Questions & Answers
I'm not a businessman and at the moment don't have plans or resources to set up a commercial production line. Instead, I'd rather see some existing or emerging manufactures awaken to the reality and the huge opportunities that the concept offers, possibly also making use of my circuits.
In general, loudspeakers have to be designed for current-drive, starting from the choice of drive units. Conventional speakers are seldom suitable as such. One design challenge is how the drivers can be kept in the current-drive mode when making passive frequency response shapings that are needed to compensate the rising response towards high frequencies due to the baffle step and horn effect of the cone. The bass resonance region also needs some treatment if the mechanical Q value of the system is not low enough.
The book shows how to accurately model the drive units and design these speakers using a circuit simulator, detailing two completed designs:
- A two-way system with tightly tuned 1st-order crossover
- A series mode 1.5-way system with passive compensation
Full-range drivers can also work, but I would not recommend very sensitive or exotic drivers. To get the best results, it is advisable (like usually) to avoid using drivers in their break-up mode region.
If one has to rely on existing speakers, they have to be closed (not reflex) and preferably with low-order crossover, and the response could be straightened e.g. by a graphic equalizer. (Speaker impedance and hence the relative response with respect to voltage drive tends to rise near the crossover frequencies.)
Almost as important as the corrent driving mode is also to fill the cabinet interior tightly by effective damping material to suppress the cabinet noise from leaking through the cone and the walls. This also helps much in keeping down the Q.
As power transistors are quite nonlinear and variable devices, distortion and other
problems can't be avoided in feedbackless design. However, I don't see any particular technical reason to avoid feedback, especially in current-drive.
In conventional voltage amplifiers, giving up negative feedback usually increases the output impedance of the amplifier remarkably. There is every reason to believe that it is mostly this increase in impedance and the consequent decrease in the EMF-derived interference currents that is behind the apparent sonic advantages in such an approach rather than the absence of the feedback in itself.
The sonic superiority of current-drive shows up mostly in the middle and treble regions. Instead, at bass frequencies, where damping issues only have significance, the driving mode is not as important as elsewhere. Therefore, despite applying current-drive for the most part of the spectrum, we still have quite free hands to use various means of damping, also electrical, to treat the fundamental resonance region.
By and large, there is much misinformation in what is commonly conceived of speaker damping and the EMFs. Notably, the significance of electrical damping and so-called damping factor has been greatly overstated in the audio community and by the marketing departments. In the book, the subject is discussed and the myths exposed from an engineering standpoint, with appropriate equivalent circuits, underlying equations, magnitude/phase diagrams and real-world examples clearly presented; as opposed to the merely verbal and vague energy flow /voltage flow jargon and even wishful thinking commonly met considering these issues.
It is important to understand that damping and the Q value of the driver-enclosure combination have effect only near the resonant frequency. Instead, at all other frequencies, from about 200 Hz up for woofers, any driver damping doesn't have any effect at all. This can also be demonstrated by basic modelling with typical driver parameters.
The motional EMF of the driver can actually be called a back-EMF only in the resonance region, where this EMF voltage acts about in phase with the applied signal and therefore reduces the flow of current on voltage drive, thus effecting the damping. Instead, when frequency rises from the resonance region, the EMF voltage soon turns perpendicular to the resistive voltage and current and at the same time decreases in magnitude, going below the resistive component typically in the whereabouts of 150 Hz. Thus, in the whole mid-frequency region, the motional EMF no more damps or controls anything but merely acts as an uncontrolled interference source between driver voltage and current, doing nothing useful.
Electrical damping can in every aspect be substituted by mechanical damping with the same end result. What electrical damping exactly does is to exert to the moving system a mechanical counter-force that is at every moment directly proportional to the instantaneous velocity of the voice coil according to the equation F = (Bl)2v/R (=constant*v), where v is the velocity and R the voice coil resistance (plus other possible series resistances). There are no other effects produced by electrical damping than this velocity-proportional counter-force and the consequent reduction in the total Q. Just the same kind of force is also introduced by mechanical resistance, that can be determined by driver materials and structure and also adjusted by cabinet stuffing. Here, the force is simply F = bv, where b is the total mechanical resistance affecting the moving system.
Thus, there is nothing indispensable in electrical damping; and in principle, there cannot be any difference in the driver's resonance behavior, neither in frequency nor in time domains, whether the damping be accomplished by a low-impedance amplifier or mechanically.
On pure current-drive, the effective Q value is determined solely by the mechanical Q of the system. As all available speaker drivers are designed to work exclusively on voltage drive, their Qm values are usually too high for current-operation as such. However, it would surely not take long to develop self-damping drivers if only some effort were put to it. Even now, there are rubbers that yield free-air Qm values of around 1.5; and according to tests with cotton cloth enclosure stuffing, the final value can yet be considerably lowered from this.
Often it is not even necessary to reach to the 0.7 since with a slightly higher value, the mild boost that develops in the 100 Hz region can be used for benefit to compensate some part of the baffle step.
The damping can also be effected by active equalization with the same end result, and the book introduces several novel circuit ideas for this.
In electrodynamic headphones, the achieved benefits of current drive are ususally very minor compared with the improvement in loudspeaker operation. This is mainly because in the impedance of headphone transducers the relative proportion of the DC resistance is generally much higher than in speaker drivers, so the interfering current components produced by the electromotive forces are left rather small even on voltage drive.
A greater problem is generally constituted by the unevenness of frequency reproduction and its dependence on the ear canal shape. As with loudspeakers, the frequency response of headphones also exhibits certain changes when moving to current-drive. These changes may, depending on the case, also result in undesirable impressions.
The operating mode in itself does not affect efficiency; but at the resonant frequency (and only there), driver efficiency is about proportional to the resulting mechanical Q value. However, even at low Qm values, that are suitable for current-drive, the limit of linear displacement is generally encountered much earlier than any power limitation.
Powers needed to drive the cone to the rated maximum excursion are actually surprisingly low. If we take as an example a middle-sized 8-inch woofer in a closed enclosure of 40 litres, with quite typical parameters: Bl = 9 N/A, Rc = 6 Ω, m = 0.025 kg, f0 = 55 Hz, and mechanical Q = 2.9, the power it takes to move the cone at +/- 5 mm amplitude is at the resonant frequency(f0) only 5 W, and only part of this is dissipated in the voice coil. With real audio signals, where the RMS value relative to peak values is considerably lower than in a sine wave, the average power needed to hit the +/- 5 mm limit mentioned is correspondingly still lower.
No connection. I discovered the benefits of current-drive before I came to know about this experiment through the internet. Their two papers mostly introduced an exceedingly complex amplification system with velocity feedback; and while there were also conducted some comparative distortion measurements, they did little to address the actual flaws of voltage drive which would have been essential to motivate the community. While they did some good work in reminding that there is a better way to go, with voltage drive appearing as the result of established practice and convenience, the articles may also have led to some discouragement to the subject by expressing that the simple current-feedback scheme would be somehow inadequate for quality application when the grounds for such a view are virtually unfounded.
At the hobbyist forum Diyaudio.com, I have had accusers with even such a fiction that the book would be only a copy of the M&H article. So, a 342-page book and 19-page paper! Though the fields overlap, even a glimpse at the contents reveals the absurdity of this claim. Also, besides that I have disagreed in several points with the paper, not one of the book's circuits or measurement or simulation results is found in M&H, nor otherwise has the article been any example.
M&H are not the inventors of current-drive, and nobody can be identified as such. On the whole, the questioning who has invented what is fruitless and moot here.
In electrodynamic (i.e. magnetic) panels, the proportions of motional EMF and inductance EMF in the total voltage of the speaker are much smaller than in voice coil drivers, due to lower flux density and the coilless structure, which is seen as nearly flat impedance. Hence, the magnitudes of the EMF-derived currents relative to the total current remain minor, and electrodynamic panels exhibit much less of the adverse effects due to voltage driving than the usual speakers do.
Electrostatic speakers, in turn, are totally free from actual electromotive forces since there are no magnetic fields involved. ESLs, however, have their own corresponding motion-related effect, that manifests itself as plate current directly proportional to the diaphragm velocity. This motional current, however, does not cause any distortion or impairment as long as it is supplied from a low-impedance signal source, keeping the voltage intact.
Open speakers are also free from any cabinet noise leakage through the diaphragm. Thus, there are found definite reasons for the reputation these systems enjoy, taken the available amplifier technology. (This is not to deny that directivity properties may also play some role.)
Yes. If we feed a speaker through a mere series resistor to increase the source impedance, most part of the amplifier's current capacity remains unutilized, and the power delivered to the driver(s) may be insufficient. In this, a step-up transformer at the amplifier output can help a lot.
For example, if we use a series resistance that is 7 times the driver impedance of 8 Ω, i.e. 56 Ω, and use a transformer with a turns ratio of 1:4, the amplifier sees the 64 Ω load at the secondary as 4 Ω at the primary, which can yet be well handled by most amplifiers. While the series resistor drops the sensitivity by 18 dB, the transformer restores back 12 dB, resulting in a net loss of just 6 dB (disregarding winding resistances).
The distortion of audio transformers decreases as the impedance of the feeding source decreases. Therefore, in this use, transformer distortion is much less a problem than e.g. in the output transformers of tube amplifiers, where the source impedance in relation to the primary inductance is higher.
Yes, if they are equipped with normal voltage feedback and are not bridged. The stability cannot, however, be guaranteed and should be checked with a scope at various load conditions.
The existing feedback resistors can be left in place, as their values are generally orders of magnitude greater than the impedances forming the current feedback. Only the common node of the feedback resistors (which is the inverting input of the power amp) needs to be located from the circuit board and tapped with a wire. A current sensing resistor of about half an ohm or less is then inserted between the speaker and the amp's ground (minus) terminal with the said tap wire connected to the joint of the resistor and speaker. To aid stability, it is advisable to also use a series RC network (e.g. 10-15 Ω and 47 nF) across the speaker, like in the second picture here.
Negative output impedance is quite the opposite of current-drive (that strives to increase the impedance) and should not be confused with or mistaken to be any alternative to it. Negative output impedance has some use only at the lowest frequencies, but elsewhere it only increases distortions for the same reasons that current-drive reduces them. Nor can any active bass control schemes, that often steal the discussion, ever substitute current-drive, that benefits almost the entire audio range.
Negative output impedance is also accomplished by current feedback which is supplied to the positive input node instead of negative. An application of the concept has been so-called ACE bass, where the output impedance is made both negative (by real part) and frequency-dependent. In theory, it has the benefit of reducing the effect of spring nonlinearity and consequent distortion below the resonant frequency.
At best, only partially. Shorting rings or cylinders in the magnetic poles are only effective in reducing magnetic distortion if they are placed next to the voice coil (in which case the high-frequency impedance becomes fairly flat), but this is a rare practice and found mostly only in some full-range transducers. Even in such cases, current-drive is still able to reduce odd-order distortion products to a fraction of what they are on voltage drive.
Current-drive also eliminates certain motional and microphonic EMF effects in the midrange that are not affected by other means.
As a downside of the shorting technique (in addition to cost), parts attached to the pole pieces also have their own characteristic frequency that makes them prone to metallic ringing.