Low power, charge injection compensated charge pump

[0002] This application is a continuation of U.S. patent application Ser. No. 09/833,296, filed on Apr. 11, 2001, now U.S. Pat. No. 6,483,358, which claims benefit of priority to U.S. Provisional Application Ser. No. 60/266,077, filed Feb. 2, 2001.

[0003] This invention is related to the field of charge pumps, and particularly their use in phase locked loops (PLLs).

[0004] Charge pumps are employed in a wide variety of circuits, such as voltage doublers, clock doublers, phase-locked loops (PLLs), delay locked loops, memory systems, etc. Generally, a charge pump is configured to provide or remove charge from an output node of the charge pump in response to input up and down signals (respectively).

[0005] Charge pumps may experience error during operation from a variety of sources. Charge injection error may occur during switching of the transistors internal to the charge pump in response to the up and down signals. Clock feedthrough error may also be experienced across the parasitic gate to source and drain capacitances. Furthermore, transistor imperfections such as body effect, channel length modulation, drain induced barrier loading (DIBL), etc. may cause errors in the charge pump's operation.

[0006] These errors affect the amount of charge provided on or removed from the output node of the charge pump in response to a particular pulse width of the up and down signals, respectively. At reduced operating voltages (which are become more and more common as integrated circuit fabrication technologies advance), these errors are magnified.

[0007] A charge pump includes at least one switching transistor for switching current on or off in response to an up or down signal, a pair of transistors (one coupled to the source and the other to the drain of the switching transistor) each having its source and drain shorted and coupled to receive a complement of the signal on the gate terminal of the switching transistor on their gate terminals, and a fourth transistor coupled to the drain of the switching transistor and a power supply. The pair of transistors may reduce charge injection error by the switching transistor. They are activated concurrent with the deactivation of the switching transistor, and thus their channels are acquiring charge at a time when the charge injection may occur from the switching transistor. The acquired charge may include the injected charge. Additionally, since the gate terminals of the pair of transistors are transitioned in the opposite direction as the gate terminal of the switching transistor, clock feedthrough from the pair of transistors may reduce clock feedthrough error from the switching transistor. In one implementation, the pair of transistors may be sized approximately ½ the size of the switching transistor. Thus, parasitic capacitances between the gate terminal of the pair and the source and drain may be approximately ½ of the parasitic capacitances of the switching transistor. However, since the source and drain of each of the pair of transistors is shorted, the parasitic capacitances in parallel add to produce approximately the same parasitic capacitance on the source or drain as the switching transistor has. The fourth transistor may provide for active shutoff of a current transistor being switched by the switching transistor, by actively charging the source of the current transistor to a voltage which is not exceeded by the gate terminal of the current transistor.

[0008] In one embodiment, the charge pump may employ dummy loads to more accurately balance the load on the up and down signals received by the charge pump, in order to more evenly balance the up and down currents on the output node of the charge pump.

[0009] Broadly speaking, a circuit is contemplated. The circuit comprises a first, second, third, and fourth transistor. The first transistor, second transistor, and third transistor are coupled in series. The first transistor includes a first terminal, a second terminal, and a first gate terminal. The second transistor includes a third terminal, a fourth terminal, and a second gate terminal. The third transistor includes a fifth terminal, a sixth terminal, and a third gate terminal. The first terminal, the second terminal, and the third terminal are to connected to a first node. The fourth terminal, the fifth terminal, and the sixth terminal are connected to a second node. The second gate terminal is coupled to receive a first signal, and the first gate terminal and the third gate terminal are coupled to receive a complement of the first signal. The fourth transistor has a seventh terminal, an eighth terminal, and a fourth gate terminal. The seventh terminal is coupled to the first node, the eighth terminal is coupled to a power supply, and the fourth gate terminal is coupled to receive the first signal.

[0010] Additionally, a method is contemplated. A first transistor having a first terminal and a second terminal is activated. The first transistor is deactivated. Concurrent with activating the first transistor, a second transistor having a third terminal and a fourth terminal each connected to the first terminal and a third transistor having a fifth terminal and a sixth terminal each connected to the second terminal are deactivated. Concurrent with deactivating the first transistor, the second transistor and the third transistor are activated. Concurrent with deactivating the first transistor, a fourth transistor having a seventh terminal coupled to the first terminal and an eighth terminal coupled to a power supply is activated.

[0011] The following detailed description makes reference to the accompanying drawings, which are now briefly described.

[0012] FIG. 1 is a block diagram of one embodiment of a phase locked loop (PLL).

[0013] FIG. 2 is a circuit diagram of one embodiment of a loop filter shown in FIG. 1.

[0014] FIG. 3 is a circuit diagram of one embodiment of a charge pump shown in FIG. 1.

[0015] FIG. 4 is a timing diagram illustrating operation of portions of one embodiment of the charge pump shown in FIG. 2.

[0016] FIG. 5 is a circuit diagram of one embodiment of a voltage controlled oscillator shown in FIG. 1.

[0017] FIG. 6 is a circuit diagram of one embodiment of a voltage regulator shown in FIG. 1.

[0018] FIG. 7 is a circuit diagram of one embodiment of a power-up control circuit shown in FIG. 6.

[0019] FIG. 8 is a circuit diagram of a gain boosted operation amplifier circuit.

[0020] FIG. 9 is a block diagram of a carrier medium.

[0021] While the invention is susceptible to various modifications and alternative forms, specific embodiments thereof are shown by way of example in the drawings and will herein be described in detail. It should be understood, however, that the drawings and detailed description thereto are not intended to limit the invention to the particular form disclosed, but on the contrary, the intention is to cover all modifications, equivalents and alternatives falling within the spirit and scope of the present invention as defined by the appended claims.

[0022] Turning now to FIG. 1, a block diagram of one embodiment of a phase locked loop to (PLL) 10 is shown. Other embodiments are possible and contemplated. In the embodiment of FIG. 1, the PLL 10 includes a voltage regulator 12, a phase/frequency detector (PFD) 14, a level shifter 16, a charge pump 18, a voltage controlled oscillator (VCO) 20, a pair of divide by two circuits 22 and 24, a divide by N circuit 26, a loop filter 28, a current generator 30, a configuration register 32, and a pair of output buffers 34 and 36. In the embodiment of FIG. 1, the voltage regulator 12 is coupled to a first power supply which supplies a first voltage (V₁). The voltage regulator 12 provides regulated output voltages V_reg1and V_reg2. The level shifter 16 is powered by V_reg2, while the charge pump 18 and the VCO 20 are powered by V_reg1. The output buffers 34 and 26, the divide by 2 circuits 22 and 24, the divide by N circuit 26, and the PFD 14 are powered by a V_ddsupply voltage. The PFD 14 is coupled to receive the output of the divide by N circuit 26 and to receive a reference clock CLK_ref. The PFD 14 provides Up and Down signals to the level shifter 16, which level shifts the Up and Down signals from the V_ddvoltage domain to the V_regvoltage domain corresponding to V_reg1and V_reg2. The charge pump 18 is coupled to receive the level-shifted Up and Down signals from the level shifter 16 as well as a reference current I_reffrom the current generator 30. The charge pump 18 provides an output voltage V_Ctrlto the VCO 20, and the loop filter 28 is coupled to the node between the charge pump 18 and the VCO 20 as well. Both the loop filter 28 and the charge pump 18 are programmable via values in the configuration register 32 to which the loop filter 28 and the current generator 30 are coupled. Particularly, the charge pump 18 is programmable by changing the reference current I_ref, as described in more detail below. The VCO 20 provides an output signal to the series connection of the divide by two circuits 22 and 24 (the output of each being buffered by output buffers 34 and 36, respectively). The output of the divide by 2 circuit 24 is fed back to the divide by N circuit 26.

[0023] Generally, the V_ddvoltage may be the voltage supplied to the circuitry within an integrated circuit also employing the PLL 10. Accordingly, the V_ddvoltage may be subject to a large amount of noise (variations in the voltage due to, e.g., digital switching in the integrated circuit, etc.). The V₁voltage may also be subject to noise. In one implementation, the V₁voltage may be the voltage supplied to input/output circuits of the integrated circuit. The V_ddand V₁voltages may be any suitable voltages, and may supply any other suitable circuitry, as desired. In one exemplary implementation, the V_ddvoltage may be nominally 1.2 volts and the V₁voltage may be nominally 3.3 volts. However, the voltages may be varied in other implementations. The V_ddvoltage may depend, at least in part, on the process technology used to form the transistors within PLL 10. The V₁voltage may depend on the process technology as well as external interface demands.

[0024] Since the V_ddand V₁voltages are noisy, they may be unsuitable for powering certain blocks of the PLL 10. Specifically, the charge pump 18 and the VCO 20 may be sensitive to power supply noise, and may function less accurately in the presence of power supply noise. Accordingly, these blocks are supplied with a regulated voltage from the voltage regulator 12 (specifically, V_reg1). In one embodiment described below, the voltage regulator 12 provides regulated voltages with a power supply rejection ratio (PSRR) greater than 40 decibels (dB), while prior voltage regulators in such contexts may have been limited to a PSRR of about 25 dB. The PSRR achieved by various embodiments of the voltage regulator 12 may vary depending upon which features of the voltage regulator as illustrated in FIG. 6 below are implemented, tradeoffs in the voltage regulator design even if the features are implemented, etc. The voltage regulator 12 may provide two regulated voltages (V_reg2and V_reg1, which may be equal in the illustrated embodiment). The second regulated voltage (V_reg2) supplies the level shifter 16.

[0025] The level shifter 16 may serve as a noise reduction device in addition to shifting the Up and Down signals from V_ddvoltage swings to V_regvoltage swings. The Up and Down signals provided by the PFD 14 may be noisy, as they are generated in the noisy V_ddvoltage domain. The level shifter 16 passes these signals through various active devices (e.g. transistors) and thus the noise level is attenuated in the active devices. Accordingly, the Up and Down signals presented to the charge pump 18 may be less noisy and thus may enhance the accuracy of the charge pump 18 and the VCO 20.

[0026] Both the loop filter 28 and the charge pump 18 may be programmable through configuration register 32. The charge pump 18 is programmed by changing the reference current I_refsupplied to the charge pump 18 (e.g. by the current generator 30 in the illustrated embodiment). In one embodiment, the current generator 30 may be supplied by the V_reg1voltage as well (or the V_reg2voltage, if desired). The loop filter 28 may be, programmed by changing the resistor and capacitor values therein. An exemplary loop filter is illustrated in FIG. 2 below.

[0027] By adjusting the loop filter 28 and the charge pump 18, the damping, loop bandwidth and lock time of the PLL 10 may be adjusted for a given operating range. Thus, the PLL 10 may be optimized for a selected operating range in a given device, and thus the flexibility of the PLL 10 over a large operating range may be enhanced. As an example, increasing the reference current I_refmay increase the rate at which the charge pump 18 may vary the V_Ctrlvoltage, and thus may decrease the lock time of the PLL 10 while decreasing the damping of the loop. Varying the loop filter parameters may change which low and high frequencies of the V_Ctrlvoltage are filtered out. These modifications may affect the lock time, damping, loop bandwidth, and jitter as well.

[0028] It is noted that, while the configuration register 32 is illustrated in proximity to the other elements of the PLL 10, the configuration register 32 may be located elsewhere. For example, in one embodiment, the PLL 10 may be included in a system on a chip design which integrates the PLL 10 with one or more processors, memory, bus bridges, etc. The configuration register 32 may be located in one of the processors, and the values stored therein may be transmitted to the PLL 10. Furthermore, other methods for programming the charge pump 18 and the loop filter 28 may be used. For example, pins on the integrated circuit including the PLL 10 may be set to appropriate voltages to provide the selected values. A combination of pins and configuration register values may to also be used.

[0029] The PLL 10 may be a low power design. For example, by operating the VCO 20 with a V_reg1supply voltage instead of the higher V₁voltage, the voltage swings of the oscillator within the VCO 20 may be limited to a maximum of approximately V_reg1, thus reducing the power consumed in the VCO 20 as compared to a VCO supplied by the V₁voltage. Additionally, in one implementation, current mirrors within the charge pump 18 and the VCO 20 may be implemented without cascoding (which may increases the device count and thus may lead to increased the power dissipation). Additionally, for VCO 20, cascoding the current mirrors reduces the maximum voltage that may be applied to the ring oscillator and thus would reduce the maximum output frequency of the VCO 20. Cascoded current mirrors are typically used to improve the operation of the current mirrors as current sources by attenuating supply noise. The non-cascoded current mirrors may provide acceptable current source operation since the voltage regulator 12 and the level shifter 16 operate to minimize the noise experienced by the charge pump 18 and the VCO 20. Still further, non-cascoded structures may more readily allow for transistors to remain in saturation, which may improve the performance of the transistors. Additionally, self-biasing may be used to ensure that voltages referenced to the V_ddand/or V₁voltages track changes in those voltages properly.

[0030] The voltage regulator 12 provides the PLL 10 with supply decoupling from the V₁power supply. Additionally, in one implementation, the PLL 10 may employ relatively large decoupling capacitance on critical nodes. For example, large decoupling capacitances may be provided on the V_reg2and V_reg1nodes. These capacitances may cause a corresponding variation on the V_reg2and V_reg1nodes in response to ground variations (“ground bounce”), thus preserving the V_reg2/V_reg1to ground potential. Furthermore, these decoupling capacitances may substantially reduce the effect of supply variations (“supply bounce”) on the corresponding nodes.

[0031] The PFD 14 generally compares the frequency and phase of the clock output by the divide by N circuit 26 to the reference clock CLK_ref. If the phase of the clock output of the divide by N circuit 26 lags the phase of the reference clock, the PFD 14 activates the Up signal during the lagging period. If the phase of the clock output of the divide by N circuit 26 leads the phase of the reference clock, the PFD 14 activates the Down signal during the lagging period. Any suitable design may be used for the PFD 14, although in one embodiment a dual D flip flop design is used.

[0032] As mentioned above, the level shifter level shifts the Up and Down outputs of the PFD 14 from the V_ddvoltage to the V_regvoltage. These level-shifted Up and Down signals (Up_Vregand Down_Vreg, respectively) are applied to the charge pump 18. In response to an active Up signal, the charge pump 18 increases the charge on the output node, thus increasing V_Ctrl. In response to an active Down signal, the charge pump 18 decreases the charge on the output node, thus decreasing V_Ctrl. Increasing V_Ctrlcauses the VCO 20 to increase its output frequency, while decreasing V_Ctrlcauses the VCO 20 to decrease its output frequency.

[0033] It is noted that the Up and Down signals may be activated, or considered to be active, at either a high voltage or a low voltage. In the embodiment of the charge pump 18 illustrated in FIG. 3 below, the Up and Down signals are activated at a high (V_reg1) voltage (and deactivated at a low, or Gnd, voltage). Additionally, in some embodiments such as the one illustrated in FIG. 3 below, the charge pump 18 may use both the true and complement of the Up and Down signals. The complement may be generated by the charge pump 18, the level shifter 16, or the PFD 14, as desired. In this context, a first signal is a “complement” of a second signal if the first signal carries a voltage representing an opposite binary state to the voltage carried by the second signal.

[0034] The VCO 20 includes an output circuit which provides a full V_ddswing on the clock output from the VCO 20. This output clock is then divided by 2 (in frequency) by the divide by 2 circuit 22 (thus producing PLL output clock Out1). The second divide by 2 circuit 24 divides the frequency by two again (thus producing PLL output clock Out2, which is also fed back to the divide by N circuit 26). Accordingly, Out1 is 2N times the frequency of CLK_refand Out1 is N times the frequency of CLK_ref(once the PLL 10 has locked to CLK_ref). Other embodiments may produce only one output clock, or more output clocks, and each output clock may be any multiple of the frequency of the clock reference CLK_refusing appropriate divide circuits in place of circuits 22, 24, and 26. As used herein, the term “signal swing” refers to a range of voltages which a signal may take on.

[0035] It is noted that, while specific embodiments of the loop filter 28, the charge pump 18, the VCO 20, and the voltage regulator 12 are illustrated in FIGS. 2, 3, 5, and 6 below (respectively), other embodiments may use any circuit for each of these elements of PLL 10, in other embodiments. Furthermore, any combination of the embodiments shown in FIGS. 2, 3, 5 and 6 along with conventional embodiments of other elements may be used.

[0036] As used herein, the term “power supply” refers to a node which is supplied, during use, with a relatively stable voltage. Thus, in many contexts, the ground node is a power supply node according to this definition, as well as nodes labeled, e.g., V₁, V_reg1, V_reg2, V_dd, etc. Thus, coupling to a power supply refers to coupling to the node supplied with the corresponding voltage, during use. Furthermore, references to coupling to a voltage (including ground, or Gnd) or to a voltage supply herein should be interpreted as coupling to the corresponding power supply.

[0037] As used herein, a “current mirror” is a circuit having an input for receiving a current and at least one output. The current mirror provides a current on the output which is proportional to the input current. The output current may approximately equal the input current if the transistors forming the current mirror are matched.

[0038] Turning next to FIG. 2, a block diagram of one embodiment of the loop filter 28 is shown. Other embodiments are possible and contemplated. In the embodiment of FIG. 2, the loop filter 28 includes a programmable resistor R₁in series with a programmable capacitor C₁, the series connection coupled between the output node of the charge pump 18 (the V_Ctrlvoltage node) and ground, and a programmable capacitor C₂in parallel with the series connection of R₁and C₁. In the illustrated embodiment, the loop filter 28 is programmable in response to configuration register values from the configuration register 32.

[0039] In the embodiment of FIG. 2, each of the filter components are programmable, providing a high degree of flexibility. Other embodiments may employ fewer programmable components (e.g. only R₁and C₂may be programmable), as desired.

[0040] The programmable components may be realized in any suitable fashion. For example, R₁may comprise two or more resistor elements in any series and/or parallel connection. Each resistor element may be shorted (series connection) or disconnected (parallel connection) via a switch which may be controlled by the configuration register value from the configuration register 32. Similarly, each capacitor may comprise one or more capacitor elements in any series and/or parallel connection, and each capacitor element may be shorted (series connection) or disconnected (parallel connection) via a switch which may be controlled by the configuration register value from the configuration register 32.

[0041] In one implementation, the register R₁may be programmable in the range of 0 to 33.2 kilo-ohms (kohms), in steps of 8.3 kohms (under worst case process conditions, or 25.0 kohms in steps of 6.25 kohms nominally); the capacitor C₁may be implemented as a parallel combination of 4 capacitors (each included or excluded programmably) having capacitances of approximately 20 picoFarads, 20 picoFarads, 10 picoFarads, and 5 picoFarads, respectively; and the capacitor C₂may be implemented as a parallel combination of a capacitor of approximately 1.25 picoFarads and a capacitor programmably included or excluded and of approximately 0.5 picoFarads. In one specific implementation, the resistor may be implemented of P+ polysilicon material using a unit cell approach, where each unit cell is approximately 1.184 kohms. The array of unit cells may be surrounded by dummy cells. The dimensions of the polysilicon may be material and/or process dependent, as is known to those of skill in the art. Alternatively, transistors may be used to construct the resistors. The capacitors may, in one implementation, be constructed from transistors with the gate terminal coupled as one capacitor input and the source and drain terminals and the bulk terminal shorted to the other capacitor terminal (e.g. ground).

[0042] Turning next to FIG. 3, a circuit diagram of one embodiment of the charge pump 18 is shown. Other embodiments are possible and contemplated. In FIG. 3 and other circuit diagrams herein, PMOS transistors are illustrated with an arrow pointing into the transistor on the source terminal (e.g. transistor M_d1is a PMOS transistor) and NMOS transistors are illustrated with an arrow pointing out of the transistor on the source terminal (e.g. transistor M_d4is an NMOS transistor). Each of the NMOS and PMOS transistors include a gate terminal, a source terminal, and a drain terminal.

[0043] In the embodiment of FIG. 3, the charge pump 18 includes current mirroring transistors M_m1, M_m2, and M_m3which serve to provide a current proportional to the reference current I_refthrough transistors M_c1and M_c2as up and down currents I_upand I_down, respectively. The current mirror is non-cascoded. Additionally, the charge pump 18 includes a series connection of transistors M_d1, M_s1, M_d2, and M_c2between the V_Ctrloutput node and the V_reg1power supply. The source and drain terminals of the M_d1and M_d2transistors are shorted. Coupled between the drain terminal of the M_s1transistor (and the source terminal of the M_c1transistor) and the Ground (Gnd) is the transistor M_x. The gate terminals of the M_d1and M_d2transistors are coupled to the Up signal, and the gate terminals of the M_s1and M_xtransistors are coupled to the complement of the Up signal to (referred to as “Up bar” herein and shown as the word “Up” with a bar over it in FIG. 3). Shown in dotted form between the source terminal of the M_c1transistor and the V_reg1power supply is a parasitic capacitance C_p1. The parasitic capacitance C_p1represents the total capacitance present at the source of transistor M_c1. The parasitic capacitance C_p1may include the source to supply parasitic capacitance of transistors M_c1, M_s1, and M_d2, as well as the drain to supply parasitic capacitance of transistors M_d2and M_x. Additionally, the capacitance may be affected by the gate to drain parasitic capacitance of M_s1and the gate to source parasitic capacitance of M_c1. The gate terminal of the transistor M_c1is coupled to the gate terminal of transistor M_m3.

[0044] Furthermore, the charge pump 18 includes a series connection of transistors M_c2, M_d3, M_s2, and M_d4between the V_Ctrloutput node and Gnd. The source and drain terminals of the M_d3and M_d4transistors are shorted. Coupled between the drain terminal of the M_s2transistor (and the source terminal of the M_c2transistor) and the V_reg1power supply is the transistor M_y. The gate terminals of the M_d3and M_d4transistors are coupled to the complement of the Down signal (referred to as “Down bar” herein and shown as the word “Down” with a bar over it in FIG. 3), and the gate terminals of the M_s2and M_ytransistors are coupled to the Down signal. Shown in dotted form between the source terminal of the M_c2transistor and Gnd is the parasitic capacitance C_p2. The parasitic capacitance C_p2represents the total capacitance present at the source of transistor M_c2. The parasitic capacitance C_p1may include the source to supply parasitic capacitance of transistors M_c2, M_s2, and M_d3, as well as the drain to supply parasitic capacitance of transistors M_d3and M_y. Additionally, the capacitance may be affected by the gate to drain parasitic capacitance of M_s2and the gate to source parasitic capacitance of M_c2. The gate terminal of the transistor M_c2is coupled to the gate terminal of transistor M_m2. Additionally shown in dotted form are the parasitic drain to gate capacitance of transistor M_s2(C_p3) and the parasitic drain to gate and source to gate capacitances (C_p3/2) of transistor M_d3. It is noted that transistors M_s1and M_d2have similar parasitic capacitances as transistors M_s2and M_d3, respectively, not shown in FIG. 3.

[0045] Finally, the charge pump 18 includes: (i) a set of transistors similar in connection to M_d1, M_s1, M_d2, M_c2, and M_xexcept the gate terminals are connected to Down or Down bar (as shown in FIG. 3) and the drain terminal of the transistor similar to M_c2is connected to a V_dummynode (reference numeral 40); and (ii) a set of transistors similar in connection to M_c2, M_d3, M_s2, M_d4, and M_yexcept the gate terminals are connected to Up or Up bar (as shown in FIG. 3) and the drain terminal of the transistor similar to M_c1is connected to the V_dummynode (reference numeral 42).

[0046] Generally, transistors M_c1and M_c2are the current devices in the charge pump 18, providing the I_upand I_downcurrents (respectively) to the output node V_Ctrl. The transistors M_s1and M_s2are the switching transistors used to switch the currents on and off responsive to the Up and Down signals, respectively. The switching of transistors frequently generates noise which may disturb the output of the charge pump 18, and thus reduce its accuracy and effectiveness. For example, when a transistor is switched from on (conducting) to off (not conducting), charge injection may occur. Generally speaking, charge injection occurs if charge in the channel of the switching transistor does not recombine in the channel when the switching transistor is switched to the off state. This charge travels out of the channel onto the source and/or drain nodes of the switching transistor, changing the voltage on these nodes. Since this voltage change is unintentional, the change represents error (noise) and may affect the output voltage V_Ctrl. Additionally, since the Up and Down signals may transition rapidly to active and inactive states, clock feedthrough (charge transmitted from the gate to the drain or source of the transistor via the parasitic capacitances between the gate and the drain or source) may occur, which also represents error.

[0047] The switching transistors M_s1and M_s2are placed on the source side of the current transistors M_c1and M_c2(i.e. the side opposite the output node V_Ctrl). Accordingly, any switching error which may occur in switching transistors M_s1and M_s2is attenuated through M_c1and M_c2. As described in more detail below, the charge pump 18 includes several mechanisms for reducing switching error from transistors M_s1and M_s2. Any switching error not removed via these mechanisms may be attenuated, further reducing its effect on the output node V_Ctrl.

[0048] The mechanisms used in the charge pump 18 for reduction of charge injection errors and clock feedthrough errors will be described next with respect to transistors M_s2, M_d3, and M_d4. Similar operation occurs with respect to transistors M_s1, M_d1, and M_d2(although the voltages that cause switching on and off differ since transistors M_s1, M_d1, and M_d2are PMOS transistors while transistors M_s2, M_d3, and M_d4are NMOS transistors).

[0049] Transistors M_d3and M_d4have gate terminals coupled to the complement of the signal on the gate terminal of transistor M_s2. In the illustrated embodiment, for example, the Down signal is received at the gate terminal of transistor M_s2, while the Down bar signal is received at the gate of transistors M_d3and M_d4. Accordingly, during times that transistor M_s2is being switched on (Down transitioning to a V_reg1voltage), transistors M_d3and M_d4are being switched off (Down bar transitioning to a Gnd voltage) and vice versa.

[0050] The aforementioned structure may reduce charge injection error. When M_s2is being switched off, and injecting charge onto its drain and/or source, M_d3and M_d4are being switched on. Thus, the channels of M_d3and M_d4acquire charge during approximately the same time period that transistor M_s2is injecting charge. The injected charge (are a large portion thereof) may be drawn into the channels of M_d3and M_d4as these channels acquire charge, thus reducing the amount of charge actually injected onto the drain and source nodes of transistor M_s2.

[0051] The transistors M_d3and M_d4may be carefully sized with respect to transistor M_s2in order to reduce the charge injection error appropriately. If the transistors are not sized appropriately, the charge injection reduction may not be minimized or may be overcompensated (making the effective charge injection negative). The assumption may be made that approximately half of the charge injected by transistor M_s2is injected onto the source and the remaining charge injected by transistor M_s2onto the drain. According, transistors M_d3and M_d4may be sized approximately half of the size of transistor M_s2, thus being capable of drawing the ½ charge injection from the corresponding source or drain terminal of transistor M_s2. The mobility of the devices affects the amount of charge injection (since the mobility may affect the recombination of charge in the channel), and the mobility may be factored into the sizing of the transistors M_d3and M_d4. For example, the transistors M_d3and M_d4may be sized plus or minus 10% of ½ the size of transistor M_s2.

[0052] It is noted that the “size” of a transistor may be the ratio of the channel width to the channel length. Thus, a first transistor may be ½ the size of a second transistor if the ratio of the channel width and channel length of the first transistor is ½ the ratio of the channel width and channel length of the second transistor. In general, the channel length of the transistors may nominally be the same for most of the transistors in FIG. 3. Particularly, it may be desirable for the channel length of transistors M_s1, M_s2, M_d1, M_d2, M_d3, and M_d4to be nominally the same since charge injection is related to the transit time across the channel.

[0053] The structure of transistors M_s2, M_d3, and M_d4may also reduce clock feedthrough error. Illustrated in FIG. 3 is the gate to drain parasitic capacitance of transistor M_s2(C_p3) as well as the gate to drain and gate to source parasitic capacitance of transistor M_d3(C_p3/2 each). The parasitic capacitances of transistor M_d3are approximately ½ the parasitic capacitance of transistor M_s2. Additionally, since the source and drain terminals of transistor M_d3are shorted, the parasitic capacitances are in parallel. Accordingly, the sum of the parasitic capacitances of transistor M_d3is approximately C_p3.

[0054] Since the gate terminal of transistor M_d3receives the complement of the signal on the gate terminal of transistor M_s2, the clock feedthrough from transistor M_d3to the drain of transistor M_s2(through a total parasitic capacitance of C_p3) may be approximately the same as the clock feedthrough from transistor M_s2to the drain of transistor M_s2, but of the opposite polarity. Summing these two clock feedthrough errors may produce a total clock feedthrough error on the drain of transistor M_s2of approximately zero, thereby reducing the clock feedthrough error. A similar discussion applies to the gate to source and drain parasitic capacitances of transistor M_d4and the gate to source parasitic capacitance of transistor M_s2(and to the transistors M_d1, M_d2, and M_s2).

[0055] It is desirable to match the currents I_upand I_down, so that equal width pulses of the Up and Down signals produce equal charge transferred to or drawn from the V_Ctrloutput node (respectively). The transistors M_c1and M_c2may be sized similarly to nominally provide the same current. Some effects (such as channel length modulation, drain induced barrier loading (DIBL), etc.) can temporarily affect this balance and thus contribute to error on the V_Ctrloutput node. Many of these effects can be reduced by increasing the channel length of the M_c1and M_c2transistors (i.e. these transistors may not have the minimum channel length in a given process technology). For example, in one specific implementation, a ratio of the channel length of transistors M_c1, and M_c2to the other transistors (which may have the minimum channel length for the process) may be in the range of approximately 3-7, and preferably about 5.6, although other ratios may be used. However, increasing the channel length (and the diffusion area width) of these devices leads to larger parasitic capacitances (e.g. the parasitic capacitance C_p2between the source of transistor M_c2and ground and the parasitic capacitance C_p1between the source of transistor M_c1and V_reg1). The parasitic capacitances C_p1and C_p2may act to cause additional current to flow through transistors M_c1and M_c2, respectively, after the corresponding switch M_s1and M_s2has switched off the current flow. This additional current flow is a source of error in the output voltage V_Ctrl. The larger parasitic capacitances may also require larger switch transistors M_s1and M_s2to switch the parasitic capacitances, which may exacerbate clock feedthrough and charge injection errors.

[0056] The charge pump 18 employs an active shutoff of the current transistors M_c1and M_c2using transistors M_xand M_y, respectively. The active turnoff of M_s2by transistor M_ywill be described, and the active turnoff of M_c1by transistor M_xmay be similar, with an appropriate change in polarities of voltages to account for the different transistor types.

[0057] Transistor M_yis a PMOS transistor coupled between the V_reg1power supply and the source of transistor M_c2. When transistor M_s2switches off in response to the down signal switching to a ground voltage, transistor M_yactivates and actively charges the source of M_c2to the V_reg1voltage. This quick charging of the source of transistor M_c2ensures that, regardless of the gate voltage of transistor M_c2(which is less than or equal to V_reg1), the current flow through M_c2stops quickly.

[0058] Another source of mismatch in the I_upand I_downcurrents for a given Up or Down signal pulse width may be the uneven loading of the Up and Down signals. By providing the sets of transistors illustrated at reference numerals 40 and 42, the load on the Up and Down signals may be balanced. For example, the Up signal is coupled to the gates of transistors M_d1, M_d2, a transistor similar to M_s2and a transistor similar to M_y. The Down signal is coupled to the gates of a transistor similar to M_d1, a transistor similar to M_d2, transistor M_s2, and transistor M_y. Accordingly, over process variation and process mismatch, the Up and Down signals should be approximately evenly loaded (and thus edge rates on these signals may be approximately equal). Transistors are said to be similar in this context if the transistors are nominally matched in characteristics (e.g. the same diffusion area width, channel length, etc.).

[0059] As mentioned above, transistors M_m1, M_m2, and M_m3comprise current mirroring for mirroring the reference current I_refto the I_upand I_downcurrents of M_c1and M_c2. A non-cascoded current mirror is used in this embodiment, which may reduce power consumption, increase headroom, and ease the task of keeping transistors in saturation. The noise suppression may be provided via the level shifter and voltage regulator, as described above. Other embodiments may employ cascoding of the current mirrors (e.g. for use in embodiments not having the voltage regulator of FIG. 1 or having a different voltage regulator).

[0060] FIG. 4 is a timing diagram further illustrating the operation of transistors M_s2, M_d3, M_d4, and M_yfor an exemplary set of transitions of the Down and Down bar signals. The Down and Down bar signals are represented by waveforms 50 and 52, respectively.

[0061] When Down transitions from high to low, charge injection may occur from transistor M_s2(arrow 54). However, at approximately the same time the down bar signal transitions from low to high, activating transistors M_d3and M_d4. Thus, these transistors may draw charge (including the injected charge), offsetting the injected charge (arrow 56).

[0062] The clock feedthrough from Down and Down bar is also shown in FIG. 4 (wave forms 58 and 60. Each of wave forms 58 and 60 is shown having approximately the same amplitude (A), since the Down and Down bar signals have the same swing and the parasitic capacitance through which they travel is approximately the same. Since the wave forms 58 and 60 have opposite polarities, the voltages sum to approximately zero at any given point (e.g. arrows 62 and 64).

[0063] Finally, a wave form 66 is illustrated in FIG. 4 illustrating an exemplary voltage on the node between the transistor M_s2and the transistor M_c2. The voltage may have some value between V_reg1and ground while the transistor M_s2is on, but than rapidly transitions to V_reg1when the transistor M_s2is turned off. The rapid transition occurs in response to the transistor M_yactively charging the node.

[0064] Turning next to FIG. 5, a circuit diagram of one embodiment of the VCO 20 is shown. Other embodiments are possible and contemplated. In the embodiment of FIG. 5, the VCO 20 includes an input transistor M_i1, a current mirror including transistors M_m4, M_m5, and M_m6, a ring oscillator 70, an output amplifier 72, an inverter 74, and a self-biasing circuit 76. The input transistor M_i1has a gate terminal coupled to receive the V_Ctrlvoltage and is coupled between the current mirror and ground. The current mirror transistor M_m5is coupled to the V_reg1power supply and to the power supply terminal for the inverters in the ring oscillator 70. The current mirror transistor M_m6is coupled between the V_reg1voltage and to the power supply terminal of the inverter 74. A first input to the amplifier 72 is coupled to the output of the ring oscillator 70. The output of the ring oscillator 70 is also coupled to the input of the inverter 74, which is coupled to the second input of the amplifier 72. The self-biasing circuit 76 provides a bias voltage to the amplifier 76, generated from the V_ddpower supply. The output of the amplifier 72 is coupled to a pair of series-coupled inverters 78.

[0065] Generally, the input transistor M_i1develops a current in response to the V_Ctrlvoltage input to the gate terminal. The current mirror transistor M_m4mirrors the current to the current mirror transistors M_m5and M_m6. The transistor M_m5provides charge to the power supply terminals of the inverters in the ring oscillator 70. Thus, as the current in the transistor M_m5increases (reflecting an increase in, the control voltage V_Ctrl), the power supply to the ring oscillator 70 increases and thus the oscillation frequency of the VCO 20 increases. Similarly, as the current in the transistor M_m5decreases, the oscillation frequency of the VCO 20 decreases. However, the signal swing of the ring oscillator 70 varies with the supply voltage as well, and may be small.

[0066] The output amplifier 72 is a differential amplifier providing an amplified, non-differential output to the output buffers 78. The output buffers 78 complete the transition of the output signal to a full V_ddswing. Since the output swing of the ring oscillator 70 is between ground and a voltage which may be small (the power supply voltage provided by the transistor M_m5), the amplifier employs PMOS input transistors (which react to low voltages well). The amplifier requires a differential input, however, and the combination of the inverter 74 supplied with a power supply through transistor M_m6in combination with the output of the ring oscillator 70 may closely approximate a differential input. If an inverter powered by the V_reg1power supply were used to provide the second input, the swing of the output signal of that inverter would not match the output of the ring oscillator 70. On the other hand, the power supply voltage for the inverter 74 may be close to the power supply voltage of the ring oscillator 70. Accordingly, although the output of the inverter 74 is slightly delayed with respect to the output of the ring oscillator 70, the output of the inverter 74 may more closely be a complement of output of the ring oscillator 70. Accordingly, the common mode noise rejection of the VCO 20 may be improved using the structure shown in FIG. 5.

[0067] It is noted that, in one implementation, the size of transistor M_m6may be approximately ⅓ the size of the transistor M_m5to provide approximately the same power supply voltage for the inverter 74 and the ring oscillator 70. Other embodiments may change to size ratio to match the number of inverters in the ring oscillator 70. For example, if an embodiment of the ring oscillator 70 includes 5 inverters, the transistor M_m6may be approximately ⅕ the size of the transistor M_m5.

[0068] The self-biasing circuit 76 operates to provide a bias voltage for amplifier 72 which is a relatively constant ratio of V_ddover the range of values that V_ddmay take on (especially due to switching noise, etc.) and over the range of process variation, operating temperature, etc. The ratio of the bias voltage to V_ddis set by the ratio of the sizes of each transistor in self-biasing circuit 76. The self-biasing may help ensure a gain in the amplifier 72 over the range of V_ddvalues which may be experienced during operation.

[0069] Additionally, the VCO 20 includes a non-cascoded current mirror including transistors M_m4, M_m5, and M_m6. Use of the non-cascoded current mirror (similar to the charge pump 18) allows for power consumption reduction and saturation of devices, as mentioned before. Furthermore, the non-cascoded current mirror results in a lower voltage drop between V_reg1and the power supply to the ring oscillator 70, allowing a wider range of operation for the ring oscillator 70 than in cascoded current mirroring were used.

[0070] It is noted that, while the ring oscillator 70 illustrated in FIG. 5 is shown as three series coupled inverters (with the output of the last inverter connected to the input of the first inverter), any odd number of inverters may be used. Furthermore, if fully differential oscillator stages are used, an even number of stages may be used. As used herein, the term “power supply terminal” refers to a terminal of a circuit designed to be connected to a power supply. In the ring oscillator 70 and inverter 74, the power supply terminal is coupled to another device to allow the voltage supplied, during use, to be varied.

[0071] Turning next to FIG. 6, a circuit diagram of one embodiment of the voltage regulator 12 is shown. Other embodiments are possible and contemplated. In the embodiment of FIG. 6, the voltage regulator 12 includes transistors M_sf1, M_sf2, and M_sf3coupled in a source follower configuration. Each of the transistors M_sf1, M_sf2, and M_sf3have a gate terminal coupled to a node between a resistor coupled to the V₁power supply and capacitor coupled to ground (e.g. the gate terminal of the transistor M_sf1is coupled to the node between the resistor R₃and the capacitor C₄). A bandgap generator 80 is used to generate a reference voltage, and an output transistor 82 is coupled to receive the reference voltage and provide it to an operational amplifier (op-amp) 84. The output transistor 82 is coupled to the transistor M_sf3as well, and a decoupling capacitor C_Decis coupled between transistor 82 and ground. The op-amp 84 amplifies the reference voltage from the bandgap generator 80 to produce the first regulated voltage V_reg1to be output by the voltage regulator 12. The output of the op-amp 84 passes through the common source transistor C_cs1to the output V_reg1and to an input of a second op-amp 86. The output voltage V_reg1is passed though a feedback network to the second input of the op-amp 84, where the ratio of the resistors in the feedback network, in conjunction with the gain of op-amp 84, divides the desired output voltage V_reg1back down to the reference voltage output from the bandgap generator 80. For example, the reference voltage may be 1.2 volts and the output voltage V_reg1may be 2.0 volts. Additionally, a transistor M_x1is coupled between the V_reg1output node and ground, and has its gate terminal coupled to the output node of the op-amp 84 (the gate terminal of the transistor M_cs1). A capacitor C_f1is coupled between the V_reg1output node and the output of the op-amp 84. Finally, a decoupling capacitor C_Decis coupled between the V_reg1output node and ground. The second op-amp 86 is connected in a unity gain configuration (the feedback network is a wire) and employs an output transistor M_cs2coupled between the transistor M_sf2and the V_reg2output node of the voltage regulator 12. Alternatively, the second op-amp 86 may be coupled in the same fashion as op-amp 84 and in parallel with op-amp 84 to produce the V_reg2voltage on the V_reg2output node. Similar to the V_reg1output node, a decoupling capacitor C_Decis coupled between the V_reg2output node and ground, a transistor M_x2is coupled between the V_reg2output node and ground with a gate terminal coupled to the output of the op-amp 86, and a capacitor C_f2is coupled between the V_reg2output node and the output of the op-amp 86. The gate terminals of transistors M_sf1and M_sf2are further coupled to a power-up control circuit 88.

[0072] The source follower transistors M_sf1, M_sf2, and M_sf3are used to provide current to the output transistors M_cs1, M_cs2, and 82, respectively. Generally, the source follower connection provides a low output impedance (the output being the terminal of the transistor M_sf1, M_sf2, and M_sf3coupled to the output transistor M_cs1, M_cs2, and 82, respectively), and thus the source follower connection may be an good voltage source. The transistors M_sf1, M_sf2and M_sf3may be sensitive to noise on the V₁power supply, particularly noise transmitted to the gate terminals of these transistors. The resistor/capacitor series connection with the gate terminal of these transistors coupled to the node connecting the resistor and the capacitor forms a low pass filter which may limit the amount of noise from the V₁power supply which is transmitted to the gate terminal of the corresponding transistor. The low pass filter effectively decouples the gate terminal from the V₁power supply and couples the gate terminal to ground through the capacitor within the low pass filter. In one implementation, for example, using a resistance of 5 Mega-ohms and a capacitance of 100 picoFarads, at least 15 dB of noise rejection may be experienced. In this manner, the transistors M_sf1, M_sf2, and M_sf3may act as better current sources even in the presence of noise. However, other embodiments of the voltage regulator 12 may not employ the resistor/capacitor connections to the gate of the source follower transistors, as desired, opting for a more conventional source follower connection or diode configuration.

[0073] While a low pass filter is used in the illustrated embodiment, any filter could be used depending on the frequencies of noise desired to be filtered. Furthermore, a combination of filters may be used.

[0074] Having the source follower transistors M_sf1, M_sf2, and M_sf3in series with another device may also limit the amount of voltage available below the series connection. Connecting a source follower to the V₁power supply may induce body effect in the source follower transistors which increases the threshold voltage of the transistors and thus the voltage drop across the transistors. If the voltage drop across the source follower transistors increases, there is less remaining voltage (between the voltage at the output of the source follower and ground) available for use in the remaining circuits (and, in the case of the transistors M_sf1and M_sf2, less voltage available for output as V_reg1and V_reg2). In one example, V₁is 3.3 volts and V_reg1/V_reg2is 2.0 volts, leaving only 1.3 volts for drops across M_sf1and M_sf2and the corresponding series devices M_cs1and M_cs2.

[0075] The body effect in the transistors M_sf1and M_sf2may be offset (at least partially), by using transistors with lower threshold voltages for transistors M_sf1and M_sf2. However, such transistors may also limit the voltages which may be tolerated between any two terminals. For example, in one implementation, the lower threshold transistors may be limited to 1.5 volts across any two terminals. During operation, the lower amount of voltage may not present a problem since most of the voltage between V₁and ground is used for V_reg1/V_reg2(e.g. in the above example, 1.3 volts separate V₁and V_reg1/V_reg2). During power up of the voltage regulator circuit 12, these transistors may be protected from over-voltage conditions using power-up control circuit 88. Particularly, the gate terminals of the transistors M_sf1and M_sf2, during power up, may initially be zero volts (in the absence of power-up control circuit 88). As the power supply V₁is powered to its operating level, the gate terminals of transistors M_sf1and M_sf2charge more slowly to the V₁voltage level (at a rate determined by the RC time constant of the resistor and capacitor coupled to the gate terminals). Thus, during power up, the full V₁voltage (or substantially the voltage, as the RC network charges) may be present between the gate and drain terminals of M_sf1and M_sf2. Power-up control circuit 88 provides a voltage (V_g1and V_g2in FIG. 6) during power up to ensure that the over-voltage condition does not occur, protecting the M_sf1and M_sf2transistors. For example, in one embodiment, the power-up control circuit 88 may provide a V₁voltage level at V_g1and V_g2. An example of the power-up control circuit 88 is shown in FIG. 7 below. For embodiments in which the transistors M_sf1and M_sf2are not implemented with lower threshold voltages and/or voltage tolerances, the power-up control circuit 88 may be omitted.

[0076] As used herein, the term “power up” refers to the application of a power supply voltage or voltages to a circuit after a period of time in which the circuit has been unpowered by the voltages.

[0077] The transistors M_x1and M_x2are provided to maintain a relatively constant current load at the V_reg1and V_reg2output nodes, respectively (in other words, relatively constant current through transistors M_cs1and M_cs2, respectively) even though the current demands of the PLL 10 generally vary during operation. By keeping the current through M_cs1and M_cs2relatively stable, the regulation of the V_reg1, and V_reg2voltages may be improved as well as stability, phase margin, and frequency response. It may generally be easier to regulate a voltage for a relatively constant current load, or a current load that varies within a small range. In the context of the PLL 10, stabilizing the current load may have the most effect for output frequencies of the PLL 10 at the lower end of the operating range of PLL 10.

[0078] The operation of M_x1will be described, and the operation of M_x2may be similar in the context of V_reg2and M_cs2. If the current flowing out of the voltage regulator 12 through the V_reg1terminal decreases, the voltage V_reg1tends to rise (since the current though M_cs2changes in proportion to its gate voltage, which hasn't yet changed). The increased voltage at V_reg1leads to an increased voltage at the positive input of the op-amp 84, which leads to an increase in the output voltage of the op-amp 84. The gate terminal of M_x1is coupled to the output of the op-amp 84, and is an NMOS transistor. Thus, as the output voltage of the op-amp 84 rises, the current in transistor M_x1increases. Accordingly, the current though transistor M_x1increases as the current traveling out of the V_reg1output node (to the PLL 10) decreases. Similarly, as the current traveling out of the V_reg1output node increases, the current through transistor M_x1decreases.

[0079] Viewed in another way, transistor M_x1is a current source which provides current inversely proportional to the current exiting the voltage regulator 12 through the V_reg1, output node. Any current source providing such a current may be used, in other embodiments. Furthermore, other embodiments of the voltage regulator 12 may not employ the current sources such as the transistors M_x1and M_x2, as desired.

[0080] It is noted that, while the common source connection of the output transistors M_cs1and M_cs2may exhibit a smaller voltage drop from the source of the transistors M_sf1and M_sf2to the output nodes V_reg1and V_reg2than a source follower connection, the output impedance may be higher. Decoupling capacitances C_Decare included on each node, mitigating the effect of the higher output impedance. Each capacitance C_Decmay be of any suitable size, but one embodiment may employ approximately 0.6 nanoFarads (nF) of decoupling capacitance.

[0081] It is noted that the stability of the op-amp/common source output transistor circuit may be affected by the existence of a dominant pole in the frequency response at both the output node of the op-amp and the output of the common source stage. Compensation capacitors C_f1and C_f2may be used to improve the stability be separating the poles. A resistor may also be placed in series with each of the capacitors C_f1and C_f2to remove a right half-plane zero in the response.

[0082] It is further noted that the use of a low pass filter on the gate terminal of a source follower circuit for noise reduction may be used in other contexts than a voltage regulator. The circuit may be used in any circuitry in which noise rejection is desirable. Similarly, the power-up control circuit 88 may be used to establish a voltage on any node during power up in order to protect a circuit from damage.

[0083] It is still further noted that other embodiments of the voltage regulator 12 may employ any combination of the above highlighted features, as desired.

[0084] Turning next to FIG. 7, a block diagram of one embodiment of the power-up control circuit 88 is shown. Other embodiments are possible and contemplated. In the embodiment of FIG. 7, the power-up control circuit 88 includes a resistor 90 coupled in series with a capacitor 92 between the V₁power supply and ground. The resistor may be implemented with transistors, in one embodiment, or with P+ poly, as desired. The gate terminal of a transistor 94 is coupled to the node between the resistor and the capacitor, and has a source coupled to the V₁power supply and a drain coupled to the V_g1output. A similar circuit including a resistor 96, a capacitor 98 and a transistor 100 is also included, with the transistor 100 having a drain coupled to the V_g2output.

[0085] The circuit including the resistor 90, the capacitor 92, and transistor 94 will be described. The circuit including the resistor 96, the capacitor 98, and the transistor 100 operates similarly. During power up, the voltage on the V₁power supply rapidly charges to the V₁voltage. However, the gate terminal of the transistor 94 charges from ground at a rate determined by the time constant of the resistor 90 and the capacitor 92. Accordingly, the transistor 94 is turned on and charges the V_g1output node to V₁rapidly. In this manner, the gate terminal of the transistor M_sf1is held at V₁during power up, reducing the gate to drain voltage on the transistor M_sf1to a tolerable level. As the gate terminal of the transistor 94 charges to V₁, the transistor 94 turns off and thus the node V_g1is not actively charged by the power-up, control circuit 88 during normal operation. Thus, the power-up control circuit 88 is active during a time period commencing at power up and ending at a later time determined by the time constant of resistor 90 and capacitor 92. The time period ends when the difference between the voltage of the gate terminal of the transistor 94 and the V₁voltage is less than a threshold voltage of the transistor 94.

[0086] While the embodiment of FIG. 7 includes two circuits producing two output voltages, other embodiments may include more circuits. Furthermore, an embodiment in which one circuit is used to supply gate voltages to both M_sf1and M_sf2from a single output node is contemplated.

[0087] Turning now to FIG. 8, a block diagram of a gain boosted op-amp 110 is shown. Other embodiments are possible and contemplated. The gain boosted op-amp 110 may be used as the op-amp 84 and/or 86 in FIG. 6, as desired. Furthermore, gain boosting may not be implemented in other embodiments.

[0088] Generally, the transistors 112 and 114 are intended as constant current devices. However, if the voltage at nodes 116 and 118 changes during operation (due to the positive and negative differential input voltages In_p and In_n changing), then the current in the transistors 112 and 114 may change. The voltage at nodes 116 and 118 is fed back through circuit 120 to the gate terminals of transistors 122 and 124, respectively. The current through the transistors 122 and 124 may thereby be altered to keep the current through the transistors 112 and 114 approximately constant. The gain boosting may ensure some gain from op-amp 110 over a large range of input voltage, power supply voltage, process, temperature, etc.

[0089] Turning next to FIG. 9, a block diagram of a carrier medium 300 including a database representative of PLL 10 is shown. Generally speaking, a carrier medium may include storage media such as magnetic or optical media, e.g., disk or CD-ROM, volatile or non-volatile memory media such as RAM (e.g. SDRAM, RDRAM, SRAM, etc.), ROM, etc., as well as transmission media or signals such as electrical, electromagnetic, or digital signals, conveyed via a communication medium such as a network and/or a wireless link.

[0090] Generally, the database of PLL 10 carried on carrier medium 300 may be a database which can be read by a program and used, directly or indirectly, to fabricate the hardware comprising PLL 10. For example, the database may, be a behavioral-level description or register-transfer level (RTL) description of the hardware functionality in a high level design language (HDL) such as Verilog or VHDL. The description may be read by a synthesis tool which may synthesize the description to produce a netlist comprising a list of gates from a synthesis library. The netlist comprises a set of gates which also represent the functionality of the hardware comprising PLL 10. The netlist may then be placed and routed to produce a data set describing geometric shapes to be applied to masks. The masks may then be used in various semiconductor fabrication steps to produce a semiconductor circuit or circuits corresponding to PLL 10. Alternatively, the database on carrier medium 300 may be the netlist (with or without the synthesis library) or the data set, as desired.

[0091] While carrier medium 300 carries a representation of PLL 10, other embodiments may carry a representation of any portion of PLL 10, as desired, including any combination of a voltage regulator 12, a charge pump 18, a VCO 20, a loop filter 28, a configuration register 32, a level shifter 16, a PFD 14, a current generator 30, and divide circuits 22, 24, and 26, etc. or portions thereof.

[0092] Numerous variations and modifications will become apparent to those skilled in the art once the above disclosure is fully appreciated. It is intended that the following claims be interpreted to embrace all such variations and modifications.