LOW RF INTERFERENCE SWITCHING AMPLIFIER AND METHOD

CROSS-REFERENCE TO RELATED APPLICATIONS) 01] This application claims the benefit of U.S. Provisional Application

No. 60/887,394, filed January 31, 2007. The entire disclosure of the prior application is considered to be part of the disclosure of the instant application and is hereby incorporated by reference therein.

TECHNICAL FIELD

[001] This invention relates to switching amplifiers and methods, and, more particularly, to a system and method for reducing the electromagnetic interference of switching amplifiers.

BACKGROUND OF THE INVENTION

[002] Switching amplifiers provide far greater efficiency than their analog counterparts, primarily because transistors used to switch voltages to a load are either turned ON, so that the voltage across the transistor is relatively low, or turned OFF, so that the current through the transistor is relatively low. With either a low voltage across the transistor or a low current through the transistor, the power dissipated by the transistor is relatively low.

[003] Although conventional switching amplifiers are widely used, they can produce excessive distortion in their output signals because of capacitive coupling between the gates of respective switching transistors used by such amplifiers. Another limitation of conventional switching amplifiers is that they can sometimes generate excessive electromagnetic radio frequency ("RF") interference, that can interfere with the operation of the amplifier as well as with other electronic devices in the vicinity of the amplifier. This RF interference can be attenuated to some extent by coupling the

load driven by the amplifier to filters formed by inductors and/or capacitors. However, the remaining RF interference can still be too high in some applications.

[004] Attempts have been made to minimize the distortion of signals output from switching amplifiers by operating them in a balanced manner so that capacitive coupling to one side of a load is matched by capacitive coupling to the other side of the load. While this approach is successful in minimizing signal distortion, it actually increases the RF interference generated by the amplifier since the number of transistors switching must be increased.

[005] Attempts have been made to minimize the frequency at which the peak amplitude of the RF interference occurs by varying to "dithering" the switching times of the transistors, but doing so tends to increase the distortion of the output signal since a signal input to the amplifier is not sampled at regular intervals. In addition, spread spectrum EMI reduction is limited by audible distortion products at higher deviations. This distortion is normally greater at higher output amplitudes, so it clearly is a distortion since it increases with amplitude.

[006] There is therefore a need for a system and method for operating switching amplifiers in a manner that minimizes the magnitude of RF interference generated by the amplifier, and does so without introducing significant distortion.

SUMMARY

[007] A switching amplifier and method includes first and second output terminals to which a load may be connected. An input signal is applied to switching amplifier. The switching amplifier applies a plurality of periodic first pulses to the first output terminal and it adjusts the widths of the first pulses as a function of the magnitude of the input signal. The switching amplifier also applies a plurality of periodic second pulses to the second output terminal. The second pulses are asynchronous with, but substantially equal in number to, the first pulses. If the input signal has positive and negative polarities, the switching amplifier may obtain positive and negative samples, respectively. In response to the positive samples, the first pulses

are periodic and their widths are modulated, and the second pulses have a constant width. In response to the negative samples, the second pulses are periodic and their widths are modulated, and the first pulses have a constant width. The widths of the modulated pulses may be greater than the width of the unmodulated pulses by a magnitude corresponding to the amplitudes of the input signal.

BRIEF DESCRIPTION OF THE DRAWINGS

[008] Figure 1 is a block diagram of a prior art switching amplifier that can be operated according to embodiments of the invention. [009] Figure 2 is a timing diagram showing an example of how the switching amplifier of Figure 1 has been operated in the prior art. [010] Figure 3 is a timing diagram showing an example of how the switching amplifier of Figure 1 has been operated in the prior art in an attempt to minimize output signal distortion. [011] Figure 4 is a timing diagram showing an example of how the switching amplifier of Figure 1 has been operated in the prior art in an attempt to minimize electromagnetic RF interference. [012] Figure 5 is a timing diagram showing the operation of the switching amplifier of Figure 1 according to one embodiment of the invention.

DETAILED DESCRIPTION

[013] A typical prior art switching amplifier 100 is shown in Figure 1. The switching amplifier 100 includes a pulse-width modulation ("PWM") modulator 102 that receives a signal through an input line 101. The PWM modulator 102 generates four outputs A-D, which drive the gates of respective transistors 103, 104, 105, 106, which are arranged in a bridge or H configuration. The transistors 103, 104 are connected in series between a supply voltage V+ and ground, and the transistors 105, 106 are similarly connected in series between V+ and ground. A load 114 is connected

between a first output node E formed by the junction between the transistors 103, 104 and a second output node F formed by the junction between the transistors 105, 106.

[014] As shown in Figure 1, the load 114 is formed by a resistive load 110 connected in parallel with a capacitor 109, both of which are connected between a pair of inductors 107, 108. As is well known in the art, the inductors 107, 108 and capacitor 109 form an LC filter that reduces the amount of electromagnetic energy generated by switching the transistors 103-106 ON and OFF. Although Figure 1 shows the load 114 formed by a resistive load 110 and LC filter, the load 114 may instead be inductive, such as a MRI coil, capacitive, such as a piezoelectric acoustic transducer, or some combination on these impedance elements. Also, although transistors 103-106 are shown as being the devices used to switch various voltages to the load 114 at various times, it will be understood that other switching devices may be used. Finally, configurations of switching devices other than that shown in Figure 1 may also be used to apply various voltages to the load 114 at various times.

[015] The operation of the amplifier 100 is shown in Figure 2 in which six waveforms A-F have been labeled to show an example of the signals present at the corresponding nodes A-F when operating the amplifier 100 in a conventional manner. The signal A is initially driven high to turn ON the transistor 103 for a period corresponding to the amplitude of the signal applied to the PWM modulator 102. At the same time and for the same period, the signal B is driven low to turn OFF the transistor 104. The transistor 106 is ON during this time. As a result, the voltage V+ is connected to ground through the inductor 108, resistive load 110 and inductor 107, thereby causing current to start flowing through the resistive load 110. Were it not for the filter formed by the inductors 107, 108 and capacitor 109, the sudden increase in current through the transistors 103, 106 resulting from turning the transistor 103 ON would result in substantial RP interference. However, this RF interference is attenuated to some extent by the filter formed by the inductors 107, 108 and capacitor 109.

[016] At the end of the period corresponding to the amplitude of the signal applied to the PWM modulator 102, the signal A transitions low to turn OFF the

transistor 103, and the signal B transitions high to turn ON the transistor 104. Again, if the inductors 107, 108 and capacitor 109 were not present, substantial RF interference might be generated by switching the transistor 103 OFF.

[017] A predetermined period later, the above sequence is repeated except that the transistor 103 is turned ON and the transistor 104 is turned OFF for a period that is shorter than the period that the transistor 103 was previously turned ON because the amplitude of the signal applied to PWM modulator 102 is lower. As a result, a current flows through the resistive load 110 for a shorter period. The above sequence is then repeated again for an even shorter period corresponding to the lower amplitude of the signal applied to the PWM modulator 102.

[018] Next, the polarity of the signal applied to the resistive load 110 is reversed by repeating the above sequence except that the transistor 105 is turned ON instead of the transistor 103, and the transistor 106 is turned OFF instead of the transistor 104. Thus, when the input signal to the PWM modulator 102 is positive, the pulse width of the signals applied to the transistors 103, 104 are modulated, and when the input signal to the PWM modulator 102 is negative, the pulse width of the signals applied to the transistors 105, 106 are modulated.

[019] The switching amplifier 100 operating as described above provides adequate performance in may cases, but it results in distortion that can be excessive when low distortion amplification is required. The reason for this distortion is essentially the capacitive coupling between the gates of the transistors 103-106 and the nodes E, F, which effectively distorts the width of the pulses generated at the nodes E, F. When this capacitive coupling occurs, the width of these pulses no longer correspond to the amplitude of the signal applied to the PWM modulator 102.

[020] One approach to reducing the distortion of the amplifier 100 operating as described above is to operate the amplifier 100 as shown in Figure 3 in which the same signal is applied to the PWM modulator 102 as in the example shown in Figure 2. As shown in Figure 3, when the input signal to the PWM modulator 102 is positive, the pulse width of the signals applied to the transistors 103, 104 are still modulated, and

when the input signal to the PWM modulator 102 is negative, the pulse width of the signals applied to the transistors 105, 106 are still modulated. However, when applying pulsewidth modulated signals to a first side of the load 114, unmodulated signals are applied to the first side of the load as well as second side of the load 114. Applying unmodulated signals to both sides of the load, mitigates differential charge injection. The transistors 103, 104 are switched ON and OFF respectively for a period that is constant when the input signal to the PWM modulator 102 is positive, and the transistors 105, 106 are switched ON and OFF, respectively, for a period that is constant when the input signal to the PWM modulator 102 is negative. When both transistors 103, 105 are turned ON, the effect is as if neither transistor 103, 105 was turned ON. However, the capacitive coupling from the gate of the transistor 103 to the node E is matched by the capacitive coupling from the gate of the transistor 105 to the node F. The periods during which the transistor 103 is turned ON while the transistor 105 is OFF are identical to the periods during which the transistor 103 is turned ON in the example of Figure 2, as can be seen by the signal E'-F\ As a result, the currents through the resistive load when operating as in Figure 3 are the same and of the same duration as when operating as shown in Figure 2. Moreover, the capacitive coupling of the signal A to the node E occurs the same number of times that the signal C is capacitively coupled to the node F. Similarly, the capacitive coupling of the signal B to the node E occurs the same number of times that the signal D is capacitively coupled to the node F. As a result, the capacitive coupling to the node E cancels out the capacitive coupling to the node F, thereby preserving the width of the pulses generated at the nodes E, F. Therefore, operating the amplifier 100 as shown in Figure 3 results in very little distortion. 1] An additional problem with operating the amplifier 100 as shown in

Figure 2 is the electromagnetic RF interference resulting from switching the transistors 103-108 as explained therein. Operating the amplifier 100 as shown in Figure 3 does not solve this problem. To the contrary, operating the amplifier 100 as shown in Figure 3 can actually exacerbate the problem of RF interference because both

transistors 103, 105 switch at the same time compared to the operation as shown in Figure 2 in which only one of the transistors, 103 or 105, switch. Similarly, in Figure 3, both transistors 104, 106 switch at the same time. Since RF interference is generated each time one of the transistors 103-106 is switched, doubling the number of transistors being switched increases the magnitude of the RF interference.

[022] One approach to reducing the magnitude of the RF interference is to vary or "dither" the timing (but not the duration) at which the transistors 103, 105 are turned ON as shown in Figure 4 in which the signal applied to the PWM modulator 102 is shown in the top line, the signal at the node E is shown in the next line, and the signal at the node F is shown in the bottom line. As shown therein, the operation of the circuit is similar to the operation shown in Figure 3. The unmodulated pulses again have a constant width, and they again are provided to compensate for the capacitive coupling through the transistors 103, 105 receiving the pulse-width modulated signals from the PWM modulator 102. The operation of the switching amplifier 100 shown in Figure 4 differs from the operation shown in Figure 3 in that the time between switching the transistors 103, 105 is not constant. Instead, for example, the duration of the period between the transistor 103 first being turned ON and the transistor 103 being turned ON a second time is different from the period between with transistor 103 being turned ON a second time and then turning ON a third time. By varying the conductive times of the transistors 103-106 in this manner, the frequency at which the peak amplitude of the RF interference spectrum occurs is varied from pulse-to-pulse, thereby distributing the RF interference over a range of frequencies. In contrast, the frequency at which the peak amplitude of the RF interference spectrum occurs when operating as shown in Figure 3 is the same from pulse-to-pulse. As a result, the peak amplitude of the RF interference at output sample rate is significantly higher when operating the amplifier 100 as shown in Figure 3 compared to operating as shown in Figure 4.

[023] Although operating the amplifier 100 as shown in Figure 4 significantly reduces the magnitude of the RF interference, the amplifier 100 can nevertheless generate RF interference that can be excessive in some instances. Additionally,

operating the amplifier 100 as shown in Figure 4 can produce excessive distortion because the signal applied to the PWM modulator 102 is not sampled at a constant rate. The technique disclosed in this embodiment causes low level constant distortion, such as tape hiss, which is not perceived to be distortion by most listeners. Dithering can therefore be pushed higher (reducing measured EMI) before encountering customer complaints.

[024] According to one embodiment of the invention, the switching amplifier is operated as shown in Figure 5. The signal applied to the PWM modulator 102 is again shown in the top line, the signal at the node E is shown in the next line, and the signal at the node F is shown in the bottom line. When the input signal is positive, the width of the pulses at node E are again modulated, and when the input signal is negative, the width of the pulses at node F are again modulated. The unmodulated pulses continue to have a constant width, and they again are provided to compensate for the capacitive coupling through the transistors 103, 105 receiving the pulse-width modulated signals from the PWM modulator 102.

[025] The operation of the switching amplifier 100 according to one embodiment of the invention as shown in Figure 5 differs from the operation shown in Figure 4 by varying or "dithering" the times between only the unmodulated pulses. The times between the PWM modulated pulses are constant. As a result, the sampling times of the input signal can be constant, thus avoiding distortion in the output signal from the amplifier 100. Yet, by varying or "dithering" the switching times of the unmodulated pulses used to compensate for the capacitive coupling of the modulated pulses, the frequency of the peak amplitude of the RF interference is varied, thus minimizing the peak amplitude of the RF interference. Operating the amplifier 100 as shown in Figure 5 thus produces relatively little RF interference in a manner that does not result in output signal distortion. The dithering of the unmodulated pulse switching times may vary in a pseudo-random manner.

[026] From the foregoing it will be appreciated that, although specific embodiments of the invention have been described herein for purposes of illustration,

various modifications may be made without deviating from the spirit and scope of the invention. Accordingly, the invention is not limited except as by the appended claims.