) is 1.8V and the interface runs out of dynamic range.
If the AOPdBSPL is specified as an rms value, the peak AOP signal level is 3dB higher. For example, if the rms AOP
is 130dBSPL, the peak AOP is 133dBSPL. If the sensitivity of the microphone is -38dBV like in the example above,
the maximum peak-to-peak voltage swing increases from 1.6V to over 2.2V. This is no longer compatible with a
1.8V single-ended interface.
In example case (1) (see Figure 5) below, the sound pressure coming into the microphone is too high for the
microphone and/or the microphone signal chain so the signal gets distorted, clipped. The problem can be
mitigated by:
(a) lowering microphone sensitivity
(b) increasing signal path rail voltage or
(c) switching from a single-ended to differential interface
It should be noted that lowering the sensitivity of the microphone may affect important parameters such as
SNR and the microphone system’s ability to capture low level / distant sounds.
Figure 5
Effect of high SPL input on the microphone output
The noise floor of the whole microphone signal chain must be low enough to support the noise performance of
the microphone. In some cases, some parts of the microphone signal path, such as an input of a codec, may not
support very low signal levels. Signal levels may be low because, for example, the sensitivity of the microphone
is low and the incoming sound pressure level is low (for example, if the sound source is relatively quiet and/or
far away). The result may be that the microphone signal gets buried in the noise floor of the signal path.
In example case (2) (see Figure 6) below, the incoming sound is very quiet (roughly equivalent of a person
talking from an 8-meter distance) and the SNR of the output of the microphone system is very low due to the
noisiness of the signal path. The problem can be mitigated by:
(a) choosing a microphone with a higher sensitivity (and preferably the same noise floor)
(b) amplifying the microphone output signal before the noisy parts of the signal chain or
(c) reducing the noise floor of the signal chain by choosing lower-noise components and making sure
disturbances do not contaminate the lines
Figure 6
Effect of low SPL input on the microphone output
Application Note
5
V1.0
2018-01-16
MEMS microphone electrical implementation
3 Electrical implementation of digital microphones
In a smartphone, low microphone sensitivity may make it more difficult to pass send distortion tests that are
part of the type approval process. On the other hand, increasing the signal level may cause signal headroom
problems (see Figure 5). As can be seen from the examples here, the sensitivity of the microphones used in a
system is often a compromise between high and low SPL capturing capabilities.
2.3
Device microphone signal path requirements - bandwidth
The DC filtering capacitors in the microphone output signal path must be sized so that they don’t limit signal
bandwidth. ≥1µF components are typically good for this purpose. Also, all the other components in the signal
chain must support the microphone output bandwidth. This is especially noteworthy if the intended bandwidth
goes beyond 20kHz (ultrasound).
3
Electrical implementation of digital microphones
The same guidelines on protecting and filtering signal connections apply to digital microphones as for analog
ones. However, due to the robustness against disturbances of digital signals, there is typically less need for
protecting and cleaning the microphone output signal. In many cases, digital microphones enable a reduction
in the amount of design work, prototyping, redesigns and filtering components while still achieving a similar or
better signal quality as an analog microphone. Digital interfaces are especially beneficial in challenging
electromagnetic environments in wirelessly connected devices such as smartphones, tablets, smart speakers
and IoT devices. Digital signals also enable using longer trace lengths without problems so they are well suited
for large devices, such as laptops, that can have long signal connections.
A drawback of digital interfaces is the higher current consumption of the microphones and the interface as
compared to analog interfaces. However, the current consumption for the whole system is not necessarily
higher since the analog to digital conversion has to be done at some point in the signal chain also for analog
microphones.
Even though implementing digital microphones is in many cases easier than analog microphones, the system
designer still has to know what she or he is doing. Analog signals (Vsupply, ground) must be kept stable and
clean and digital signal connections must be implemented properly to ensure seamless operation.
3.1
PDM interface
The most common digital interface in MEMS microphones is PDM (Pulse Density Modulation). It is a 1-bit
interface. A PDM interface is simpler and cheaper than other digital alternatives. A key simplicity factor is that a
PDM interface does not require having a decimator in the microphone. This saves chip area, cost and current
consumption in the microphone. The delay caused by the analog to digital conversion is very small in PDM
microphones.
A PDM interface consists of two interface signals: Clock and Data. The L/R (left/right) select pins enable using
two microphones in the same data line by connecting the pins to either ground or Vsupply. With supply voltage
and ground, the total microphone pin count is 5.
3.2
Signal connections
As explained in Electrical implementation of analog microphones, signal connection circuit and trace designs
are critical for analog microphones. They are equally important for digital microphones, but for different
reasons. All traces on a circuit board have impedance. The very high frequencies (up to hundreds of megahertz)
caused by sharp digital signal edges in the clock and data signals of digital microphones are affected
significantly by those impedance. Also, the outputs and inputs of the components that send and receive signals
to and from each other play a big role in maintaining the signal quality. The principles for designing and
executing signal connections that do not degrade digital signal quality are relatively straightforward. The key
principles are as follows.
The output of the component sending a signal (source) has to match the signal line it drives. A mismatch can
cause issues like overshoot, undershoot, reflections and ringing. These can lead to unpredictable signal levels
and bit errors as well as emitted disturbances that can affect adjacent systems. Bit errors lower the
performance of the microphone system. Ringing, overshoot and undershoot can cause minimum and maximum
Application Note
6
V1.0
2018-01-16
MEMS microphone electrical implementation
3 Electrical implementation of digital microphones
signal levels to be significantly lower or higher than anticipated. The resulting levels may violate the absolute
maximum ratings specified for the microphone (or other parts of the system).
To avoid problems, the impedance of the clock and data traces should be controlled. The impedance of a trace
depends on its dimensions and the materials used in the circuit board. Circuit design tools can be used to
design controlled impedance traces, for example between 50 and 100 Ohms.
Source Termination Resistors
• Source termination resistors are used to match the impedance of the source and the trace
•
- The resistors are placed in series with the trace, close to the source
-
The component values of the termination resistors depend on the properties of the trace they are on;
typical values range from 50 to 100 Ohms
•
Clock line source termination resistors (RTC) (1) keep the signal edges clean of overshoot and ringing and
help avoid interface timing issues, emitted interference and increased current consumption
•
Termination resistors are added also on the data lines (RTD) (2) to avoid timing issues and bit errors
Filtering Capacitors
• Vsupply interference filtering capacitors (CPS) (3); a typical value is 0.1µF but up to 10µF is possible
•
RF disturbance filtering capacitors (CRF) (4) for Vsupply
- Several may be needed, each to filter out a specific interference frequency
- The capacitors are typically small value, 10 – 500pF
• The smallest value filtering capacitor should be placed closest to the microphone
• The trace lengths connecting the capacitors to the power plane should be as short as possible (5); avoiding
traces altogether and using vias directly from pads is recommended
• Inductance should be minimized in the capacitor traces (5) to maximize filtering performance
• Some capacitor types (e.g. those based on class 2 dielectrics) can generate acoustic noise
- They should not be used in close proximity to microphones
Figure 7
Example of digital microphone and audio codec interface
Other Guidelines
•
A series resistor or ferrite (R/F) (6) can be added to the power supply line for each microphone to suppress
or filter the amount of high frequency electromagnetic interference (EMI) noise
•
Traces from the L/R pins on the microphones to Vdd and Ground (7) should be short
•
Microphone grounds (8) should be connected to the circuit board ground plane with vias that are directly
on the microphone ground pad; traces should be avoided
• Microphone data and clock lines should be kept separate from high-speed transmission lines
Application Note
7
V1.0
2018-01-16
MEMS microphone electrical implementation
3 Electrical implementation of digital microphones
In the configuration shown above, both microphones are powered on or switched to other power modes (e.g.
Normal mode <-> Low power mode, controlled by the clock frequency) simultaneously. Switches or AND-gates
can be used together with GPIO outputs (general-purpose IO) of the codec to control the power and clock to
each microphone individually.
3.3
Digital clock and bandwidth of a PDM interface
Digital microphones need a clock signal from the host device. The clock sets the timing for the binary data
output of the microphone. The allowed clock frequencies are specified in component data sheets.
The bandwidth of a PDM interface is limited by its clock frequency because of the increase in quantization noise
at higher microphone output signal frequencies. A high clock frequency enables noise shaping, i.e. pushing the
noise out of the audio frequencies to higher frequencies that are made available by the higher clock frequency.
Oversampling ratio (OSR, decimation factor) is the ratio of the PDM interface clock frequency to the baseband
sampling rate. A typical OSR is 64 but there is no defined standard. According to the Nyquist theorem, the
sampling rate must be twice the audio bandwidth (BW, the highest used audio frequency). The OSR and the
bandwidth determine the required clock frequency:
The minimum clock frequency needed for the full audio band (20Hz – 20kHz) is 2.56MHz (with OSR 64). The
commonly used 2.4MHz clock enables an 18.75kHz audio bandwidth (with OSR 64).
Some standard audio sampling rates exist and example can be found in Table 4.
Using high ultrasonic frequencies requires the PDM interface clock frequency to be significantly higher than for
audible frequencies. A 40kHz bandwidth (with OSR 64) requires a 5.12MHz or higher clock frequency.
3.4
L/R channel multiplexing
Left/right channel multiplexing is done by
using the rising and falling clock signal
edges to drive two microphones (channels).
The multiplexing works so that at on each
clock edge one microphone is transmitting
and the other is in a high-impedance state.
For example, on a rising clock signal, DataL
channel asserts (writes) data onto the data
line and DataR goes into a high-impedance
(HiZ) state. Similarly, on a falling clock
signal, channel DataR asserts data and DataL
goes into a high-impedance state.
When in the high-impedance state, the
microphone is electrically invisible to the
output data line. This allows each
microphone to drive the contents of the data
line while the other one is waiting quietly for
its turn (in HiZ state).
Figure 8
Note:
L/R channel multiplexing for using two
microphones on a single DATA line
The clock frequency of a PDM interface must be high enough not to limit the frequency band of the
microphone. For a system with OSR = 64 the clock frequency must be 128 times the desired
microphone output bandwidth.
Application Note
8
V1.0
2018-01-16
MEMS microphone electrical implementation
3 Electrical implementation of digital microphones
3.5
Digital timing
When
implementing
different
microphone models or microphones
from several suppliers into the same
device, it is very important to make sure
their interface properties allow them to
work together seamlessly. For example,
when using two PDM microphones on
one data line, their timing parameters
must be compatible. Naturally, also the
timings of the microphone system and
the microphones must match.
Table 1 shows an example set of timing
parameters from Infineon’s IM69D130
microphone data sheet.
Figure 9
Table 1
Digital timing for implementing two microphones
on the DATA line
Common digital timing parameters
Parameter
Symbol
Values
Min.
PDM clock
frequency
fclock
Clock duty cycle
Delay time for
DATA driven
Delay time for
DATA high-Z1)
Application Note
Max.
0.35
3.3
40
60
48
52
Clock rise/fall
time
Unit
Note or
Test
Condition
Defines the sampling rate and the rate at
which bits are transmitted on the data line
MHz
%
fclock<2.75
MHz
fclock≥2.75
MHz
13
tHZ
40
5
80
The ratio of clock high to one clock period,
given as a percentage (%).
Different duty cycles may be specified for
different clock frequencies.
The maximum time it takes for the clock
signal to rise and fall.
The exact voltage levels from which the
time is measured may be specified
together with the rise / fall time.
ns
tDD
Description
ns
Delay time Delay time from when the clock edge is at
for CLOCK 50% of supply voltage (0.5 x VDD) to when
edge (0.5 x data is driven on the data line.
VDD) to
DATA
driven
ns
Delay time
for CLOCK
edge (0.5 x
VDD) to
DATA high
impedance
state
30
9
Delay time from when the clock edge is at
0.5 x VDD to when the data output of the
microphone switches into a high
impedance (HiZ) state.
In the high-impedance state the
microphone allows the other microphone
to drive the data line.
V1.0
2018-01-16
MEMS microphone electrical implementation
3 Electrical implementation of digital microphones
Table 1
Common digital timing parameters (continued)
Parameter
Symbol
Values
Min.
Delay time for
DATA valid2)
Unit
Max.
tDV
100
ns
Note or
Test
Condition
Description
Delay time
for CLOCK
edge (0.5 x
VDD) to
DATA valid
(<0.3 x VDD
Delay time from when the clock edge is at
0.5 x VDD to when the data driven by the
microphone on the Data line is valid
(accurately readable).
The voltage levels for valid 0 and 1 should
be specified; e.g. 0 is < 0.30 x VDD and 1 is >
0.70 x VDD.
or >0.7 x VDD
3.6
Digital I/O levels
In addition to timing parameters, also the IO (input / output) voltage levels of microphones must be specified to
make sure the microphones work as intended within the system together with other microphones. Choosing
the correct levels helps minimize the amount of data errors (cases where the output of the microphone is
interpreted as a 1 when the microphone is actually giving 0 as an output, or vice versa).
The properties of the Data line affect the functionality
of the interface. Therefore, the maximum output load
capacitance on the Data line (Cload) that the
microphone is capable of driving is typically also
specified in microphone data sheets. The data line in
the device should be compatible with the required max
Cload.
The Data pin of the microphone is typically specified to
be in a high-impedance state when the microphone is
in standby mode.
Figure 10
Hysteresis in microphone logic level
shift
The hysteresis in microphone logic level shifts from 0 to 1 and back is shown in the illustration on the right. Key
IO parameters for PDM microphones are listed in the table below. The table on the next page shows an example
set of IO parameters from Infineon’s IM69D130 microphone data sheet.
Table 2
Common IO level parameters
Parameter
Symbol
Values
Min.
Unit
Note or
Test
Conditio
n
Max.
Description
Input logic level
VIL
-0.3
0.35 x
VDD
V
The range of voltages the microphone
interprets as a 0 (logic low).
Input logic high
level
VIH
0.65 x
VDD
VDD
V
The range of voltages the microphone
interprets as a 1 (logic high).
Hysterisis width
Vhys
0.1 x
VDD
0.29 x
VDD
V
The difference between the low->high
and high->low thresholds (VLH / VHL)
(see drawing).
Application Note
+ 0.4
10
V1.0
2018-01-16
MEMS microphone electrical implementation
3 Electrical implementation of digital microphones
Table 2
Common IO level parameters (continued)
Parameter
Symbol
Values
Min.
Output logic low
level
VOL
Output logic high
level
VOH
Output load
capacitance on
DATA
Cload
3.7
Unit
Note or
Test
Conditio
n
Description
V
Iout=2mA
The range of microphone output logic
voltages that represent a 0 (logic low).
The output current Iout should also be
specified.
V
Iout=2mA
The range of microphone output logic
voltages that represent a 1 (logic high)
The output current Iout should also be
specified
Max.
0.3 x
VDD
0.7 x
VDD
200
pF
Device microphone signal path requirements - dynamic range
The output of a PDM MEMS microphone is converted into an analog signal for human ears (by low-pass filtering)
or a PCM (pulse code modulated) digital signal that is passed on in the device system. For example, DSP systems
accept PCM as their input (but cannot handle PDM).
MICROPHONE
94dBSPL
Cal. point =
94dBSPL
25dBSPL
Figure 11
Application Note
11
AOP
Noise
Floor
0dBFS
Dynamic Range = 105dB
130dBSPL
SNR
Conversion to PCM is done with a decimator that uses digital
filtering downsample the signal according to the oversampling
rate. The sample rate reduction increases the word length of
the digital signal so the resulting baseband rate PCM signal is
no longer a 1-bit signal. Typical resulting word lengths are 16,
20 or 24 bits. The aliasing to audible frequencies of the noise at
ultrasonic frequencies in the PDM signal must be prevented by
using decimation filters. The decimator system can be included
in the microphone for PCM output. PCM/I2S microphones have
not gained as much popularity as microphones with PDM
outputs.
Two key parameters of a microphone are Signal to Noise Ratio
(SNR) and Acoustic Overload Point (AOP). These parameters
reflect the quietest sound a microphone can handle (SNR) and
the loudest sound a microphone can detect while maintaining
a reasonable distortion level (AOP; the sound pressure level at
which THD rises above 10%). By combining these two
parameters we can determine the entire range of sound levels
which a microphone can detect, from the noise floor to the
overload point; see illustration on the right.
-36dBFS
-105dBFS
Dynamic range and SNR
representation of Infineon
digital microphone
V1.0
2018-01-16
MEMS microphone electrical implementation
3 Electrical implementation of digital microphones
For PCM-based audio system components, such as ADCs and codecs, Dynamic Range and SNR are measured in
a significantly different way to a microphone SNR. Microphone SNR is defined as the difference (in dB[A])
between the microphone output when the acoustic input is a 1kHz sine wave at 94dBSPL and when there is no
acoustic input to the microphone (noise floor). This is because microphone SNR is a measurement of the
audibility of the microphone self-noise, so it is referred to a defined acoustic “real world” signal.
ADC or codec SNR is generally defined as the difference (in dB) between the maximum input signal level and the
minimum input signal level. This performance is usually reflected in the bit depth of the digital encoding
scheme used, typically 16-bit, 20-bit or 24-bit. This means that the SNRs of audio components are actually more
closely related to microphone Dynamic Ranges than microphone SNRs.
It is necessary to select audio components and a digital encoding scheme which have equivalent or greater
SNR / Dynamic Range than the Dynamic Range of the microphone being used. If not, the noise floor of these
audio components will dominate the system and the full microphone performance will not be realized.
The number of bits (N = word length) determines the dynamic range (DR) of PCM-based ADCs and codecs:
For example, inserting 16 as the N in the equation above for a 16-bit system gives a dynamic range of about
98dB. In practice, dithering reduces the SNR by about 4 dB (dithering is a deliberate addition of noise used to
randomize quantization error and thereby linearize the system and eliminate noise modulation). For easy
estimation the equation above can be reduced to: DR[dB] = N * 6 (not taking dithering into account).
All digital signals are measured in dBFS (dB relative to full scale). Full scale is the maximum number which can
be represented in the digital numbering system. In PDM, this is represented by 100% 1’s in the output. In a
digital MEMS microphone, the AOP typically coincides approximately with 0dBFS (full scale). All other output
levels are specified as –xdBFS, signifying that they are lower than full scale. The Dynamic Range of the codec
specifies the lowest dBFS level which it can pass through.
Note:
25dBSPL
Noise
Floor
-36dBFS
-105dBFS
DIG. ENCODING
Full scale
0dBFS
24 bit - SNR/DR = 144dB
SNR
94dBSPL
Cal. point =
94dBSPL
0dBFS
Dynamic Range = 105dB
130dBSPL
AOP
20 bit - SNR/DR = 120dB
MICROPHONE
16 bit - SNR/DR = 96dB
See the illustration on the right for an example of a
microphone and some accompanying audio
components with different dynamic ranges
As illustrated here, the full-scale levels of the
microphone and downstream components will
generally align regardless of the signal chain
dynamic range. However, the noise floors may not.
This is because 0dBFS (full scale) is the same for
both systems, but the digital numbering system
used by the audio signal chain may not be able to
represent levels as low as the microphone noise
floor. This means that if a digital microphone is
used with a digital encoding scheme which does
not use enough bits to represent the noise floor of
the microphone, the system noise will not reflect
the microphone’s true performance.
-96dBFS
Noise
Floor
-120dBFS
Noise
Floor
Select audio components and a digital
encoding
scheme
which
have
equivalent or greater SNR / Dynamic
Range than the Dynamic Range of the
microphone being used.
-144dBFS
Noise
Floor
Figure 12
16bit encoding does not have sufficient
Dynamic range for this microphone. The
system performance will be limited by
the audio encoding scheme
If a high SNR microphone is to be used in an audio signal chain with low dynamic range (for example, 16 bits),
the system is not able to preserve the SNR performance of the microphone (see illustration above). One option
to mitigate this is to use a microphone that has a lower AOP. This can be achieved, for example, by calibrating
Application Note
12
V1.0
2018-01-16
MEMS microphone electrical implementation
3 Electrical implementation of digital microphones
the microphone to an increased sensitivity, for example -26dBFS instead of -36dBFS. This will reduce the AOP
from 130 to 120 dBSPL. The dBFS value of the noise floor will increase but since the full-scale value (0 dBFS) was
reduced by the same number of decibels (on the SPL scale), the noise floor of the microphone is preserved. This
is the difference between Infineon’s IM69D130 and IM69D120 microphones (see details in the table below);
IM69D130 cannot be used in a 16-bit system without SNR degradation but IM69D120 can. If both high AOP
(130dBSPL) and high SNR (69dB) are required, a 20-bit or 24-bit audio signal chain must be used.
130dBSPL
AOP
IM69D120
120dBSPL
0dBFS
AOP
-36dBFS
25dBSPL
25dBSPL
Noise
Floor
-105dBFS
0dBFS
Dynamic Range = 95dB
10dB shift up
SNR = 69dBA
SNR = 69dBA
94dBSPL
Dynamic Range = 105dB
94dBSPL
DIG. ENCODING
Noise
Floor
Full scale
0dBFS
-26dBFS
-95dBFS
16 bit - SNR/DR = 96dB
IM69D130
Noise
Floor
-96dBFS
Figure 13
Reduction of the AOP allows high SNR in a 16 bit system
Table 3
Important Infineon digital microphone parameters and corresponding system bit
requirement
Microphone
Model
Sensitivity
SNR
AOP
DR
Required Bits in
System
IM69D130
-36dBFS
69dB
130dBSPL
105dB
20
IM69D120
-26dBFS
69dB
120dBSPL
95dB
16
3.8
Device microphone signal path requirements - bandwidth
The sampling rate (Fs) determines the bandwidth of a PCM system:
Common audio bandwidths and the corresponding sampling rate requirements are listed in the table below. To
enable an audio system with a 20kHz bandwidth, the sampling rate must be 40kHz or higher. 48kHz and
44.1kHz (used in CDs), are typical rates. In communication systems, full band audio is enabled by, for example,
VoIP (Voice over Internet Protocol) and VoLTE (Long-Term Evolution) technologies.
Application Note
13
V1.0
2018-01-16
MEMS microphone electrical implementation
3 Electrical implementation of digital microphones
Table 4
Common codecs used and their corresponding parameters
Codec / System
Bandwidth (Hz)
Min. Sample Rate (Hz)
Standard telephone bandwidth
(AMR)
300 – 3.4k
8k
“Wideband Audio” / “HD-Voice” /
AMR-WB
50 – 7k
16k
“Superwideband Audio”
50–14k
32k
“Fullband Audio” / “Full-HD
Voice” / VoIP / VoLTE
20 – 20k *
44.1k / 48k
Ultrasound (excluding frequencies
>48kHz)
Up to 48kHz
96k
* depends on codec used
Another often available sampling rate is 96kHz. It may be needed in microphone systems used for capturing
ultrasound with frequencies up to 48kHz. Using a 96kHz sampling rate is very unlikely to improve audio quality
in practice in the audible frequency range.
Even a 32kHz or 16kHz sampling rate can be high enough if the goal is not to cover the whole 20kHz audible
sound bandwidth. The lower sampling rates can be considered in order to gain benefits such as lower
transmission bit rates, lower system current consumption, simpler system or lower price. Even a 16kHz
sampling rate and the corresponding 8kHz audio bandwidth enable “HD Voice” quality and using the AMR-WB
(Adaptive Multi-Rate Wideband; ITU-T / 3GPP) codec used in GSM telephony
Note:
Design the system to have an audio bandwidth that enables fulfilling the audio quality requirements
in all use cases; the sampling rate must be 2x the bandwidth.
3.9
Current consumption
The current consumption of digital microphones is typically higher than the consumption of analog interface
microphones with a comparable performance level. The difference is due to the analog to digital conversion
done in the microphone instead of doing it later in the signal chain. There are also other factors that affect the
current consumption of digital microphone systems. The consumption depends on the supply voltage level,
clock frequency and the capacitive load in the system. The higher the clock frequency, the faster the clock and
data lines have to be driven back and forth from one state to another. The higher the capacitive load, the more
current is consumed to drive those lines.
The current consumption of the microphone system depends also on the power supply arrangement. A
microphone with high power supply rejection performance (PSR/PSRR) can be used with a power supply that is
readily available in the system that is maybe not absolutely clean of disturbances. A microphone with lower
rejection capabilities may have to be used with a very clean power supply that is created with the help of an
LDO (low-dropout regulator) from a higher voltage source. The LDO solution is likely to consume significantly
more current. This applies also to analog microphones.
3.10
Multi-mode PDM microphones
The current consumption of high-performance digital microphones can be too high for some applications or
use cases. There can also be other reasons for wanting to change the characteristics of microphones. Multimode microphones address this need for microphone versatility. The most common available alternative use
mode in PDM MEMS microphones is the Low Power Mode, that typically compromises the performance of the
microphone to enable reaching a lower current consumption.
In PDM interface microphones, the mode is usually controlled by changing the frequency of the microphone
clock. This, of course, means that the device system (/ codec) must have the needed clock frequencies available
Application Note
14
V1.0
2018-01-16
MEMS microphone electrical implementation
4 Electromagnetic compatibility (EMC)
and a way to switch from one frequency to another. For example, Normal Mode can be in use at 2.4 or 3.072 MHz
and Low Power Mode at 768kHz.
The system should also take into account that the switching from one mode to another may not be completely
glitch-free. To avoid any unwanted pops or clicks in the output of the microphone system, the microphone
signal may have to be muted temporarily during mode switching.
4
Electromagnetic compatibility (EMC)
Electromagnetic compatibility (EMC), describes the ability of a microphone.
• to operate in a device without being disturbed by the electromagnetic environment
•
to not disturb other systems in the device
An EMC issue with a microphone can manifest itself in different ways:
•
The microphone gets disturbed by the radiated or conducted disturbances in the device
•
A poorly designed digital microphone (for example, too fast signal rise/fall speeds, compromised
grounding) can emit disturbances that can affect antennas located very close to the microphone
•
The microphone – effectively a relatively large grounded metal box – passively disturbs the functionality of
adjacent antennas
- This can be mitigated by moving the microphone further from the antenna or by improving grounding
There are many noise sources in connected devices such as smartphones:
•
Wireless connectivity antennas (cellular, wi-fi, etc.) output both electrical and magnetic fields
•
Other signal lines from which the disturbances couple to the microphone lines
•
Indirect coupling: for example, radiated RF disturbances that originate either in the device itself or in
external sources get coupled into signal traces and from there to the microphone
•
Noisy grounds
•
Electrically noisy components (such as RF power systems) may add noise into microphone signal traces
Radio frequency interference (RFI) appears when RF disturbances get coupled to the microphone signal lines or
directly into the microphone itself. The disturbances can propagate to the microphone output signal and cause
an audible disturbance, ‘TDMA noise’. GSM cellular devices use time-division multiple access (TDMA) technology
at 800 to 900MHz and 1800 to 1900MHz. The transmission pulses at an audible 217Hz frequency and the power
levels can be high, causing the 217Hz pulsing to couple into the microphone output signal.
4.1
RF disturbance mitigation
Microphone implementation has to be well executed so that the microphone is well protected from all the
radiated and conducted disturbances that are present in wireless connected devices.
Microphone signals should be filtered with, for example, capacitors and inductors
•
A capacitor (C) passes high frequencies, depending on its capacitance value, so it can be used for shorting
unwanted high frequencies to the ground of the device
• An inductor (L) allows low frequencies to pass and blocks high frequencies so it can be used in series on a
signal line to filter out radio frequency disturbances
•
Combinations of capacitors and inductors may yield the best filtering results; for example, the so-called pi
filter (see illustration on the right)
The optimal filtering circuit design, layout and filtering components depend on
• the design of the microphone and possible filtering systems built into it
•
the filtering built into the system that receives the signal (e.g. codec)
Proper circuit design and layout are key enablers for disturbance free
microphone signals. The guidelines for sensitive signal wiring are as follows:
Application Note
15
Figure 14
Pi filter
V1.0
2018-01-16
MEMS microphone electrical implementation
4 Electromagnetic compatibility (EMC)
•
•
•
•
Signal trace lengths should be minimized to
minimize disturbance pick-up (1)
Signal lines should be separated to prevents
cross talk (2)
Signals should be shielded properly from
electromagnetic radiation and cross talk from
other signal lines (3)
- Often the safest place for the signal lines
is inside a circuit board between two
ground planes (layers) with protective
ground traces running on both sides of Figure 15
the signal trace
Filtering should be used to remove
disturbances from the signals (4)
Guidelines
for
microphone signals
disturbance
free
•
Microphones or other audio systems should not be placed close to RF systems
- It is good practice to separate microphones and antennas onto different planes in the device
• Audio signals or grounding should not be routed parallel to RF traces (RF signals or RF grounding)
- Signals should be routed perpendicular to each other and the distance between the microphone
signals and disturbance sources should be maximized
- RF traces or other disturbance sources should not be located under the microphones
In addition to the signal cleanliness advice above, the methods listed in the digital microphone implementation
section apply also generally (see more details in section ‘Signal Connections’)
• Use capacitors (CPS, CRF) in Vsupply to filter low and high frequency interference from the power lines
- Place the smallest value filtering capacitor closest to the microphone
- Minimize trace lengths from the filtering capacitors to the power plane
- Minimize inductance in the capacitor traces
- Avoid capacitors based on class 2 dielectrics close to microphones (due to risk of audible noise)
• Add a series resistor or ferrite to each power supply line for additional EMI suppression
• Connect microphone grounds to the circuit board ground plane with vias that are directly on the
microphone ground pads; avoid traces
A digital (or differential analog) interface is recommended; especially if the lines are long and the environment
challenging. See guidelines for digital microphone RF implementation is section ‘Electrical Implementation of
Digital Microphones’.
Note:
RF disturbances can be mitigated by, for example, using filtering and shielding, minimizing signal
trace lengths and preventing disturbance coupling between signal lines and disturbance sources.
The inside of the microphone is protected against radiated disturbances by the metal package, however high
quality grounding is also essential for high disturbance immunity.
• A high-quality ground must be provided for the microphone so that disturbances have a low impedance
path to bypass the microphone
• To optimize disturbance protection, the ground should be designed to have minimal potential (voltage)
variation throughout the ground area (ΔV = 0)
- A solid board-wide ground plane provides the lowest impedance for RF signals and thereby a steady
ground
• Using a flexible circuit board may degrade ground quality; ground stability and disturbance immunity may
be compromised
- Ground traces on flex boards should be made wide to improve ground quality
- Tying the microphone ground to the ground on the main circuit board (typically the same as e.g.
codecs) is recommended
Application Note
16
V1.0
2018-01-16
MEMS microphone electrical implementation
5 Power supply disturbances
-
Paying attention to ground quality on flex boards is especially important when trace lengths are long
and close to disturbance sources
5
Power supply disturbances
Immunity to disturbances in the power supply line is a key factor in a microphone’s electrical robustness and its
immunity to electromagnetic and other disturbances in the device. A microphone with good power supply
rejection capabilities can save a device designer / manufacturer time and money but it is still also up to the
device design to make sure the disturbances in the power line stay at a level at which the microphones can cope
with them without degradation in output signal quality. Power Supply Rejection (PSR) and Power Supply
Rejection Ratio (PSRR) are the parameters that indicate the rejection capabilities of the microphone.
The power supply line must be kept clean of disturbances
•
Filter disturbances (see ‘RF Disturbance Mitigation’ on the previous page for more details)
•
Shield power supply traces
•
Avoid the proximity of power supply disturbance sources such as data buses and switching regulators
•
Use a regulated low noise power supply if necessary
As mentioned in the ‘Current consumption’ section, a microphone with good power supply interference
rejection performance can help reduce system current consumption by enabling simpler powering solutions.
6
Electrostatic discharges (ESD)
The way a MEMS microphone is implemented into a device can have a significant effect on the microphone’s
immunity to electrostatic discharges (ESD). A discharge can harm a MEMS microphone in two different ways:
electrically and mechanically (due to high-pressure pulses caused by the sparks).
The electrical damage caused by the high voltage levels associated with ESD can be prevented by preventing
the spark from reaching the sensitive semiconductor parts of the microphone
• A low-impedance path to ground that bypasses the microphone must be provided for the discharge
• MEMS microphone packages are typically shielded and grounded but it is preferable not to let the spark get
grounded through the package
- It is important to keep the extremely loud spark as far away as possible from the acoustic sensor
• The key implementation factor is that the quality of the microphone ground must be high, i.e. the
impedance to the ground and inside the ground must be low
7
Wind
Wind is a challenging environmental factor for microphones. There are few effective ways to mitigate wind
noise. In most types of consumer electronics devices, traditional wind noise mitigation tactics such as foam or
fur wind screens are not feasible due to their sizes, appearances and challenging mass manufacturability.
Electrical solutions are relatively ineffective. Wind (noise) is typically arbitrary and uncorrelated from one
location to another and therefore difficult to cancel with, for example, multi-microphone software solutions.
Wind noise also tends to have a wide frequency band that overlaps with the wanted sounds so electrical signal
filtering cannot be used effectively either. Most wind noise energy is at low frequencies so some improvement
can be achieved with high pass filtering but this is likely to impact the low frequencies of the wanted audio
signal.
Due to the often-turbulent nature of wind, the noise can often sound like bad distortion but this does not
necessarily mean that the microphone (system) is actually saturated. However, some types of wind, such as
bursts of non-stationary noise can cause the signal level to rise high enough to cause saturation. High acoustic
overload point of the microphone and the system may help with this kind of wind noise.
8
Further information
Read more about PDM and PCM here:
http://users.ece.utexas.edu/~bevans/courses/rtdsp/lectures/10_Data_Conversion/
AP_Understanding_PDM_Digital_Audio.pdf
Application Note
17
V1.0
2018-01-16
Trademarks
All referenced product or service names and trademarks are the property of their respective owners.
Edition 2018-01-16
Published by
Infineon Technologies AG
81726 Munich, Germany
© 2018 Infineon Technologies AG
All Rights Reserved.
Do you have a question about any
aspect of this document?
Email: erratum@infineon.com
Document reference
IFX-hgf1512738861367
IMPORTANT NOTICE
The information contained in this application note is
given as a hint for the implementation of the product
only and shall in no event be regarded as a description
or warranty of a certain functionality, condition or
quality of the product. Before implementation of the
product, the recipient of this application note must
verify any function and other technical information
given herein in the real application. Infineon
Technologies hereby disclaims any and all warranties
and liabilities of any kind (including without limitation
warranties of non-infringement of intellectual property
rights of any third party) with respect to any and all
information given in this application note.
The data contained in this document is exclusively
intended for technically trained staff. It is the
responsibility of customer’s technical departments to
evaluate the suitability of the product for the intended
application and the completeness of the product
information given in this document with respect to such
application.
WARNINGS
Due to technical requirements products may contain
dangerous substances. For information on the types
in question please contact your nearest Infineon
Technologies office.
Except as otherwise explicitly approved by Infineon
Technologies in a written document signed by
authorized representatives of Infineon Technologies,
Infineon Technologies’ products may not be used in
any applications where a failure of the product or
any consequences of the use thereof can reasonably
be expected to result in personal injury
Download PDF
Similar pages