5.9. Dual-diaphragm combination of pressure and pressure-gradient microphones
A versatile microphone that is popular in recording studios and for recording ensembles on location is the dual-diaphragm variable-pattern capacitor microphone, the schematic of which is shown in Fig. 5.31. It has two diaphragms: one mounted in front (F) of and the other mounted at the back (B) of a common central plate (P). An array of holes in the central plate provides a mixture of resistance and reactance. When the slider of the potentiometer is at position “k,” the polarizing voltages on both diaphragms are equal so that they behave like a pair of back-to-back pressure microphones. Hence the
resulting directivity pattern is omnidirectional. The low compliance of the air in the holes of the plate ensures that the resonance frequencies of the diaphragms are high to provide a suitably wide working bandwidth. As with all pressure microphones, it is stiffness-controlled and displacement-sensitive in this mode.
When the slider is at the other end of its range in position “i,” the polarizing voltages on the two diaphragms are of equal magnitude but opposite polarity. Therefore, the microphone is only sensitive to the difference in pressures at the two diaphragms so that they behave more or less as a single diaphragm. Hence the resulting directivity pattern is figure 8. The resonance frequency is now determined by the diaphragm tension instead of the stiffness of the trapped air, which now travels back and forth through the holes providing a high viscous damping resistance. It is this damping resistance that determines the bandwidth of the microphone, which is resistance-controlled and velocity-sensitive in this mode. Without this resistance, the frequency response would be just one sharp resonant peak.
When the slider is at position “j,” no polarizing voltage is supplied to the diaphragm at the back (B), which in turn no longer contributes to the output voltage. Because it has low mass and high compliance, the back diaphragm is also acoustically transparent so that we have essentially the same configuration as an acoustic combination of pressure and pressure-gradient microphones shown in Fig. 5.7.
The analogous circuit for this kind of microphone is shown in Fig. 5.32, where p˜1 and p˜2 are the pressures on the outer surfaces of the front and back diaphragms, U˜1 and U˜2 are their respective volume velocities, E1 and E2 are the front and back polarizing voltages, and e˜ is the microphone output voltage. Also, CE0 is the static capacitance of each diaphragm when blocked, S=πa2 is the surface area of each diaphragm, d is distance between each diaphragm and the central plate, CAP is the compliance of the air in the holes of the plate, CAG is the compliance of the air in the gap between each diaphragm and the plate, and ZAD is the impedance of each diaphragm that includes the
mass, compliance, radiation mass, and resistance. The holes in the plate are represented by the T-circuit impedances ZAP1 and ZAP2 using the tube model shown in Fig. 4.45 except that the holes are assumed to be so narrow that the pressure variations are effectively isothermal and hence RT can be ignored. The elements CE0 and CAG are defined by
CE0=ε0Sd,
(5.69)
CAG=Sdρ0c2,
(5.70)
where ε0=8.85×10−12 is the permittivity of free space, ρ0=1.18kg/m3 is the static density of air, and c=345m/s is the speed of sound in free space. From Fig. 5.32 we can write the following equations:
p˜1=p˜a+(ZAD+1jωCAG)U˜1−1jωCAGU˜3,
(5.71)
p˜2=p˜b−1jωCAGU˜4+(ZAD+1jωCAG)U˜2,
(5.72)
0=−1jωCAGU˜1(ZAP1+ZAP2+1jωCAG)U˜3+ZAP2U˜4,
(5.73)
0=−1jωCAGU˜2+ZAP2U˜3+(ZAP1+ZAP2+1jωCAG)U˜4,
(5.74)
e˜1=SdE1CE0p˜a=(1−jωCE0+12jωCE0)i˜1+12jωCE0i˜2,
(5.75)
e˜2=SdE2CE0p˜b=12jωCE0i˜1+(1−jωCE0+12jωCE0)i˜2,
(5.76)
U˜1=SdE1CE0i˜1,
(5.77)
U˜2=SdE2CE0i˜2,
(5.78)
e˜=i˜1+i˜22jωCE0.
(5.79)
Firstly we solve Eqs. (5.73) and (5.74) for U˜3 and U˜4 and insert these into Eqs. (5.71) and (5.72). If we then insert p˜a,p˜b,U˜1, and U˜2 from Eqs. (5.75–5.78) respectively into Eqs. (5.71) and (5.72), and solve for i˜1 and i˜2 before inserting the latter in Eq. (5.79), we obtain an expression for the output voltage e˜. Furthermore, we use the expression for p˜2 given by Eq. (5.17).
Omnidirectional performance
If E2=E1, which corresponds to the slider being at position “k” in Fig. 5.31, we have a pressure microphone and the sensitivity is given by
in accordance with Eq. (5.15) for a resilient disk. Using this formula, the exact on-axis response with the switch in position “i” for an omnidirectional pattern is plotted in Fig. 5.33. We see from this formula that the directivity pattern is essentially omnidirectional provided that
f<<c3a+πl
(5.82)
so that the cos θ term becomes insignificant. The impedances can be expanded as follows:
ZAD=jωMAD+1jωCAD+11jωMAR+1RAR,
(5.83)
ZAP1=RAP2+jωMAP2,
(5.84)
ZAP2=1jωCAP−RAP6−jωMAP6,
(5.85)
where dynamic mass MAD and compliance CAD of the membrane are given by Ref. [1].
MAD=4ρDh3S,
(5.86)
Table 5.2
Dual-diaphragm condenser microphone parameters
Membrane
Radius
a
12.6mm
Thickness
h
2.5μm
Density
ρD
1400kg/m3
Tension
T
50N/m
Air
Density
ρ0
1.18kg/m3
Absolute viscosity
μ
17.9μNs/m2
Mean free path
λ
60nm
Accommodation coefficient
α
0.9
Adiabatic sound speed
c
345m/s
Specific heat ratio
γ
1.403
Gap
d
50μm
Permittivity
ε0
8.85pF/m
Polarizing voltage
E1
100V
Plate
Hole radius
ap
6μm
Hole pitch (center to center)
p
18μm
Depth
l
1.15mm
CAD=S28πT,
(5.87)
where ρD is the density of the membrane material, h is its thickness, S=πa2 is its surface area, a is its radius, and T is its tension. The acoustic radiation mass and resistance are given by
RAR=ρ0cS,
(5.88)
MAr=ρ04a.
(5.89)
From Chapter 4 the plate mass, compliance, and resistance are given by
RAP=8μl(1+4Bu)a2pSff,
(5.90)
MAP=1+3Bu1+4Bu4ρ0l3Sff,
(5.91)
CAP=lSffρ0c2,
(5.92)
where l is the thickness of the plate, ap is the hole diameter, μ=18.6×10−6N·s/m2 is the viscosity of air, and ff is the fill factor, which for a rectangular hole grid is given by
ff=πa2pp2,
(5.93)
where p is the hole pitch. Also Bu is the boundary slip factor that is given by
Bu=(2αu−1)λmap,
(5.94)
where αu=0.9 is the accommodation coefficient and λm=60nm is the mean free path length of an air molecule between collisions. It is assumed that the holes are so narrow that the pressure variations within them are isothermal because of heat conduction through the walls. Hence the specific heat ratio γ is absent from the expression for CAP. Because the membrane is flexible as opposed to rigid, the radiation mass MAR is that of a resilient disk in free space as derived in Chapter 13. If we ignore MAP, MAR, RAR, CAD, and ω3 in Eq. (5.80), we obtain the following approximate formula for the sensitivity
Using this formula, the approximate on-axis response with the switch in position “k” for an omnidirectional pattern is plotted in Fig. 5.33 and also in Fig. 5.34 along with the 180degrees off-axis response. We see that at low to mid frequencies, where ω→0, the reference sensitivity is given by
e˜k(ref)=E1(CAG+CAP2)Sdp˜1.
(5.96)
The upper limit of the working range is roughly determined by the resonance frequency
fU=12πMAD(CAG+CAP2)√.
(5.97)
Bidirectional performance
If E2=−E1, which corresponds to the slider being at position “i” in Fig. 5.31, we have a pressure-gradient microphone and the exact sensitivity is given by
which gives a bidirectional directivity pattern at all frequencies. Using this formula, the on-axis response is plotted in Fig. 5.33 and also in Fig. 5.34 along with the 180degrees off-axis response. If we ignore MAP, MAR, and RAR, we obtain the following approximate formula for the sensitivity
e˜i≈E1Δlcosθ2cSd(RAP2+jωRAPCAG+jωMAD+1jωCAD)p˜1,
(5.99)
which is also plotted in Fig. 5.33. We also see that at mid frequencies, where jωCAD>2/RAP but jωMAD<RAP/2 and jωRAPCAG<2, the reference sensitivity is given by
e˜i(ref)=E1ΔlcSdRAPp˜1.
(5.100)
The lower cut-off frequency is
fL=1πRAPCAD
(5.101)
and the upper limit of the working range is roughly determined by the resonance frequency:
fU=12πMADCAG√.
(5.102)
Unidirectional performance
If E2=0, which corresponds to the slider being at position “j” in Fig. 5.31, we have a combination of pressure and pressure-gradient microphones and the exact sensitivity is given by
which is plotted in Fig. 5.33. The expression in the numerator is a pure summation of the pressure and pressure-gradient responses obtained with the switch in positions “k” and “i” respectively. In those positions, all terms containing CE0 are balanced out, but in position “j” there is no such balance, which explains the presence of the complicated denominator term in Eq. (5.103). However, the denominator only contributes at low frequencies so that after removing all the high-frequency terms we can make the following approximation:
e˜j=2RAPCAD+jω2RAP(1CAD−CE0E214S2d2)+jω⋅e˜i+e˜k2,
(5.104)
which produces a shelf at low frequencies that is determined by the amount of negative stiffness produced by the force of electrostatic attraction. This in fact helps to equalize the amount of low-frequency attenuation, which would otherwise be 6dB because of taking the half sum of the pressure and pressure-gradient responses, where the former is flat at
low frequencies and the latter rolls-off at a rate of 6dB per octave. The low-frequency shelf starts to rise at the upper frequency of
fSU=1πRAPCAD
(5.105)
and levels off at the lower frequency of
fSL=1πRAP(1CAD−CE0E214S2ds).
(5.106)
The approximate on-axis response with the switch in position “j” for a unidirectional pattern is plotted in Fig. 5.33 where the approximate expressions for e˜i and e˜k obtained from Eqs. (5.95) and (5.99) respectively are inserted into Eq. (5.104).
Condition for equal sensitivity in all three switch positions
Ideally we would like the sensitivity of the microphone to be the same at all three switch positions “i”, “j”, and “k”. It turns out that this is also the condition for obtaining the optimum cardioid directivity pattern in position “j,” which is met by setting e˜i=e˜k in Eqs. (5.95) and (5.99) to yield
RAP=Δl(CAG+CAP2)c,
(5.107)
which after inserting the path length difference from Eq. (5.15) gives
RAP=l+πa4(CAG+CAP2)c.
(5.108)
Quite a large resistance is needed to meet this condition so that the holes through the plate have to be very narrow.
Condition for stability
Another condition that must be met is for the restoring force of the membrane tension to be greater than the force of electrostatic attraction towards the plate. We see from the schematic that this is met if
CA0=(SdCE0E1)2CE0>CAD
(5.109)
or using the expressions from Eqs. (5.69) and (5.87), we obtain the minimum tension value:
T>ε0a2E218d3.
(5.110)
Typically the tension should be about three times the minimum value to allow for slackening through age and environmental conditions.