The invention relates to electro-dynamic moving-coil loudspeakers which are designed to be used with current-output amplifiers. It discloses a protection circuit by which the power assumed by a speaker driver can be limited dynamically so that in an overload condition the driver is prevented from overheating while still continuing its operation within the set power limit. The invention is applicable to protect loudspeakers that are used essentially as current-driven, whereupon the signal feeding the speaker is essentially a current signal and not voltage signal as usually.

Background: the Qualitative Superiority of Current-Drive

Today all commercially available audio amplifier and loudspeaker equipment works on voltage drive principle without significant exceptions. This means that the amplifier acts as a voltage source and therefore has small output impedance. However, both technical aspects and listening experiences indicate plainly that voltage drive is the wrong choice if sound quality is considered important. The reason is that the electromotive forces (EMF) generated by the speaker driver disturb the voltage-to-current conversion that in the voltage drive principle is left as the job of the loudspeaker.

The driving force (F), that sets the diaphragm in motion, is proportional to the current (i) flowing through the voice coil according to the formula F = Bli, where the product Bl is called force factor (B = magnetic flux density; l = wire length in the magnetic field). In a voltage-controlled driver, however, this current becomes corrupted in a plurality of ways.

The electrical equivalent circuit of a moving-coil driver unit can be presented as in Figure 1. Rc represents voice coil resistance, voltage source em represents the motional EMF (back-EMF) of the driver and is calculated by em = Blv (v = voice coil velocity), and voltage source ei represents the inductance EMF, that is generated by the lossy inductance of the voice coil.

In general, em dominates at the low end of the driver's operation range, and ei respectively dominates at the high end of the operation range. In a typical cone or dome driver, the sum of the magnitudes of em and ei is at least of the same order than the voltage drop in Rc throughout the whole operation frequency range.

Both em and ei are subject to several disturbance factors, for which reason these voltages do not follow the applied signal linearly but include very manifold distortion components. When feeding the equivalent circuit of Fig. 1 by a conventional low-output-impedance amplifier, that functionally corresponds to a voltage source, sources em and ei and all the indefinite distortion components contained by these modulate the voltage that is left across resistance Rc and hence the current flowing in the circuit, with consequences that are very detrimental to sound quality all over the world.

Instead, by feeding the equivalent circuit of Fig. 1 by a high-output-impedance amplifier, that functionally corresponds to a current source, sources em and ei and the indefinite distortion components contained by these are not able to modulate the current flowing in the circuit, which current is now determined directly by the feeding current source, that follows the applied signal without interference.

The following interference factors muddle voice coil current under voltage drive but naturally not under current-drive:

1) Voice coil acting as a microphone for sound waves reflecting from inside the cabinet and passing through the diaphragm
2) Voice coil acting as a microphone for sound waves from adjacent drivers
3) Mechanical and pneumatic non-idealities of the moving parts causing uncontrollable EMF effects
4) Bl-variation causing fluctuation in impedance's angle and hence phase modulation of current at middle frequencies
5) Position-dependent inductance of voice coil causing both amplitude and phase modulation in current
6) Voice coil inductance depends strongly on signal level causing nonharmonic distortion
7) Resistance changes due to temperature variations and manufacturing tolerances
8) Program-signal-dependent contact resistance variations in connectors and switches

Detailed information about these serious interference mechanisms and current-drive in general is found in the book: Esa Merilšinen: Current-Driving of Loudspeakers, CreateSpace 2010, USA (ISBN: 1450544002).

A speaker driver operates essentially in the current-drive mode when the impedance seen by the driver is high compared to the driver's own impedance at those frequencies that the driver is intended to reproduce. In other words, if an impedance meter was connected in place of the driver, the meter should show as high as possible impedance magnitude. To achieve this, the output impedance of the amplifier must, of course, be high; but in addition, possible crossover filters also have to be designed for high impedance, whereupon it is mostly 1st-order filters that are suitable. (It must be recognized that totally corresponding restrictions would also apply to conventional voltage drive systems if in them one would correspondingly hold on to the requirement of low impedance seen by the driver. In practice however, this matter is just not cared about; and so the actual operation mode of the drivers is often found in some middle ground between voltage- and current-drive, generally, however, closer to voltage drive.)

The current-drive principle is thus applicable for use in both active and passive speakers equally with the voltage drive principle. It is thus very unfortunate that the interference mechanisms listed above are allowed to affect and mar sound quality in all contemporary sound reproduction systems, where voltage drive is held as the standard practice regardless of the realities of physics.

The introduction of current-drive has perhaps been hampered by a prejudice according to which the fundamental resonance of the bass driver could not easily be damped, as so-called electrical damping, that pertains to voltage drive, is absent. However, the needed resonance damping is established by using a driver of low mechanical Q value, by filling the enclosure with effective damping material, and by using an appropriate active or passive attenuation network, by which the bass response can be fine-tuned in place.

According to a widely spread but ill-justified notion, the amplifier's output impedance should be as low as possible in order to 'control' cone movement and to somehow suppress the EMF:s generated by the driver. The physical fact is that cone movement (more precisely the acceleration) can be governed only by current, and the EMF:s themselves can never be suppressed without suppressing the whole speaker; simply because the law em = Blv can never cease to hold, no matter what impedances or circuits are used. Instead, it is the harmful EMF-derived currents that can and must be suppressed, and this is only done by careful current-drive design.

Existing Protection Techniques

Currently used loudspeaker protection means are mostly current limiters or breakers connected in series with the driver. The breaking operation is accomplished by a fuse or relay that disconnects the driver when the average signal level exceeds a specified limit. For current limiting, there has also been used various devices that exhibit temperature dependent resistance, like PTC resistors, light bulbs, and temperature-sensitive elements specifically designed for the purpose ('posistor').

Instead, protection devices connected merely in parallel with the driver are not possible to be used in voltage-driven speakers since in these the current conducted past the driver does not decrease the current through the driver but only causes extra load for the amplifier.

Using fuses and relays is also possible in current-drive speakers; but due to the slowness of fuses and inaccuracy of relays, setting of the activation limit is troublesome. A sudden break-off of the signal is also often not desirable.

In patent application GB2115628A has been presented a protection circuit intended for an ordinary voltage drive speaker, applying a rectification bridge and a voltage-sensitive switch. The rectified signal is, however, not filtered in any way, and so the switch becomes triggered already from the first threshold-crossing wave top.

General Scheme of the Invention

Current-drive offers the possibility to also use transistors as the protection devices since current conducted past the driver is away from the driver itself. This invention describes a means to protect a current-driven speaker driver from excessive power by using a protection circuit scheme shown in Fig. 2.

The voltage across the object to be protected 1 is rectified by a full wave rectification bridge 2a-d, producing a voltage that is low-pass filtered by a leveling filter 3, whose time constant approximately corresponds to the warming time constant of the voice coil of the object to be protected 1. The voltage so produced by the leveling filter 3, being indicative of the temperature rise in the voice coil to be protected, is used through a driving circuit 4 to drive a power transistor 5, that turns conductive when the voltage produced by the leveling filter 3 exceeds a set limit value. When the power transistor 5 is conductive, the current and power assumed by the object to be protected 1 remain within permitted limits although the current fed 6 would be much greater than the current withstood by the voice coil to be protected.

Detailed Description of the Invention

Figure 2 shows a simple embodiment of the invention, using a power MOSFET transistor as the limiting device.

Figure 3 shows an otherwise correspondent protection circuit than Fig. 2, but in addition there has been introduced simple frequency equalization.

Figure 4 shows an otherwise correspondent protection circuit than Fig. 2; but to sharpen the operation, a bipolar drive transistor has been employed.

Figure 5 shows an otherwise correspondent protection circuit than Fig. 4, but using so-called Darlington pairs as both the drive transistor and power transistor.

In Fig. 2, resistor 7 and capacitor 8 make up a leveling filter 3, that produces from the rectified voltage an approximate short-term average. Resistor 9, which is much lower in value than resistor 7, is needed in order that capacitor 8 would be charged towards the average of said rectified voltage, instead of the peak value. An averaging operation is better because the warming of the voice coil is determined by the RMS value; and with practical waveforms, the ratio of peak values to RMS value varies more than the ratio of RMS value to average value (the RMS value being always greater than the average).

Resistance 10 is higher than resistance 7, so that capacitor 8 would not be needlessly loaded. The reaction speed of the protector is then determined by the time constant RC, where R is the value of resistor 7 and C is the value of capacitor 8.

The voltage drop developed in the diode chain 11 is set to correspond to the difference between the voltage across capacitor 8 that is needed to activate the protector and the threshold voltage of MOSFET 5.

By using diodes 11, the firing of the protector is made sharper, and at the same time the dependence on the threshold voltage deviation of MOSFET 5 decreases. However, when dealing with lowest powers, diodes 11 cannot be used because the required voltage drop would become diminutive.

For transistor 5, it is advisable to choose a type of at least about 10 amperes, even if the currents to be expected were smaller, because the higher the transconductance of transistor 5, the more discriminatingly the protector functions. The threshold voltage should be low and its deviation moderate. MOSFETs are, however, quite inexpensive, so it is possible to do some selection with the threshold voltages.

The cooling need of MOSFET 5 can be estimated by the amplifier's maximum current and the maximum voltage of the driver to be protected 1. After the protector turns active, most part of the power is, however, consumed in the amplifier, in whose thermal design the possibility of a short circuit must be regarded anyway.

When the impedance of the object to be protected 1 is strongly dependent on frequency, the voltage across this impedance does not correspond very well the voltage drop in the voice coil resistance, which voltage drop determines the warming of the voice coil. Then, the monitored voltage can be filtered before the rectification and averaging. An example of such a circuit is shown in Fig. 3.

Resistors 12, 13, 14, and capacitor 15 make up a symmetrized correction filter 16, by which it is possible to attenuate high frequencies 0-6 dB, e.g. to compensate voice coil inductance. During one half-cycle, the leveling circuitry 17 gets its voltage from the correction filter 16 via diodes 18 and 2c, and during the other half-cycle via diodes 19 and 2d. Diodes 18 and 19 may be ordinary small-signal diodes, as also those ones 11 leading to the gate of the MOSFET 5.

The leveling circuitry 17 must not load the correction filter 16 too much; and filter 16, again, must not establish too low an impedance in parallel with the driver to be protected 1. If resistances 12 and 14 are, say, some hundreds of ohms, resistance 9 should thus be at least some kilo-ohms and resistance 7, respectively, still higher.

The operation of the protector may yet be sharpened and the threshold voltage dependence lessened by using a drive transistor in front of the power transistor 5.

An N-type MOSFET can be driven with a PNP transistor, as in Fig. 4. The leveling filter 3 is now referenced to the positive node 20, and the protector turns on when the PNP transistor's 21 base-emitter voltage, i.e. the voltage across resistor 22 reaches a certain limit (about 0.6 V), so that resistor 23 begins to gain current and opens the MOSFET 5.

Deviation in the base-emitter voltage of bipolar transistors is minute, so the operation accuracy is good even at low powers. The current gain factor is not of much significance here because the base is driven mostly by voltage.

The base current 24 required for activation should be, however, at least an order of magnitude smaller than the current through resistor 22, which current, in turn, should be much smaller than the current charging the capacitor 8, which current again, has to be small relative to the current in resistor 9. Therefore, the base current 24 can be, in practice, only a few microamperes; and thus, to avoid interferences, high-current wires may not be good to be situated very close to the drive transistor 21 though we are not dealing with any precision circuit.

The waveform of the limited voltage is rather different between the circuits of Figures 4 and 2 because the drive transistor's 21 collector current and the MOSFET's 5 gate voltage proportional to it depend a little on the instantaneous signal value. The circuit of Fig. 4 works by cutting softly the signal peaks exceeding a certain limit, whereas the circuit of Fig. 2 tends to cut the bottom part of waves, leaving the peak shape quite intact.

Using a drive transistor also enables the use of a bipolar Darlington pair as the power transistor, as in Fig. 5. To get enough base current for this pair, a Darlington transistor can be applied as the drive transistor also. The base voltage required for activation is then, however, 2-fold compared with Fig. 4.

The correction filter 16 shown in Fig. 3 can also be annexed to the circuits of Figures 4 and 5 if only the loading capability of the different stages is minded.