15.17. Summary of electrostatic loudspeaker design
We can see from Eqs. (15.19) and (15.20) that to maximize the radiated sound pressure, the field strength EP/d should be as close to its maximum allowable value as possible, which is around 2kV/mm. Similarly, we generally set the maximum peak signal voltage 12e˜in to the same value as the polarizing voltage EP because the maximum sound pressure is proportional to the product of the two, but their sum should not exceed 4kV/mm.
If the gap distance d is reduced, we can reduce both EP and e˜in while keeping the sound pressure constant. Hence the sensitivity is increased. Although the capacitance is increased proportionately, the input current does not change because of the reduced voltage. Hence the reactive power is reduced, while the efficiency is increased. With a smaller gap, some very low bass is sacrificed, but this is usually in a region where there is little output anyway because of rear-wave cancellation.
A good place to start the design process is with the highest frequency of interest because we can use this to define the membrane thickness h independently of its radius a. From this we can work out the maximum tension T that it can withstand. Maximizing the tension enables a smaller gap d to be employed without the membrane touching the electrodes at the maximum input voltage at low frequencies, where the displacement is highest. We have shown (see Eqs. 15.84 and 15.86) that making the tension large enough to avoid contact automatically meets the condition for stability under the electrostatic force of attraction.
Knowing the tension, we can then define the fundamental resonance frequency f0 for any given radius a or choose a fundamental resonance frequency to find the required radius. Having found the radius, we can define the gap d, the polarizing voltage EP, the specific resistance of the dust screen Rs (for the desired Q factor), and the sound pressure level.
1. Referring to Fig. 15.18, choose the highest frequency of interest f2. By rearranging Eq. (15.110), calculate the membrane thickness h using
h=ρ0ρD(cπf2−2t+0.85ah(1−1.28ϕh√)ϕh)
(15.111)
or we can calculate the cut-off frequency for a given thickness
f2=ρ0cπ(ρDh+2ρ0(t+0.85ah(1−1.28ϕh√))/ϕh).
(15.112)
2. Let the tension be 60% of the maximum value, so that
T=0.6σmaxh.
(15.113)
3. Using this tension, choose the lowest frequency of interest (or fundamental resonance frequency) f0, which is given by Eq. (15.101) and repeated here for convenience
f0=13.26aTaρ0−−−√
(15.114)
From this we obtain the membrane radius a in terms of the fundamental resonance frequency f0
a=12.2Tf20ρ0−−−−√3.
(15.115)
When building an electrostatic loudspeaker, it is easier to adjust the tension by measuring the resonance frequency, like tuning a drum skin, rather than by measuring the tension directly, as Eq. (15.114) gives a direct relationship between the two.
4. Assuming the maximum peak input signal is equal to the polarizing voltage such that 12e˜in=EP, we obtain from Eq. (15.93) the maximum rms sound pressure at f0 assuming Q=1
so thatζ≈1
SPL0=94+20log10(ε0πa2f02√rc⋅E2Pd2)dBatf0.
(15.116)
If this is not loud enough, then it is necessary to lower f0 by increasing a. Note that this equation ignores the attenuation due to the resistance RS of a dust screen.
5. Choose a suitable Q value of 0.7–1.0. Again, using T=0.6·σmax·h, we obtain the optimum specific acoustic resistance of the covering cloth from Eq. (15.103) as follows
RS=2Qρ0Ta−−−√.
(15.117)
6. Assuming EP/d=2000 V/mm and T=0.6·σmax·h, we obtain from Eq. (15.86) the gap distance between the membrane and each stator
d=ε0a2T(EPd)2.
(15.118)
7. Then we obtain the polarizing voltage using simply
EP=(EPd)d.
(15.119)
8. If a delay line is used, this may be designed using the formulas given in Sections 15.6and 15.8
. Some experimentation will be needed to find the maximum value of the termination resistance RT that does not result in negative capacitor values yet guarantees optimum efficiency. If the push–pull transformer shown in Fig. 15.1 has a turns ratio of 1:τ+τ, and the nominal input impedance is Znom, then the turns ratio is given by
τ=RT2Znom−−−−√.
(15.120)
Once the design has been arrived at, the frequency response with normal directivity may be calculated using Eq. (15.93). Constant directivity may be modeled by including the factor ω1/(jω+ω1), where ω1=2c/a.
Example 15.1. We now use the equations of Section 15.17, to derive the tension, resistance, gap width, and polarizing voltage required for the 280mm diameter membrane for which we derived the on-axis pressure and directivity pattern in the previous sections while assuming an ideal membrane with no mass or tension. If we use a stator with a thickness of t=1mm perforated with holes of radius ah=1.5mm and pitch bh=5mm in a triangular array, the porosity is given by
ϕh=2πa2h3b2h√=2×3.14×1.523×52√=0.33.
Then if we use a membrane with a thickness of h=12μm and density of ρD=1400kg/m3, the upper cut-off frequency is given by
We let Q=1 so that we can obtain the optimum specific acoustic resistance of the dust screen
RS=2Qρ0Ta−−−√=211.18×4320.14−−−−−−√=121rayls.
Referring to Table 4.1 on p. 126, this value may be made up from using a 75rayls dust screen on the front and a 47rayls screen on the back, for example. The gap between the membrane and each stator is given by
These values are used in Eq. (15.93) to calculate the on-axis response shown in Fig. 15.25 for an electrostatic loudspeaker without a delay line. Also shown is the on-axis response of the same loudspeaker calculated from rigorous equations [10] from Section 14.10 based on an analytical model. This reveals an excellent correlation between the lumped-element model and the analytical one, except that the latter shows a 2.7dB dip at 330Hz, which is smoothed out by the simpler lumped-element model. Because the simple model cannot reproduce any harmonics above the fundamental resonance, this correlation is dependent on the membrane being well damped through the use of a dust screen. The simple model appears to overestimate the peak displacement by around 10% at the lowest frequencies, whereas the rigorous model keeps it just within the gap width of 1.6mm.
Note that the plot of Walker's voltage equation shown in Fig. 15.25 uses the following version of Eq. (15.20) which is modified to include the resistance RS of the dust screen
Next, we combine the on-axis response of a real membrane, based on the lumped-element model, from Eq. (15.93) with Eq. (15.52) for the far-field pressure of an ideal electrostatic loudspeaker with a delay line to yield that of a real membrane with a delay line
p˜(r,θ)=−jka2p˜02S2DZAR2S2DZAR+ZMD⋅e−jkr4rD(θ),
(15.122)
where D(θ) is given by Eq. (15.53). This is plotted in Fig. 15.26 for θ=0 (on-axis). Also, the reference sensitivity is given by Eq. (15.19), modified to include the resistance of the dust screen
Assuming 12e˜in=EP=3200V, d=1.6mm, a=160mm, and RS=121rayls, this gives a pressure of 4.31Pa rms or 106.7dB SPL with an input voltage of 2×3200V rms.
Although the high-frequency response appears to roll-off prematurely, in practice this can be compensated for by omitting some of the stray neutralizing capacitors.
Problem 15.1. A simple electrostatic loudspeaker with no delay or any kind of directivity control whatsoever has a diameter of 10cm (i.e., radius of 5cm). Assuming a maximum field strength of 2000 V/mm, calculate the maximum rms sound pressure in Pa and dB SPL re 20μPa at 1m on-axis at 1kHz. Assume that 12e˜in=EP.
Hint: From Eq. (15.20), prms(r,0)=2πfε0a2(EP/d)2/(22–√rc) and use Eq. (1.18) to calculate the sound pressure level in dB SPL.
Problem 15.2. Suppose that the loudspeaker has a gap width of d=0.224mm. Calculate the polarization voltage EP, minimum tension T needed for stability (from Eq. 15.86), resonance frequency f0, and specific resistance RS needed for unity Q. Assuming the maximum membrane stress before breaking is 55MPa, calculate the maximum allowable tension according to Eq. (15.113).