BACKGROUND OF THE INVENTION

This invention pertains to amplifier circuits suitable for fabrication in CMOS-type technology, and wherein such amplifiers use both bipolar and field-effect transistors.

It is desirable to combine complementary symmetry metal-oxide-semiconductor (CMOS) and bipolar technology using the relatively simple CMOS process to fabricate both MOS transistors an bipolar transistors on the same semiconductor substrate. Such combination offers several important advantages.

One advantage in combining CMOS and bipolar technology is the ability to use both MOS transistors and bipolar transistors together in a linear amplifier to obtain the benefits of the best characteristics of each type of transistor. For example, MOS transistors can provide exceptionally high input impedance, while bipolar transistors can provide much higher current capability than MOS transistors for a given substrate area. Therefore, by using a matched pair of MOS transistors arranged as a differential amplifier input means to provide improved input characteristics, and by using bipolar transistors arranged in current mirror amplifier configurations to provide high performance linear signal gain, the performance level of an operational amplifier circuit using both types of transistors can be improved beyond that available by using MOS or bipolar transistors alone.

Another advantage of combining CMOS and bipolar technology is that digital functions may be advantageously combined with linear functions on the same substrate. For example, the operational amplifier portion of an analog-to-digital converter may be constructed using MOS and bipolar transistors, while the digital control section thereof may be constructed using CMOS transistors as logic elements. Such mixed linear and digital capability leads to still higher levels of integration and more flexible system partitioning between large scale integrated chips.

There are other mixed-technology integrated circuits using FET and bipolar transistors. One such type is a combination of bipolar and junction FET (BiFET). Another type is bipolar and MOS (BiMOS). But BiFET has no suitable logic family, and BiMOS has poor linear packing density. The combination of CMOS and bipolar technology, however, provides both a suitable logic family and improved packing density. Digital logic is provided by conventional CMOS logic circuits. Furthermore, linear packing density is improved by using conventional CMOS guard band techniques to isolate various transistors from each other, to avoid having to devote substantial areas of the chip for the isolation function, as is required in BiMOS technology. Finally, the CMOS fabrication process is generally less complex than either the BiFET or BiMOS processes.

SUMMARY OF THE INVENTION

In accordance with the present invention, linear amplifier circuits comprising bipolar and MOS transistors are fabricated using transistor structures available in a CMOS-type structure. One type of bipolar structure available is a vertical transistor having high gain, but a collector region integral to the substrate and committed to the substrate potential. Another type of bipolar structure available is a lateral transistor having low gain, but an uncommitted collector. A primary aspect of the present invention is the cascading of vertical and lateral transistors to form a composite transistor with high current gain and uncommitted collector. The lateral/vertical composite transistor is then combined with other MOS transistors to form novel amplifier arrangements in further aspects of the invention.

BRIEF DESCRIPTION OF THE DRAWINGS

FIG. 1 is a schematic diagram of an amplifier using lateral and vertical transistors.

FIGS. 2, 3, and 4 are each schematic diagrams illustrating alternate output stages for the amplifier of FIG. 1.

FIG. 5 is a cross-sectional view illustrating various transistor structures available on an integrated circuit substrate embodying the present invention.

DESCRIPTION OF A PREFERRED EMBODIMENT OF THE INVENTION

A basic amplifier circuit embodying the present invention is shown in FIG. 1. Transistors Q13 and Q14 are each N-channel MOS devices, while transistors Q15 through Q20 are each P-channel MOS devices. Transistors Q1 through Q5, and Q10 through Q12 are each NPN bipolar transistors. Terminals 18 and 20 are arranged for receiving an operating potential therebetween. For the polarity of transistors shown, the V+ potential applied to terminal 18 is positive relative to the V- potential applied at terminal 20.

FIG. 1 further includes a legend defining different symbols for two types of NPN bipolar transistors used in the FIGURE. In particular, transistors Q1 through Q5 are each bipolar NPN transistors are realized in a lateral structure; transistors Q10 through Q12 are each bipolar NPN transistors as realized in a vertical structure.

As will become clear from a description hereinafter, concerning FIGURES of the drawing, vertical and lateral transistors available in CMOS have different characteristics owing to their respective different structures. For example, the common emitter forward current gain, or beta, of a vertical transistor, β_V, is relatively higher than that of a lateral transistor. A lateral transistor can have a beta, β_L, typically between 30 and 50. A vertical transistor, on the other hand, has a beta in the range of 100 to 400, with a typical value of about 200. However, the higher gain vertical transistors are limited in their connection flexibility because the substrate of the integrated circuit also serves as the collector region thereof. Hence, vertical transistors Q10, Q11, and Q12 in FIG. 1 all have their collector electrode connected to terminal 18 via a common connection to the substrate. In comparison, lateral transistors (Q1-Q5) all have uncommitted collector, emitter and base electrodes.

Vertical and lateral transistors are used in various combinations resulting in a composite transistor structure having both high gain and connection flexibility. A basic circuit configuration using lateral and vertical transistors, shown in FIG. 2 for example, is a lateral transistor Q6 having the emitter electrode thereof connected to a point of more negative potential, such as terminal 20, and a vertical transistor Q12 having a connection from the emitter electrode thereof to the base electrode of the lateral transistor. The resulting composite transistor includes an emitter follower (comprising the vertical transistor) followed in cascade by a common emitter amplifier (comprising the lateral transistor). The input or "base" connection of such composite transistor is the base electrode of the vertical transistor; the output or "collector" conection of such composite transistor is the collector electrode of the lateral transistor; and the node to which the emitter of the lateral transistor connects is the common or "emitter" connection of the composite transistor. The beta of the resulting composite transistor is (β_V +1)β_L. Since β_V is much larger than 1, the beta of the composite transistor is essentially equal to the product of the respective betas of the lateral and vertical transistors. Note that the composite transistor has both a high gain and an uncommitted collector output.

A high impedance input stage may be added to the composite transistor, thus described, by combining an MOS transistor having its principal current conduction channel connected between the base electrode of the vertical transistor and a point of relatively positive potential, such as terminal 18. The input to such MOS/bipolar combination is then the gate electrode of the MOS transistor.

An embodiment of vertical and lateral transistors for use in the present invention is illustrated in FIG. 5. A plurality of P type regions (26, 28, 36, 38, 40, 42, 48, 50, 54, 56, 62, 72) are formed on the surface of an N type substrate 22, by means well known to those skilled in the art. A second plurality of N type regions (30, 32, 34, 58, 64, 70) are further formed within a number of the first plurality of P type regions by means well known to those skilled in the art. A vertical NPN transistor structure is thereby comprised of N+ region 58 as the emitter E_V, P- region 56 as the base B_V and the substrate 22 as the collector C_V. A lateral NPN transistor structure is comprised of N+ region 32 as the emitter electrode E_L, P- region 28 as the base electrode B_L, and N+ regions 30,34 as the collector electrode C_L.

It will be recognized that the type of integrated circuit structure illustrated in FIG. 5 is a CMOS-compatible structure wherein PMOS and NMOS transistors are available for use in complementary symmetry circuits. In particular, the source and drain electrodes S_P, D_P of a PMOS transistor are formed on the surface of the substrate 22 by respective P+ regions 42 and 48. The gate electrode G_P is formed by metallization 44 over an insulating layer 46. The source and drain electrodes S_N, D_N of an NMOS transistor are formed on the surface of P- well 72 by respective N+ regions 64 and 70. The gate electrode G_N is formed by metallization 66 over an insulating layer 68.

Referring back to FIG. 1, a bias current as determined by impedance 12 (which may for example be a resistor) flows through diode-connected transistor Q20. Transistor Q20 is the master path of a current mirror having a plurality of slave paths. Each of the plural slave paths of the current mirror through the respective source-to-drain channels of Q17, Q18, and Q19, will source a respective output current that is a substantially constant multiple of the input current through Q20, as determined by the ratio of the respective relative sizes of Q17, Q18 and Q19 to Q20. If Q17, Q18, and Q19 each have a channel width-to-length ratio equal to that of Q20, then each will act as a current source providing a current substantially equal to the channel current through Q20.

Transistors Q15 and Q16 are a pair of PMOS transistors having matched transfer characteristics which essentially track each other over process and temperature variations. Transistors Q15 and Q16 are connected in differential amplifier configuration forming a differential input stage, responsive to the potential difference between input terminals 14 and 16 to divide the current from Q17 (i₁) into two portions, i₂ and i₃, differentially varied according the potential between the gates of Q15 and Q16. The differential varying currents i₂,i₃ are received by a balanced-to-single ended converter means comprising transistors Q1, Q2, Q3, and Q10 to provide a single ended output current i₅. A balanced-to-single ended converter here is a current mirror amplifier which receives one of the balanced differential currents (i₂) through the master path thereof and the other of the balanced differential currents (i₃) through the slave path thereof the master path being the collector-to-emitter path of Q1 and the slave path being the collector-to-emitter path of Q2. The collector-emitter path of the current mirror amplifier transistor Q1 accepts all of i₂, except for small base current error, i.sub. 4. The collector-emitter path of the current mirror amplifier through Q2 therefore demands i₂ also, assuming the base-emitter junction areas of Q1 and Q2 are essentially equal since they have the same emitter-to-base voltage. The output current i₅ is therefore equal to the difference between i₃ and the current demanded by Q2 (which equals i₂) so that i₅ equals i₃ -i₂.

The base current required by Q1 and Q2, as well as the current flowing through diode-connection transistor Q3, is supplied via emitter follower Q10. The purpose of vertical transistor Q10 is to reduce the base current error of the bipolar mirror Q1, Q2, Q3 by the beta of that transistor so that i₄ is small compared to i₂.

Current i₅ output from the differential input stage flows through an intermediate amplification stage comprising transistor Q11, Q4, Q5 and current source Q18. Transistors Q4 and Q5 are arranged as a conventional current mirror amplifier with Q4 being a diode-connected master path thereof and Q5 being an output slave path thereof. Vertical transistor Q11 provides high beta for signal amplification. The voltage output at the collector of Q5 is fed back to the input at the base of Q11 via compensation capacitor C which may be integrated on the same substrate with the amplifier. The value of capacitor C determines the dominant small signal frequency response for the amplifier. Particularly, the open loop unity gain frequency (which is also the gain-bandwith product for the amplifier under closed loop conditions) is determined by appropriate choice of value for C.

In this embodiment, the output signal from the collector electrode of Q5 is connected to an output stage comprising transistor Q12, current source Q19, and current mirror amplifier Q13, Q14. The current provided by slave path Q19 is inverted in current mirror amplifier Q13, Q14 for providing a dc pull-down (idling current) of the output at the emitter of Q12. Alternate output stages for the amplifier of FIG. 1 are shown in FIGS. 2, 3, and 4. Transistor Q13 is represented in FIGS. 2, 3, and 4 as a substantially constant current source. In FIG. 2, a lateral transistor Q6 provides an uncommitted collector output at terminal 10. In FIG. 3, a pair of vertical transistors Q12,Q22 are arranged in a Darlington amplifier configuration to provide beta multiplication. In FIG. 4, the output stage is somewhat modified using an NMOS transistor Q12' in lieu of the bipolar transistor Q12. The high imput impedance of Q12' reduces the loading effect on the output of the previous amplifier stage (from the collector of Q5).

A comprehensive analysis of operational amplifier construction and operations including thermal considerations, slew rates and frequency analysis, as well as second order effects regarding the operation of amplifier circuits of the type presented here may be found in an article by J. E. Solomon, published December 1974 in the IEEE Journal of Solid State Circuits, Volume SC9, No. 6, entitled "The Monolithic Op-Amp: a Tutorial Study."

A particular feature of the amplifier in FIG. 1 is the balance achieved by appropriate configuration of transistor Q3. As mentioned above, a current through the collector-emitter path of Q1 is i₂ minus the small base current error i₄ through Q10. Similarly, the current through the collector-emitter path Q2 is i₃ minus the small base current i₅ through Q11. Under balanced conditions, i₂ will more nearly equal i₃ if the small base currents i₄ and i₅ withdrawn by transistors Q10 and Q11 respectively, are also equal to each other. To this end, the size of transistor Q3 is chosen so that i₄ and i₅ tend to be equal, which improves the quiescent dc balance provided by current mirror amplifier pair Q1, Q2.

For example, assume that i₁ and i₆ from constant current sources Q17 and Q18 respectively are each 200 microamperes. At quiescent dc balance, i₂ and i₃ are therefore 100 microamperes each. Further, assume that the size of transistor Q5 is four times that of Q4, that Q1 and Q2 are each two times the size of Q4, and that Q3 is equal in size to Q4 as indicated in FIG. 1.

The emitter current of Q11 is the collector current of Q4 (plus the small respective base currents of Q4 and Q5). Since the emitter-base junction area of Q4 is about one fourth that of Q5, Q4 will draw about one fourth as much collector current as Q5 for the same base-emitter voltage. Therefore, the emitter current of Q11 is essentially 1/4×200, or 50 microamperes.

The emitter current of Q10 is the collector current drawn by Q3 (plus the small respective base currents of Q1, Q2, and Q3). Since Q3 draws essentially one-half the collector current of Q1, then Q10 emitter current is essentially 1/2×100, or 50 microamperes. The emitter currents, and hence the base currents, of Q10 and Q11 are therefore essentially equal.

Note that generally the base electrode of transistor Q3 need not be connected to its collector but can be alternatively connected to a source of bias potential of a magnitude so that Q3 draws sufficient collector current to balance the quiescent dc collector-emitter currents drawn by Q1 and Q2 respectively. However, the connection of the base electrode of Q3 to the base electrode interconnection of Q1 and Q2 is the preferred embodiment, not only because of its convenience, but because the small base current errors of Q1-Q5 also tend to contribute equally to the respective emitter currents of Q10 and Q11. In particular, the emitter current of Q11 includes 5i_b where i_b is the base current of Q4 and 4i_b is the base current of Q5. By inspection, the emitter current of Q10 also includes 5i_b, because the base current of Q1 and Q2 is substantially 2i_b each and the base current of Q3 is substantially i_b.

As previously described in connection with FIG. 5, a CMOS-type of structure is used with the present invention to form vertical and lateral transistors. Further, conventional CMOS guard band techniques are used to isolate the different types of transistors from each other. A PMOS device in FIG. 5, for example, is guarded by P- well 40,40a which is connected to the V- potential, and forms a reversed bias diode with respect to the substrate 22 which is connected to V+. Similarly, an NMOS device is guarded by P+ band 62,62a which is connected to V- potential, and which forms a reversed bias diode with respect to N+ region 60, which is held at V+ substrate potential. The vertical transistor in FIG. 5 is similarly isolated by a reverse bias diode formed by P+ band 54,54a and substrate N+ region, 52,60. The lateral transistor contained in P- well 28 is isolated by reversed bias diode formed by bands 26,26a and the substrate region 24a,24b.

The lateral transistor contained in P- well 28 is further formed with a P+ region 36 directly below the emitter region 32 thereof for the purpose of reducing the effect of an undesirable parasitic transistor. The unwanted parasitic transistor is a vertical device formed by emitter region 32, base region 28, and the substrate 22 as a collector region. When forward-bias potential is applied between electrodes B_L and E_L, transistor action in such vertical parasitic transistor tends to produce current flow in a vertical direction from the substrate 22 to the emitter region 32 which undesirably increases circuit power dissipation. The P+ region 36, being more heavily doped than the P- base region 28, reduces the injection efficiency of the emitter-base junction formed by respective regions 32 and 28 in the vertical plane. The effect of such P+ region 36 is to reduce the beta of the vertical parasitic transistor so that the parasitic current drain is thereby reduced. The beta of the lateral transistor, however, is not adversely affected because region 36 does not extend to the lateral sides of emitter region 32. It is noted that an N- region may also be used in lieu of P+ region 32, which also is reasoned to have a similar effect of reducing the injection efficiency of the emitter-base junction for the vertical parasitic transistor.