HIGH-SENSITIVITY ELECTRONIC RELAYS

M. S. ROITMAN, V. I. ORLOV

(Presented by the scientific seminar of the Faculty of Automation and Computer Engineering)

At present there is an acute need for the creation of simple, highly reliable electronic relays with an operating voltage below 50 mV at a current of less than 10 μA.

Progress in recent years in the field of modulators with low zero drift makes it possible to create electronic relays with an operating voltage on the order of 20 μV at \(I_{\mathrm{op}} \leqslant 10^{-8}\) A. However, such devices are complex (they are a combination of a modulator, an amplifier, and an output relay stage), do not always meet reliability requirements, and have considerably lower speed of response (by several times) than electronic relays without signal modulation. The principal drawback of existing electronic relays without preliminary amplification is the substantial magnitude of the operating voltage and current.

The latter is due to the fact that existing switching circuits with a linear load in the collector circuit require, in order to transfer the triode from the “cutoff” mode to the “saturation” mode, the supply of a significant current to the base.

The effort to create simple electronic relays with a minimum operating voltage prompted the authors to carry out a detailed investigation of the possibility of operation of semiconductor triodes in the relay mode with small input signals. This possibility is of interest because, if it can be realized, it is possible to create simple electronic relays with a very low operating power. The latter follows directly from consideration of the dependence of the signal power required to change the collector current by \(m\%\) on the operating mode of the triode (Fig. 1)\(^{1}\).

It is known that, in order to bring a semiconductor triode into the saturated state, a base current is required

\[ I_{\mathrm{bn}} \geqslant \frac{U_k}{R_{\mathrm{e}}\cdot \beta}, \]

where \(U_k\) is the supply voltage of the circuit, \(R_{\mathrm{e}}\) is the equivalent resistance in the collector circuit; in the simplest case it is equal to the load resistance \(R_{\mathrm{n}}\), and \(\beta\) is the current gain. To reduce the value of \(I_{\mathrm{bn}}\) saturating the triode, it is necessary to increase \(R_{\mathrm{n}}\). However, this obvious way of reducing the input signal is practically inapplicable because triodes have a through

\(^{1}\) Such a course of the curve is explained by the fact that at very small signals charges flow out of the base, at large ones they are injected into the base, and at a certain input voltage the currents compensate one another.

collector current \(I'_{k0}\) and its sharp dependence on temperature. For example, for a germanium triode with \(\beta = 50\) and \(I'_{k0}=2\,\mu\text{A}\) at \(U_k = 10\text{ V}\) and \(t^\circ = 20^\circ\text{C}\), \(I_{bn}\) can be reduced to \(5\,\mu\text{A}\) by increasing \(R_n\) to

\[ \frac{U_k}{I_{bn}\cdot \beta}=40\,\text{k}\Omega . \]

In this case, even in the absence of a signal, a voltage \(I'_{k0}\cdot R_n \approx 4\text{ V}\) drops across \(R_n\). When the ambient temperature is increased by only \((8\text{–}10^\circ\text{C})\), this voltage drop will double, and almost the entire source voltage will be applied to the load resistance \(R_n\). It is practically impossible to substantially reduce this disadvantage by using, as the load, a linear temperature-dependent resistance. Consequently, in order to reduce the effect of the through collector current \(I'_{k0}\), it is necessary to use comparatively small load resistances \((R_n \leq 10\,\text{k}\Omega)\).

Fig. 1.

In order to reduce the actuation power and increase the nonlinearity between the input and output voltages, it is desirable to use, as the load, a nonlinear element with small static \((r_{\text{st}})\) and large dynamic \((R_{\text{dyn}})\) resistances. In this case, in the absence of an input signal \(U_n = I'_{k0}\cdot r_{\text{st}}\), while when a signal is applied \(U_n = I'_{k0}\cdot r_{\text{st}} + \Delta I_k\cdot R_{\text{dyn}}\). Obviously, the actuation power will be the smaller, and the relay character of the characteristic the greater, the larger the ratio \(R_{\text{dyn}}/r_{\text{st}}\). It is easy to see that a nonlinear element can be used successfully only if changes in \(I_k\) caused not by the input signal (for example, a change in ambient temperature) lead to \(\Delta U_k = \Delta I_k\cdot r_{\text{st}}\). Thus, the optimal element would be one that had:

1. Low static resistance.
2. High dynamic resistance.
3. Temperature characteristics coinciding with the characteristics of the triode.

Low static resistance makes it possible to have a small initial voltage drop across the load, which remains insignificant even when the temperature rises, when the through collector current increases sharply.

High dynamic resistance makes it possible to obtain high switch sensitivity.

When the temperature rises, the static resistance of the load element must decrease, which will lead to compensation of the effect of the increasing collector current of the triode.

The above requirements are more or less satisfied by a \(p\)-\(n\) junction (in the simplest case, a junction diode). A circuit of a voltage relay with a diode as the load resistance is shown in Fig. 2. The volt-ampere characteristics of the relay can be obtained by superposing the characteristic of the diode in the nonconducting direction and that of the triode (Fig. 3),

It is not difficult to see that the relay will operate normally only under the condition that the triode current \(I_{k0}\) is less than the reverse current of the diode \(I_d\). In this case, the voltage drop across the diode will be less than \(0.3\ \text{V}\) when there is no signal at the base of the triode. Applying an opening voltage to the base leads

Fig. 2. Relay circuit with a diode used as the load.

Fig. 3.

to an increase in the collector current \(I_k\), and as soon as \(I_k\) becomes equal to \(I_{k1}=I_d\), the triode will begin to switch into saturation mode. From Fig. 3 it is clear that, in order to switch the triode into saturation mode when a diode is used as the load resistance, the current \(I_{bk2}\) required is much smaller than with a constant (linear) resistance \(R_n\) \((I_{k2} \gg I_{k2}')\). We tested about 50 relays assembled according to the circuit of Fig. 2. The averaged experimental characteristics are shown in Fig. 4.

Fig. 4.

The temperature instability of the circuit is determined by a somewhat greater value than that of the diode, by the temperature dependence of the triode cutoff current when the resistance of the base circuit is above \(R_{b\max}\). Thermal compensation of the circuit can be accomplished by shunting the triode with a resistance. The steepness of the output characteristic, in this case, naturally decreases.

The main disadvantages of circuits with diodes are determined by the uncontrollability of their characteristics, and also by the small values and considerable scatter of the reverse currents of the diodes \(I_t\). A more flexible circuit is one using a semiconductor triode as the load (Fig. 5).

Fig. 5. Relay circuit with a triode as the load.

By changing the value of the resistance \(R_b\), one can regulate, over fairly wide limits, the collector current of the upper triode from \(I_k = I_{k0}(1+\beta)\) with \(R_b\) short-circuited to \(I_k = I_{k0}\beta\) at \(R_b=\infty\), which makes it possible, without selecting triodes, to obtain voltage relays with identical characteristics.

By using triodes of the same type, one can obtain good temperature stability of the circuit.

A characteristic feature of the operation of switching circuits with a nonlinear load is that none of the triodes operates in the cutoff mode. When the switch operates, one of the triodes (the load triode) leaves the saturation mode, and the other enters it. As a result, the entire source voltage, initially applied to the lower (operating) triode, becomes applied to the upper (load) triode. Thus, in order to accomplish switching it is sufficient to increase the collector current of the triodes, according to Fig. 6, by the amount

\[ \Delta I_k \simeq 2\frac{U_k}{r_k^*} \tag{1} \]

where \(U_k\) is the supply voltage of the switch;
\(r_k^*\) is the differential resistance of the collector junction for the common-emitter circuit.

Analyzing formula (1), one can determine the limiting sensitivity of the switch circuit with a nonlinear load.

Let us use the known relations [1]

\[ \Delta I_k = \Delta I_b \cdot \beta \quad \text{and} \quad r_k^* = \frac{r_k}{1+\beta}, \]

where \(\Delta I_b\) is the increment of the base current;
\(\beta\) is the current transfer coefficient in the common-emitter circuit;
\(r_k\) is the differential resistance of the collector junction in the common-base circuit.

Substituting into (1), we have

\[ \Delta I_b = \frac{2U_k}{r_k \cdot \alpha}, \tag{2} \]

where

\[ \alpha = \frac{\beta}{1+\beta} \]

is the current transfer coefficient in the common-base circuit.

In deriving the expression for \(r_k\), we assume that the magnitude of the differential resistance of the collector junction depends on two phenomena in the triode:

Fig. 6.

1) modulation of the base width;
2) impact ionization in the region of the collector junction.

The magnitude of the collector current in the CB circuit is determined as

\[ I_{\mathrm{k}}=\alpha_{\mathrm{m}} I_{\mathrm{e}}+I_{\mathrm{ko}}, \tag{3} \]

where \(\alpha_{\mathrm{m}}\) is the current transfer coefficient with allowance for impact ionization, equal to

\[ \alpha_{\mathrm{m}}=M\alpha=M\gamma\left(1-\frac{w^{2}}{2L^{2}}\right), \]

here

\[ M=\frac{1}{1-\left(\frac{U_{\mathrm{k}}}{U_{\mathrm{m}}}\right)^{n}} \]

is the impact-ionization coefficient,

\[ \begin{aligned} \gamma &\text{ is the injection coefficient,}\\ w &\text{ is the base width,}\\ L &\text{ is the diffusion length.} \end{aligned} \]

Substituting the value of \(\alpha_{\mathrm{m}}\) into (3), we have

\[ I_{\mathrm{k}}=M\gamma I_{\mathrm{e}}\left(1-\frac{w^{2}}{2L^{2}}\right)+I_{\mathrm{ko}}. \tag{4} \]

It is known that \(\dfrac{dw}{dU_{\mathrm{k}}}\) is determined through the physical parameters as [1]

\[ \frac{dw}{dU_{\mathrm{k}}} = -\sqrt{\frac{\varepsilon_{0}\varepsilon}{2qN_{\mathrm{d}}}}\cdot\frac{1}{\sqrt{U_{\mathrm{k}}}}. \tag{5} \]

Differentiating (4) with respect to \(U_{\mathrm{k}}\) and using expression (5), we obtain

\[ \frac{dI_{\mathrm{k}}}{dU_{\mathrm{k}}} = \frac{1}{r_{\mathrm{k}}} = \gamma I_{\mathrm{e}} \left[ \frac{ \dfrac{nU_{\mathrm{k}}^{\,n-1}}{U_{\mathrm{m}}^{\,n}} }{ \left(1-\dfrac{U_{\mathrm{k}}^{\,n}}{U_{\mathrm{m}}^{\,n}}\right)^{2} } + \frac{ \dfrac{w}{L^{2}} \sqrt{\dfrac{\varepsilon_{0}\varepsilon}{2qN_{\mathrm{d}}}} \dfrac{1}{\sqrt{U_{\mathrm{k}}}} }{ \left(1-\dfrac{U_{\mathrm{k}}}{U_{\mathrm{m}}}\right)^{n} } \right]. \]

After substituting the expressions for \(\alpha_{\mathrm{m}}\) and \(\dfrac{1}{r_{\mathrm{k}}}\) into formula (2), we obtain

\[ \Delta I_{\mathrm{b}} = \frac{2nI_{\mathrm{e}}}{\left(\dfrac{U_{\mathrm{m}}}{U_{\mathrm{k}}}\right)^{n}-1} + \frac{ 2I_{\mathrm{e}}\sqrt{U_{\mathrm{k}}} }{ \sqrt{\dfrac{2qN_{\mathrm{d}}}{\varepsilon_{0}\varepsilon}} \left( \dfrac{L^{2}}{w}-\dfrac{w}{2} \right) }. \tag{6} \]

The first term in formula (6) reflects the influence of impact ionization on the magnitude of \(\Delta I_{\mathrm{b}}\), while the second reflects the influence of modulation of the base width.

In order to perform calculations using formula (6), it is necessary to know the value of the avalanche-breakdown voltage \(U_{\mathrm{m}}\), which is determined by the formula

\[ U_{\mathrm{m}}=a\rho_{\mathrm{b}}^{m}, \]

where \(\rho_{\mathrm{b}}\) is the specific resistance of the base.

The coefficients \(a, m, n\) for various base types are given in Table 1.

Table 1

As an example, let us estimate the value of \(\Delta I_b\) for triodes of type P26 at \(U_k = 30\) V and \(I_e = 0.1\) mA,

\[ \rho_b = 3\ \Omega\cdot\text{cm}, \]

\[ N_d = 10^{15}\ \text{cm}^{-3}, \]

\[ L = 0.1\ \text{mm}, \]

\[ w = 40\ \mu\text{m}. \]

The value \(\Delta I_b\) is found to be equal to 6 μA.

Formula (6) makes it possible to establish the dependence of the switch sensitivity on the operating mode of the triode and on temperature.

The sensitivity of the switch decreases in proportion to an increase in \(I_e\), and, at small values of \(U_k\), in proportion to \(\sqrt{U_k}\). However, with a further increase in \(U_k\), the phenomenon of impact ionization begins to be observed (especially at elevated temperature), and therefore the dependence of sensitivity on \(U_k\) becomes sharper.

The dependence on temperature is rather complex and has a minimum in the region of temperatures close to zero.

When the temperature is raised, the main influence on the sensitivity is exerted by the first term of formula (6), or, more precisely, by the decrease in the avalanche-breakdown voltage \(U_m\), which leads to a reduction in sensitivity.

As the temperature is lowered, the current gain begins to decrease, which, naturally, leads to a reduction in sensitivity.

The current gain at reduced temperature and at a small current \(I_e\) decreases mainly because of an increase in the surface recombination rate [2].

Fig. 7. Dependence of the relay operating threshold on temperature.

Since formula (6) was derived without taking into account the phenomenon of surface recombination, it cannot be used to establish the dependence of the reduction in sensitivity on decreasing temperature.

Figure 7 presents the experimental dependence \(U_{\text{op}}(\Delta I_b)\) on temperature for a relay using triodes of type P26 at \(U_k = 30\) and \(I_e = 0.1\) mA, which satisfactorily confirms the conclusions stated above concerning the dependence of sensitivity on temperature.

It is known that, in order to improve temperature stability, the current gain should be increased by raising the current level \(I_{\mathrm{e}}\) [2]; however, according to (6), this will lead to a decrease in sensitivity. An increase in the current \(I_{\mathrm{e}}\) causes additional heating of the junctions in the triodes due to the dissipated power. Since the voltages applied to the triodes differ substantially, the additional heating of the junctions in the triodes will also differ substantially, which will lead to additional instability of the circuit as the temperature rises; however, the range of stable relay operation in the region of negative temperatures expands somewhat when the current \(I_{\mathrm{e}}\) is increased.

Consequently, in selecting the operating mode of the circuit, one should proceed from the specific requirements, which limit the temperature range, specify the required sensitivity, and the permissible instability of the relay operating threshold.

An electronic voltage relay acquires a number of additional advantages when it operates together with an indication circuit (Fig. 8).

In the indication circuit, an ordinary relaxation generator using a neon lamp is employed.

The operation of the generator up to the moment when the lamp ignites is described by the equation [3]

\[ C\frac{du}{dt}=\frac{E+U_{\mathrm{out}}(U_{\mathrm{in}})-U}{R+R_{\mathrm{out}}(U_{\mathrm{in}})}. \]

When integrated over the limits from \(U_{\mathrm{g}}\) to \(U_{\mathrm{z}}\), the period of the relaxation oscillations is obtained:

\[ T=[R+R_{\mathrm{out}}(U_{\mathrm{in}})]\cdot C\cdot \ln\left[ \frac{E+U_{\mathrm{out}}(U_{\mathrm{in}})-U_{\mathrm{g}}} {E+U_{\mathrm{out}}(U_{\mathrm{in}})-U_{\mathrm{z}}} \right], \tag{7} \]

where

\[ U_{\mathrm{out}}(U_{\mathrm{in}})\text{ is the voltage at the relay output,} \]

\[ R_{\mathrm{out}}(U_{\mathrm{in}})\text{ is the output resistance of the relay,} \]

\[ U_{\mathrm{g}}\text{ is the lamp extinction voltage,} \]

\[ U_{\mathrm{z}}\text{ is the lamp ignition voltage.} \]

Fig. 8. Electronic relay together with an indication circuit.

As can be seen from formula (7), the frequency of the relaxation oscillations does not remain constant, but is a function of the input parameter \(U_{\mathrm{in}}\) or \(I_{\mathrm{in}}\).

The voltage of the additional source \(E\) must be less than the lamp extinction voltage \(U_{\mathrm{g}}\), while the total voltage of the source \(E\) and the source \(U_k\) must be greater than the ignition voltage. When

\[ E+U_{\mathrm{out}}(U_{\mathrm{in}})\approx U_{\mathrm{z}}, \]

relaxation oscillations begin.

As the input parameter increases, the oscillation period decreases; and when the relay has fully operated, a constant oscillation frequency is established, or, for certain values of \(R\) and \(C\), the oscillations cease and the lamp burns continuously.

The presence of such a dependence of the oscillation period on the input parameter makes it possible to regard this voltage relay—of course, with a certain degree of approximation—as a multiposition relay.

The second advantage of the combined circuit is as follows. The voltage relay retains high sensitivity only when operating—

for a high-impedance load. This makes it difficult to couple the relay stage with other circuits using semiconductor devices. Connecting the relaxation oscillator does not reduce the sensitivity of the voltage relay, since, until the neon lamp ignites, the oscillator does not shunt the relay input at all; and at the instant the lamp ignites it should be assumed that the information has already arrived.

Finally, the third advantage is that a comparatively low-resistance resistor \(R_{c1}\) or \(R_{c2}\), or both together, can be placed in the discharge circuit of capacitor \(C\), and voltage pulses arising at the moments of capacitor discharge, with an amplitude approximately equal to \(U_z - U_2\), can be taken from them.

These pulses can be used to operate subsequent pulse circuits, including circuits with a low-impedance input.

Fig. 9. Schematic diagram of the ER-1 module.

Fig. 10. Schematic diagram of the ER-2 module.

We have developed and put into series production two practical relay circuits:

1. For negative polarity of the input signal, the ER-1 module (Fig. 9).

2. For positive polarity of the input signal, the ER-2 module (Fig. 10).

References

1. I. P. Stepanenko. Fundamentals of the Theory of Transistors and Transistor Circuits. Gosenergoizdat, 1963.

2. N. S. Yakovchuk, V. E. Chelnokov, M. P. Geifman. Junction Transistors. Sudpromgiz, 1961.

3. I. M. Kapchinsky. Methods of Oscillation Theory in Radio Engineering. Gosenergoizdat, 1954.

4. M. S. Roitman, V. I. Orlov, S. A. Gofman. Electronic relays based on complex optoelectronic systems. Proceedings of the Fifth All-Union Conference on Automatic Control. Novosibirsk, 1963.