Ratio independent cyclic A/D and D/A conversion using a reciprocating reference approach

BACKGROUND OF THE INVENTION

The present invention relates to analog-to-digital conversion and digital-to-analog conversion using a switch capacitor technique.

BACKGROUND OF THE ART

Previously described approaches to digital-to-analog and to successive-approximation analog-to-digital conversion have relied on circuit techniques which required the matching of on-chip precision passive components to an accuracy comparable with the integral non-linearity required in the converter. R-2R ladders, resistor strings, and capacitor array digital-to-analog converters (DACs) fall into this category. The realization of the required matching has typically required laser trimming or other form of trimming in order to achieve non-linearity smaller than 0.2 percent. Recently a new technique has been described which allows the automatic calibration of a capacitor array-based analog-to-digital converter. While very promising, this approach requires considerable complexity in implementation.

Traditional cyclic or algorithmic conversion involves voltage comparisons, reference voltage subtraction if applicable, and multiplication of the result by a factor of two. For an analog-to-digital conversion, the first step in the process is the decision as to whether the signal is in the upper or lower half of the full range. If the signal is in the upper range, the reference is subtracted from the signal, and the difference is doubled. The most significant bit of the digital output is set to one. If in the lower half, the input signal is simply doubled, and the most significant bit is set to zero. In the next cycle, the remainder of the previous cycle is used as the input, and the process is repeated. Using this process, a signal can be decoded into twelve bits if twelve iterations are used. In the digital-to-analog operation of the prior art, the least significant bit is decoded first. The reference voltage is added, depending on the digital input code, and the sum is divided by two each time. At the end of the twelve cycles, the analog output is derived from the digital code.

The principal source of error in these two conversions is offset in the loop and a loop gain which is not precisely two or one-half. If the loop gain is not precisely 2 or 1/2, both integral and differential nonlinearity is introduced into the converter transfer characteristic.

These nonlinearities can be explained as follows. In the traditional switch capacitor implementation of cyclic conversion, as previously indicated, an amplifier with a gain of two or one-half is required. This is accomplished through the use of two capacitors and an amplifier. The operation of sampling and transferring the signal twice will result in multiplication of the signal by a factor of the ratio of the capacitors. If the capacitors are equal, then the gain is two. However, deviations from the equal values of the capacitors will result in conversion error. The present invention is directed to overcoming this capacitor ratio inaccuracy and other higher order errors in the conversion loop.

SUMMARY OF THE INVENTION

The present invention contemplates a new approach to the implementation of high-speed analog-to-digital and digital-to-analog conversion which achieves an absolute linearity which is independent of the matching accuracy of passive components used to implement the converter. This is achieved through the use of a modified form of algorithmic converter in which the reference voltage is circulated around the loop as well the signal, thereby cancelling the gain error in the loop in contrast to the prior art.

The reference recirculation technique can be implemented so as to give both ratio-independent analog-to-digital conversion and digital-to-analog conversion. If the reference voltage is circulated as well as the input voltage signal, the reference voltage will experience the same gain error as the signal voltage did, and thus compensation will be accomplished for the gain error.

It is to be understood that because of the ratio-independent aspect, very small capacitor values can be used, and as a result, the die area on the chip required for this circuitry can be scaled downwardly.

An embodiment of the invention includes an apparatus for analog-to-digital signal conversion and for digital-to-analog signal conversion which comprises a first means for sampling, holding and amplifying a signal and a second means for sampling, holding and amplifying a signal. The apparatus further includes a means for communicating the output of the first means to the input of the second means and for communicating the output of the second means to the input of the first means. Further, the apparatus includes means for introducing an input signal to be converted to one of said first and second means, and means for introducing a reference signal to be combined with the input signal to one of said first and second means.

The apparatus further includes control means for circulating said input signal to be converted and said reference signal to be combined with the input signal between the first means and the second means, and means for combining said signals.

BRIEF DESCRIPTION OF THE FIGURES

FIG. 1 is a schematic of an embodiment of the converter of the invention.

FIG. 2 is a flow diagram for analog-to-digital conversion.

FIG. 3 is a flow diagram for digital-to-analog conversion.

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENT

With reference to the figures and in particular to FIG. 1, an embodiment of the A/D and D/A converter 10 of the invention is depicted. The converter 10 is comprised of differential amplifiers 31, 32, comparator 33, pairs of capacitors C₁, C₁ ', C₂, C₂ ', C₃, C₃ ', C₄, C₄ ', C₅, C₅ ', control unit 34 and switches S₀, S₁, S₂, S₃, S₄, S₅, S₆, S₇, S₈, S₉, S₁₀, S₁₁, S₁₂, S₁₃, S₁₄, S₁₅, S₁₆. In a preferred embodiment these components are implemented on a chip in MOS technology with the switches being MOSFETs. The control unit 34 can be implemented in ways that are well known in the art and is used to control the opening and closing of the switches and to control a series to parallel conversion for a digital output as is described below.

In FIG. 1, a reference voltage signal (which in this embodiment is half full scale or V_hfs) is provided as V_ref /2 and -V_ref /2 on lines 40, 42 when switches S₂ are closed, which lines 40, 42 are connected to capacitors C₁, C_1' respectively and can be connected to ground by switch S₄. Capacitors C₁ and C₁ ' are connected respectively to the non-inverting (positive) and inverting (negative) terminals of differential amplifier 31 and capacitors C₂, C₂ ' by lines 44, 46 through switches S₈ which are provided in lines 44, 46. Lines 44, 46 are connected to ground through switches S₅. Lines 48, 50 communicate respectively with the negative and positive output terminals of amplifier 31. Capacitors C₂, C₂ ' are connected across amplifier 31 between lines 44 and 48, and lines 46 and 50. Capacitors C₂, C₂ ' can be shunted and cleared when switches S₉, which are provided across capacitors C₂, C₂ ' are closed.

Capacitors C₅, C₅ ' can be communicated with capacitors C₂, C₂ ' respectively when switches S₈ ' are closed, and with ground when switches S₆ are closed. Further, closing switches S₇ communicate capacitors C₅, C₅ ' with lines 52, 54 respectively.

Output lines 48, 50 can selectively communicate through switches S₁₀ to comparator 33 for a digital output. Additionally, lines 50, 48 can communicate respectively with lines 60, 62 with switches S₁₁ closed and respectively with lines 62, 60 with switches S₁₂ closed.

Lines 60, 62 communicate with capacitor C₃, C₃ ' respectively and can communicate with ground through switch S₁₃. Capacitors C₃, C₃ ' communicate through lines 64, 66 with the non-inverting (positive) and inverting (negative) terminals of differential amplifier 32 when switches S₁₅ provided in lines 64, 66 are closed. Switches S₁₄ can communicate lines 64, 66 with ground.

An analog input signal V_X /2, -V_X /2 (for A/D conversion) is communicated to lines 60, 62 through switches S₀. Another digital reference signal V_Y /2, -V_Y /2 (equivalent to V_hfs /2, -V_hfs /2, which represents the mid-rise half least significant bit (LSB) is provided to lines 60, 62 by switches S₁.

Lines 68, 70 communicate with the negative and positive output terminals of amplifier 32 respectively. Lines 68 and 70 communicate with lines 52 and 58 respectively through switches S₁₈.

Capacitors C₄, C₄ ' are communicated between lines 64 and 68 and lines 66 and 70 respectively across differential amplifier 32. Switches S₁₆ are provided across capacitors C₄, C₄ ' and can shunt and thus clear capacitors C₄, C₄ '.

For analog-to-digital conversion, the ratio C₄ /C₃ is equal to one, and for digital-to-analog conversion, the ratio C₄ /C₃ is equal to two. Otherwise all the capacitors have the same value.

For analog-to-digital conversion, the output is read from the output of comparator 33. Control unit 34 has a series-to-parallel converter which converts a series of bits to a word output. For digital-to-analog conversion, the output is read from the operational amplifier 31 at the end of the conversion cycle.

INDUSTRIAL APPLICABILITY

Flow charts showing the cyclic or algorithmic conversion technique of the present invention are depicted in FIGS. 2 and 3. For analog-to-digital conversion (FIG. 2), first the sign of the analog input voltage V_in is detected by block 80 by methods known in the art, and V_in is designated V_x for purposes of the algorithm. If the sign is positive (V_x >0) the left hand portion of the algorithm is selected and the sign bit is set to one. Then the reference voltage which is designated V_ref in FIG. 1 and which is equal to the half full scale voltage, or V_hfs, is subtracted from V_x (Block 84). If the result is greater than zero the most significant bit is set to one (Block 86) and the difference is multiplied by two (Block 90). If the difference is less than zero the reference voltage V_vfs is added back to the difference to arrive at the original V_x (Block 88), which original V_x is multiplied by two (Block 90). The bit position is decremented (Block 92) until all the positions including the last position, which is the least significant bit (LSB), are filled. The right hand portion of the flow chart shows a similar technique when the sign of V_in is negative. In this situation, -V_vfs is used as V_ref.

For digital-to-analog conversion (FIG. 3), after the sign bit is detected by techniques known in the art (Block 100), and assuming it is positive, the reference voltage V_hfs is divided by two and designated V_Y (Block 102). If the bit in the least significant position is set to one (Block 104), then the reference voltage V_hfs is added to V_Y (Block 108) and the result is divided by two (Block 106). If the bit is not set, then V_Y is divided by two (Block 106). This algorithm repeats itself (Block 108) until the most significant bit is processed. Again if the sign bit is negative, the right portion of the algorithm of FIG. 3 is used. Again -V_hfs is used as V_ref in this situation.

Incorporating the algorithm of FIG. 2 with FIG. 1 and the control sequence for analog-to-digital conversion in Table I, the operation of the analog-to-digital converter 10 (FIG. 1) of the invention for positive values of V_in is as follows.

Initially the V_in /2 analog signal and the inverted V_in /2 signal are sampled on capacitors C₃, C₃ ' by closing switch S₁₄ to ground and closing switch S₀. V_in is designated V_X. After this is accomplished, capacitor C₄, C₄ ' is cleared by closing switch S₁₆ (cycle 1). The reference signal V_ref /2, -V_ref /2 (which is equivalent to the voltage of half full scale) is sampled on capacitors C₁, C₁ ' respectively by closing switches S₂, S₅. Next the capacitors C₂, C₂ ' are cleared by closing switch S₉ (cycle 2). The reference voltage is transferred to capacitors C₂, C₂ ' from capacitors C₁, C₁ ' by closing switches S₄, S₈. Then the input voltage V_X is transferred to capacitors C₄, C₄ ' from capacitors C₃, C₃ ' by closing switches S₁₃ and S₁₅ (cycle 3). V_ref is then sampled on capacitors C₃, C₃ ' from capacitors C₂, C₂ ' by closing switches S₁₁, S₁₄. The V_X is sampled on capacitors C₁, C₁ ' from capacitors C₄, C₄ ' by closing switches S₁₈, S₃, and S₅ (cycle 4). Next in cycle 5 of the control sequence, capacitors C₂, C₂ ' are cleared by closing switch S₉. At the same time the top plate potential of capacitors C₅, C₅ ' are set by closing switches S₆, S₈ ' and which stay closed until cycle 7. Capacitors C₄, C₄ ' are cleared by closing S₁₆. In cycle 6, V_ref is transferred to capacitors C₄, C₄ ' from capacitors C₃, C₃ ' by closing switches S₁₃ and S₁₅. In cycle 7, V_X is transferred to capacitors C₂, C₂ ' from capacitors C₁, C₁ ' by closing switches S₄, S₈ and V_ref is subtracted from capacitors C₂, C₂ ' through capacitors C₅, C₅ ' by closing switches S₇, S₁₈, S₈ ' and opening S₆. The sign of the resulting value is then compared in comparator 33 by closing switch S₁₀. If this value is greather than zero a 1 bit is set in the most significant bit location. If this value is less than zero, the charge on C₂ is restored to its original value (cycle 8) by closing switch S₆ and opening S₇. If this occurs, and thus the value is less than zero, a zero is placed in the most significant bit location. Next in cycle 9, the V_ref on capacitors C₁, C₁ ' is sampled from C₄, C₄ ' by closing switches S₁₈, S₃, S₅. The value on capacitors C₂, C₂ ' is sampled on capacitors C₃, C₃ ' by closing switches S₁₁, S₁₅, S₁₆. The voltage on capacitors C₃, C₃ ' is then multiplied by two and transferred to C₄, C₄ ' by closing switch S₁₂, opening switch S₁₆ and opening switch S₁₁ (cycle 10). Thus a doubling of remainders is accomplished by sampling and reversing the input path on capacitors C₃, C₃ ' using a switch capacitor technique. The value on capacitors C₂, C₂ ' is cleared by closing switch S₉ (cycle 11). Next the value V_ref is transferred to capacitors C₂, C₂ ' from capacitors C₁, C₁ ' by closing switches S₄, S₈. At this point the algorithm repeats itself in order to determine the remaining bits by going back to the steps of cycle 4 and continuing until all the bits have been determined.

Incorporating the algorithm of FIG. 3 with FIG. 1 and the control sequence for digital-to-analog conversion as shown in Table 2, the operation of the digital-to-analog converter of the invention for positive values is as follows:

Initially, the mid-rise half value of the least significant bit (LSB) is sampled on capacitors C₃, C₃ ' by closing switches S₁ and S₁₄. The mid-rise half least significant bit voltage is the half full scale voltage. Then the capacitors C₄, C₄ ' are cleared by closing switch S₁₆, and the reference voltage V_ref is sampled on capacitors C₁, C₁ ' by closing swtiches S₂ and S₅. This time the V_ref voltage is equivalent to the negatives of the half full-scale voltage. Then the capacitors C₂, C₂ ' are cleared by closing switch S₉. This completes the first cycle of the digital-to-analog conversion.

In the second cycle, the reference voltage, V_ref, is transferred to capacitors C₂, C₂ ' by closing switches S₄, S₈. Then V_Y is transferred to capacitors C₄, C₄, by closing switches S₁₄ and S₁₅. V_Y is then sampled on capacitors C₁, C₁ ' from capacitors C₄, C₄ ' by closing switches S₁₈, S₃, and S₁₆. Then V_ref is then sampled on capacitors C₃, C₃ ' from capacitors C₂, C₂ ' by closing switches S₁₁, S₁₅, and S₁₆. The reference is then multiplied by two and transferred to C₄, C₄ ' by closing switch S₁₂ and opening switch S₁₆ and opening switch S₁₁. The reason that this is accomplished is that the ratio of the capacitor C₃ over the capacitor C₄ is one half, and it is desired that the reference not be halved at this point but put through at a full value which is a value of half full scale voltage.

In cycle 3, capacitors C₂, C₂ ' are cleared by closing switch S₉ and capacitor C₅, C₅ ' has its top plate potential set by closing switches S₈ ', S₆.

In cycle 4, V_Y is transferred to capacitors C₂, C₂ ' from capacitors C₁, C₁ ' by closing switches S₄, S₈. At this point if the input bit, which for the first pass through this algorithm as the least significant bit, is 1, the V_ref is subtracted from the V_Y in the capacitors C₂, C₂ ' through capacitors C₅, C₅ ' by closing switches S₇, S₁₈ (which means V_hfs is added to V_Y. If the bit is not a 1, this addition does not occur. The control unit 34 senses the value of the bits in order to decide which switches to open and close.

In cycle 5, the V_ref is sampled on capacitors C₁, C₁ ' from capacitors C₄, C₄ ' by closing switches S₁₈, S₃, and S₅. V_Y is sampled on capacitors C₃, C₃ ' from capacitors C₂, C₂ ' by closing switches S₁₂, S₁₄. Then capacitors C₂, C₂ ' and C₄, C₄ ' are cleared by closing switches S₉ and S₁₆ respectively. At this point the algorithm repeats itself by returning to cycle 2 to process the next bit. This algorithm repeats itself until the most significant bit has been processed and the digital-to-analog conversion has been accomplished. The analog output is available at the output of device 31. Again it is to be remembered that both with the control sequence shown in Table 1 and Table 2, that the control unit 34 which controls this functions could be provided by one of ordinary skill in the art.

From the above it can be seen that the present invention accounts for and corrects gain errors by circulating the reference voltage so that both the signal and the reference voltage have the same error, which error is cancelled when the signal and the reference voltage are combined.

Other advantages, and objects of the invention can be obtained from a review of the figures and appended claims. It is to be understood that other embodiments of the present invention can be developed and still come within the scope and spirit of the appended claims.

                                  TABLE I__________________________________________________________________________Analog-to-Digital Converter's Control SequenceCYCLEFUNCTION                     SWITCHES CLOSED__________________________________________________________________________1    Sample the input signal (+VX /2,-VX /2) on C3,C3 '                    S0, S14Clear the C4, C4 ' S162    Sample the Reference (+Vref /2,-Vref /2) on C1,C1 '                    S2, S5Clear the C2, C2 ' S93    Transfer Vref to C2, C2 ' from C1, C1                             S4 S8Transfer VX to C4 C4 ' from C3,                             S13 S154    Sample Vref on C3, C3 ' from C2                             S11 S14Sample VX on C1, C1 ' from C4 C4 '                             S.sub. 18 S3 S55    Clear C2, C2 '     S9Clear C4, C4 '     S16Set C5, C5 'top plate potential                             S6 S8 '6    Transfer Vref to C4, C4 ' from C3, C3                             S13 S157    Transfer VX to C2, C2 ' from C1, C1                             S4 S8and Substract Vref from C2 C2 ' through C5,C5 '                    S7 S18 S8 'Determine sign with comparator                             S108    Restore the charge if needed S69    Sample Vref on C1, C1 ' from C4, C4                             S18 S3 S510   Sample the remainder on C3, C3 ' thru (C2 , C2')                           S11 S15 S16Multiply the remainder by 2 & transfer the charge to C4,C4 '                    S12 (open S16, open                             S11)                             (inverse)11   Clear C2, C2 '     S912   Transfer Vref to C2, C2 ' from C1, C1                             S4 S8Continue the remaining bits by going back to Cycle__________________________________________________________________________4

                                  TABLE II__________________________________________________________________________Digital-to-Analog Converter' s Control SequenceCYCLEFUNCTION                    SWITCHES CLOSED__________________________________________________________________________1    Sample the mid-rise half LSB (+Vy /2, -Vy /2) on C3,C3 '(Vy =Vhfs /2)     S1 S14Clear the C4, C4 '                            S16Sample in the reference (+Vref /2, -Vref /2) on C1,C1 '                   S2 S5Clear the C2, C2 '                            S92    Transfer Vref to C2, C2 '                            S4 S8Transfer Vy to C4, C4 '                            S13 S15Sample Vy on C1, C1 ' from C4,                            S18 S3 S16Sample Vref on C3, C3 ' from C2, C2                            S11 S15 S16Multiply the reference by two and transfer to C4, C4                            S12 open S16 open S11(Since C3 /C4 is one-halfthe actual factor for the reference is unity)3    Clear C2, C2 '    S9Set C5 C5 ' top plate potential                            S8 ' S64    Transfer Vy to C2, C2 ' from C1, C1                            S4 S8If the input digital bit is oneAdd Vref to C2 C2 ' through C5                            S7 S18Otherwise do not add5    Sample Vref on C1, C1 '                            S18 S3 S5Sample the Vy on C3, C3 ' from C2, C2                            S12 S146    Clear C2               S9Clear C4               S16Continue with the remaining bits by going back to Cycle__________________________________________________________________________2