Speaker Project CS-8
This is a 1.5-way all-passive 8-litre closed-box speaker project for current-drive (Current Speaker 8). Besides the baffle step, the bass resonance has also been compensated with passive circuits, so the feeding transconductance amplifier can have flat frequency response. Detailed design principles and process with the optimized driver models and simulation results are presented in section 11.5 of the book.
The driver used is a 4½-inch, 4-ohm device Vifa PL11WH09-04, whose operation range extends on current-drive up to 10 kHz, making it thus possible to omit the tweeter. The lower −6 dB cut-off frequency obtained is 56 Hz, which is in this size class a fairly competitive result.
The driver's rated resonant frequency is 67 Hz and the mechanical Q value 1.8, which numbers also hold true surprisingly well. Two drives accrue effective area 116 cm2, which is yet fairly adequate for civilized listening.
The schematic is shown below. Driver A gets current mostly only at low frequencies, while driver B operates in the whole range. To adjust the frequency balance, A receives, however, a little current also at high frequencies via the 33 Ω resistor.
The resonance compensation has been realized with the series RCL network connected across driver B. In this case, it is sufficient to apply the compensation only for driver B because the resonance prominence of A remains quite minor even as such.
The resonant frequency of the enclosed drivers with damping material in place is about 85 Hz, while the mechanical Q value was found to be 1.5, enabling yet passive equalization without significant cost to the drivers' source impedance.
Both inductors are here with ferrite core. The coils should not be situated side by side and preferably not in the same orientation.
The 40 mH inductor can be wound by oneself. It doesn't have to be large in size because its resistance is not critical, and is best made out of 0.4-0.5 mm diameter wire. The 20 Ω series resistance can be divided between the inductor and an actual resistor.
To construct the coil, it is best to first dismantle some existing coil and count the number of turns needed for it. If the dismantled coil had inductance L0 and turns number N0, the desired inductance L is established on the same core with the turns number N0*sqrt(L/L0). Difference in wire thickness is quite irrelevant; but at first it is better to provide some extraneous turns, so that the final adjustment can be performed by lessening them. Instructions for accurate inductance measurement are given in section 13.4 of the book.
Despite their size, both capacitors are of plastic dielectric, preferably polypropylene.
For the resistors, 5 W power handling is enough if they are surrounded by free air. The tolerance of all elements should be 5%.
The simulated frequency responses of the loudspeaker and both drivers are shown below. Here, the drivers are assumed to operate in an infinite front panel and with ideal diaphragms.
The response of driver B has been modified only in the resonance region, whereupon the responses are quite well balanced at low frequencies. The response of driver A, in turn, has been shaped so as to provide the needed baffle step correction in the total response (the sloping between 100 Hz and 1 kHz). A has, however, a little share also at treble frequencies which was found necessary despite the fact that current-drive in itself already tends to reinforce this region.
At low frequencies, the total response rises due to resonance behavior a little above the 0 dB level, which represents unaccentuated and unattenuated reproduction. This relieves a little the attenuation need at high frequencies, improving efficiency.
Enclosure dimensions are shown below. With recommended 19 mm material thickness, the net air volume will be a good 8 litres. Driver B is mounted in the upper hole, which is closer to the panel center.
The air holes in the driver chassis are so narrow that they may be blocked too much when using proper board thickness. Hence, it is worth beveling or rounding the front panel holes from inside to necessary extent (as shown in the side view). The screw spots can, however, be left unbeveled to ensure sufficient thickness in them.
Embedding of the drivers can also be established with a lifting plate that is glued over the front panel and has holes of the size of the flange. This also gives a little more stiffness.
The cabinet is crammed full of cotton cloth rent into shreds (best obtained from sheets), so as to make the Q value low enough. Around the drivers is left empty space so much that they can be mounted freely. It is also important to fill the space between the drivers tightly, for this has a great significance to the resonance parameters.
For the front cover is recommended foam mask, similarly as in the CS-12 project.
The filtering circuits have been housed in a separate back box, so that one doesn't have to be careful of them when stuffing the damping material. The solution also facilitates possible servicing. One should, however, see to it that such a box is not allowed to develop resonations as it vibrates with the cabinet.
The measured current frequency responses are shown below. The scale represents sound pressure at 0.354 A current, at 1 m distance. The bass region has been measured separately from the near field.
The on-axis response is, as a general nature, slightly rising which compensates partially the fall in the power response as the directivity of the drivers increases. The measured bass response agrees well with the simulation result, and the attenuation slope comes out to be 14 dB per octave in both cases.
At the high end, reproduction extends yet to frequencies above 10 kHz, even though with only about half of the amplitude. The peak seen at 9 kHz does not have a counterpart in the driver's characteristics, so the issue is probably about unit-to-unit divergence.
As Vifa drivers are no longer available, and neither is there found a direct substitute that would offer comparable response and resonance characteristics, the schematic as such is currently not usable. However, if we accept 8 Ω drivers and the fact that the speaker's total impedance in the series-mode connection will rise accordingly, then we can use in stead the Seas type CA12RCY which is quite similar but has twice the impedance of the Vifa.
Then, all circuit impedances have to be scaled accordingly; that is, the resistances and inductances doubled and the capacitances halved. The changes needed are as follows:
3.3 mH -> 6.8 mH
33 Ω -> 68 Ω
47 uF -> 22 uF
12 Ω -> 25 Ω
40 mH -> 80 mH
20 Ω -> 40 Ω
55 uF -> 27 uF
The mounting holes also need to be a bit smaller than drawn above.