CN104252563B - Optimization method of silicon preset-capacitance microphone - Google Patents
Optimization method of silicon preset-capacitance microphone Download PDFInfo
- Publication number
- CN104252563B CN104252563B CN201410478070.8A CN201410478070A CN104252563B CN 104252563 B CN104252563 B CN 104252563B CN 201410478070 A CN201410478070 A CN 201410478070A CN 104252563 B CN104252563 B CN 104252563B
- Authority
- CN
- China
- Prior art keywords
- value
- microphone
- point
- derivative
- interval
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Landscapes
- Electrostatic, Electromagnetic, Magneto- Strictive, And Variable-Resistance Transducers (AREA)
Abstract
A kind of optimization method of silicon micro capacitor microphone includes the following steps: 1) to construct objective function according to optimization aim; 2) initial point m0=(r0, b0) is chosen, the number of iteration, set algorithm error precision is indicated with k; 3) using microphone emulation and formula, the sensitivity curve of the point is obtained, and calculate the target function value Jk of the point; 4) directional derivative of the J at mk is solved
A value rk ' is chosen near rk, keeps bk constant, and emulation finds out its sensitivity curve, and calculates Jk ', and utilizes formula
Find out objective function J at rk to r partial derivative; Same method acquires objective function J at b0 to b partial derivative; 5) it is optimal step size that next point mk+1=(rk+1, bk+1)=mk- α kdk, α k is found on the direction gradient dk, so that objective function Jk+1 is minimum; 6) judge | mk+1-mk | whether tend to 0, if it is not, then mk+1 is substituted into step 2) again, enable k=k+1, is iterated calculating; If it is, iteration stopping, optimal value m*=mk+1. The present invention makes microphone sensitivity with higher under conditions of meeting working frequency, can also largely reduce the microphone development time.
Description
Technical field
Present invention design is related to MEMS, finite element simulation, breadboardin and steepest and declines the technologies such as emulation neck
Domain, it combines optimized algorithm and device simulation, belongs to power electronics, optimized algorithm and computer simulation emulation technological synthesis
The technology of intersection.
Background technology
Micro-system (Micro Electro Mechanical System, abbreviation MEMS) technology is one and emerging grinds
Study carefully field and industry, it is a front end scientific research field comprising multiple subjects, mainly there is chemistry, optics, biology, physics
Learn, control multiple subjects such as engineering and mechanics.Typical micro-system size typically between several microns to several millimeters, including biography
The parts such as sensor, actuator, circuit module and micro mechanical structure, it is possible to achieve measurement, energy conversion, information processing and execution etc.
Function.
MEMS microphone, as an important branch of micro-system field, is also a focus of current research.MEMS wheats
Gram wind refers to the microphone manufactured based on MEMS technology, the miniature wheat made using MEMS processing technologys on a silicon microwafer
Gram wind, therefore alternatively referred to as silicon micro-microphone.Different from conventional microphone, MEMS microphone has little size, light weight, installation
Simply, easily formed array, low cost, can batch micro operations the features such as.MEMS microphone has application widely, for example
In consumer electronics field, the built-in microphone such as people's mobile phone used in everyday, mp4, notebook computer, in automobile industry
Hands-free phone, the audiphone in medical domain etc. are all based on the microphone of MEMS technology manufacture.As MEMS processes work
The continuous development of skill, the sensitivity of MEMS microphone and stability are improved constantly, its application also constantly expanding,
Progressively replacing the status of conventional microphone.Within following a period of time, MEMS microphone will substitute completely traditional
Microphone.Therefore the research to MEMS microphone has great meaning.
Silicon micro-microphone can be divided into four classes according to the difference of sensitivity principle:Pressure resistance type, piezoelectric type, condenser type and optical fiber
Formula.Pressure resistance type microphone procedure of processing is simple, and cost is relatively low, is easy to batch production.But compared to other sensors, its spirit
Sensitivity is low, and the sensitivity of general pressure resistance type microphone only has about 25 μ V/Pa.It is a kind of this Mike of lifting to increase vibrating diaphragm area
The effective measures of wind sensitivity, but this also causes vibrating diaphragm to be easily broken, and affects the stability of microphone.Piezoelectric microphone and pressure
Resistive microphone is similar to, and its sensitivity is high relative to pressure resistance type microphone, typically in 50-250 μ V/Pa.But the Mike
Wind is easily affected by the extraneous bad border such as temperature, air humidity, less stable.Optical fiber type microphone be at present for it is relative
Newer research, it has higher sensitivity, in being mainly used in the environment that other microphones cannot work, is such as subject to electromagnetism
The environment or needs of interference prevents occasion being ravesdropping etc..Too high yet with its cost, measurement means are complicated with device,
It is not widely used in industrial process.
Electret Condencer Microphone is the main flow in current MEMS microphone market, its flatness and sensitivity etc. in frequency response
The stability aspect of key property and performance has obvious advantage.In recent years, diversified MEMS microphone structure quilt
Design, have twin-wafer type, single wafer and vibrating diaphragm band cellular type, the characteristics of these microphones have each of which.Vibrating diaphragm band
Cellular type microphone relative to other types, with preferable stability, simple processing technology, cheap processing cost, it is easy to
The advantages of encapsulation.Can be used as the optimization object of the present invention.Often there was only the design of single structure in most of documents, that is, design
After going out a kind of physical dimension, emulation testing its performance is not optimized to the structure.There is document also to have to tie microphone
The optimization of structure, but they first change a structural parameters of microphone, such as vibrating diaphragm often using a kind of method tried and gather
Area, then observe the situation of change of its sensitivity curve, then change another parameter, such as the air gap, continue
Observe its change.By constantly test, certain law of the microphone parameters to performance impact is summarized.It is desired according to oneself
Performance continues to change parameter, finally draws one group so that performance meets the dimensional structure of sets target.This procedure complexity,
Less efficient, generally requiring certain practical experience could faster obtain desired structural parameters, and the structure ginseng for obtaining
Number is not necessarily one group of parameter of optimum.Not a kind of preferable algorithm to microphone structure being optimized.Greatly prolong
Construction cycle of microphone.
The content of the invention
In order to the sensitivity for overcoming the shortcomings of existing microphone is relatively low, the construction cycle is longer, the present invention is to silicon micro capacitor
Microphone structure is optimized, and is declined using steepest and is emulated, and microphone can not only had under conditions of operating frequency is met
There is higher sensitivity, can also in a large number reduce the time of microphone exploitation.
Plan steepest of the present invention declines emulation, comprises the steps:
A kind of optimization method of silicon micro capacitor microphone, the optimization method is comprised the following steps:
1), according to optimization aim, object function is builtWherein, W1And W2For power
Coefficient, W1For negative value, W2It is on the occasion of S (r, b) is sensitivity, and sensitivity is from b with regard to circular membrane radius r and acoustics pitch-row
Function, k (f, r, b) is slope of the frequency response curve at frequency f, and the frequency response curve of microphone is also to shake with regard to circle
The function of film radius r and acoustics pitch-row from b;AndRepresent bent to sensitivity of microphone between 20Hz-20kHz
The absolute value of line slope be integrated after value;
2), initial point m is chosen0=(r0,b0), r0And b0The respectively vibrating diaphragm radius and acoustics pitch-row of initial point is from using k tables
Show the number of times of iteration, for initial point, make k=0, set algorithm error precision ε=0.001
3), using microphone emulation and formula, the sensitivity curve of the point is drawn, and calculates the target function value of the point
Jk;
4) J, is solved in mkThe directional derivative at placemkFor the point after kth time iteration, rkAnd bkPoint
The vibrating diaphragm radius and acoustics pitch-row of point that Wei be after kth time iteration is from rkNearby choose value rk', keep bkIt is constant, imitate
Its sensitivity curve is really obtained, and calculates Jk', then using formulaObtain object function
J is in rkPlace to r partial derivatives;Same method tries to achieve object function J in b0Place is to b partial derivatives;
5), in gradient dkNext point m is found on directionk+1=(rk+1,bk+1)=mk-αkdk, wherein, rk+1For subsequent point
Vibrating diaphragm radius, bk+1For the acoustics pitch of holes of subsequent point, αkFor optimal step size so that object function Jk+1Minimum, it is real in two steps
It is existing:
5.1) scope that object function minimum of a value is present is chosen;
5.2) previous step has solved αkInterval range, it is designated as into αk∈[αmin,αmax], utilize in the range of this
Dichotomy is found;
6), m is obtainedk+1Afterwards, judge | mk+1-mk| whether tend to 0, if it is not, then by mk+1Again step 2 is substituted into),
K=k+1 is made, calculating is iterated;If it is, iteration stopping, optimal value m*=mk+1。
The plan steepest of the present invention declines simulation calculation process, and different from the steepest of known expression formula emulation is declined, and is solving
During the problem, emulation is declined as basic thought with steepest, and with reference to dichotomy solving to the problem.Solving
During, the sensitivity curve that software emulation draws is by means of, needed in optimization process is obtained using sensitivity curve
A little parameters.The algorithm can be optimized to microphone structure, greatly shorten the construction cycle of microphone.
Description of the drawings
Fig. 1 is the structural representation for optimizing microphone in the present invention;
Fig. 2 is diaphragm structure schematic diagram;
Fig. 3 is the flow chart that steepest declines emulation;
Fig. 4 is the sensitivity curve figure at initial point;
Fig. 5 is the sensitivity curve figure of any near initial point;
Fig. 6 is the sensitivity curve figure of microphone after optimization.
Specific embodiment
The present invention is described in detail with reference to the accompanying drawings and examples.
Reference Fig. 1~Fig. 6, a kind of optimization method of silicon micro capacitor microphone, for this structure, microphone diaphragm face
Product, acoustics pore size distribution, the air gap of two-plate are the major parameters for affecting microphone property.And according to reality processing experience,
The air gap often adopts 3 μm.Therefore, this algorithm is optimized to vibrating diaphragm area and acoustics interporal lacuna the two parameters, it is desirable to
Microphone has flatter sensitivity curve between 20Hz-20kHz frequencies, and with higher sensitivity.Algorithm flow chart
With reference to Fig. 3, the optimization method is comprised the following steps:
1), object function is set upWherein W1And W2For weight coefficient, S (r, b) is
Sensitivity, according to analysis above, it is the function with regard to circular membrane radius r and acoustics pitch-row from b, and sensitivity S is with a height of
It is good;K (f, r, b) is slope of the frequency response curve at frequency f, and the frequency response curve of microphone is also with regard to circular membrane
The function of radius r and acoustics pitch-row from b.AndRepresent between 20Hz-20kHz to sensitivity of microphone curve
The absolute value of slope be integrated after value, the integrated value reflects the uniformity of microphone frequency response, and its value is less more consistent
Property is better.In order to meet the requirement that steepest declines emulation, object function J smallers are for excellent, therefore weights W1Should be negative value, and W2Should
It is on the occasion of selecting W here1=-0.7, W2=0.3.
2), initial point m is selected0=(r0,b0), r0And b0Respectively vibrating diaphragm radius and acoustics pitch-row make r from initial point0=
250, b0=100.The number of times of iteration is represented with k, for initial point, k=0 is made.Because the algorithm is to solve for realistic model, nothing
Too high precision need to be designed.So setting the error precision of the algorithm as ε=0.001.
3), emulated using microphone, simulate the sensitivity curve of microphone at initial point, Fig. 4 is the wheat at initial point
Gram wind sensitivity curve.J at the initial point0Value be 30.66.
4) microphone directional derivative at this point, is obtained, the present invention takes and take near the point point, simulates it
Sensitivity curve, its target function value is calculated by curve, using formula:
Calculate the directional derivative of the point.For the point near initial point, r0'=251, b '0=b0=60, emulation draws
Its sensitivity curve.As shown in figure 5, target function value J ' at the point0Value be 30.57.Can be tried to achieve using same method
Object function J is in b0Place is to b partial derivatives.
5), in gradient dkNext point m is found on directionk+1=(rk+1,bk+1)=mk-αkdk, wherein, rk+1For subsequent point
Vibrating diaphragm radius, bk+1For the acoustics pitch of holes of subsequent point, αkFor optimal step size so that object function Jk+1Minimum, due to target
Function is unknown, and the present invention is found using a kind of method of similar dichotomy, and step is as follows:
5.1) scope that object function minimum of a value is present is chosen.Step-length h is selected, α is started by step-length h from 0 to increase, α=
0,h,2h,…,nh.When α=0, the object function corresponding to it is the target function value J obtained in previous stepk.Then α is made
=h, calculates target function value Jh.By JhWith JkCompare, it there are two kinds of results.The first result:If Jh> Jk, and target
There is minimum of a value in function, and most start one section for subtraction function in α ∈ (0 ,+∞) are interval, hence in so that object function is obtained most
It is α that little value is intervalk∈[0,h].Second result:If Jh< Jk, then α=2h is made, calculate J2hValue, and and JhCompare, this
Two kinds of results of Shi Huiyou:(1) if J2h> Jh, then target function value greatly-little-big form is formed at 3 points of 0, h and 2h, then say
Bright αk∈[0,2h];(2) if J2h< Jh, then continue to increase step-length h, until causing J(i+1)h> JihSo that in object function
Value forms greatly-little-big form at tri- points of (i-1) h, ih and (i+1) h.Then illustrate that minimum of a value is interval for αk∈[(i-1)
h,(i+1)h]。
5.2) previous step has solved αkInterval range, it is designated as into αk∈[αmin,αmax], find in the range of this
αkValue.Known derivativeWithOrderUsing step 3)
The middle method for solving derivative calculates the derivative value of the pointIf derivative is equal to 0, illustrate
When, object function obtains minimum of a value, if derivative is more than 0, reduces interval, and interval is changed intoIf
Derivative is less than 0, then reduce interval, and interval is changed intoAfter reducing interval range, proceed iteration,
Until finding αkSo that object function is zero in the derivative value of the point, i.e.,
6), m is obtainedk+1Afterwards, judge | mk+1-mk| whether tend to 0, if it is not, then by mk+1Again step 2, order are substituted into
K=k+1, is iterated calculating.If it is, iteration stopping.Optimal value m*=mk+1。
Table 1- tables 4 are respectively the data obtained in 4 iterative process of the algorithm.
Parameter name | (r0,b0) | J0 | d0 | α0 |
Parameter value | (250,100) | 30.66 | (-0.14,0) | 211 |
Table 1m0Locate the calculated value of each variable
Parameter name | (r1,b1) | J1 | d1 | α1 |
Parameter value | (268.6,100) | 29.43 | (0,0.62) | 42 |
Table 2m1Locate the calculated value of each variable
Parameter name | (r2,b2) | J2 | d2 | α2 |
Parameter value | (268.6,63.96) | 29.356 | (0.022,0) | 23 |
Table 3m2Locate the calculated value of each variable
Parameter name | (r3,b3) | J3 | d3 | α3 |
Parameter value | (271.4,63.96) | 29.14 | (0,0) | — |
Table 4m3Locate the calculated value of each variable
Fig. 6 is the sensitivity curve of structure after the microphone optimization, it can be seen that the microphone of this size meets
There is flatter sensitivity curve between 20Hz-20kHz frequencies, and sensitivity is higher.
Described above is the effect of optimization that the present invention is drawn with cellular type silicon micro capacitor microphone with vibrating diaphragm by example, not
Limit the practical range of the present invention, it is done on the premise of without departing from scope involved by substance of the present invention various modifications and
Improve, should not exclude outside protection scope of the present invention.
Claims (2)
1. a kind of optimization method of silicon micro capacitor microphone, it is characterised in that:The optimization method is comprised the following steps:
1), according to optimization aim, object function is builtWherein, W1And W2For weight coefficient,
W1For negative value, W2It on the occasion of, S (r, b) is sensitivity to be, sensitivity is the function with regard to circular membrane radius r and acoustics pitch-row from b,
K (f, r, b) is slope of the frequency response curve at frequency f, and the frequency response curve of microphone is also with regard to circular membrane half
The function of footpath r and acoustics pitch-row from b;AndRepresent oblique to sensitivity of microphone curve between 20Hz-20kHz
The absolute value of rate be integrated after value;
2), initial point m is chosen0=(r0,b0), r0And b0The respectively vibrating diaphragm radius and acoustics pitch-row of initial point with k from being represented repeatedly
The number of times in generation, for initial point, makes k=0, set algorithm error precision ε=0.001;
3), using microphone emulation and formula, the sensitivity curve of the point is drawn, and calculates the target function value J of the pointk;
4) J, is solved in mkThe directional derivative at placemkFor the point after kth time iteration, rkAnd bkRespectively
The vibrating diaphragm radius and acoustics pitch-row of the point after k iteration is from rkNearby choose a value r 'k, keep bkConstant, emulation is obtained
Its sensitivity curve, and calculate Jk', then using formulaObject function J is obtained in rk
Place to r partial derivatives;Same method tries to achieve object function J in bkPlace is to b partial derivatives;
5), in gradient dkNext point m is found on directionk+1=(rk+1,bk+1)=mk-αkdk, wherein, rk+1For the vibrating diaphragm of subsequent point
Radius, bk+1For the acoustics pitch of holes of subsequent point, αkFor optimal step size so that object function Jk+1Minimum, realizes in two steps:
5.1) α when choosing object function minimum of a valuekThe scope of presence, selects step-length h, α is started by step-length h from 0 to increase, α=0,
h,2h,…,nh;When α=0, the object function corresponding to it is the target function value J obtained in previous stepk;Then make α=
H, calculates target function value Jh;By JhWith JkCompare, it has two kinds of results, the first result:If Jh> Jk, and target letter
There is minimum of a value in number, and most start one section for subtraction function in α ∈ (0 ,+∞) are interval, hence in so that object function obtains minimum
It is α that value is intervalk∈[0,h];Second result:If Jh< Jk, then α=2h is made, calculate J2hValue, and and JhCompare, at this moment
Have two kinds of results:(1) if J2h> Jh, then target function value greatly-little-big form is formed at 3 points of 0, h and 2h, then illustrate
αk∈[0,2h];(2) if J2h< Jh, then continue to increase step-length h, until causing J(i+1)h> JihSo that in the value of object function
Big-little-big form is formed at tri- points of (i-1) h, ih and (i+1) h, then illustrates that minimum of a value is interval for αk∈[(i-1)h,
(i+1)h];
5.2) previous step has solved αkInterval range, it is designated as into αk∈[αmin,αmax], two points are utilized in the range of this
Method is found;
6), m is obtainedk+1Afterwards, judge | mk+1-mk| whether tend to 0, if it is not, then by mk+1Again step 2 is substituted into), make k=k
+ 1, it is iterated calculating;If it is, iteration stopping, optimal value m*=mk+1。
2. a kind of optimization method of silicon micro capacitor microphone as claimed in claim 1, it is characterised in that:The step 5.2)
In, αkInterval range be designated as αk∈[αmin,αmax], α is found in the range of thiskValue;Known derivativeWithOrderUsing step 4) the middle method for solving derivative
Calculate the derivative value of the pointIf derivative is equal to 0, illustrateWhen, target
Function obtains minimum of a value, if derivative is more than 0, reduces interval, and interval is changed intoIf derivative is little
In 0, then reduce interval, interval is changed intoAfter reducing interval range, proceed iteration, until looking for
To αkSo that object function is zero in the derivative value of the point, i.e.,
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201410478070.8A CN104252563B (en) | 2014-09-18 | 2014-09-18 | Optimization method of silicon preset-capacitance microphone |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201410478070.8A CN104252563B (en) | 2014-09-18 | 2014-09-18 | Optimization method of silicon preset-capacitance microphone |
Publications (2)
Publication Number | Publication Date |
---|---|
CN104252563A CN104252563A (en) | 2014-12-31 |
CN104252563B true CN104252563B (en) | 2017-04-19 |
Family
ID=52187451
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201410478070.8A Active CN104252563B (en) | 2014-09-18 | 2014-09-18 | Optimization method of silicon preset-capacitance microphone |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN104252563B (en) |
Citations (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103561377A (en) * | 2013-11-07 | 2014-02-05 | 美特科技(苏州)有限公司 | Optimized setting method for electroacoustic products |
Family Cites Families (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US8001515B2 (en) * | 2007-12-21 | 2011-08-16 | National Semiconductor Corporation | Simultaneous optimization of analog design parameters using a cost function of responses |
TWI480522B (en) * | 2012-10-09 | 2015-04-11 | Univ Feng Chia | Method for measuring electroacoustic parameters of transducer |
-
2014
- 2014-09-18 CN CN201410478070.8A patent/CN104252563B/en active Active
Patent Citations (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103561377A (en) * | 2013-11-07 | 2014-02-05 | 美特科技(苏州)有限公司 | Optimized setting method for electroacoustic products |
Non-Patent Citations (2)
Title |
---|
《悬臂梁膜硅微机械电容式麦克风的设计与仿真》;王伟;《中国优秀硕士学位论文全文数据库 信息科技辑》;20120315;全文 * |
具有自由悬浮敏感膜的硅微机械电容式麦克风的设计与仿真计算;王伟 等;《传感技术学报》;20091130;第22卷(第11期);1571-1575 * |
Also Published As
Publication number | Publication date |
---|---|
CN104252563A (en) | 2014-12-31 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN110400575B (en) | Inter-channel feature extraction method, audio separation method and device and computing equipment | |
JP6739571B2 (en) | Haptic transformation system using frequency shift | |
CN104272137B (en) | Systems and methods for mapping a source location | |
WO2017012321A1 (en) | Display state adjustment method, display state adjustment device and display device | |
Le et al. | Noncontact human–machine interface using complementary information fusion based on mems and triboelectric sensors | |
CN109151301A (en) | Electronic device including camera model | |
US20160299999A1 (en) | Systems and methods for power plant model optimization | |
EP3019881A2 (en) | Performance improvement of mems devices | |
WO2022062901A1 (en) | Calibration method and apparatus, monocular laser measurement device, and calibration system | |
CN103561377A (en) | Optimized setting method for electroacoustic products | |
CN106597852A (en) | MEMS gyroscope temperature compensation method based on RBF neural network | |
CN104267596A (en) | Finite-time decoupling control method of cart inverted pendulum system | |
Zhou et al. | Metal mesh as a transparent omnidirectional strain sensor | |
CN104252563B (en) | Optimization method of silicon preset-capacitance microphone | |
Zhang et al. | Bioinspired tactile sensation based on synergistic microcrack-bristle structure design toward high mechanical sensitivity and direction-resolving capability | |
Chen et al. | Modeling and simulation of a condenser microphone | |
CN102437781B (en) | Based on distributed piezoelectric actuator Active Vibration Control circuit optimization structure and method | |
Mu et al. | All-printed flexible capacitive array tactile force sensors with tunable sensitivity and low crosstalk for micro motion detection | |
CN106156429B (en) | A kind of Electrostatic deformation film antenna finite element modeling method based on information in kind | |
Zhu et al. | Fabrication of robust convex array microstructured PDMS-based capacitance sensor molded by laser patterned silicon for human-being acoustic monitoring | |
Zhang et al. | Deep Learning Enabled High-Performance Speech Command Recognition on Graphene Flexible Microphones | |
CN115081643B (en) | Confrontation sample generation method, related device and storage medium | |
WO2020048109A1 (en) | Mechanical analysis method of mems double-end fixed beam structure under bending condition of flexible substrate | |
CN103576282A (en) | Method for manufacturing static stretching thin film reflecting mirror | |
Lv et al. | Analysis of a micro pressure-sensor employing SiC-AlN-SiC structure |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |