INTERMEDIATE CLASS-D SEMICONDUCTOR AMPLIFIER FOR A VARIABLE-SPEED ELECTRIC DRIVE SYSTEM WITH MAGNETIC AMPLIFIERS

V. A. Sevastyanov, A. P. Ineshin, A. I. Esin

(Recommended by the scientific seminar of the Electromechanical Faculty of the Tomsk Polytechnic Institute)

At the present time, considerable experience has been accumulated in the introduction and operation of intermediate semiconductor amplifiers, chiefly for circuits of electric-machine automation [1, 2].

Semiconductor amplifiers with signal conversion—the so-called “AC amplifiers”—have proved themselves best, being distinguished by high operational reliability. The occurrence of practically any fault in such an amplifier leads to cessation of amplification of the carrier-frequency signal and to zero output, and consequently to shutdown of the drive.

Fig. 1. Block diagram of a class-D amplifier.

This article is devoted to the features of the development, investigation, and operation of more powerful class-D ISPA units operating on the control circuits of a self-saturating power magnetic amplifier (MA).

The operating principle of a class-D amplifier is explained by the block diagram in Fig. 1. The generator of switching pulses (GSP), whose role, in particular, may be performed by a 50-Hz voltage source, produces symmetrical rectangular voltage pulses of frequency \(f_k\), which are applied to the pulse-width modulator—PWM. The latter converts the control voltage at the amplifier input into the relative pulse duration

\[ \gamma=\frac{T_i}{T_k}, \]

where \(T_i\) is the pulse duration, and

\[ T_k=\frac{1}{f_k} \]

is the repetition period. These pulses are amplified and transformed in shape by a pulse amplifier—PA. The output voltage of the PA, rectangular in shape and with relative duration \(\gamma\), controls the operation of the amplifier output stage—the semiconductor switch \(K\)—in the load circuit, i.e., the control winding of the MA. To reduce the time constant of the CW, an additional resistance \(R_{\mathrm{add}} \gg r_{\mathrm{cw}}\) is connected in series with it; therefore

\[ T_{\mathrm{cw}}=\frac{L_{\mathrm{cw}}}{r_{\mathrm{cw}}}\gg T'_{\mathrm{cw}}=\frac{L_{\mathrm{cw}}}{r_{\mathrm{cw}}+R_{\mathrm{add}}}. \]

The PWM duration modulator is included in the intermediate semiconductor amplifier as an independent element, operates as an intermediate pulse amplifier, and performs the functions of converting and amplifying the control signal \(U_y\). The basic requirements imposed on this element (providing the main voltage gain and noise immunity) are satisfied by the PWM circuit, whose operating principle is explained in Fig. 2.

The source of the switching voltage \(U_k\) is the mains (50 Hz, 33 V), from which the signal is applied through the limiting resistance \(R_b\) to a silicon Zener diode \(d_1\) (type D 813). The silicon

Fig. 2. Schematic circuit of the PWM pulse-width modulator.

the Zener diode, having an asymmetric characteristic, asymmetrically limits the sinusoidal commutation voltage and is used as a “breakdown” resistance (of the switch type) controlling the charge–discharge processes of capacitor “C”.

The process of charging capacitor “C” takes place during the negative half-period of the commutating voltage through the circuit: source \(U_k\) — diode \(d_2\) (D2E) — capacitance “C” — resistance \(R_b\). The voltage across the capacitance during charging reaches the value of the reverse breakdown voltage \(U_{\mathrm{st.ob}}\) of the silicon Zener diode (of the order of 13 V). The time constant of this circuit is selected from the condition of ensuring full charging of the capacitance during a half-period of the commutating voltage and is equal to:

\[ T_z = C(R_b + R_{d2}) \simeq C R_b = \frac{T_k}{8}. \]

The process of discharging capacitance “C” during the positive half-period of the commutating voltage takes place through the circuit: charged capacitance “C,” which is a voltage source, — diode \(d_3\) (D2E) — collector circuit of the semiconductor triode \(T_p\) (P9) — silicon Zener diode \(d_1\), “broken down” by the commutating voltage in the forward direction. The time constant of the discharge circuit of capacitance “C” is equal to:

\[ T_{\mathrm{dis}} = C(R_{\mathrm{st.pr}} + R_{d3} + R_t) \simeq C R_t \]

and is determined mainly by the equivalent resistance \(R_t\) of the collector junction of triode \(T_p\), controlled by the voltage of the input signal \(U_y\). Owing to the large internal resistance of the triode, the discharge of the capacitance is linearized; the current during discharge changes hardly at all and is determined by the expression:

\[ i_p = B I_b, \]

where \(I_b\) is the base current, and \(B\) is the gain factor of the triode.

In the absence of a control voltage at the modulator input \((U_y = 0)\), the semiconductor triode is brought into saturation through the base circuit \(R_m\) by the bias current source \((+)U_{\mathrm{st}}\). The resistance of the collector–emitter junction in saturation is small (tens of ohms), and the discharge-circuit time constant is determined mainly by the resistances of diodes \(d_1\) and \(d_3\) in the conducting direction. Therefore the discharge of capacitance “C” proceeds in a short time, and the output voltage of the modulator \(U_m\) is minimal in amplitude and duration.

When a control voltage of the corresponding polarity \((U_y < 0)\) is present at the modulator input, redistribution of the currents of the modulator input circuit occurs: the input current \(I_{\mathrm{in}}\) increases, while the base current \(I_b\) of the triode decreases. The semiconductor triode is taken out of saturation into the amplification mode. The discharge of the capacitance proceeds with a constant current \(i_p\), inversely proportional to the magnitude of the control signal,

\[ i_p = B I_b = \frac{K_1 B}{U_y}. \]

and takes place over a certain time proportional to the control signal,

\[ T_i = K_2 U_y \]

Here \(K_1\) and \(K_2\) are proportionality coefficients. At the collector of triode \(T_p\), output-voltage pulses of sharply peaked form appear—

... whose amplitude \(U_{\mathrm{m}}\) and duration \(T_{\mathrm{r}}\) are proportional to the magnitude \(U_{y}\).

The relative duration of the output pulse \(\gamma\) varies within the limits from 0 to 0.6 (with \(\gamma_{\max}=0.6>0.5\)), since the maximum possible duration of the discharge time \(T_{\mathrm{m}} > \dfrac{T_{\mathrm{k}}}{2}\) of the charge is due to the asymmetrical characteristic of the silicon Zener diode (Fig. 2).

The initial bias of the triode, determined by the resistance \(R_{\mathrm{sm}}\), is selected from the condition of bringing it into saturation, taking into account the shunting action of the circuit of the control-signal source. Therefore \(\gamma=f(U_{y})\) has a dead zone with respect to the control signal, which is necessary for taking the triode out of the saturation mode into the amplification mode (Fig. 3).

Fig. 3. Modulator characteristics for two adjustment cases.

When the input circuit of the amplifier is open (\(I_{\mathrm{in}}=0\)), the base current \(I_{\mathrm{b}}\) increases and the semiconductor triode is brought into a deep-saturation mode, which leads to reliable blocking of the entire amplification path of the semiconductor and magnetic amplifier. The saturation mode of the modulator triode also determines the improved noise immunity of the amplifier input circuit.

During the intervals of discharge of the capacitance \(C\), the input signal with superposed interference acts on the PWM process. During the remaining time, the PWM is as if self-disconnected from the source of the input signal and interference. The principal interference is transmitted to the PWM input by the electromagnetic-correction circuit [4] and is the PPV output voltage, transformed in the correction winding, operating in the switching mode. By including a tuned filter in the correction circuit, it is possible to achieve fulfillment of the condition that the PWM and the interference voltage at its input be in phase. In this case, the action of the total input voltage on the PWM occurs only during time intervals coinciding with the value of the principal useful part of the direct-current control signal, while the periodic interference has no effect on the PWM [3].

The complete circuit of the pulse-width modulator, shown in Fig. 4, differs from the considered schematic circuit (Fig. 2) by the presence of a second analogous circuit for controlling the charge-discharge of the capacitance \(C\). The switching voltage controlling the operation of this circuit,

Fig. 4. Complete circuit of the duration modulator—PWM.

shifts the processes of charging and discharging the capacitance by a half-period in comparison with the first circuit. This brings about the alternate discharge of both capacitors through the common resistance \(R_t\), controlled by the input signal—the semiconductor triode. The voltage across the semiconductor triode is a series of pulses following at twice the frequency \(2 f_k = 100\) Hz and \(\gamma\), varying from 0 to 1 depending on the magnitude of the control voltage \(U_y\) (Fig. 5). This voltage is applied through an isolating silicon Zener diode to the output stage of the modulator, assembled according to the usual common-emitter circuit for connecting a semiconductor triode (type P-13). In the absence of the input signal \(U_y\), the output stage of the modulator is brought into saturation by the base-bias circuit. When a control signal is present at the modulator input, positive voltage pulses arrive at the base of the triode of the output stage, bringing the triode into the cutoff state for the duration of their length. At the modulator output a series of approximately rectangular pulses of negative polarity is formed, whose relative duration varies as a function of \(U_y\) from 0 to 1 (Fig. 5). The characteristic of the modulator is approximately linear and has an adjustable dead zone; the slope of the modulator characteristic is

\[ K = \frac{\Delta \gamma}{\Delta U_y} = 0.01 \ \frac{1}{\mathrm{mV}} \]

and is sufficient to provide a speed-control range of the electric drive of about 200:1 without the use of additional voltage-amplification stages.

Figure 6 shows the circuit of the output stage; its main distinguishing feature is that, in the absence of a control signal at the amplifier input \((U_y = 0)\), the output semiconductor-

output switch is on (is in saturation mode). In addition, the “normally on” state of the output triode makes it possible to combine the control and bias functions on one and the same control winding of the power magnetic amplifier.

Fig. 5. Oscillograms of the operation of the pulse-width modulator

The circuit consists of two amplification stages on semiconductor triodes (types P-13 and P4V); the collector circuits are supplied with a voltage of negative polarity \((-20\ \mathrm{V})\). The first preliminary stage, using a common-emitter circuit on a low-power P13 triode, serves to amplify and invert the control pulse signal \(U_c\); in the absence of this signal the triode is in the cutoff state.

(off) by passive locking from the saturated output transistor of the modulator.

The second, i.e., the actual output stage of power amplification (the switch), is made with a P4V-type transistor according to a common-emitter circuit and is normally

Fig. 6. Schematic diagram of the output stage of a class-D semiconductor amplifier.

open (in the absence of a control signal)—it is in the saturation mode. The output transistor forms a controlled arm of a bridge consisting, on one side, of the output transistor and its collector resistance \(R_k\), and, on the other side, of an ohmic divider on resistances \(R_{\mathrm{dop}}\) and \(R_1\). The greater part of the collector current of the saturated output transistor, flowing through the control winding and the additional resistance \(R_{\mathrm{dop}}\), forms the bias current \(I_{\mathrm{cm}}\) of the magnetic control amplifier.

The input control signal is supplied from the output of the modulator in the form of a series of pulses of negative polarity, likewise with frequency \(2 f_k = 100\) Hz. In the presence of control pulses, the first transistor opens, going into the saturation mode, while the output transistor is locked, going into cutoff under the action of the voltage source of positive polarity in the base divider \(R_{\mathrm{b2}} - R_{\mathrm{cm}}\). For large control signals \(\gamma \to 1\), the current of the control winding falls to 0 and then reverses, fully opening the power magnetic control amplifier.

The semiconductor diode \(d_4\) (type D7V) protects the output transistor from collector overvoltages.

The pulse overvoltage arising when the power circuit of the magnetic control amplifier is switched on, having “+” polarity at the beginning of the control winding, is closed in the discharge circuit: control winding—\(R_{\mathrm{dop}}\)—opening diode \(d_4\). The semiconductor diode \(d_5\) (type D7Zh) in this case prevents passage of the overvoltage pulse into other circuits of the semiconductor amplifier, interrupting the path of its passage.

In real drives with a magnetic control amplifier, a spectrum of disturbances of two clearly expressed frequencies arrives at the input of the semiconductor amplifier: the line-frequency disturbance, predominant in most cases, \(f_p = f_k = 50\) Hz, and the pulse-width-modulation frequency, 100 Hz.

The oscillogram in Fig. 7b shows the passage of the disturbance \(f_p = 50\) Hz through the PWM. Since \(f_p = \dfrac{100}{2}\), for one PWM pulse it acts as positive feedback, and for another—as

negative. The oscillogram clearly shows the indicated pairs of pulses, one of which is widened in proportion to the signal \(\Delta U_y+\Delta U_p\), while the other is narrowed in proportion to the signal \(\Delta U_y-\Delta U_p\). In this case the average value \(\gamma\) over two commutation periods remains approximately the same. Therefore, the influence of \(f_p=50\) Hz, when the amplifier operates on the linear part of the characteristic (input—output), does not affect the operation of the drive with the MAS. When the amplifier enters the saturation region, this action is reduced mainly to a decrease in the gain coefficient of the IPA. Theoretically, the most dangerous disturbances are those whose frequencies are close to the internal commutation frequency of the class-D IPA. The oscillogram in Fig. 7a shows the passage through the PWM of an artificially created disturbance of frequency \(f_p=95\) Hz. In this case, low-frequency beats are formed at the amplifier output, which are followed by the actuator motor (shown by the dashed line in Fig. 7a). In actual drive systems, however, the appearance of such disturbances is unlikely. On the oscill—

Fig. 7. Oscillogram of the passage of a disturbance signal through a class-D amplifier.
a—disturbance voltage of frequency 95 Hz;
b—disturbance voltage of frequency 50 Hz.

...the oscillograms below the zero line show the pulses of the amplifier output voltage \(U_{\text{out}}\) and the interference \(U_{\text{i}}\) with frequency \(f_{\text{i}}\). At the bottom of the oscillogram, time marks are given with \(T = 0.02\) sec.

The amplifier output pulses were additionally amplitude-modulated at a frequency of 300 Hz by the pulsating output voltage of the rectifier.

The semiconductor amplifier considered was manufactured and investigated in the sectoral laboratory of the Ulyanovsk Polytechnic Institute and at UZTS.

Conclusions

1. The class-D semiconductor amplifier is distinguished by high reliability and increased output power, which is necessary when matching the semiconductor amplifier with a magnetic amplifier system.

2. The developed version of the pulse-width modulator circuit, with the corresponding adjustment of the filter in the correction circuit, has satisfactory noise immunity.

3. The positive results obtained in factory testing of the class-D amplifier make it possible to recommend it for use as a preliminary amplification stage in the electric drive of heavy machine tools.

References

1. Collection 1, Adjustable Electric Drives and Contactless Automation in Mechanical Engineering, Moscow House of Scientific and Technical Propaganda named after F. E. Dzerzhinsky, 1962.

2. Leningrad House of Scientific and Technical Propaganda, T. A. Glazenko, Class-D Semiconductor Amplifiers for DC Electromechanical Automation Systems, Leningrad, 1963.

3. Ya. Z. Tsypkin, Theory of Pulse Systems, Fizmat, 1958.

4. M. A. Rozenblat, Magnetic Amplifiers, Gosenergoizdat, 1960.