Common mode shift in downstream integrators of high order delta sigma modulators

[0002] 1. Field of the Invention

[0003] The present invention relates to the field of converter circuits and, more particularly, to high order modulator circuits.

[0004] 2. Background of the Related Art

[0005] The use of analog-to-digital converters (A/D converters or ADCs) to convert analog signals into digital signals is well known in the art. State-of-the-art practice is to fabricate an ADC on a single integrated circuit chip. Generally, an ADC samples an analog signal at a sampling rate which is twice the highest frequency component being sampled. This is known as the Nyquist rate. The sampled signal is processed and converted into a digital format for output from the converter.

[0006] A particular type of ADC employs an oversampling technique in which the analog input signal is sampled at a much higher rate than the Nyquist rate. The higher sampling rate improves the performance of the ADC for signal conversion and processing. One commonly used oversampling type of ADC uses a delta-sigma (ΔΣ) modulator for oversampling the analog input. The oversampled output of the ΔΣ modulator is coupled to a decimator, which employs a low-pass filtering technique to extract the lower frequency components to generate a converted digital output signal at the Nyquist rate. The ΔΣ modulators generally employ multiple integrator stages for providing the signal conversion.

[0007] It is appreciated that a typical high order modulator for a mixed-signal circuit utilizes a large number of digital switches. For example, when switched capacitor sampling circuits are utilized to sample the input signal, four switches are typically used with the sampling capacitor for each stage of the modulator. In a fifth-order ΔΣ modulator, there are five integrator stages present, each with a switched capacitor input sampling circuit. When differential integrator stages are used, at least eight switches are typically required for the sampling operation of each integrator stage.

[0008] In order to reduce power consumption in a chip, devices are designed to operate at lower supply voltages. Low power devices are desirable for applications where power drain is a concern. A lap-top (or notebook) computer, which uses a battery power source, is an example. However, the switches may not turn on at these lower voltages, if the gate voltage is not appreciably different from the source/drain voltage. To alleviate this turn-on problem, prior art modulators use a voltage booster (such as a voltage doubler) to drive the gates of the switches to ensure that the switches turn on. However, boosting the voltage for switch activation increases the power consumption in the device, as well as increasing the area on the device to accommodate the boosting circuitry.

[0009] The present invention implements a technique which allows certain stages of the modulator to operate at lower voltages in order to reduce power consumption in a mixed-signal modulator.

[0010] A modulator comprised of a plurality of differential stages for converting an input signal is described. A range of the input to the modulator is such that the first stage requires a common mode voltage that requires a boosted supply voltage to clock the input transistor switches. The outputs of the downstream integrators are typically smaller so as to enable a common mode that is closer to either the supply voltage or its return. Accordingly, a non-boosted voltage can be utilized for the gate drive of the downstream integrator stages, which results in lower power consumption and less chip area for the voltage boost circuitry of the modulator, thus making the modulator implementation more power/area efficient at lower supply voltages.

[0011] FIG. 1 is a block schematic diagram of a high order modulator incorporating the present invention.

[0012] FIG. 2 is a circuit schematic diagram of a first integrator stage of the modulator illustrated in FIG. 1.

[0013] FIG. 3 is a circuit schematic diagram of a second integrator stage of the modulator illustrated in FIG. 1.

[0014] FIG. 4 is a circuit schematic diagram of a third and subsequent integrator stage of the modulator illustrated in FIG. 1.

[0015] FIG. 5 is a block schematic diagram of a CODEC which implements the modulator incorporating the present invention.

[0016] Referring to FIG. 1, a high-order modulator 9 for use in signal conversion is shown. In the particular example, modulator 9 is a delta-sigma (ΔΣ) modulator used in an analog-to-digital converter (ADC). The ΔΣ modulator shown is a fifth-order modulator having five integrator stages 1-5, which are also identified as stages I₁-I₅. The first stage I₁receives an analog input signal and commences the sampling procedure. A reference voltage V_REF, typically coupled to an external filtering capacitor, is used as a reference voltage for the analog input signal at the first stage. The integrator stages are coupled serially, so that a differential output from a given stage (noted as OUT_P1and OUT_N1for the positive and negative differential outputs from stage I₁, OUT_P2and OUT_N2for stage I₂, etc.) is fed as an input to the next stage. Each stage output is also coupled to a summation network 8 to generate a final output from a comparator, which is included within the summation network 8. The output D from the network 8 is either a 1 or a 0 (the digital states). Thus, for each sampled input of the analog signal, a digital output D is provided. It is appreciated that the basic concept of a ΔΣ modulator is known in the art.

[0017] Although the concept of a ΔΣ modulator is known, the modulator 9 of FIG. 1 implements the present invention as described below in reference to the circuitry shown in FIGS. 2-4. FIG. 2 shows the integrator for stage 1, FIG. 3 shows the integrator for stage 2 and FIG. 4 shows the integrator for each of the last three stages 3-5. It is appreciated that there are other circuit components within each of the stages 1-5, but only the circuitry relevant to understanding the present invention are illustrated in FIGS. 2-4.

[0018] Also shown in FIG. 1 are a clock generation circuit 6 and a supply voltage booster circuit 7. The clock generation circuit 6 receives a reference clocking signal CLK and generates the various clocking signals required by the integrators 1-5. The various clocking signals shown are discussed below in reference to the other Figures. The voltage booster circuit 7 is utilized to boost (such as double) the supply voltage Vdd coupled to the device. The boosted supply voltage (higher voltage) is noted as “hv”, such as Vdd−hv. In the example, φ₁and φ₂signals switch between Vdd and ground. The other clock signals having −hv appended, switch between Vdd−hv and ground.

[0019] The summation network 8 receives the outputs from each of the integrator stages 1-5 to provide a summation which is compared to the common mode voltage in the included comparator to derive a digital signal, which corresponds to the analog signal sampled. The digital output D is also used as a feedback signal to stage 1 to control the polarity of the +V_REFand −V_REFsignals being applied to the input of the stage 1 integrator.

[0020] Shown in FIG. 1 are two common mode voltages V_CM1and V_CM2. V_CM1is coupled to stage 1 and 2. V_CM2is coupled to stage 2 and beyond. As will be described below, V_CM1is used with stage 1 where the drive voltages are all boosted voltages. Stage 2 receives and operates with both V_CM1and V_CM2. The remaining stages operate having the common mode of V_CM2. As will be described, the shifting of the common mode voltage at stage 2 allows circuitry using V_CM2for the common mode to operate at a lower (non-boosted) supply voltage drive.

[0021] In FIG. 2, an integrator circuit 10 is shown having two sets of switched capacitor input sampling circuitry, which are coupled to a summation node at the input of the integrator 12. The switched capacitor integrator circuit 10 is a fully-differential integrator in which a switched capacitor input circuitry 11 is used to sample a differential input (+IN and −IN). An integrator 12, having integrating capacitors C_INT1in each of the feedback paths from the differential output (OUT_P1and OUT_N1) to the input, provide the integration of the sampled analog input. The input circuit is comprised of a sampling capacitor C_S1for each input line of the differential input for sampling the analog signal. The frequency of the sampling is determined by the operation of transistor switches, which closing/opening is synchronized to clocking signals φ₁and φ₂to switch in/out the sampling capacitors. During time φ₁the input is sampled by the sampling capacitors, and during φ₂the sampled signal is coupled to the input of the integrator 12. The common connection is noted as V_CM1to denote the common mode voltage of the differential amplifier.

[0022] The example circuit 10 utilizes transistors 13a,b-15a,b(also referred to as switches, due to the switching action of the transistors) to perform the switching operation on the sampling capacitor C_S1. The φ₁and φ₂clock signals on the gate of the transistor switches control the activation of these transistors. The “hv” designation denotes the higher voltage applied to the gate of these transistors during each of the phase periods φ₁or φ₂. Therefore in FIG. 1, the gate drivers of the transistors 13a,b-16a,bare φ₁−hv or φ₂−hv clocking signals. The “a” and “b” designations are used to differentiate the two sides of the differential circuit. Therefore, during φ₁switches 13a,band 14a,bclose to sample the analog input signal (switches 15a,band 16a,bare open). During φ₂switches 15a,band 16a,bclose to transfer the sampled input to the integrator 12 (switches 13a,band 14a,bare open).

[0023] It is to be noted that an alternative technique is to provide delayed clocking phases of φ₁and φ₂to some of the switches in order to improve the performance of the switching circuitry. For example, φ₁d−hv (delayed φ₁−hv) is used to clock switches 13a,band φ₂d−hv (delayed φ₂−hv) is used to clock switches 15a,b. The delay is a short delay, typically in the nanosecond range. Furthermore, similar delayed clocking scheme can be applied with the input switches 31a,band 33a,bin FIG. 3 and 41a,band 43a,bin FIG. 4.

[0024] Furthermore, the integrator stage 1 is generally the most critical stage in a ΔΣ modulator. To obtain better performance, a completely symmetric and differential setup is preferred. For example, V_CM1should equal Vdd/2, so that OUT_P1and OUT_N1can have the largest symmetrical swing.

[0025] In order to provide the sampling, the input signal is sampled and compared to a reference voltage V_REF. Typically, the +V_REFand the −V_REFterminals are coupled to an external capacitor C_R, in ordered to provide a filtered voltage as a reference source. The +V_REFand the −V_REFterminals are coupled to a pair of sampling capacitors C_SRduring φ₁, when switches 19a,band either switches 17a,bor 18a,bclose. During φ₂the sampled signal is coupled to the input of the integrator 12, when switches 20a,band either switches 21a,bor 22a,bclose, where it is summed with the sampled input signal at the integrator input nodes. The +V_REFand the −V_REFlines are coupled to the integrator 12 input by the operation of switches 21a,band 22a,b, according to an one-bit data output D of the modulator 9.

[0026] For example, if the feedback is “1”, then the +V_REFis coupled to the minus input of the integrator 12 and if the feedback is “0”, the +V_REFis coupled to the plus input. That is, switches 21a,bclose during time period φ₂when D=1 and switches 22a,bclose when D=0. In the example circuit 10, a logic circuit 25 is used to provide the AND logic function of D and φ₂−hv, as well as D# (# denoting a complement of D and shown as a bar over D in the Figures) and φ₂−hv to generate the D·(φ₂−hv) and (D#)·(φ₂−hv) to activate switches 21a,band 22a,brespectively.

[0027] During φ₁the switches 19a,bclose. Also during φ₁switches 17a,band 18a,bclose, but not at the same time. The switches 17a,band 18a,bare operated by the clocking signals φ_1Rand φ_1F. The φ_1Rsignal transitions high and closes switches 17a,bfirst, allowing the input summing nodes to charge/discharge toward Vdd/GND. Then switches 17a,bopen after a selected time delay and switches 18a,bclose with the transition to φ_1F. This technique allows rough (R) charging/discharging of the reference capacitor C_Rnodes initially, followed by a fine (F) leveling of the voltage +V_REF/−V_REFat the reference capacitor nodes. The coarse or rough charging/discharging (by the closing of switches 17a,b) allows DC voltages to be used to bring the integrator input nodes faster toward the appropriate V_REFvalue (which is Vdd), so that when the actual filtered V_REFpotential is switched in, only a fine adjustment is required. This technique inhibits large load swings to be impressed on the V_REFreference lines, which swings could cause performance degradation and harmonic distortion in the modulator. As noted earlier, the “hv” designation denotes a “high voltage” signal. Typically, the signals φ₁and φ₂are clocking signals switching between the supply voltage Vdd and its return Vss (which is ground in this instance). When the clocking signal is in the high state, Vdd is placed on the gate of the transistor and the transistor is turned on to pass the signal through. When the clocking signal is in the low state, ground is placed on the gate and the transistor is turned off. The common mode voltage V_CM1is generally set at Vdd/2.

[0028] When the supply voltage operates at a higher level (such as at 5 V), the transistors have no difficulty turning on when Vdd is placed on the gate. Since the common mode is set at 2.5 V, a 1 V rms (1.4 V peak) signal variation implies a source/drain voltage of 1.09 V on a n-channel (3.91 V on a p-channel) transistor. With a gate drive of 5 V (0 V for p-channel), the transistor is turned on sufficiently. However, at some lower threshold, Vdd on the gate is too low in value to activate the transistor. For example, if the device is designed for operation at a supply voltage of 1.8 V, the common mode voltage drops to 0.9 V. Even half of the previous input signal variation of 0.707 V about the common mode voltage results in an input signal variation in the range of 0.193-1.607. With 1.607 V at the input and only 1.8 V on the gate, the transistor will not turn on. Thus, a voltage booster circuit is used to boost the clock signals which gate the transistor switches. With a voltage doubler, the Vdd of 1.8 V is boosted to provide a clocking signal of 3.6 V to turn on the transistor. In actual practice, a boost to about 3.0 volts is achieved to turn on the transistors.

[0029] Thus, as noted in FIG. 2, the clocking signals to the gates of the transistor switches have their signal levels boosted. This allows a Vdd of 1.8 volts as the supply voltage for the device (chip), but the voltage boosting is used where necessary to ensure activation of the transistor switches. However, it is appreciated that voltage boosting increases the power consumption by the transistor switches.

[0030] In the practice of the present invention, the very first stage 1 utilizes boosted clocking voltages on the gate of the transistor switches 13a,b-22a,b. The boosting is desirable since the input signal as well as the output of the integrator stage 1 may fluctuate substantially, as noted in the example above and boosting may be required to turn on the transistor switches. However, the subsequent stages 2-5 do not fluctuate as much as the first stage. Rather, the outputs from the subsequent stages 2-5 remain much closer to the common mode, so that any variations in signal output is not as drastic as the first stage. Accordingly, a technique of the present invention is used to permit later stages of the modulator 9 to have a different common mode voltage than the first stage. This allows the normal supply voltage value Vdd to drive the gates of the transistors of the later stages. The voltage boosting is not required to drive the gates of the transistor switches of these remaining stages.

[0031] In FIG. 3, an integrator circuit for stage 2 is shown. An input sampling circuit, comprised of transistor switches 31a,b-34a,band sampling capacitors C_S2along with integrator 35, operate equivalently to the input switched capacitor sampling circuit of FIG. 2, but with a notable difference. The common mode voltage for the circuitry to the input side of the C_S2capacitor has the common mode voltage of V_CM1. The common mode for the circuitry to the integrator side of C_S2has a different common mode voltage V_CM2. Furthermore, the switches which operate with the V_CM1common mode voltage (which are switches 31a,band 33a,b) still have their gates driven by the boosted clocking signal voltage of φ₁−hv or φ₂−hv. However, the switches which operate with the V_CM2common mode voltage (which are switches 32a,band 34a,b) now utilize the normal Vdd value for the clocking signal. Thus gates of switches 32a,band 34a,boperate on the normal (non-boosted) clocking signals φ₁and φ₁(and not the φ₁−hv and φ₂−hv). As noted previously, switches 31a,band 33a,bcan be delayed.

[0032] Since the input signal swing remains very close to the common mode voltage, the circuitry past the sampling capacitor C_S2of stage 2 can operate at the normal unboosted gate drive on the transistor switches. However to ensure that the transistor switches are turned on, the V_CM2value is set to some value below V_CM1. In the example, since V_CM1is set to Vdd/2, V_CM2is set to a value somewhere near Vdd/4.

[0033] Referring to FIG. 4, an integrator circuit for use in subsequent integrator stages is shown. The circuit of FIG. 4 is equivalent to the circuit in FIG. 3, except that now all of the transistors 41a,b-44a,bare driven by the normal unboosted clocking signals φ₁or φ₂. Again, φ₁and φ₂signals to the gates of the transistors 41a,band 43a,bcan be delayed. However, in this instance the delayed signals are the unboosted φ₁d and φ₂d (not φ₁−hv or φ₂−hv). Thus, stages 3-5 utilize the circuit of FIG. 4, in which unboosted clocking signals drive the gates of transistors 41a,b-44a,b. The circuitry uses the V_CM2common mode voltage, which in the example is in the vicinity of Vdd/4.

[0034] It is appreciated that the invention shifts the common mode voltage to a lower value at one of the integrator stages, so that subsequent integrator stages can use the lower common mode voltage. In the example, the common mode shift occurs in the integrator stage 2. Due to the DC blocking property of capacitors, the sampling capacitor at stage 2 is used as the demarcation for the common mode shift. Stage 2 is selected for initiating the common mode shift, since integrator outputs from stage 2 and beyond remain very close to the common mode value. The shifting of the second common mode voltage to a value below the first allows the transistor switches to be driven by the normal unboosted supply voltage. Only the gate drives of the transistors having the higher common mode require the voltage boost. Accordingly, by reducing the gate drives of some of the stages, power consumption is decreased and chip area required for the voltage boosting circuitry of the modulator is also decreased.

[0035] It was noted above that low-power mixed-signal devices (on semiconductor chips) operating at low supply voltages need boosted clocks (or gate drive control signals) to turn on n-channel transistors or switches adequately. These boosted signals can either be derived from boosted clock doubler circuits or by having clocks or control signals level-shifted up to the boosted supply in which case, a separate boosted supply generator is required. Regardless of the implementation, by having fewer switches that need boosted supply drive, power savings (due to a smaller capacitive load off the boosted supply and less crank-through current in the level-shifters as well) can be achieved. In the second implementation, by having fewer switches and hence a smaller load capacitance, smaller filtering and pump capacitors are required in the voltage boost circuitry, resulting in less chip layout area. Overall, the fewer switches that have to be run off the boosted supply, the less power and area are required for the voltage boost circuit.

[0036] The invention is described above implementing NMOS transistor devices. However, the invention can be implemented using PMOS devices as well. In that instance, the drive signals and voltage values will be patterned for differential circuits using PMOS transistors. Thus, for example, the V_CM1voltage would still be Vdd/2, but V_CM2will now be Vdd-(Vdd/4). The PMOS transistors will be driven when the drive is low (ground in the example) and the clock phases will be inverted. Otherwise, the circuitry operates equivalently to the NMOS circuitry described above.

[0037] It is appreciated that the modulator comprising the present invention can be utilized in a variety of ways and implemented in a variety of systems. In FIG. 5, a low-voltage, stereo audio coder-decoder (CODEC) integrated circuit chip 70 is shown in which digital input/output is at the left and the analog input/output is at the right. A serial port receives a digital input and separates the right (R) and left (L) channels of the stereo signal. The signals are coupled to an input processing circuit 72. The circuit 72 provides de-emphasis, volume control, base/treble control, boost, compression limiting etc. and sends the signal to a digital filter 73, prior to signal conversion. The filtered digital signals are converted to analog signals by a pair of DACs 74a,b. The DACs 74a,butilize ΔΣ modulators for providing the conversion. The converted outputs are filtered by a pair of analog filters 75a,bto remove the high sampling frequencies and retrieve the lower frequency audio analog signals.

[0038] The R and L analog outputs are coupled to respective volume controls 76a,b, which when amplified by a headphone amplifier 78, provides stereo audio outputs (HP_Rand HP_L) to a headphone coupled to the chip 70. The analog outputs are also coupled to a gain compression circuit 77 before being amplified by a line amplifier 79 to provide line outputs (L_Rand L_L).

[0039] In the opposite direction for signal processing, stereo analog inputs are received at L1/L2 and R1/R2 terminals. The R and L analog inputs are coupled to a pair of multiplexers (MUXs) 81a,b, for selecting between different inputs. After the gain is adjusted by a pair of gain amplifiers 82a,b, the analog signals are coupled to a pair of ADCs 83a,b. The ADCs 83a,balso utilize ΔΣ modulators for high frequency sampling to provide the conversion. A digital filter 84 is used to remove the high frequency component and retrieve the converted filtered signal. The digital signal is processed by an automatic gain control (AGC) circuit by respective AGC circuits 85a,b, then coupled to the serial port 71 for providing a digital output from the chip 70.

[0040] Also, noted is a control circuit 89, which receives control signals from an external circuitry and generates necessary on-chip signals. It is appreciated that other signal lines and terminals are not shown, but are understood to be present for operation of the chip 70. Such signals include, clocking signals, reset signals, supply voltages and grounds.

[0041] In the particular example, the modulator 9 incorporating the invention is implemented in the ADCs 83aand 83b. The CODEC 70 is capable of operating with a supply voltage in the approximate range of 1.6-3.6 volts. Accordingly, the CODEC 70 is suitable for use in portable devices, such as portable audio CD (compact disk) players, MPs and MD recorders and other portable systems that require extremely low power consumption.

[0042] It is appreciated that an ADC is described above to illustrate an implementation of the present invention. However, the invention need not be limited to such application only. The shifting of the common mode voltage for the downstream integrator stages allows for operation of those stages at a lower supply drive voltage than the first stage. The lower voltage operation of the switches results in lower power consumption for the device.

[0043] Thus, a common mode shift in downstream integrators of high order modulators is described. The particular example described is in reference to a delta-sigma modulator used in an analog-to-digital converter. However, the invention can be readily used in other applications which utilize differential circuitry and/or where the common mode of the signal at different points are such that the common mode can be translated to avoid the need for boosted clock generators.

[0044] Furthermore, it is to be noted that the invention is described in reference to a fully-differential integrator circuit. However, single-ended circuitry can be utilized to obtain equivalent differential results. For example, the circuit of FIG. 2 can be adapted for single-ended input by connecting the +IN and −IN terminals together to receive a single-ended input, but having the clock drives to the gates of switches 13band 15breversed. Therefore, switch 13bis now gated by φ₂−hv and switch 15bis now gated by φ₁−hv. The resulting operation of this circuit is equivalent to the differential circuit described in FIG. 2. Thus, other circuit implementations can readily implement the present invention.