Weighted Capacitor Analog/Digital Converting Apparatus and Method

This invention relates to weighted capacitor analog/digital conversion circuitry and techniques and more particularly to such conversion techniques utilizing metal oxide semiconductor circuitry providing conversion with either a linear relationship between the analog and digital signals or a nonlinear relationship therebetween.

Conventional techniques for performing analog/digital conversion generally require sophisticated analog circuitry including operational amplifiers, and digital circuitry for counting, sequencing, and data storage. Consequently, hybrid circuits have resulted including one or more bipolar analog circuit elements together with metal oxide semiconductor circuit elements for performance of the digital functions while remaining within reasonable economic limits. Applications exist for coding analog signals in nonlinear digital codes and for subsequent decoding to an analog form. Voice signals for telephone transmission and subsequent reception are often subjected to such nonlinear coding and decoding. Present coder/decoder installations are complex and expensive, therefore requiring multiplexing for sharing of the installations so that they may accommodate multiple telephone transmissions simultaneously. A low cost configuration of minimal complexity if desirable for use in both linear and nonlinear conversion between analog and digital signal forms, while maintaining low error and distortion levels through the conversions.

In general, the electronic circuitry disclosed herein for converting between analog and digital signal forms includes an array of weighted capacitors having capacitance values corresponding to individual digits within a predetermined range of digits from a most to a least significant digit in the range. An additional capacitor having a capacitance value equivalent to the value of the capacitor in the array corresponding to the least significant digit is connected to the array. All of the capacitors in the array and the additional capacitor have one terminal connected to a common point and to one input of a voltage comparator. Another input on the voltage comparator is connected to a comparator reference voltage and the other terminals on the array of capacitors are connected to individual ones in a plurality of switches. Each of the plurality of switches is capable of selecting one of several terminals to which are connected the comparator reference level, an analog signal level, and an analog signal reference level. The plurality of switches are operated in predetermined sequence by a digital control circuit coupled to receive a serial digit output from the voltage comparator. Thus, an analog signal connected to one of the several terminals produces a digital combination from the digital control circuit and a digital combination connected to the digital control circuit produces an analog signal at another of the several terminals.

It is an object of the present invention to provide conversion between analog and digital signals in minimal time with a high degree of accuracy using simple easily fabricated circuitry.

Another object of the invention is to provide conversion between analog and digital signal forms wherein parasitic capacitance is removed as an error factor.

Another object of the present invention is to provide conversion between analog and digital signal forms utilizing circuitry having inherent sample and hold functions.

Another object of the present invention is to provide conversion between analog and digital signal forms wherein component offset voltages are compensated prior to the conversion.

Another object of the present invention is to provide conversion between analog and digital signal forms wherein there is a nonlinear relation between the analog and digital signals.

Another object of the present invention is to provide conversion between analog and digital signal forms in a device which may be used as a coder or a decoder.

Another object of the present invention is to provide conversion between analog and digital signal forms which is inexpensive enough to allow elimination of multiplexing and institution of per channel coding and decoding.

Additional objects and features of the invention will appear from the following description in which the preferred embodiment has been set forth in detail in conjunction with the accompanying drawings.

FIG. 1 is a block diagram of the weighted capacitor analog/digital converter.

FIG. 2 is a schematic diagram of the weighted capacitor array and a plurality of switches included in the digital control and sequencing circuit.

FIG. 3 is a timing diagram showing a typical example of successive approximation accomplished by the embodiments of FIGS. 1 and 2.

FIG. 4 is a chart showing the positions of the switches of FIG. 2 in sequence for obtaining the timing diagram of FIG. 3.

FIG. 5 is an electrical schematic diagram of one stage of a comparator used in the invention.

FIG. 6 is a combination block and electrical schematic diagram demonstrating one form of offset voltage compensation.

FIG. 7 is a block diagram illustrating a nonlinear analog/digital converter.

FIG. 8 is an electrical schematic diagram illustrating the arrangement of the weighted capacitor arrays and switches of the digital control and sequencing circuit of FIG. 7.

FIG. 9 is a chart showing nonlinearity between the digital and analog signals.

FIG. 10 is a detail view of the segment A--B of FIG. 9.

FIG. 11 is a timing diagram showing a sequence of analog signals produced by a sequence of digital signals with nonlinear relationship therebetween.

FIG. 12 is a chart showing the switching sequence for the circuit of FIG. 8 to obtain the timing diagram of FIG. 11.

FIG. 13 is a timing diagram showing conversion of an analog signal to a digital signal utilizing the circuit of FIG. 8.

FIG. 14 is the standard pulse code modulation presentation for the companded signal associated with the timing diagram of FIG. 13.

FIG. 15 is a chart showing the switch sequencing in the circuit of FIG. 8 for obtaining the successive approximation of FIG. 13.

Referring to FIG. 1 of the drawings, a block diagram of one embodiment of the device for converting between analog and digital signal forms is shown. A weighted capacitor array 11 is shown receiving an analog reference signal V.sub.REF and an analog signal V.sub.IN. Weighted capacitor array 11 produces a successive approximation signal V.sub.X which is connected to a comparator 12. Comparator 12 produces a serial digital output which is coupled to a digital control and sequencing circuit 13. A plurality of digital data output terminals 14 is available at digital control sequencing circuit 13. Digital control sequencing circuit 13 receives a clock input for providing sequence dwell time and produces a series of switching functions on a plurality of switching function terminals 16 coupled to weighted capacitor array 11 and 17 coupled to comparator 12.

FIG. 2 shows the weighted capacitor array 11 in circuit with a plurality of switches 18 included in digital control and sequencing circuit 13 for controlling charge placed on individual ones of the capacitors in the capacitor array 11 and for connecting thereto predetermined ones of the analog signal V.sub.IN, analog reference signal V.sub.REF and the reference signal for comparator 12, which is shown in FIG. 2 as ground. One terminal of all of the capacitors in the array is connected to a common point and in turn connected to one input of comparator 12. The voltage at the common point or the input to comparator 12 is designated as V.sub.X. A switch S1 is provided for selectively placing the common point for the capacitors in the array at ground potential. Another switch S7 is provided for selecting the analog signal V.sub.IN or the analog reference signal V.sub.REF. It should be noted that digital control and sequencing circuit 13 includes sequencing and control logic which controls all of the switches in the circuit on the time base provided by the clock input to the digital control and sequencing circuit 13. The capacitors in the weighted capacitor array 11 have values as indicated in the example of FIG. 2 corresponding to a predetermined number of binary bits ranging from a most to a least significant bit. The capacitor representing the most significant bit is marked C and that representing the least significant bit is marked C/8. An additional capacitor 19 is provided having a capacitance value equivalent to that of the capacitor representing the least significant bit, which is in this example C/8. It is seen in FIG. 2 that one terminal on additional capacitor 19 is also connected to the common point at one input to comparator 12.

The plurality of switches 18 are designated S2 through S6 in this example, each having terminals marked 1 and 2. Switches S2 through S6 selectively place the other terminal on individual capacitors in the capacitor array 11 to ground potential in position 1 or to switch S7 in position 2. Switch S7 in turn has two positions 1 and 2 for selecting the analog signal V.sub.IN or the analog reference signal V.sub.REF respectively. As indicated in FIG. 2, a digit serial output is provided by comparator 12 in accordance with the successive approximation signal level V.sub.X. The digit serial output is connected to digit control and sequencing circuit 13 as mentioned above for controlling the sequence and control logic and the switching sequence of switches S1 through S7.

Turning to FIG. 3 the manner in which the device of FIGS. 1 and 2 operates will be described. The successive approximation signal level V.sub.X is shown on the ordinate of the chart of FIG. 3 and the abscissa is divided into time periods as indicated and as provided by the clock input to digital control and sequencing circuit 13. In this example to be presented a conversion from an analog signal V.sub.IN to a digital signal will be followed. The sequence begins at time t.sub.0 and during the first period t.sub.1 -t.sub.0, switch S1 is closed to ground and switches S2 through S6 are also placed in position 1 to ground. In this fashion, all of the capacitors in the array and additional capacitor 19 are discharged. It will be noted that the total of all of the capacitance values in the array 11 and additional capacitor 19 is 2C. Consequently when switch S7 is placed in position 1 to select V.sub.IN and switches S2 through S6 are placed in position 2, a charge is stored in the capacitors of 2CV.sub.IN at the common point or plates of the capacitors. Thus, the analog signal V.sub.IN is sampled during the second period t.sub.2 -t.sub.1. During the third period t.sub.3 -t.sub.2, switch S1 is opened, removing the common capacitor plates from ground. During the fourth period, t.sub.4 -t.sub.3, switches S2 through S6 are connected to ground, and since the charge on the common plate is conserved, V.sub.X assumes a potential of -V.sub.IN in the absence of parasitic capacitance to ground. Switch S7 is placed in position 2 during the fourth period, preparatory to applying portions of the analog reference voltage V.sub.REF to the common plate of the capacitors for modifying the voltage V.sub.X.

During the fifth period, t.sub.5 -t.sub.4, switch S2 is placed in position 2 thereby connecting V.sub.REF to one terminal of capacitor C. It may be seen that the circuit arrangement existing in the fifth period is such as to divide V.sub.REF in the series capacitance voltage divider between V.sub.REF and ground consisting of capacitance C in series with the remainder of the capacitance, which is also equivalent to C. Therefore, V.sub.REF /2 is applied at the common plate of the capacitors during the fifth period and V.sub.X is the combination of -V.sub.IN and V.sub.REF /2. Comparator 12 provides an output signifying V.sub.X as being in excess of the ground reference to the comparator 12 and the serial digit output produced thereby and connected to digital control and sequencing circuit 13 causes switch S2 to be repositioned in position 1 during the sixth period t.sub.6 -t.sub.5.

During the seventh period, t.sub.7 -t.sub.6, switch S3 is placed in position 2 to select V.sub.REF to be applied to one terminal of the capacitance C/2. In the same fashion as explained above, a series capacitance voltage divider is in existance during the seventh period wherein V.sub.REF /4 is added to V.sub.X, so that the successive approximation signal assumes a value which is the combination of -V.sub.IN and V.sub.REF /4. This produces a serial digit output from comparator 12 to digital control and sequencing circuit 13 which retains switch S3 in position 2 and, during the eighth period t.sub.8 -t.sub.7, places switch S4 in position 2 to provide 3/8 of V.sub.REF in successive approximation signal V.sub.X. Thus, during the eighth period the input to comparator 12 is the combination of -V.sub.IN and 3/8 V.sub.REF. The serial digit output from comparator 12 connected to digital control and sequencing circuit 13 retains S4 in position 2 as well as S3 and during the ninth period, t.sub.9 -t.sub.8, switch S5 is placed in poistion 2. It may be seen by reference to FIG. 3 that the combination voltage produced during the ninth period is in excess of the comparator reference voltage and the serial digit output directs the digital control and sequencing circuit 13 to reposition switch S5 in position 1 during the tenth period, t.sub.10 -t.sub.9. The successive approximation at the input of comparator 12 is complete and the analog voltage V.sub.IN is presented at the data output terminals 14 of digital control and sequencing circuit 13 as the four bit binary number 0110.

Remaining with FIG. 3 it may be shown that parasitic capacitance existing at the input of comparator 12 will not affect the accuracy of the conversion as it has in prior conversion schemes. A property of the conversion technique is that, since the successive approximation signal V.sub.X is nulled back toward the zero or comparator reference position as the bit decisions are made, the parasitic capacitance from the common plate of the capacitors to ground in this example does not affect accuracy. When the comparator input voltage V.sub.X is nulled back to the comparator threshold, which is ground in this example, the charge on the parasitic capacitor is identical to that which exists at the end of the sample mode during the third period t.sub.3 -t.sub.2. Therefore the error charge placed on the parasitic capacitor by sampling is conserved and error contributed thereby during the ensuing redistribution periods is zero. The parasitic capacitance increases the total capacitance in the array of capacitors. This decreases the voltage division characteristics of each capacitor in the array. Thus, smaller fractions of the analog reference signal V.sub.REF are subtracted from the analog signal V.sub.IN, but the analog signal V.sub.IN appears as a lower signal level at the common point for the capacitors, so the effect is inherently compensated. The dashed lines signifying the level of the successive approximation signal V.sub.X in FIG. 3 represent a situation which is a gross case and is presented here for illustrative purposes only. Where, as shown, the level of the analog voltage V.sub.IN is cut in half, the parasitic capacitance is equivalent to 2C. Thus the capacitance array actually contains 4C, but only the 2C containing charge distributed and redistributed by the digital control and sequencing circuit 13 enters into the successive approximation. While the successive approximation signal levels V.sub.X are lower in this latter case, the comparative levels have the same sense and the serial digital output from comparator 12 for sequencing the operative switches S2 through S5 for redistribution of charge is the same.

Another error source in the disclosed circuitry derives from what is known as comparator offset. Comparator offset is that value of comparator output which exists for zero differential comparator input. The comparator output is proportional to comparator input, but the output is amplified and the slope of the output as a function of input signal is steep. Therefore, the output from comparator 12 is such as to be either near zero or at some large value thereby providing the serial digital output indicative of an input which is either in excess or not in excess of the comparator reference. Offset is the departure from this ideal situation wherein some small output exists at the comparator when the input terminals are both placed at the same input potential. Referring to FIG. 5, a differential amplifier circuit 21 is shown wherein the offset characteristic is caused by a mismatch of field effect transistor characteristics. The mismatch which primarily produces the offset is between transistors Q3 and Q4 of FIG. 5. Metal oxide semiconductor contruction of all of the circuit elements necessary to this invention is made possible by the concept of this invention. However, metal oxide semiconductor (MOS) construction of the comparator of FIG. 5 is especially susceptible to offset voltage errors. The offset exists as a DC differential voltage at the output of the comparator when the input differential is zero. Described in an alternate manner, a signal differential occurring between the input terminals of the comparator may produce an output modified by the offset voltage of the comparator which is not a true indication of the differential between the input terminals. Consequently, an error in the serial digital output could occur and an error would accure in the conversion.

One manner of compensating for the offset voltage in the comparator 12 is to couple the offset output through a buffering element and switch to the common point for the capacitor array 11 during the sampling mode to thereby compensate for the offset voltage in the signal at the input of the comparator prior to beginning the redistribution of charge and successive approximation sequencing.

Referring to FIG. 6 of the drawing the voltage comparator 12 is shown as a multistage circuit containing, in this example, three stages of the differential amplifier circuit 21 of FIG. 5 utilizing MOS field effect transistors (FETS) Q1 through Q4 and having input and output terminals as shown. The offset voltage may be stored in the circuit of comparator 12 by means of the capacitors C1 through C4 and switches S8 through S11 shown in FIG. 6. Prior to the sampling mode while switch S1 of FIG. 2 is closed, all switches S8 through S11 are also closed. In this fashion, any offset voltage existing at the output of the first stage of comparator 12 is stored as a difference voltage between capacitors C1 and C2. When switches S8 and S9 are open together with S1 of FIG. 2, a compensation voltage is stored at the input to the second stage of comparator 12 which is equivalent to the offset of the first stage.

In FIG. 6 it should be noted that compensation is made only for the offset voltages of the first and second stages. The offsets in the multistage circuit are divided by the gains of the previous stages. Therefore assuming a gain of ten in each stage in the circuit of FIG. 6 and a 50 millivolt offset for the first stage, a 0.5 millivolt offset would be the residual offset therefrom at the input to the third stage. Thus correction for each of the three stages is unnecessary and a practical circuit is seen in FIG. 6 having no offset correction in the third stage. It should be noted that the inherent offset characteristic obtained in MOS circuitry technology is conveniently compensated through the use of MOS circuit technology to provide the circuit seen in FIG. 6. While offset errors are less severe in bipolar circuitry, such a correction technique could not be utilized in bipolar circuitry because of the discharge paths presented in the bipolar elements which would prevent storage of the offset compensation signals on capacitors in the bipolar circuitry.

The initial voltage V.sub.X established at the common plate of the capacitor array by closing switch S1 need not be ground as shown, but must only be the threshold voltage of the comparator 12. With this in mind, a method has been disclosed for converting between an analog and a digital indication which includes the steps of weighting an array of capacitors so that they assume capacitance values which are in digital relation from a most to a least significant digit. Further steps include connecting one end of each of the capacitors in the array to a common point and discharging the array of capacitors so that zero charge is stored thereon. Subsequently the step of charging the array of capacitors with an analog signal is undertaken. A reference signal for the analog signal is divided in the array of capacitors thereby imparting a reference charge to the common point by sequentially connecting ones of the array of capacitors between the reference signal and the common point while the remainder of the array of capacitors is in series connection between the common point and a discharge reference. A trial signal is thereby constructed at the common point which is thereafter coupled to a comparator having the discharge level as a reference input. In this fashion, a serial digital output is provided as the output of the comparator in accordance with serial comparisons between the discharge level and the trial signal. The step of connecting a reference charge in a predetermined sequence to the capacitors in the array operates to cause the trial signal to converge toward the discharge reference level. In this fashion, any charge imposed on parasitic capacitance in the converter appearing between the common point and the discharge reference level converges toward the initial charge on the parasitic capacitance imposed during the sampling period. The method is completed by sequencing the steps of discharging the capacitor array, sampling and charging the capacitor array, and redistributing the charge on the capacitor array so that the serial digital output from the comparator is in a predetermined digit sequence in accordance with the analog signal.

The block diagram of FIG. 7 shows the major elements in a nonlinear converter between analog and digital signals. The nonlinear converter will be disclosed in terms of a device which finds utility as a companded pulse code modulation voice coder/decoder and which uses monolithic arrays of binary weighted MOS capacitors providing a standard 15 segment approximation to the 255.mu. compression law in a coder for pulse code modulation PCM telephony. The level of expense and complexity of the nonlinear converter disclosed herein suggests that a per channel voice coder/decoder is realizable in a single MOS chip. The arrangement of FIG. 7 will allow coding of voice signals in nonlinear digital codes for digital telephone transmission while maintaining an exceptably high ratio of signal to noise and low distortion over a wide dynamic range without requiring an excessive digital transmission bandwidth. Widely used in the United States is the fifteen segment approximation of the .mu. = 255 compression law, which is illustrated in FIG. 9 and 10 to be discussed hereinafter. Returning to FIG. 7, a weighted capacitor array 22 similar to array 11 above is attached to terminals for connection to an analog voltage V.sub.IN and an analog voltage reference V.sub.REF. Weighted capacitor array 22 provides a successive approximation voltage V.sub.X which is connected to a comparator 23, which provides a serial digital output connected to a digital control and sequencing circuit 24. A plurality of sequencing outputs 26 are directed to weighted capacitor array 22. A plurality of digital data output terminals 27 are provided on digital control and sequencing circuit 24. An additional weighted capacitor array 28 is provided which receives the analog signal reference voltage V.sub.REF as well as a plurality of sequencing outputs 29 from digital control and sequencing circuit 24. Additional weighted capacitor array 28 provides an output signal V.sub.Y connected to a buffer element 31. Buffer element 31 provides an output signal V.sub.Z for coupling to weighted capacitor array 22. A plurality of sequencing outputs 32 are connected from digital control and sequencing circuit 24 to comparator 23. The time base for the sequencing outputs 26, 29 and 32 are provided by a clock input to digital control and sequence circuit 24.

The foregoing description is for illustrative purposes, and it should be understood that the additional weighted capacitor array 28 and buffer amplifier 31 may be replaced by any type of digital-to-analog conversion device.

FIG. 8 shows a circuit diagram which discloses the interconnections between weighted capacitor array 22, comparator 23, buffer 31, additional weighted capacitor array 28 and certain of the switches included in the digital control and sequencing circuit 24. FIG. 8 does not include the sequencing and control logic utilized for controlling the switches seen in FIG. 8, as that control logic is well known and not part of this invention. Weighted capacitor array 22 includes capacitors C13 through C21, having capacitance values of C, C/2, C/4, C/8, C/16, C/32, C/64, C/128 and C/128 respectively. One terminal of all of the capacitors C13 through C21 is common and is connected to one input on the comparator 23. Another input on the comparator 23 is connected to the comparator reference which is ground in this example. A switch S12 is provided for connecting the common plate of the capacitors to the comparator reference voltage level, ground in this example, when closed. It then may be seen that the total of all of the capacitances C13 through C21 is 2C.

A plurality of switches 33 are included in digital control and sequencing circuit 24 and are designated as S13 through S20 in FIG. 8. Switches 33 are configured to connect the noncommon terminal on each one of the capacitors C13 through C20 to one of three terminals. The first of the three terminals selected by any of switches S13 through S20 is connected to a switch S21 having positions 1 and 2 thereon, for selecting the voltage comparator reference, ground in this example, or an analog voltage V.sub.IN on position 2. The second terminal on each of the switches S13 through S20 is connected to a switch S22 having positions 1 and 2 thereon for selecting a positive or negative analog reference signal level V.sub.REF respectively for application to ones of the noncommon terminals of capacitors C13 through C20. The third position selectable at switches S13 through S20 connects a buffered step signal V.sub.Z to individual ones of the noncommon terminals of capacitor C13 through C20. The additional weighted capacitor array 28 has included therein an array of capacitors C24 through C28 having one plate connected to a common point and to one input of unity buffer amplifier 31. Capacitors C24 through C28 are weighted in this example to assume capacitance values of C', C'/2, C'/4, C'/8 and C'/8 respectively. The common point between capacitors C24 through C28 is connected to one terminal of a switch S23 which connectes the common point to the reference voltage level for comparator 23, ground in this instance. The noncommon terminals of capacitors C24 through C27 are connected to an additional plurality of switches 34 designated S24 through S26 having positions 1 and 2 thereon for alternately selecting the reference signal level for comparator 23 on position 1 or a connection to switch S28 on position 2. Switch S28 has positions 1 and 2 for selecting either positive or negative analog reference voltage V.sub.REF respectively.

The two binary weighted capacitor arrays 22 and 28 are the key precision elements. Other elements as mentioned above include a binary comparator 23, a near unity gain buffer amplifier 31 with high input impedance, numerous analog switches S12 through S28 and binary sequencing and control logic totaling approximately sixty gates and flip-flops. These elements are all realizable on a single MOS chip. A conversion utilizing the circuitry of FIG. 8 proceeds as follows. Initially switches S12 through S21 and S23 through S27 are thrown to the reference voltage level for comparator 23, ground in this example. Next, switches S13 through S20 are retained in position 1 and switch S21 is thrown to position 2 to store V.sub.IN on the capacitor array 22. Next, S12 is opened and S21 is returned to position 1 to select ground while switches S13 through S20 are retained in the number 1 position. The sign of the successive approximation voltage V.sub.X is negative in this example which determines the sign bit of the PCM code. Thus, the analog signal V.sub.IN is sampled and the sample is held on the capacitor array 22. Referring to FIG. 9, the segment in the line representing the relationship between the analog and digital signals is determined by successively throwing switches S20 through S13 in inverse order starting with switch S20 from ground position 1 to analog reference voltage position 2 until the sign of successive approximation voltages V.sub.X changes. Switch S22 is thrown to position 1 to select the positive analog reference voltage if the sign of analog voltage V.sub.IN was previously determined to be positive and to position 2 to select the negative analog reference voltage if the sign of V.sub.IN was determined to be negative. If, for example, the output from comparator 23 changes after switch S18 is thrown, this indicates that the analog input voltage V.sub.IN lies within the segment A-B shown in FIG. 9.

The final stage in the conversion is the determination of the step within the segment. The capacitor array 28 is used in this process. Initially, as described above all switches S23 through S27 are thrown to the reference level for comparator 23, ground in this example. To follow the example entered into above, actuation of any of the switches S13 through S20 applies a trial voltage to the common point for the capacitor array 22 and the trial voltage applied by actuating switch S18 to position 2 provided the first trial voltage which caused comparator 23 to change output states at the serial digital output thereof. Consequently, switch S18 is thrown to position 3 to select a buffered subtrial voltage V.sub.Z provided at the output of unity gain buffer 31. Next S23 is opened and switches S24 through S27 are thrown in sequence from position 1, ground, to position 2 to connect the noncommon terminals of individual ones of capacitors C24 through C27 to switch S28 and to the appropriate analog voltage reference as provided at either terminal 1 or 2 on switch S28. The successive approximation algorithm is used for the "in chord" coding which determines within which of the sixteen steps of the chord A-B of FIG. 10 the analog sample signal lies. Thus, the subtrial voltage V.sub.Y is obtained at the input to unity gain buffer amplifier 31 which is provided as the buffered subtrial voltage V.sub.Z connected to terminals 3 for switches S13 through S20, and through switch S18 in this example to be combined with the trial voltage V.sub.X. The combined trial and subtrial voltage levels provide the appropriate serial digital output from comparator 23 which is connected as shown in FIG. 7 to digital control and sequencing circuit 24 for providing the corresponding digital data output therefrom.

Referring to FIG. 13 a timing diagram is shown illustrating the successive approximation described above for the circuit of FIG. 8. The successive approximation voltage V.sub.X is plotted along the ordinate of the chart of FIG. 13 and the time base is plotted along the abscissa thereof. V.sub.X is shown at the reference level for the comparator 23, zero potential or ground in this example, through the first period, t.sub.1 -t.sub.0, when S12 is closed. S21 is set to position 2 to select the analog sample to be applied to the noncommon terminals of the capacitor array C13 through C21 simultaneously. Switch S12 is opened during the second period, t.sub.2 -t.sub.1, and during the third period, t.sub.3 -t.sub.2, switch S21 is returned to position 1 to select the reference for comparator 23 so that -V.sub.IN appears at the common terminal for the capacitor array 22. Next, S22 is placed in position 1 to select positive V.sub.REF at position 2 on switches S13 through S20. During the period t.sub.5 -t.sub.4, switch S20 is thrown to position 2 to determine if the analog voltage V.sub.IN is within the first positive chord in the curve of FIG. 9. FIG. 13 shows that the analog voltage V.sub.IN is greater than that portion of the analog reference voltage applied as a trial voltage to the common point for the capacitor array 22. Switch S20 being retained in position 2, switch S19 is thrown to position 2 during the sixth period to see if a trial voltage has been obtained at the common point of the capacitor array 22 which indicates the chord has been reached wherein the analog voltage V.sub.IN lies. FIG. 13 shows a failure during the sixth period, t.sub.6 -t.sub.5, and switch S18 is therefore sequenced to position 2 to present a greater predetermined portion of the analog reference voltage V.sub.REF to the common point of the capacitor array 22. The level of the trial voltage during the seventh period is seen to exceed the reference level for the comparator 23 and the sequencing and control circuit 24 directs switch S18 to position 3 thereby dropping the trial voltage to the level obtained by the positioning of switches S19 and S20 to position 2 and indicating that the third chord on the positive side of the curve of FIG. 9, designated A-B, contains the analog signal V.sub.IN. During the ninth period, switch S24 is sequenced to position 2 to select a portion of the analog voltage reference to be presented at the common point of capacitor array 28 as the subtrial voltage for subsequent application to the common point of capacitor array 22 as described above. The combination of trial and subtrial voltages during period nine may be seen to be in excess of the reference voltage for comparator 23 and switch S24 is returned to position 1 during the tenth period of the conversion as seen in FIG. 13. Switch S25 is subsequently thrown to position 2 to provide a subtrial voltage as described above to the common point of capacitor array 22 which may be seen in FIG. 13 to be less than the reference voltage level for comparator 23. Consequently, switch S25 is left in the position 2 and switch 26 is thrown to position 2 during period twelve producing the signal shown in FIG. 13 in excess of the comparator reference voltage. Switch S26 is returned to position 1 and switch S27 is thrown to position 2 providing a subtrial voltage for combination with the trial voltage at the common point of capacitor array 22 which is less than the reference voltage level for comparator 23. The conversion sequence is ended for the companded analog to digital conversion and the standard PCM representation of the companded digital signal is seen in FIG. 14. The first digit represents the sign or sense of the analog to digital relation. The second, third and fourth digits represent the chord in the analog/digital relation curve as determined by the trial signal. This may be seen to be the third chord A-B in FIG. 9. The fifth through the eighth digits represent the step within the third chord through which the analog signal V.sub.IN passes as determined by the combination of the trial and subtrial signals. This may be seen to be the fifth step in the chord A-B as seen in FIG. 10 of the drawings. The switching sequence directed by the digital control and sequencing circuit 24 may be seen in the chart of FIG. 15. The chart shows that fourteen clock periods have been utilized for the determination of the companded digital code, but some operations may be combined so that the coding may take place in fewer clock periods than shown in the chart of FIG. 15.

The device shown in the block diagram of FIG. 7 and the electrical schematic diagram of FIG. 8 may be used as a decoder or a digital to analog converter as well. The timing diagram of FIG. 11 shows the two analog voltages V.sub.Y and V.sub.X being generated in sequence according to digital control and sequencing provided by the circuit 24 including the switches of FIG. 8 in accordance with the sequence shown in the chart of FIG. 12. FIG. 11 shows the analog voltages V.sub.X and V.sub.Y on the ordinate and the sequential time period on the abscissa. Note that the first sequence shows switches S12 and S23 closed and all other switches in the plurality of switches 33 and 34 as well as switch 21 placed in the number 1 position for selecting the reference signal level for the comparator 23, grouond in this example. Thus, all capacitors in arrays 22 and 28 are discharged initially. Next, switches S12 and S23 are opened in the second sequence. In the third sequence switch S28 is connected to the analog reference voltage on position 1, switch S27 is selected to position 2 and switch S20 is selected to position 3. In this fashion what has previously been referred to as the subtrial voltage is transferred through the voltage division provided by the additional capacitor array 28, through unity gain buffer amplifier 31, through switch S20 and the voltage division provided by capacitor array 22 to appear at the input of comparator 23 as V.sub.X seen in the third sequence of FIG. 11. The plurality of switches 34 are sequenced through the entire sixteen bit succession to provide the sixteen steps in the first chord shown in sequences two through seventeen. Corresponding switching combinations for the plurality of switches 34 are shown in the sequences two through seventeen of FIG. 12. The variation of V.sub.Y is shown for these corresponding switching combinations in FIG. 11.

Switch S20 is placed in position 2 during the eighteenth sequence and switch S19 is placed in position 3 during the ninteenth sequence. The plurality of switches 34 are sequenced through the entire sixteen bit succession as described above providing the sixteen steps in the second chord of FIG. 11. Switch S19 is returned to position 2 in the thirty-fourth sequence and switch S18 is placed in the number 3 position in the thirty-fifth sequence. The plurality of switches 34 are sequenced through the entire sixteen bit succession as before, providing the sixteen steps between the letters A--B as seen in FIG. 11. The analog voltage output decoded with the standard PCM representation of FIG. 14 is shown as V.sub.IN in FIG. 11 intercepting the fifth step of the third chord in the digital/analog relationship. It should be noted that FIG. 11 represents a decoded analog voltage for a successive series of digital inputs. When a specific standard PCM representation is applied to the decoder it is not necessary to generate the analog values in sequence when using the circuit as a digital to analog device. By one set of switch closures any analog output voltage may be generated for a particular PCM companded digital input.

A device for conversion between an analog and a digital indication is provided which is all MOS contruction, which contains compensation for parasitic capacitance and comparator offset errors and which may be applied on a per channel basis in nonlinear analog/digital coding and decoding. The basic converting circuit structure for conversion between analog and digital forms of signal is utilized in both the linear and nonlinear conversion structures.