Apparatus and Method for Current Steering Digital-to-Analog Converter Units

1. Field of the Invention

The present invention relates generally to digital-to-analog converter units and, more particularly, to digital-to-analog converter units based on current mirror techniques. In the conventional implementation of a digital-to-analog converter unit, a constant current flows through the unit while the digital input signals steer or divide the current between two conductors, the difference in the divided current resulting in a current mode representation of the digital word or data group.

2. Description of the Related Art

Referring to FIG. 1, the general structure of an digital-to-analog converter unit 10 based on current mirror techniques, according to the related art, is shown. Digital input signals are applied to current mirror 11. Current mirror 11 provides two output currents, I.sub.0 and I.sub.1. The sum of these two output currents is a constant. The digital input signals determine the division of the current between I.sub.0 and I.sub.1. The currents I.sub.0 and I.sub.1 flow through resistors 12 and 13, respectively. The detection unit 14 provides an analog output signal determined the voltages across resistors 12 and 13. This analog output signal is a result of the digital input signals applied to the current mirror 11.

The current mirror-based digital-to-analog converter units rely on the property of a current mirror circuit to replicate a selected unit current. Well designed current mirror units generally have an accuracy of 1%. The current matching between two similar metal-oxide-semiconductor (MOS) transistors operated with similar gate voltages and drain voltages is determined by variations in the diffusion widths, in the poly-silicon widths, in the gate oxide thickness, and ground potential. The present accuracy in current matching was achieved by reducing these variations during the fabrication of the converter units. In spite of improvements, digital-to-analog converter units with even greater accuracy are needed.

A need has therefore been felt for a digital-to-analog converter unit and a related method with a higher accuracy than has been possible with conventional techniques.

The aforementioned and other features are accomplished, according to the present invention, by providing an array of current mirror components which are divided into groups having components which are mutually symmetric with respect to the center of the component array, by providing a resistive network which supplies weighted gate voltages used to the bias the current mirror components, and by implementing the current steering portions of the current mirror components with an annular transistor configuration.

These and other features of the present invention will be understood upon the reading of the following description in conjunction with the Figures.

FIG. 1 is a block diagram of a current mirror-based digital-to-analog converter unit according to the prior art.

FIG. 2 is schematic circuit diagram of a current mirror component according to the present invention.

FIG. 3A is a block diagram showing the physical placement of a 4.times.4 array of current mirror components on a substrate, while FIG. 3B is representative diagram showing the physical placement of an 8.times.8 arrangement of current mirror components on a substrate.

FIG. 4 is a schematic diagram of a resistor grid used to provide a weighted gate voltage for a 4.times.4 array of current mirror components.

FIG. 5 is a top view of a layout of a current mirror component on a semiconductor substrate implementing the schematic diagram of a current mirror component shown in FIG. 2.

1. DETAILED DESCRIPTION OF THE DRAWINGS

Referring to FIG. 2, a schematic circuit diagram of a current mirror component 20 used in the preferred embodiment of the present invention is shown. The current mirror 20 has a negative channel field effect transistor 22 acting as a bias transistor. The source terminal 22S of transistor 22 is coupled to the ground potential. The gate terminal 22G of transistor 22 is coupled to a network or grid of resistors 47 which will be described below. The drain terminal 22.sub.D of transistor 22 is coupled to the source terminals 25.sub.S of a plurality of n-channel field effect transistors 25.sub.0 -25.sub.N-1 and to the source terminals 21.sub.s of a plurality of n-channel field effect transistors 21.sub.0 -21.sub.N-1. The gate terminals of transistors 25.sub.0 -25.sub.N-1 are coupled to terminal 25.sub.G, while the gate terminals of transistors 21.sub.0 -21.sub.N-1 are coupled to terminal 21.sub.G. The drain terminals 25.sub.D of transistors 25.sub.0 -25.sub.N-1 are coupled to an I.sub.0 terminal while the drain terminals 21.sub.D of transistors 21.sub.0 -21.sub.N-1 are coupled to an I.sub.1 terminal. The group of transistors 25.sub.0 -25.sub.N-1 and the group of transistors 21.sub.0 -21.sub.N-1 complementary signals applied to the gate terminals 25.sub.G and 21.sub.G, respectively, thereby determining whether the current through transistor 22 is applied to the I.sub.0 terminal or to the I.sub.1 terminal.

In the preferred embodiment, each current mirror has 16 transistors coupled to the I.sub.0 terminal and 16 transistors coupled to the I.sub.1 terminal.

Referring to FIG. 3A, the physical placement of current mirror components 20 and 20' on a substrate 30 is shown. Each of the current mirror components 20 has gates of transistors 25.sub.0 -25.sub.N-1 coupled together and the gates of transistors 21.sub.0 -21.sub.N-1 coupled together. The converter unit has the digital (logic) signals D.sub.256, D.sub.128, D.sub.64, D.sub.32, and D.sub.16, D.sub.8, D.sub.4, D.sub.2, and D.sub.1 applied thereto. One of the digital input signals D.sub.16 -D.sub.256 is applied to one set of the coupled gates of each of the current mirror components 20, while the logical complement signal is applied to the other set of coupled gates in each current mirror component. For the current mirror component 20' shown at array location (1, 2) in FIG. 3A, a modified transistor gate coupling arrangement is implemented. The arrangement includes one transistor 21.sub.0 receiving the D.sub.1 signal (i.e., the least significant logic signal) at the gate terminal, while one transistor 25.sub.0 receives the complement of the D.sub.1 signal at its gate terminal in current mirror component 20'. Two transistors 21.sub.1-2 have the D.sub.2 signal applied to their gate terminals, while two transistors 25.sub.1-2 have the complement of the D.sub.2 signal applied to the gate terminals in current mirror component 20'. The D.sub.4 signal is applied to four of the transistor 21.sub.3-6 gate terminals, while the complement of the D.sub.4 signal is applied to four of the transistor 25.sub.3-6 gate terminals in current mirror component 20'. The D.sub.8 signal is applied to eight of the gate terminals of transistors 21.sub.7-14, while the complement of the D.sub.8 signal is applied to eight of the gate terminals of transistors 25.sub.7-14 Thus, the four least significant digital signals are applied to the mirror current component 20' at array position (1,2). The next most significant digital input signal D.sub.16 is applied to the mirror current component 20 at array position (-1, -2) and, in the preferred embodiment, controls the current through 16 transistor pairs. The next most significant logic signal D.sub.32 (controlling the current through 32 transistor pairs) is applied to the 2 mirror current components 20 at array positions (-1, 2) and (1, -2). The next most significant digital input logic signal D.sub.64 (64 transistor pairs)is applied to the four mirror current components 20 at matrix positions (.+-.1, .+-.1). The most significant input logic signal D.sub.128 (128 transistor pairs) is applied to the eight mirror current components 20 at the matrix positions (.+-.2, .+-.2) and (.+-.2, .+-.1). As will be clear, each digital signal D.sub.N directs the current to the I.sub.0 or the I.sub.0 conductor through the number of transistors equal to the number represented by the digital input signal.

FIG. 3B illustrates the same axial symmetric positioning of the current mirror components except that, instead of a 4.times.4 array shown in FIG. 3A, an 8.times.8 array is shown. As in FIG. 3A, the reflection symmetry, also referred to an centroid symmetry, through axis 31 is clearly indicated.

Referring to FIG. 4, a first terminal of signal generator 41 is coupled the ground potential. A second terminal of signal generator 41 is coupled to a drain terminal and to gate terminal of p-channel field effect transistor 42, to gate terminal of p-channel field effect transistor 431, to a gate terminal of p-channel field effect transistor 441, to a gate terminal of p-channel field effect transistor 451, and to a gate terminal of p-channel field effect transistor 461. The source terminals of transistors 42, 431, 441, 451, and 461 are coupled to the supply terminal V.sub.cc. The drain terminal' of transistor 431 is coupled to a drain and a gate terminal of n-channel field effect transistor 432. The drain terminal of transistor 441 is coupled to a drain terminal and a gate terminal of n-channel field effect transistor 442. The drain terminal of transistor 451 is coupled to a drain terminal and a gate terminal of n-channel field effect transistor 451. The drain terminal of transistor 461 is coupled to a drain terminal and a gate terminal of n-channel field effect transistor 462. The source terminals of transistors 432, 442, 452, and 462 are coupled to the ground potential. A three-by-three grid 40 of resistive elements 47 is formed such that the corners of the grid of resistors coupled to the drain terminals of transistors 432, 442, 452, and 462. The three-by-three grid 40 of resistive elements provides a four-by-four grid of nodes N where the resistive elements are coupled together. Each of the nodes N is coupled to a gate terminal of one of the current mirror component bias transistor.

Referring to FIG. 5, a semiconductor layout implementing the current mirror component of FIG. 2 is shown. The realization of the elements in of FIG. 2 in FIG. 5 is indicated by the use of equivalent elements being numbered identically. The gate 22.sub.G of bias transistor 22 demonstrates the origin of the term annular transistor, the gate terminal surrounding the transistor source terminal 22.sub.s. The annular transistor drain terminal 22.sub.D is identical with the source terminals 21.sub.S of transistors 21.sub.i and is identical with the source terminals 25.sub.S of transistors 25, the annular transistor drain terminal 22.sub.D being coupled to the source terminal of transistors 21 and 25 by metal layers.

2. Operation of the Preferred Embodiment(s)

The digital-to-analog converter of the present invention involves a combination of several features which result in increased accuracy. First, a standard current mirror component is implemented using an annular transistor. In the preferred embodiment each component represents 16 unit currents and an array of such components forms a binary weighted current source. A differential approach is implemented with current steering for reduced voltage swings (i.e., higher speeds) as well as reduced ground voltage drops. In the annular transistor, the width-to length ratio is determined by poly-silicon length and is therefore relatively immune to dimensional variations. Because control of the poly-silicon length is tighter other process controls, the annular device reduces a major variation resulting from the fabrication of the device. (Other transistor implementations that define transistor width with the gate poly-silicon region can be used. However, the present approach will be best for length/width definition using only poly-silicon regions.) Symmetry with respect to all angles tends to reduce the V.sub.t mismatch as well.

Second, as illustrated in FIG. 3, the current mirror components (20) are physically positioned to provide a reflection symmetry through the axis at the center of the component array. The symmetry provides a compensation across the current mirror array for process variations during the fabrication of the converter unit.

In addition, instead of coupling all the bias transistor gates to a common potential, the two dimensional resistive element network, illustrated in FIG. 4, is employed. The two dimensional resistive element network is activated by at the corners of the network by diode-connected MOSFET transistors fed by equal currents. The corner potentials of the resistive element network are a function of the systematic gate-oxide thickness, the poly-silicon gate length, the ground potential drop, and the V.sub.t variations. The gate potential at a node of the resistive element network or grid is a weighted average of the voltage at the corners of the network. Therefore, the fraction of the undesired process variations that can be expressed as a linear function can be compensated for by the resistive element network.

(As will be clear, the diode-connected transistors can be positioned at more and/or at different locations to provide compensation for a more complicated gate potential profile.) The resistive elements can be the poly-silicon gate conductor itself, or any other suitable layer such as an n-well structure. In order to achieve a dense layout, small transistors operating in their linear region are preferred. A further improvement, not shown, is to implement the four current sources in p-channel technology. The small mismatch can be compensated for by scrambling the currents. The uniform current is thus achieved at the cost of additional noise.

While the invention has been described with particular reference to the preferred embodiment, it will be under stood by those skilled in the art that various changes may be made and equivalents substituted for elements of the preferred embodiment without departing from the invention. In addition, many modifications may be made to adapt a particular situation and material to a teaching of the present invention without departing from the essential teachings of the present invention.

As is evident from the foregoing discussion, certain aspects of the invention are not limited to the particular details of the examples illustrated, and it is therefore contemplated that other modifications and applications will occur to those skilled in the art. It is accordingly intended that the claims shall cover all modifications and applications as do not depart from the spirit and scope of the invention.