Low Voltage Current Mirror

This invention relates to the field of current mirror circuits and, more particularly, to the provision of an accurate current mirror operating on a minimum supply voltage.

Current mirrors are well known in the art as a means for supplying a known current as, for example, in biasing transistor gain stages. While a "current source" may be as simple as a resistor, current mirrors have become increasingly used for several reasons. First, they offer improved circuit performance and more accurate current control than do resistors, and may also require less area on an integrated circuit chip. The most conventional and simple circuit for a current mirror requires high beta transistors, or at least transistors having consistent values of beta, in order to approach the desired correlation between output and reference current. In IC processing, this is, of course, difficult and thus expensive as beta values tend to have a wide range of values and restriction of the usable range of values makes for a low IC production yield.

More complicated current mirror circuits have been devised which are less sensitive to beta variations. One such is known as the Wilson current mirror which decreases the sensitivity to beta value by means of an additional buffer transistor which supplies the base current for the mirror circuit transistors without significantly disturbing the reference current. In such a circuit, however, a high supply voltage is required, due to the series-connected emitter-base junctions. This limitation precludes the use of such a circuit in many of today's miniature electronic devices such as hearing aides, pagers, etc.

It is therefore an object of the present invention to provide a highly accurate current mirror circuit for use with low voltage supplies.

It is a particular object to provide such a circuit which is highly independent of the beta values in an integrated circuit application.

These objects and others which will become apparent are provided in a current mirror circuit wherein the two matched transistor devices are biased by an automatically controlled biasing network. A sensing circuit compares the current in one of the transistors with the reference current and adjusts the biasing voltage, forcing the collector current of that transistor to be equal to the reference current. In this mirror circuit, there are no bias paths containing more than one diode drop, thus the circuit can operate with supply voltages ranging down to 1.0 volts.

FIG. 1 is a schematic diagram of a conventional prior art current mirror circuit.

FIG. 2 is a schematic diagram of a Wilson prior art current mirror circuit.

FIG. 3 is a schematic drawing of the current mirror circuit of the present invention.

FIG. 4 is a schematic diagram of an application of the circuit of FIG. 3.

FIG. 1 shows a schematic diagram of a typical current mirror circuit as is well known in the art, and is given here to serve as a reference only. It consists of matched transistors Q1, Q2 and a current reference source I.sub.REF. Q1 is diode-connected and is coupled to the reference current. The collector current of Q2 (I.sub.C2) is the output current of the circuit. Since the emitter current in a bipolar transistor is a function of the emitter-base junction saturation current density, the area of that junction, electronic charge, the base-emitter voltage, Boltzmann's constant, and the absolute temperature, it is apparent that, for transistors of the same conductivity type on one IC chip, the only variable between emitter currents is the emitter-base junction area; i.e., I.sub.E2 =NI.sub.E1, where N is the ratio of the two emitter-base junction areas.

Since I.sub.B1 =I.sub.E1 /(.beta.+1), I.sub.B2 =NI.sub.E1 /(.beta.+1) and .beta.=I.sub.C /I.sub.B, then I.sub.REF =I.sub.C1 +I.sub.B1 +I.sub.B2 =.beta.I.sub.B1 +I.sub.B1 +NI.sub.B1, and I.sub.C2 =.beta.I.sub.B2 =.beta.NI.sub.B1. Therefore, ##EQU1## Thus, I.sub.C2 =NI.sub.REF /[1+(N+1)/.beta.], and only for large values of beta does I.sub.C2 closely approximate NI.sub.REF and the error increases as N increases.

In the current mirror of FIG. 2, commonly known as the Wilson current mirror, ##EQU2## the variation of I.sub.C2 with beta has been reduced by the buffering action of Q3, (replacing the Q1 diode connection with the Q3 base-emitter junction) which supplies the base current for Q1 and Q2 without significantly disturbing the reference circuit. However, since the two diode drops are in series, the use of this mirror structure requires a supply voltage of at least 2 volts, which is more than is available in many miniature electronic devices.

The circuit of the present invention as shown in the preferred embodiment of FIG. 3, provides the advantages of both the two previously described circuits and the disadvantages of neither; i.e., it provides the same accuracy as the latter circuit with the low voltage requirement of the former.

Again, transistors Q1 and Q2 (within dashed line 10) are two matched devices, i.e. having the same conductivity type and characteristics, preferably on a single chip. Both Q1 and Q2 are initially biased by a biasing network 12 consisting of R, Q4 and a current source I.sub.BIAS. Q1 is coupled to the reference source I.sub.REF as before. A balance sensing circuit 14 such as transistor Q5, compares I.sub.C1 and I.sub.REF and adjusts the mirror circuit bias voltage until the two currents are equal. Since Q1 and Q2 have matching characteristics, I.sub.C2 is forced to equal Ni.sub.REF, N being the ratio of the two-base emitter junction areas.

More specifically, I.sub.BIAS establishes a voltage drop across the combination of the diode-connected transistor Q4 and the resistor R that biases the base-emitter junctions of transistors Q1, Q2. The bias network parameters, namely the value of the current source I.sub.BIAS, the value of R and the voltage drop across Q4 at the bias current level are selected so that, if transistors Q1 and Q2 require no base current (if they had infinite beta), the base to emitter voltage of transistors Q1, Q2 would correspond to the value needed to establish an emitter current of I.sub.REF in Q1 and NI.sub.REF in Q2.

For finite values of transistor beta, transistors Q1, Q2 will draw base current from the current source I.sub.BIAS, with the result that less current will flow through Q4 and resistor R. In consequence, the collector current of Q1 will be less than I.sub.REF. Transistor Q5 serves to measure this imbalance between the mirror current, the excess currents is the base current for Q5, which is multiplied by the beta of Q5 and applied to the resistor R of the bias network 12.

The current through R that is applied by transistor Q5 serves to increase the base-emitter voltage of Q1 and Q2, and raises the value of the mirror current until it very closely approximates the value of the reference current. It can be shown then that, ##EQU3## It will be noted that this is the same expression given above for the Wilson current mirror.

FIG. 4 shows a typical use of a current mirror in an IC design where the mirror circuit of FIG. 3 is used to transfer a current reference level from an NPN transistor to a series of PNP transistors. With prior art circuits, separate current reference circuits would have been required for biasing various chains of interconnected current source transistors due to the beta sensitivity problem.

In this application, a current reference at a terminal 16, which is used in Circuit No. 1 is also needed for Circuit No. 2. As may be seen in FIG. 4, the Q2 of FIG. 3 may actually be any desired number of transistors Q2, Q2', etc. The current sources supplying I.sub.REF and I.sub.BIAS are here shown as PNP transistors Q6 and Q7, respectively, but Q6 functions with Q8 as a conventional current mirror (similar to that of FIG. 1), using the current reference from terminal 16. I.sub.REF may be made any desired fraction of I.sub.C8 by ratioing the areas of Q6 and Q8, or by the use of Q9 and R2, diverting a portion of I.sub.C8 from the base-emitter junction of Q8 and thus reducing I.sub.REF accordingly (this control including ON-OFF type control). This current diversion may be accomplished via a control terminal 18. The current in each individual current source transistor is accurately controlled and separate current reference circuits are no longer required. The elimination of the extra reference circuits results in significant savings in IC chip area and, therefore, in IC cost.

Thus, there has been shown and described a current mirror circuit for integrated circuit application particularly which will provide highly accurate reference currents with voltage supplies lower than 2 volts. While specific circuits and transistors have been shown as preferred, it will be obvious to those skilled in the art that other choices of transistors and arrangements thereof are possible and it is intended to cover all such as fall within the spirit and scope of the appended claims.