Cross-Reference to Related Applications

[0001] This application is a continuation of and claims benefit of and priority to United States Non-Provisional Patent Application Serial no. 14/454,342 filed August 7, 2014, entitled "METHODS AND APPARATUS FOR LOW INPUT VOLTAGE BANDGAP REFERENCE ARCHITECTURE AND CIRCUITS," the entire contents of which is incorporated by reference herein in its entirety.

Background

[0002] Some embodiments described herein relate generally to methods and apparatus for generating a temperature insensitive bandgap voltage reference using an input (supply) voltage that is lower than the base-emitter voltage (VBE) of a bipolar junction transistor (BJT).

[0003] Portable electronic/electrical systems that operate from a battery and/or from power harvested from the internal local environment typically consume small amounts of energy to prolong the system lifetime for a given amount of available energy. The energy budget for a portable system affects a widening set of applications due to a combination of requirements for smaller size (less battery volume, and hence less energy available), longer lifetimes (energy has to last longer), and/or more functionality (increased number of applications to implement with the same amount of energy). Many sensing applications use integrated circuits (ICs) or systems on chip (SoCs) to perform the sensing, computation, and communication functions that are used by a variety of applications.

[0004] In many cases, the time between sensor measurements can be relatively long such that the IC or SoC spends a substantial fraction of its lifetime in a standby mode. Known techniques reduce power consumed by the IC or SoC during standby mode, for example, by power gating unused circuit blocks. A subset of circuit blocks remains powered up during all times of device operation including, for example, a DC-DC regulator remains powered up to supply a stable operating voltage, VDD, which in turn involves a voltage reference to set the correct value for VDD- Typically, the most commonly used voltage reference is a bandgap reference that uses the silicon bandgap voltage to generate a temperature independent voltage reference.

[0005] An ideal voltage reference is independent of variation of power supply or temperature. A voltage reference is often included in many circuits, such as analog-to-digital converters, DC-DC converters, energy harvesting circuits, timing generation circuits, or other voltage regulators. Known implementations of bandgap reference typically involve the use of bipolar junction transistors (BJT) and large resistors to provide generate the bandgap voltage reference. Known conventional bandgap reference circuits, however, are limited to using input voltages higher than the base-emitter voltage (VBE) of a BJT because they inject a current into the BJT using a current source, current mirror, resistor, or switched capacitor network at a voltage higher than VBE-

[0006] Accordingly, for severely energy constrained electronic/electrical systems, a need exists for bandgap reference circuits with a low input voltage to allow for compatibility with energy harvesting and sub-threshold digital logic voltage levels. Additionally, a need exists to minimize power consumption for the bandgap reference circuit.

Summary

[0007] In some embodiments, an apparatus includes a bandgap reference circuit having a first bipolar junction transistor (BJT) that can receive a current from a node having a terminal voltage and can output a base emitter voltage. The terminal voltage of the first BJT substantially corresponds to or is lower than the base emitter voltage of the first BJT for at least a time period. In such embodiments, the apparatus also includes a second bipolar junction transistor (BJT) having a device width greater than a device width of the first BJT. The second BJT can receive a current from a node having a terminal voltage and output a base emitter voltage, where the terminal voltage of the second BJT substantially corresponds to or is lower than the base emitter voltage of the second BJT for at least a time period. In such embodiments, the apparatus also includes a reference generation circuit operatively coupled to the first BJT and the second BJT, where the reference generation circuit can generate a bandgap reference voltage based on the base emitter voltage of the first BJT and the base emitter voltage of the second BJT.

Brief Description of the Drawings

[0008] FIG. 1 is a block diagram of an integrated system used for feeding an input voltage to a bandgap reference circuit used in known portable electrical systems.

[0009] FIG. 2 is a schematic diagram representing a bandgap reference circuit generating a constant voltage reference across varying temperatures, according to an embodiment.

[0010] FIG. 3 is a schematic illustration of a bandgap reference circuit system that uses an input voltage less than the base-emitter voltage of a bipolar junction transistor, according to an embodiment.

[0011] FIG. 4 is a schematic illustration of a bandgap reference circuit that uses switched capacitor charge pumps to drive an input voltage less than the base-emitter voltage of a bipolar junction transistor, according to an embodiment.

[0012] FIGS. 5A-C are schematic illustrations showing the charging of a switched capacitor charge pump circuit associated with the bandgap reference circuit shown in FIG. 4.

[0013] FIG. 6 is a schematic illustration of the charged switched capacitor charge pump circuit shown in FIG. 5A driving an input current into a base emitter voltage clamp.

[0014] FIGS. 7A-7B present simulation results of the variation of VBE and AVBE as a function of temperature that is generated from the bandgap voltage reference circuit of FIG. 4.

[0015] FIGS. 8A-C are schematic illustrations of different scaling circuits to scale AVBE, according to different embodiments. [0016] FIGS. 9A-C are schematic illustrations of different configurations of a scaling circuit to scale VBE, according to an embodiment.

[0017] FIGS. lOA-C are schematic illustrations of a reference generation circuit for generating the bandgap reference voltage, according to an embodiment.

[0018] FIG. 1 1 shows the block diagram of a clock signal generation scheme for the bandpass reference voltage circuit, according to an embodiment.

[0019] FIG. 12 is a schematic illustration of an oscillator shown in FIG. 1 1 that can be used to generate a clock signal for a bandgap reference circuit, according to an embodiment.

[0020] FIGS. 13A-B are schematic illustrations of an implementation of switches for the bandgap reference circuit shown in FIG. 4.

[0021] FIGS. 14A-C are schematic illustrations of the steps involved in implementing a clock doubling technique to generate clock signals at different phases, according to an embodiment.

[0022] FIGS. 15A-B present the results of simulations of an example of a clock doubler circuit that sends boosted clock phase signals to a bandgap voltage reference circuit.

[0023] FIG. 16 shows the annotated lay out of a bandgap reference circuit, according to an embodiment.

[0024] FIG. 17 is a graphical display of an example of the transient behavior of a bandgap reference circuit at startup.

[0025] FIG. 18 shows the simulated variation of an embodiment of a bandgap reference circuit output for a temperature range of -20°C to 100°C.

[0026] FIG. 19 presents the results of a Monte-Carlo simulation that shows an example of the change in bandgap reference output with respect to process and mismatch variation.

[0027] FIG. 20 presents the results of a simulation that shows an example of the change in bandgap reference voltage with respect to variation with input voltage (Vi„). Detailed Description

[0028] In some embodiments, an apparatus includes a bandgap reference circuit having a first bipolar junction transistor (BJT) that can receive a current from a node having a terminal voltage and can output a base emitter voltage. The terminal voltage of the first BJT substantially corresponds to or is lower than the base emitter voltage of the first BJT for at least a time period. In such embodiments, the apparatus also includes a second bipolar junction transistor (BJT) having a device width greater than a device width of the first BJT. The second BJT can receive a current from a node having a terminal voltage and output a base emitter voltage, where the terminal voltage of the second BJT substantially corresponds to or is lower than the base emitter voltage of the second BJT for at least a time period. In such embodiments, the apparatus also includes a reference generation circuit operatively coupled to the first BJT and the second BJT, where the reference generation circuit can generate a bandgap reference voltage based on the base emitter voltage of the first BJT and the base emitter voltage of the second BJT.

[0029] In some embodiments, an apparatus includes a base emitter voltage generation circuit having a bipolar junction transistor (BJT) configured to receive, in a voltage clamp configuration, a current from charge pump circuit and at a node having an input voltage and to output a base emitter voltage, where the input voltage substantially corresponds to or is lower than the base emitter voltage.

[0030] In some embodiments, an apparatus includes a clock circuit that is operatively coupled to a bandgap reference circuit, where the clock circuit has a first circuit portion that can receive from an on-chip clock a clock signal having an input voltage. The first circuit portion can produce (1) a first clock phase signal having a minimal voltage and a maximum voltage, and (2) a second clock phase signal non-overlapping with the first clock phase signal and having a minimal voltage and a maximum voltage. In such embodiments, the clock circuit also has a second circuit portion that is operatively coupled to the first circuit portion, where the second circuit portion includes a set of capacitors and a set of invertors that can collectively output a third clock phase signal and a fourth clock phase signal, the third clock phase signal and the fourth clock phase signal each having a minimal voltage greater than the minimum voltage of the first clock phase signal and the minimal voltage of the second clock phase signal. The third clock phase signal and the fourth clock phase signal each also has a maximum voltage greater than the maximum voltage of the first clock phase signal and the maximum voltage of the second clock phase signal. In such embodiments, the clock circuit also has a third circuit portion operatively coupled to the second circuit portion, where the third circuit portion includes a set of transistors that can output a fifth clock phase signal and a sixth clock phase signal. The fifth clock phase signal and the sixth clock phase signal each has a minimal voltage substantially equal to the minimum voltage of the first clock phase signal and the minimal voltage of the second clock phase signal. The fifth clock phase signal and the sixth clock phase signal each also has a maximum voltage substantially equal to the maximum voltage of the fourth clock phase signal and the maximum voltage of the fifth clock phase signal.

[0031] As used in this specification, the singular forms "a," "an" and "the" include plural referents unless the context clearly dictates otherwise. Thus, for example, the term "a transistor" is intended to mean a single transistor or a combination of transistors.

[0032] FIG. 1 is a block diagram of an integrated system used for feeding an input voltage to a bandgap reference circuit used in known portable electrical systems. The integrated system 100 is typically associated with larger electrical systems and, for example, can obtain energy from an external energy source 110 (e.g., a battery) using any number of energy harvesting mechanisms and in some instances, a boost converter 120. The boost converter 120 typically enhances or boosts the voltage obtained from the energy harvesting source 1 10 to a value above VBE- This can further be stabilized by the DC-DC regulator 130 before being sent to the bandgap reference circuit 140. Typical known bandgap reference circuits such as the bandgap reference circuit 140 are limited to using input voltages higher than VBE of a BJT because such known bandgap reference circuits inject a current into the BJT using a current source, current mirror, resistor, or switched capacitor network at a voltage higher than VBE- Achieving a lower operational output voltage from the bandgap reference circuit 140, however, is desirable for ultra-low-power (ULP) devices that include complex ICs, SoCs, body sensor nodes (BSNs) and wireless sensors for the internet of things. The output voltage from the bandgap reference circuit 140 determines the voltage at which the ULP device can turn on and operate because the reference voltage is used to turn on the power supplies of the ULP device. A lower bandgap reference voltage will reduce the turn-on voltage for the ULP device, reduce power loss, and increase the operational lifetime of the ULP device. Additionally, a lower bandgap reference voltage can also assist in the miniaturization of ULP devices.

[0033] FIG. 2 is a schematic diagram representing a bandgap reference circuit generating a constant voltage reference across varying temperatures, according to an embodiment. The bandgap reference circuit 200 includes a BJT base emitter voltage (VBE) generated by a complementary-to-absolute-temperature (CTAT) voltage generation circuit 205. The CTAT voltage generation circuit 205 includes a BJT (not shown in FIG. 2) connected to a power source (not shown in FIG. 2) in a diode configuration. The CTAT voltage corresponds to the VBE of the BJT transistor. The value of VBE decreases with increasing temperature because of the generation of increased number of carrier with increased temperature. Because the number of carriers increases with temperature, the conductivity of the transistor (i.e., BJT) increases, thus decreasing the value of VBE- In the example of FIG. 2, the VBE decreases with increasing temperature with a slope given by -2.2mV/°C. The voltage V _t is the output of the proportional-to-absolute-temperature (PTAT) voltage generation circuit 210. Unlike the CTAT voltage generation circuit 205, here the output voltage increases in magnitude with increasing temperature. In the example of FIG. 2, the voltage V _t increases with increasing temperature with a slope given by 0.085mV/°C. The voltage V _t is multiplied with a constant K at the multiplier 215 and added to the CTAT voltage (VBE) at the adder 220 to generate the bandgap reference voltage VREF (where VREF = VBE + KV _t ) that is temperature independent. The value of the constant K at the multiplier 215 is chosen such that the temperature dependence of the CTAT portion and PTAT portion of the bandgap reference circuit 200 cancel each other and VREF becomes a temperature independent voltage reference (typically in the range of less than 10ppm/ °C).

[0034] FIG. 3 is a schematic illustration of a bandgap reference circuit system that uses an input voltage less than the base-emitter voltage of a bipolar junction transistor. The bandgap reference circuit system 300 includes a bandgap reference circuit 305 operably coupled to a clock circuit 335. The bandgap reference circuit 305 includes a first charge pump circuit 310, a second charge pump circuit 320, a first base-emitter voltage clamp 315, a second base- emitter voltage clamp 325, and a reference generation circuit 330. It is to be noted that the BJT in the second base-emitter voltage clamp 325 has a device width greater than the device width of the BJT in the first base-emitter voltage clamp 3 15. The bandgap reference circuit system 300 can generate a temperature insensitive bandgap reference voltage (VREF) using an input (supply) voltage that is lower than the base-emitter voltage (VBE) of a BJT. In such instances, the first charge pump circuit 3 10 (e.g., a boost circuit such as a switched capacitor circuit) drives a current into the first base-emitter voltage clamp 3 15 (e.g., including a first bipolar junction transistor (BJT) connected in parallel to a first load capacitor) from a voltage that is lower than the VBE of the BJT in the first base-emitter voltage clamp 3 15. This causes the first base-emitter voltage clamp 3 15 to clamp its base-emitter voltage at VBEI- Similarly, the second charge pump circuit 320 drives a current into the second base-emitter voltage clamp 325 (e.g., also including a second BJT connected in parallel to a second load capacitor) from a voltage that is lower than the VBE of the BJT in the second base-emitter voltage clamp 325. This causes the second base-emitter voltage clamp 325 to clamp its base-emitter voltage at a different voltage VBE2- The reference generation circuit 330 can include, for example, a programmable switched capacitor circuit can generate a temperature insensitive bandgap reference voltage (VREF) from VBEI and AVBE (VBEI - VBE2), which can be any fractional multiple of the silicon bandgap voltage. In some configurations, the reference generation circuit 330 can include a capacitor that can store the voltage AVBE _. In such configurations, the reference generation circuit 330 can also include a summing circuit that can generate various constants for VBEI and AVBE, which are then added to generate the desired temperature insensitive bandgap reference voltage (VREF).

[0035] It is to be noted that the process of generating constants for VBEI and AVBE can be, for example, a time-gated process where clock phase signals with different time intervals (non- overlapping) are used to open and close various switches in charge pump circuits 3 10 and 320 and the reference generation circuit 330. Such clock phases are defined by discrete clock signals that are sent by the clock circuit 335 that is operably coupled to the bandgap reference circuit 305. The clock circuit 335 can provide clock signals of different frequencies from, for example, an on-chip oscillator, a crystal oscillator or any other clock source. Additionally, the clock circuit 335 also includes a clock doubler circuit that is used to double the swing of the output clock signal to enable switches that can pass at least a voltage level of VBE- The clock circuit 335 will be discussed in greater detail below in relation to FIGS. 1 1- 16. [0036] FIG. 4 is a schematic illustration of a bandgap reference circuit that uses switched capacitor charge pumps to drive an input voltage less than the base-emitter voltage of a bipolar junction transistor, according to an embodiment. The bandgap reference circuit 405 includes switched capacitor charge pumps 410 and 420 (that each include capacitors Cf), base-emitter voltage clamp 415 (that includes BJT transistor Ql and capacitor CL), base- emitter voltage clamp 425 (that includes BJT transistor Q2 and capacitor CL), and the reference generation circuit 430 that includes the summing circuit 432 and the capacitor Q, that stores the voltage AVBE- The switched capacitor charge pump 410 typically generates voltages from the source Vi„. The output of the switched capacitor charge pump 410 is connected to the BJT Ql, which in turn clamps its output voltage to VBEI- Similarly, the switched capacitor charge pump 420 also generates voltages from Vi„. The output of the switched capacitor charge pump 420 is connected to the BJT Q2, which in turn clamps its output voltage to VBE2- The use of the charge pumps 410 and 420 to drive current into the BJTs Q l and Q2 enables low voltage operation of the bandgap reference circuit 405. Additionally, the clock circuit (e.g., clock circuit 335 shown in FIG. 3), which is used to supply the clock signals for the two clock phases (pi and φ ₂ used in the operation of the switched capacitor charge pumps 410 and 420, can be made to operate at lower frequencies and input voltage (Vi„) to reduce power consumption. The lower Vi„ and the lower clock frequency for the switched capacitor charge pumps 410 and 420 enables lower power consumption when compared to known bandgap voltage reference generators. Each of the sub-components (e.g., the charge pumps 410 and 420 and the reference generator circuit 430) of the bandgap reference circuit 405 shown in FIG. 4 is described below.

[0037] For the bandgap reference circuit 405 shown in FIG. 4, in some instances, the first BJT Q 1 can receive a current from a node (marked as A) having a first terminal voltage and can output a first base-emitter voltage (VBEI), where the first terminal voltage (i.e., voltage at node A) substantially corresponds to or is lower than VBEI- In such instances, the second BJT Q2 can receive a current from a node (marked as B) having a second terminal voltage and can output a second base-emitter voltage (VBE ₂ ), where the second terminal voltage (i.e., voltage at node B) substantially corresponds to or is lower than VBE2- Note that the second BJT Q2 has a device width greater than the first BJT Ql (as seen by 1 representing Ql and M representing Q2 in FIG. 4, where M > 1). Additionally, in such instances, the bandgap reference circuit 405 also includes a reference generation circuit 430 that is operatively coupled to the first BJT Ql and the second BJT Q2, where the reference generation circuit 430 can generate a bandgap reference voltage (VREF) based on the base emitter voltage of the first BJT Ql (VBEI) and the base emitter voltage of the second BJT Q2 (VBE2).

[0038] In the configuration of the bandgap reference circuit 405 shown in FIG. 4, the first BJT Ql can receive the terminal voltage for the first BJT Ql (at node A) from a supply (e.g., Vin) without generation of an intermediate voltage that is higher than the base emitter voltage of the first BJT Ql (VBEI). Similarly, the second BJT Q2 can receive the terminal voltage for the second BJT Q2 (at node B) from a supply (e.g., Vin) without generation of an intermediate voltage that is higher than the base emitter voltage of the second BJT Q2 (VBE2). Note that the first BJT Q 1 receives the current for the first BJT Q 1 from the first charge pump circuit 410 via at least one capacitor Cf. Similarly, the second BJT Q2 receives the current for the second BJT Q2 from the second charge pump circuit 420 via at least one capacitor Cf.

[0039] Referring to FIGS. 3 and 4, the first charge pump circuit 410 is operatively coupled to the first BJT Ql and a clock circuit (e.g., clock circuit 335 in FIG. 3). The first charge pump circuit 410 can receive an input voltage (Vi„) and can output the terminal voltage of the first BJT Ql at node A, where Vi„ is less than the terminal voltage at node A. Similarly, the second charge pump circuit 420 is operatively coupled to the second BJT Q2 and a clock circuit (e.g., clock circuit 335 in FIG. 3). The second charge pump circuit 420 can receive an input voltage (Vi„) and can output the terminal voltage of the second BJT Q2 at node B, where Vi„ is less than the terminal voltage at node B. Note that the frequency of the clock signal send by the clock circuit 335 varies inversely with the terminal voltage for the first BJT Ql (i.e., voltage at node A).

[0040] The clock circuit 335 sends a clock signal having a first clock phase (pi and a second clock phase q¾. The first charge pump circuit 410 has a first configuration when receiving the first clock phase (pi signal and a second configuration when receiving the second clock phase φ ₂ signal (as discussed in greater detail in relation to FIGS. 5- 6 below). The first charge pump circuit 410 can output the terminal voltage of the first BJT Ql (i.e., voltage at node A) based on a charge stored at a first capacitor (Cf) during the first configuration and the second configuration of the first charge pump 410 (as discussed in greater detail in relation to FIGS. 5- 6 below). Similarly, the first charge pump circuit 420 has a first configuration when receiving the first clock phase (pi signal and a second configuration when receiving the second clock phase φ ₂ signal. The second charge pump circuit 420 can output the terminal voltage of the second BJT Q2 (i.e., voltage at node B) based on a charge stored at a first capacitor (Cf) during the first configuration and the second configuration of the first charge pump 420.

[0041] FIGS. 5A-C are schematic illustrations showing the charging of a switched capacitor charge pump circuit associated with the bandgap reference circuit shown in FIG. 4. The switched capacitor charge pump 410 (also known as charge pump circuit) shown in FIGS. 4 and 5A-C can boost the input voltage Vin by a factor of two (i.e., 2*Vin) and can also be used to output a voltage value of lower than Vi„. The unloaded charge pump circuit 410 shown in FIG. 5A uses non-overlapping clock phases (pi and φ ₂ , respectively. During operation in clock phase (pi as shown in FIG. 5B, node 1 is connected to Vin, and node 2 (shown in FIG. 5B) is connected to ground, charging the top plate of the capacitor Cf to Vi„ and the bottom plate of the capacitor Cf to ground. During operation in clock phase φ ₂ as shown in FIG. 5C, node 2 is connected to V; _n and node 1 to the output capacitor CL. Because the top plate of the capacitor Cf was charged to V; _n during clock phase (pi, charging the bottom plate of capacitor Cf to Vin in clock phase φ ₂ allows the voltage at node 1 to go to 2*Vi„ because the voltage across the capacitor Cf is Vin. The capacitor CL eventually charges to a voltage of 2*Vm after a given number of switching cycles at startup. Hence, the unloaded charge pump circuit 410 shown in FIG. 5A can generate a voltage that is twice the input voltage Vi„.

[0042] FIG. 6 is a schematic illustration of the charged switched capacitor charge pump circuit shown in FIG. 5A driving an input current into a base emitter voltage clamp. The output of the charged switched capacitor charge pump circuit 410 is connected to the BJT Ql of the base emitter voltage clamp 415. Note the similar charged switched capacitor charge pump circuit 420 can be used to drive the base emitter voltage clamp 425 that includes the BJT Q2 (that in the example of FIG. 4 is M times bigger than Ql). In the absence of the BJT transistor Ql, the output of the base emitter voltage clamp 415 would go to 2*Vin. The presence of the BJT transistor Ql, however, restricts the output voltage of the base emitter voltage clamp 415 to VBEI- A significant advantage of the circuit shown in FIG. 6 is that the voltage Vin involved in generating VBEI is smaller than VBE (where VBE = VBEI for the case of transistor Ql and VBE = BE2 for the case of transistor Q2). The minimum voltage for the bandgap to be operational V _m i„ is given by the following equation:

Vmin > - ^V f (1)

[0043] Where N =2 is applicable for a voltage doubling switched capacitor charge pump as described in FIGS. 4-6. Eq. 1 shows that in some other configurations, if a voltage tripler or a higher order (i.e., N) switched capacitor charge pump is used, even lower values of Vin can be obtained.

[0044] FIGS. 7A-7B present simulation results of the variation of VBE and AVBE as a function of temperature that is generated from the bandgap voltage reference circuit of FIG. 4. FIG. 7A shows the temperature dependence of VBEI and VBE2 where a CTAT behavior of both VBEI and VBE2 with respect to temperature is observed. Conversely, FIG. 7B shows the temperature dependence of AVBE where a PTAT behavior of AVBE with respect to temperature is observed. The voltages of VBEI, VBE2 and AVBE have been simulated using a Vin of 0.4V. The weights of the voltages VBEI and AVBE are added to generate the bandgap reference voltage. In some instances, the bandgap reference circuit shown in FIG. 4 can generate a bandgap reference voltage (VREF) given by the following equation:

VREF = a(V _BE1 + bAV _BE ) (2)

[0045] Where the constants a and b are involved in generating the weights for VBE and AVBE to generate VREF- Note that in other instances, a different summing circuit (e.g., summing circuit 432 shown in FIG. 4) using different values of VBEI, VBE2 and AVBE can generate a different value for VREF- The constants a and b in Eq. 2 above are defined or established by employing switched capacitor circuit techniques as opposed to the use of resistors that are typically used in known methods. In such known methods, the use of resistors increases the area of the circuit for low power or ULP devices. The power consumption of the bandgap reference circuit typically depends on the value of the resistor with typically larger resistors leading to lower power consumption. For example, the size of resistors typically involved in the design of a 200nW bandgap reference circuit is approximately 14ΜΩ. Resistors in the ΜΩ-sized range typically occupy a large physical area, a feature undesirable for low power or ULP devices. Additionally, for low power applications, large resistors are used in known bandgap reference circuits and such large resistors also increase the thermal and flicker noise for the bandgap reference circuit. The use of switched capacitor circuits, however, can define or establish such constants (e.g., a and b as shown in Eq. 2) with a significantly lower area.

[0046] The different voltage parameters described above (e.g., VBEI, BE2 and AVBE) can be scalable, particularly for dynamic voltage scaling (DVS) applications. The bandgap reference voltage VREF discussed in Eq. 2 is also scalable where a and b are the constants used to produce a scalable bandgap reference voltage. In Eq. 2, one of the constants can be a natural number while the other constant a rational number. Note that the circuits used for physically scaling the different voltages VBEI, VBE ₂ and AVBE are included within the summing circuit (e.g., summing circuit 432 shown in FIG. 4) of the reference generation circuit.

[0047] FIGS. 8A-C are schematic illustrations of different scaling circuits to scale AVBE, according to different embodiments. As seen in FIG. 8A, the capacitor Q, is connected between the nodes with the voltage of VBEI and VBE ₂ , respectively, that are generated from the switched capacitor charge pump based bandgap reference circuit as shown in FIG. 4 (i.e., voltage across capacitor Cb is AVBE). For generating different bandgap reference voltages (VREF), AVBE has to be multiplied (or scaled) by different constants. The scaling circuits 800 presented in FIGS. 8A-C present ways to generate three alternate constants for AVBE, namely one (FIG. 8A), two (FIG. 8B) and three (FIG. 8C). FIG. 8A shows the circuit for generating 1 *AVBE, which is simply the portion of the charge-pump-based bandgap reference circuit shown in FIG. 4 with no additional signal modifications performed by the reference generation circuit. FIG. 8B shows the scaling circuit 800 for generating 2*AVBE that uses the two non-overlapping clock phases (pi and φ ₂ . In phase φ ₂ , the voltages VBEI and VBE2 are connected across the capacitors CM and Q, ₂ . In phase (pi, the connection of the capacitors are re-arranged and the top plate of CM is connected to the bottom plate of C _b2 are shown in FIG. 8B. So the voltage appearing on the top plate of C _b2 is 2*AVBE- This is a depiction of the voltage doubling scheme. Similarly, FIG. 8C shows the scaling circuit 850 for generating 3*AVBE that also uses the two non-overlapping clock phases (pi and φ ₂ . The functioning of the voltage tripling circuit 850 in FIG. 8C is similar to the voltage doubling circuit 800 shown in FIG. 8B. Note that varying the scaling circuit can allow scaling or multiplication of AVBE by any integer value.

[0048] In some instances, the generation of multiple bandgap reference voltages can be involved for SoC applications to generate multiple VDDS values. In such instances, a AVBE voltage can be selected based on the transistor Q2 as shown in FIG. 4. Subsequently, multiple scaled values of AVBE can be generated as described above. This can complete half of the scaling involved in generating the appropriate VREF values according the Eq. 2. Subsequently, different fractional constant multipliers of VBE also can be generated to obtain the appropriate bandgap reference voltages (VREF) for the SoC applications.

[0049] FIGS. 9A-C are schematic illustrations of different configurations of a scaling circuit to scale VBE, according to an embodiment. Note that the scaling circuit 900 shown in FIGS. 9A-C will scale or multiply VBE with a fractional number (and not an integer). The scaling circuit 900 for VBE also includes switched capacitor circuits with non-overlapping clock phases (pi and φ ₂ . FIG. 9A shows the unloaded scaling circuit 900 for scaling VBE before clock phase signals have been applied. During operation in clock phase φ ₂ as shown in FIG. 9B, the capacitor C2 is connected to VBE, while the capacitor Ci is connected to ground. Therefore the charge stored on capacitor C2 is given by:

Q ₂ = ^BE ^2 (3)

[0050] In contrast, the charge stored on the capacitor Ci is zero. During operation in clock phase (pi as shown in FIG. 9C, the capacitors CI and C2 are connected together and so the total charge on the capacitors remains the same. Therefore:

Q ₂ = Q v, x (4)

[0051] So,

[0052] Therefore V _X is given by:

V _y = V B, E (6) [0053] Hence, by selecting the appropriate values of the capacitors Ci and C ₂ , a value of Vx is obtained that is a fraction of VBE as given by Eq. 6. The discussion presented herein in relation to FIGS. 8A-C and FIGS. 9A-C relate to scaling the voltages V _BE and AV _BE , respectively. Next, adding the scaled voltages VBE and AVBE in the reference generation circuit to achieve the desired bandgap reference voltage value VREF is discussed.

[0054] FIGS. lOA-C are schematic illustrations of a reference generation circuit for generating the bandgap reference voltage, according to an embodiment. The reference generation circuit 1000 includes the circuits used for generating constants for VBE and AVBE as discussed in FIGS. 8A-C and FIGS. 9A-C and also uses the switched capacitor scheme to generate the desired bandgap reference voltage value VREF- FIG. 10A shows the reference generation circuit 1000 (or summing circuit) with the appropriate signals. During operation in clock phase φ ₂ , the switches connected with the (clock phase) signal φ ₂ are closed and the reference generation circuit 1000 is configured as shown in FIG. 10B. The capacitor C _a i is discharged to the ground while the top plate of the capacitors C _a 2, CM, 0,2, and Oe are connected to VBEI- The bottom plate of the capacitor Ca2 is connected to ground, while the bottom plate of CM, Οΰ, and 0,3 are connected to VBE2- SO, the voltage across C _a 2 is VBEI, while the voltage across CM, Cb2, and Oe is AVBE- During operation in clock phase (pi, the switches are reconfigured and the reference generation circuit 1000 is arranged as shown in FIG. IOC. First, the capacitors C _a i and C _a 2 are connected and charge shared to generate the VBE component of the bandgap reference voltage. The voltage at node 1 is given by:

Vi = VBEI ¾- (?)

^L al ^{+ L} a2

[0055] Additionally, during operation in clock phase (pi, the capacitors CM, 0,2, and Oe are rearranged to generate 3 *AVBE between nodes 1 and 2 that leads to the generation of the desired bandgap reference voltage VREF as shown by:

V _REF = V _BE1 -^ — + 3AV _BE (8)

[0056] Equation 8 shown above shows the generation of the proposed temperature independent bandgap reference voltage. It is to be noted that other values of VREF can be generated (or obtained) different values for the capacitors C _a i and C _a 2 and different scaling factors (or weights) for AVBE-

[0057] The bandgap reference circuit described in FIGS. 1- 10 uses switched capacitor circuits that use two non-overlapping phases of a clock signal having a first clock phase (pi and a second clock phase φ ₂ . The clock signal is generated by a clock circuit (e.g., clock circuit 335 shown in FIG. 3) for the proper functioning of the bandgap reference circuit. The temperature independent bandgap reference voltage (VREF) as described by Eq. 8 above is independent of clock frequency in the embodiments of the bandgap reference circuits presented in FIGS. 1 - 10. Hence, the power consumption of the clock circuit used to achieve VREF can be reduced or minimized by operating the clock circuit at a very low frequency. The frequency of the clock signal should, however, be high enough to maintain the bias voltage of BJT Q l (VBEI) and BJT Q2 (VBE2) against leakage. Additionally, the frequency of the clock signal sent by the clock circuit varies inversely with the terminal voltage for the first BJT (e.g., Q l in FIG. 4). Hence, a low frequency, low power clock circuit can be used to generate the desired temperature independent bandgap reference voltage (VREF).

[0058] The different switches used in the bandgap reference circuits can pass a voltage equivalent to at least VBE, which is a voltage higher than Vi _n . Therefore, the clock signals associated with clock phases φι and Φ2 can sweep from 0 to > VBE- If not, the voltage input at the gate terminal of a switch (e.g., an NMOS switch) is lower than the voltage value (or voltage level) that the switch has to pass, and the switch cannot pass the full voltage. Accordingly, because the switches in the bandgap reference circuit (e.g., switches in the summing circuit and the switched capacitor charge pumps) pass voltages up to VBE, the clock signals (that drives the gate terminals of such switches) have voltages substantially equal to or higher than VBE.

[0059] FIG. 1 1 shows the block diagram of a clock signal generation scheme for the bandpass reference voltage circuit, according to an embodiment. The clock circuit 1 105 is operably coupled to a bandgap voltage reference circuit 1 140. The clock circuit 1 105 includes an oscillator 1 120 to provide the initial clock signal. The oscillator 1 120 can be, for example, a current-controlled ring oscillator (e.g., that can produce clock signal of approximately 30kHz at 0.4V Vjn and consume approximately 2nW of power). In other configurations, the initial clock signal can be generated by, for example, an on-chip oscillator, a crystal oscillator (that is an electronic oscillator circuit that uses the mechanical resonance of a vibrating crystal of piezoelectric material to define an electrical signal with a very precise frequency), or any other appropriate clock source. The clock circuit 1 105 also includes a PTAT current source 1 1 10 and a clock doubler 1 130. The PTAT current source 1 1 10 can be the same source that supplies Vin for the bandgap voltage reference circuit 1 140. The clock doubler 1 130 is used to double the voltage sweep range of the output clock signal to enable switches in the bandgap voltage reference circuit 1 140 to pass at least a voltage level of VBE as discussed above. It is to be noted that the output clock signals from the clock doubler 1 130 occur in two non-overlapping clock phases φι and φ ₂ .

[0060] FIG. 12 is a schematic illustration of an oscillator shown in FIG. 1 1 that can be used to generate a clock signal for a bandgap reference circuit, according to an embodiment. In the example of FIG. 12, the oscillator is represented by a current-controlled ring oscillator circuit 1200. Referring to FIGS. 1 1- 12, the current-controlled ring oscillator 1200 uses the current from the PTAT source 1 1 10. This current increases with temperature but does not change with Vi„. Because the power consumption of the PTAT current source 1 1 10 increases with increasing Vin, the architecture of the current-controlled ring oscillator 1200 is such that the frequency of the of the clock signals decreases with increasing Vin to keep the power consumption of the clock circuit 1 105 low. This is because the delay of one inverter cell (TRO) in the current-controlled ring oscillator is given by:

T — ^C ° ^Vin (Q\

[0061] Therefore, the frequency of the ring oscillator is given by:

/„= ^ = ^ 00)

b ^l Ro ^L O ^v in

[0062] Eq. (10) gives the expression of the output frequency (fo) for the current controlled ring oscillator. The current Io used in Eq. 9 and 10 above comes from a PTAT current source (e.g., PTAT current source 1 1 10 in FIG. 1 1), which remains constant with Vi _n because of the high power supply rejection. Because the current I _p within the current-controlled ring oscillator remains constant with Io, Eq. (1 1) shows the output frequency of the current controlled ring oscillator (f ₀ ) decreases with increasing Vi„, which helps keep the power consumption of the bandgap voltage reference circuit low with increasing Vi„.

[0063] Note that the current-controlled clock source (implemented by using a ring oscillator and the PTAT current source) as described in FIGS. 1 1-12 is a satisfactory choice to cater to a widely varying Vi„ voltage to reduce or restrict power consumption. If, however, in some configurations, a clock source such as a crystal oscillator, a system clock, or a real time clock is already available on the device chip for other applications, then overall system power can be reduced by using such existing internal clock sources instead of generating a clock source for the bandgap voltage reference circuit as described above.

[0064] As described above, the clock circuit sends clock signals associated with clock phases (pi and q¾ that sweep from 0V to a voltage greater than VBE to pass a voltage equivalent to at least VBE (which is a voltage higher than Vin) through a set of switches in the bandgap reference circuit (e.g., switched capacitor charge pump circuits, reference generation circuit, etc.) to generate the desired bandgap reference voltage (VREF). This is because closing a switch to pass a voltage involves inherent voltage loss within the source-drain of the transistors of the switch. Hence, for passing a voltage of VBE through a switch, the clock signal has to sweep to a voltage value greater than VBE. Otherwise if the input voltage at the gate terminal of a switch (e.g., an NMOS switch) is lower than the voltage value (or voltage level) that the switch has to pass, the switch cannot pass the full voltage (VBE). AS a result, in some instances, the clock signal being generated from the oscillator (e.g., oscillator 1120 in FIG. 1 1) undergoes signal boosting or enhancement (e.g., via a clock doubler) before being sent to the bandgap reference circuit as discussed in greater detail below.

[0065] FIGS. 13A-B are schematic illustrations of an implementation of switches for the bandgap reference circuit shown in FIG. 4. FIG. 13A shows the switched capacitor charge pump circuit 410 in electrical connection with the base-emitter voltage clamp circuit 415 (that includes BJT Ql and the capacitor CL). FIG. 13B shows an implementation of one of the switch 417 associated with the clock phase signal q¾. The switch 417 is implemented using a transmission gate including transistors (metal-oxide field effect transistors (MOSFETs)) MNS and M _PS . In some embodiments, the voltage VBE2 is typically clamped by the BJT Ql around 0.7-0.8V. In some embodiments, a clock phase signal q¾ running on the magnitude V; _n cannot be used to close the switch 417. In such embodiments, the clock phase signal φ ₂ swings to a magnitude of at least 2*Vi„ to enable the transmission gate to pass the terminal voltage VD properly into VBE2 (because of inherent losses within source-drain of the transistors MNS and Mps within the transmission gate). Therefore, in such instances, a clock doubling circuit is implemented to convert a clock phase signal that swings from 0 to Vi„ into a clock phase signal that swings from 0 > VBE2 (e.g., 2*Vi„ in this example).

[0066] FIGS. 14A-C are schematic illustrations of the steps involved in implementing a clock doubling technique to generate clock signals at different phases that swings from 0 to 2Vi„, according to an embodiment. The steps involved in clock doubling as shown in FIGS. 14A-C are implemented in the clock doubler of the clock circuit (e.g., clock doubler 1130 shown in the FIG. 1 1). FIG. 14A shows a first circuit portion 1410 that can generate non-overlapping clock phase signals. In FIG. 14A, the first circuit portion 1410 receives from an on-chip clock a clock signal (e.g., CLK) having an input voltage. The first circuit portion 1410 produces a first clock phase signal (e.g., pi) having a minimal voltage (e.g., 0) and a maximum voltage (e.g., Vin). Similarly, the first circuit portion 1410 also produces a second clock phase signal (e.g., P2) that is non-overlapping with the first clock phase signal and having a minimal voltage (e.g., 0) and a maximum voltage (e.g., Vin). Said in another way, the first circuit portion generates two non-overlapping signals that swing from 0 to Vin. The signals pi and P2 can be seen as being non-overlapping because at any time (i.e., during any T) when the signal pi has an amplitude of zero, the signal p ₂ has an amplitude of V; _n .

[0067] The signals pi and p2 will be used to generate new signals that swing from V; _n to 2V; _n using the second circuit portion as shown in FIG. 14B. In FIG. 14B, a second circuit portion (represented in FIG. 14B as two sub-portions 1430 and 1435) is operatively coupled to the first circuit portion 1410, where the second circuit portion 1430 and 1435 includes a set of capacitors and a set of invertors that are collectively configured to output a third clock phase signal (e.g., signal represented at xi) and a fourth clock phase signal (e.g., signal represented at x ₂ ). The third clock phase signal (e.g., xi) and the fourth clock phase signal (e.g., x ₂ ) each has a minimal voltage (e.g., V; _n ) that is greater than the minimum voltage of the first clock phase signal (e.g., 0) and the minimal voltage of the second clock phase signal (e.g., 0). Additionally, the third clock phase signal (xi) and the fourth clock phase signal (x ₂ ) each has a maximum voltage (e.g., 2Vi„) that is greater than the maximum voltage of the first clock phase signal (Vm) and the maximum voltage of the second clock phase signal (Vi„). In FIG. 14B, the node xbi (shown in sub-portion 1430) and the node xb ₂ (shown in sub-portion 1435) are the output of inverters running on Vi„ and thus the voltage at nodes xbi and xb ₂ swing from 0 to Vin. Node xi (in sub-portion 1430) and node X2 (in sub-portion 1435) are connected through diode-connected NMOS transistors to a capacitor. The transistors used are low threshold voltage (L _V T) transistors, and hence in the absence of a load, nodes xi and X2 will charge to Vi„, because the L _V T transistors have high leakage. Furthermore, the bottom plate of the capacitors connected to node xi and X2 swing from 0 to Vi„. Therefore, the top plate of such capacitors will swing from Vin to 2Vi„ resulting in the signals represented at xi and at x ₂ respectively in the chart of FIG.14B.

[0068] The signals represented at xi and at X2 respectively in FIG.14B are transformed into signals that can swing from 0 to 2*Vi„ using the third circuit portion shown in FIG. 14C. In FIG. 14C, a third circuit portion (represented in FIG. 14C as two sub-portions 1450 and 1455) is operatively coupled to the second circuit portion (1430 and 1435 in FIG. 14B). The third circuit portion 1450 and 1455 includes a set of transistors that can output a fifth clock phase signal (e.g., represented as (pi) and a sixth clock phase signal (e.g., represented as 9 ₂ ). Furthermore, the fifth clock phase signal ((pi) and the sixth clock phase signal (9 ₂ ) each has a minimal voltage substantially equal to the minimum voltage of the first clock phase signal (0) and the minimal voltage of the second clock phase signal (0), and the fifth clock phase signal ((pi) and the sixth clock phase signal (9 ₂ ) each have a maximum voltage (2*Vi„) substantially equal to the maximum voltage of the third clock phase signal (xi) (2*Vi„) and the maximum voltage of the fourth clock phase signal (x ₂ ) (2*Vi„). In FIG. 14C, in the third circuit sub- portion 1450, when the voltage at pi is high, the voltage at x ₂ is also high, and thus the net voltage of the phase signal ((pi) is pulled down to ground. When the voltage at pi is zero, the voltage at X2 is low at Vi _n . At this time, the voltage at xi is at 2*Vi _n . At this time the PMOS transistor turns on and passes the xi voltage level to the clock phase signal (pi. As a result, the clock phase signal (pi swings from 0 to 2*Vi„. Similarly, the clock phase signal φ ₂ also swings from 0 to 2*Vi„ in a non-overlapping manner as shown in the chart in FIG. 14C.

[0069] FIGS. 15A-B present the results of simulations of an example of a clock doubler circuit that sends boosted clock phase signals to a bandgap voltage reference circuit. FIG. 15A shows that the signal p ₂ (similar to the phase signal p2 in FIG. 14A) swings from 0 to 400mV in time (i.e., swings from 0 to Vi„). FIG. 15A also shows that the signal xi (similar to the phase signal xi in FIG. 14B) swings from 350mV to 750mV in time (i.e., approximately swings from Vin to 2*Vi„). FIG. 15 B shows that the signal phi ₂ (similar to the phase signal ( 2 in FIG. 14C) swings from 0 to 750mV in time (i.e., approximately swings from 0 to 2*V _m ).

[0070] Referring to FIGS. 3, 4 and 14, in some configurations of a bandgap voltage reference circuit system, a first switched capacitor charge pump (e.g., switched capacitor charge pump 410 in FIG. 4) (or simply a first charge pump) is operatively coupled to the clock circuit (e.g., clock circuit 335 in FIG. 3) and a first BJT of the bandgap reference circuit (e.g. BJT Ql in FIG. 4). In such configurations, the first switched capacitor charge pump can receive the fifth clock phase signal (e.g., clock phase signal ( i in FIG. 14C) and the sixth clock phase signal (e.g., clock phase signal q¾ in FIG. 14C) and output a voltage driving the terminal of the first BJT (e.g. BJT Ql in FIG. 4). Similarly, in such configurations, a second switched capacitor charge pump (e.g., switched capacitor charge pump 420 in FIG. 4) (or simply a second charge pump) is operatively coupled to the clock circuit (e.g., clock circuit 335 in FIG. 3) and a second BJT of the bandgap reference circuit (e.g. BJT Q2 in FIG. 4). In such configurations, the second switched capacitor charge pump can receive the fifth clock phase signal (e.g., clock phase signal ( i in FIG. 14C) and the sixth clock phase signal (e.g., clock phase signal q¾ in FIG. 14C) and output a voltage driving the terminal of the first BJT (e.g. BJT Ql in FIG. 4).

[0071] Also referring to FIGS. 3, 4 and 14, the clock circuit (e.g., clock circuit 335 in FIG. 3) sends a clock signal with a specific frequency to the bandgap voltage reference circuit (e.g., bandgap voltage reference circuit 305 in FIG. 3). In such configurations, a first switched capacitor charge pump (e.g., switched capacitor charge pump 410 in FIG. 4) (or simply a first charge pump) is operatively coupled to the clock circuit (e.g., clock circuit 335 in FIG. 3) and a first BJT of the bandgap reference circuit (e.g. BJT Ql in FIG. 4). In such configurations, the first switched capacitor charge pump can output a voltage (i.e., voltage at node A in FIG. 4) driving the terminal of the first BJT based on the fifth clock phase signal (e.g., clock phase signal (pi in FIG. 14C) and the sixth clock phase signal (e.g., clock phase signal ½ in FIG. 14C), where the frequency of the fifth clock phase signal and the sixth clock phase signal varies inversely with the input voltage of the first BJT (i.e., voltage at node A in FIG. 4). Similarly, in such configurations, a second switched capacitor charge pump (e.g., switched capacitor charge pump 420 in FIG. 4) (or simply a second charge pump) is operatively coupled to the clock circuit (e.g., clock circuit 335 in FIG. 3) and a second BJT of the bandgap reference circuit (e.g. BJT Q2 in FIG. 4). In such configurations, the second switched capacitor charge pump can output a voltage (i.e., voltage at node B in FIG. 4) driving the terminal of the second BJT based on the fifth clock phase signal (e.g., clock phase signal ( i in FIG. 14C) and the sixth clock phase signal (e.g., clock phase signal φ ₂ in FIG. 14C), where the frequency of the fifth clock phase signal and the sixth clock phase signal varies inversely with the input voltage of the second BJT (i.e., voltage at node B in FIG. 4). .

[0072] FIG. 16 shows the annotated lay out of the complete bandgap reference circuit, according to an embodiment. The bandgap voltage reference circuit shown in FIG. 16 has an area of 0.0264mm ² and can be implemented, for example, in a commercial bulk 130nm complementary metal-oxide-semiconductor (CMOS) process or other types of suitable technologies. The capacitors are implemented using nMOS (or n-channel MOSFET) capacitors and metal-insulator-metal (MIM) capacitors. The load capacitors for the VBE generation circuit and the VBE fraction generation switched capacitor circuit (see circuits in FIG. 9) were implemented using nMOS capacitors, whereas the load capacitors for the bandgap output generation (see circuit in FIG. 10) and the AVBE doubling circuit (see circuit in FIG. 8) were implemented using MIM capacitors to avoid bottom plate capacitor parasitics. The total area of the bandgap voltage reference circuit as shown in FIG. 16 is significantly smaller than known low power bandgap reference circuits because the bandgap voltage reference circuit shown in FIG. 16 does not use large resistors. The bandgap voltage reference circuit shown in FIG. 16 also consumes 19.2nW of power at 0.4V V _m , which is an order of magnitude lower than the power used in known non-duty-cycled bandgap reference circuits.

[0073] Because the bandgap reference circuit is a switching capacitor circuit, the bandgap reference circuit has a settling time at startup. FIG. 17 is a graphical display of an example of the transient behavior of a bandgap reference circuit at start-up. FIG. 17 shows the bandgap reference circuit takes 15 msec to settle at a 0.8V Vm. At 0.4V, the settling time is 90 msec. The settling time is directly dependent on the clock frequency and the power supply V _m . In some configurations, the settling time for the bandgap reference circuit can be large. In such configurations, a fast start-up mode for the bandgap reference circuit can be implemented. In such configurations, during the fast start-up mode, the clock frequency can be made several times faster than during a normal operational mode, which can reduce the settling time of the bandgap reference circuit. This can be done during power on the fast start-up mode, where the current source of the clock source (e.g., clock circuit 335 in FIG. 3) is increased several times which then increases the clock frequency. A settling time of 20 μ8 during startup of the bandgap reference circuit can be used in the fast start-up mode.

[0074] An embodiment of the bandgap reference circuit was verified for proper functionality in the temperature range of -20°C to 100°C. While this range is quite large for the intended ULP applications, the performance of the bandgap reference circuit in this range is relevant as it compares with known state-of-the-art bandgap reference circuits. FIG. 18 shows the simulated variation of an embodiment of a bandgap reference circuit output for a temperature range of -20°C to 100°C. The bandgap reference circuit can provide an output voltage of 500mV and the output voltage varies by 3mV over a temperature variation of 120°C, thus achieving a performance of 50 ppm/ °C. The performance of such a bandgap reference circuit with temperature as shown in FIG. 20 is in line with known technologies and an improved performance can be achieved at a higher output voltage (i.e., output voltage > 500mV).

[0075] FIG. 19 presents the results of a Monte-Carlo simulation that shows an example of the change in bandgap reference output with respect to process and mismatch variation. FIG. 19 shows the untrimmed output of the bandgap reference circuit, where the output achieves a mean (μ) of 508 mV and a standard deviation (σ) of 5mV. The untrimmed output of the bandgap reference circuit also shows a 3σ variation of < 3%. The variation in the output (voltage) shown in FIG. 19 can be reduced by trimming the bandgap output using the capacitors used in the switched capacitor circuits (see FIGS. 8-10) to generate the appropriate constants for the bandgap reference output.

[0076] FIG. 20 presents the results of a simulation that shows an example of the change in bandgap reference voltage with respect to variation with input voltage (V _m ). FIG. 20 shows the variation of input voltage (V _m ) from two separate sources, namely an external clock and an on-chip clock. FIG. 20 shows that the bandgap reference voltage varies by approximately 4% when an external constant clock source is used to deliver Vm, and the bandgap reference voltage varies by approximately 2% when an on-chip cock is used to deliver Vi„. Thus the use of an on-chip clock as discussed in the specifications thus far reduces the bandgap reference circuit output variance by approximately 50%.

[0077] The bandgap reference circuit discussed herein operates from a minimum input voltage of 0.4V, thus improving over two-fold from the known bandgap reference circuits. The power consumption of the proposed bandgap reference circuit is 19.2nW, which is over nine- fold lower than achieved without duty cycling in known bandgap reference circuits. Known bandgap reference circuits typically achieve a low power of 170nW by sampling the reference voltage on a capacitor by periodically turning it on and off. Duty cycling can be applied to one or more bandgap reference circuit embodiments described herein as well to further lower power. The power supply variation can be higher in the one or more bandgap reference circuit embodiments described herein because the architecture does not use external current sources, which are typically used in known architectures. The lower area of the bandgap reference circuit (0.0264mm ² ) is also achieved because large resistors are not used.

[0078] Note that the BJT's used in the bandgap reference circuit discussed above has been shown to be a PNP BJT as an example only, and not a limitation. In other configurations, the BJT's used in the bandgap reference circuit can be an NPN BJT(s). In such configurations (i.e., during use of an NPN BJT(s)), the bandgap reference circuit can generate a temperature insensitive bandgap reference voltage (VREF) using an input (supply) voltage that is lower than the base-emitter voltage (VBE) of the NPN BJT. Note the term base-emitter voltage (VBE) is intended to cover both the base-emitter voltage for an NPN BJT and the emitter-base voltage for a PNP BJT. The bandgap reference circuits described thus far can be implemented using both PNP BJT's as well as NPN BJT's. Furthermore, the bandgap reference circuits using PNP BJT's can be fabricated using a CMOS process, and the bandgap reference circuits using NPN BJT's can be fabricated using biCMOS or other processes.

[0079] While various embodiments have been described above, it should be understood that they have been presented by way of example only, and not limitation. Where methods described above indicate certain events occurring in certain order, the ordering of certain events may be modified. Additionally, certain of the events may be performed concurrently in a parallel process when possible, as well as performed sequentially as described above. Likewise, the various diagrams may depict an example architectural or other configuration for the invention, which is done to aid in understanding the features and functionality that can be included in the invention. The invention is not restricted to the illustrated example architectures or configurations, but can be implemented using a variety of alternative architectures and configurations. Additionally, although the invention is described above in terms of various exemplary embodiments and implementations, it should be understood that the various features and functionality described in one or more of the individual embodiments are not limited in their applicability to the particular embodiment with which they are described, but instead can be applied, alone or in some combination, to one or more of the other embodiments of the invention, whether or not such embodiments are described and whether or not such features are presented as being a part of a described embodiment. Thus the breadth and scope of the present invention should not be limited by any of the above- described exemplary embodiments.