1. Field of the Invention
Embodiments in accordance with the present disclosure are directed to integrated circuits containing voltage regulators.
2. Description of the Related Art
A voltage regulator is an electronic circuit that provides a regulated voltage to a load. One problem with voltage regulators is that if the load changes, the voltage regulator may have difficulty maintaining the voltage at a target level. For purposes of illustration, the following example is provided of the difficulties that can occur when using a voltage regulator to supply voltages to a memory array of non-volatile storage elements. In the following example, the non-volatile storage elements are of a variety that store information based on a detectable change in state (state change elements).
Materials having a detectable level of change in state, such as a resistance or phase change, are used to form various types of non-volatile semiconductor based memory devices. For example, some materials (e.g., carbon) can be switched between low and high resistance states. These types of materials can be used to form re-writable memory elements. Multiple levels of detectable resistance in materials can further be used to form multi-state devices which may or may not be re-writable.
In many implementations, a memory array is arranged as a set of word lines and bit lines that are substantially perpendicular to each other with a memory element at the intersection of each word line and bit line. Thus, two-terminal memory elements can be constructed at the intersections with one terminal (e.g., terminal portion of the cell or separate layer of the cell) in contact with the conductor forming the respective word line and another terminal in contact with the conductor forming the respective bit line.
Programming and reading of the memory elements typically involves applying certain voltages to the word lines and bit lines. For example, a relatively large program voltage may be applied across a memory element by applying appropriate voltages at a certain word line and bit line.
Reading the memory elements typically involves applying a read voltage that is smaller than the program voltage. The program state of the memory element can be sensed by the amount of current that flows through the memory element in response to the read voltage. In some implementations, the memory element current causes a drop in voltage at a reference node based on the capacitance of the reference node. After a period of time, the voltage at the reference node is compared with a reference voltage to determine the program state of the memory element.
A power management circuit in a memory array has a linear voltage regulator that provides the voltages to read (and program) the memory elements. However, under some circumstances the load current on the voltage regulator changes very rapidly. The change in load current causes a ripple of the output voltage of the voltage regulator. Unfortunately, this voltage ripple interferes with the ability to accurately sense the state of the memory element. In some implementations, the voltage ripple causes a capacitive current due to one or capacitances in the memory sense circuit. Note that this capacitance may be a parasitic capacitance and is not the aforementioned capacitance of the reference node. However, this capacitive current may lead to an error when comparing a reference current with the memory element current. That is, the capacitive current can add to or subtract from the memory element current, which alters the rate at which the voltage at the reference node changes.
In one implementation, the problem is overcome by delaying the sensing of the memory element until the voltage ripple settles down. However, this adds to the time needed to sense the state of the memory elements.
Another possible solution is to add an amplifier to the circuit to provide a large pre-charging current prior to sensing the state of the memory element. The pre-charging current may prevent the voltage ripple from occurring, or at least substantially reduce the voltage ripple. However, the amplifier that is needed to adequately reduce the voltage ripple needs to be a fast amplifier that consumes considerable power to eliminate the voltage ripple.
A voltage regulator is disclosed. The voltage regulator may be used to supply a voltage to non-volatile storage elements as a part of reading the state of the non-volatile storage elements. Even if a voltage regulator in accordance with an embodiment of the present invention is subjected to a large change in load current, the output of the voltage regulator settles down relatively quickly. Therefore, when used to provide a voltage to read non-volatile storage elements the voltage regulator output settles down quickly for a quick read operation. Further, the solution does not require the addition of a power consuming amplifier to prevent the voltage ripple.
One embodiment is a voltage regulation circuit comprising the following elements. The voltage regulation circuit has a voltage generation circuit that outputs a regulated voltage and a load current. The voltage regulation circuit has a sensing circuit that senses a peak magnitude of the load current. The sensing circuit stores a peak signal that is based on the peak magnitude of the load current. The voltage regulation circuit has a current generation circuit that generates a compensation current that has a magnitude that is proportional to the peak magnitude of the load current. The current generation circuit generates the compensation current based on the peak signal. The current generation circuit generates the compensation current during a time interval that is defined by at least one signal that is input to the voltage regulation circuit. In one implementation, the sensing circuit senses the peak magnitude of the load current during a time interval that is defined by the at least one signal.
One embodiment is a memory array having non-volatile storage elements and sense amplifiers that sense conditions of the non-volatile storage elements. The memory array has a voltage regulation circuit that comprises the following elements. The voltage regulation circuit has a voltage generation circuit that outputs a regulated voltage. The voltage regulation circuit establishes the regulated voltage at the non-volatile storage elements prior to the sense amplifiers sensing conditions of the non-volatile storage elements. The voltage regulation circuit has a sensing circuit that senses a peak magnitude of a load current that results from the voltage generation circuit establishing the regulated voltage at the non-volatile storage elements. The sensing circuit stores a peak signal that is based on the peak magnitude of the load current. The voltage regulation circuit has a current generation circuit that generates a compensation current that is proportional to the peak magnitude of the load current. The current generation circuit generates the compensation current based on the peak signal. The current generation circuit generates the compensation current during at least a portion of an interval in which the sense amplifiers sense conditions of the non-volatile storage elements.
One embodiment is a method that comprising the following. A regulated voltage is generated at an output node. The output node has a load current associated therewith. The regulated voltage is applied to non-volatile storage elements in a memory array. A peak magnitude of the load current is sensed in response to at least one signal that is derived from timing signals that are used when reading the non-volatile storage elements. A compensation current that is proportional to the peak magnitude of the load current is generated. A condition associated with at least a first non-volatile storage element of the non-volatile storage elements is determined. The compensation current is provided to the output node while determining the condition.
The voltage converting circuit 102 provides a regulated voltage “VOUT” to a load (e.g., memory element). The voltage converting circuit 102 provides a load current “ILOAD” that depends on the characteristics (e.g., impedance) of the load. Thus, if the impedance of the load changes, the load current will change. For example, if a switch closes to connect the voltage regulator 100 to memory elements to charge them to a known voltage prior to reading the memory elements, the load abruptly changes. The voltage regulator 100 attempts to keep the output voltage at a target voltage. However, abrupt changes to the load current can cause VOUT to deviate from the target voltage for a period of time. Specifically, VOUT may oscillate above and below the target voltage in a ripple that decays over time.
The maximum load current determination circuit 104 determines the maximum load current “ILOADMAX”. In this implementation, the maximum load current determination circuit 104 inputs a signal that defines the interval during which the maximum load current should be determined. As an example, the maximum load current when charging the memory elements could be determined.
The proportional current generation circuit 106 generates a compensation current “ICOMP” that is proportional to the maximum load current ILOADMAX. The compensation current is provided to the output of the voltage regulator 100. The purpose of the compensation current is to help to stabilize VOUT after an abrupt change to the load current ILOAD. For example, after an abrupt change to the load current causes the output voltage to deviate from the target voltage, the compensation current ICOMP helps the voltage regulator 100 stabilize VOUT back to the target level.
The proportional current generation circuit 106 inputs a signal that defines an interval during which the compensation current ICOMP is to be provided to the output. Note that there may be a gap in time between sensing the maximum load current ILOADMAX and providing the compensation current ICOMP. Also note that the compensation current ICOMP may be significantly smaller in absolute magnitude than the maximum load current ILOADMAX.
As previously mentioned, the voltage regulator 100 may be used in a non-volatile storage device. The following is an example of memory elements that may be used in a non-volatile storage device and an example architecture for a non-volatile storage device.
Examples of suitable materials for state change element 204 include, but are not limited to doped semiconductors (e.g., polycrystalline silicon, more commonly polysilicon), transition metal oxides, complex metal oxides, programmable metallization connections, phase change resistive elements, organic material variable resistors, carbon polymer films, doped chalcogenide glass, and Schottky barrier diodes containing mobile atoms that change resistance. The resistivity of these materials in some cases may only be switched in a first direction (e.g., high to low), while in others, the resistivity may be switched from a first level (e.g., higher resistance) to a second level (e.g., lower resistance), and then switched back toward the first resistivity level.
In one embodiment, the state change element 204 is Ge2Sb2Te5 (GST). GST has a property of reversible phase change from crystalline to amorphous-allowing two levels per cell. However, quasi-amorphous and quasi-crystalline phases may also be used to allow additional levels per cell with GST.
In some embodiments, the state change element 204 is formed from a carbon material. A state change element 204 that is formed from carbon may comprise any combination of amorphous and graphitic carbon. In one aspect, the carbon is deposited as a carbon film. However, it is not required that a carbon state change element be a carbon film. In one aspect, the state change element 204 is a carbon nanotube (CNT).
By assigning logical data values to the various levels of resistance that can be set and read from resistance change element 204, memory element 200 can provide reliable data read/write capabilities. Anti-fuse 206 can further provide resistance state change abilities that can be exploited for non-volatile data storage. An anti-fuse is manufactured in a high resistance state and can be popped or fused to a lower resistance state. An anti-fuse is typically non-conductive in its initial state and exhibits high conductivity with low resistance in its popped or fused state. As a discreet device or element may have a resistance and different resistance states, the terms resistivity and resistivity state are used to refer to the properties of materials themselves. Thus, while a resistance change element or device may have resistance states, a resistivity change material may have resistivity states.
Anti-fuse 206 can provide benefits to memory element 200 beyond its state change ability. For example, an anti-fuse can serve to set the on-resistance of the memory element in at an appropriate level relative to the read-write circuitry associated with the cell. These circuits are typically used to pop the anti-fuse and have an associated resistance. Because these circuits drive the voltages and current levels to pop the anti-fuse, the anti-fuse tends to set the memory element in an appropriate on-resistance state for these same circuits during later operations.
A range of resistance values can be assigned to a physical data state to accommodate differences amongst devices as well as variations within devices after set and reset cycling. The terms set and reset are typically used, respectively, to refer to the process of changing an element from a high resistance physical state to a low resistance physical state (set) and changing an element from a low resistance physical state to a higher resistance physical state (reset). Embodiments in accordance with the present disclosure can be used to set memory elements to a lower resistance state or to reset memory elements to a higher resistance state. While specific examples may be provided with respect to set or reset operations, it will be appreciated that these are mere examples and that the disclosure is not so limited.
Various types of suitable state change elements are described in U.S. Pat. No. 6,034,882 entitled “Vertically Stacked Field Programmable Non-volatile Memory and Method of Fabrication.” Various other types of state change elements may be used, including those described in U.S. Pat. No. 6,420,215 entitled “Three Dimensional Memory Array and Method of fabrication,” and U.S. Pat. No. 6,631,085, entitled “Three-Dimensional Memory Array Incorporating Serial Chain Diode Stack,” all hereby incorporated by reference in their entirety.
It will be appreciated that other types of two-terminal non-volatile memory elements can be used in embodiments. For example, one embodiment does not have an anti-fuse 206 and merely includes state change element 204 and steering element 202. Other embodiments may include additional state change elements in place of or in addition to the anti-fuse.
Conductors 210 and 212 are typically orthogonal to one another and form array terminal lines for accessing an array of memory elements 200. The array terminal lines (also called array lines) at one layer may be termed word lines or X-lines. The array lines at a vertically adjacent layer may be termed bit lines or Y-lines. A memory element can be formed at the projected intersection of each word line and each bit line, and connected between the respective intersecting word line and bit line as shown for the formation of memory element 200. A three-dimensional memory array which has at least two levels of memory elements (i.e., two memory planes) may utilize more than one layer of word lines and/or more than one layer of bit lines. A monolithic three dimensional memory array is one in which multiple memory levels are formed above a single substrate, such as a wafer, with no intervening substrates.
In an alternative embodiment, an inter-level dielectric can be formed between adjacent memory levels. In this alternative, no conductors are shared between memory levels. This type of structure for three-dimensional monolithic storage memory is often referred to as a non-mirrored structure. In some embodiments, adjacent memory levels that share conductors and adjacent memory levels that do not share conductors can be stacked in the same monolithic three dimensional memory array. In other embodiments, some conductors are shared while others are not. For example, only the word lines or only the bit lines are shared in some configurations. A first memory level L0 can include memory elements between a bit line level BL0 and word line level WL0. The word lines at level WL0 can be shared to form cells at a memory level L1 that connect to a second bit line level BL1. The bit line layers are not shared so the next layer can include an interlayer dielectric to separate bit lines BL1 from the next level of conductors. This type of configuration is often referred to as half-mirrored. Memory levels need not all be formed having the same type of memory element. If desired, memory levels using resistive change materials can alternate with memory levels using other types of memory elements, etc.
In one embodiment, word lines are formed using word line segments disposed on different word line layers of the array. The segments can be connected by a vertical connection to form an individual word line. A group of word lines, each residing on a separate layer and substantially vertically-aligned (notwithstanding small lateral offsets on some layers), may be collectively termed a row. The word lines within a row preferably share at least a portion of the row address. Similarly, a group of bit lines, each residing on a separate layer and substantially vertically-aligned (again, notwithstanding small lateral offsets on some layers), may be collectively termed a column. The bit lines within a column preferably share at least a portion of the column address. An example of such a configuration is described in U.S. Pat. No. 7,054,219, entitled, “Transistor Layout Configuration for Tight Pitched Memory Array Lines, which is hereby incorporated by reference in its entirety.
Voltage regulator 100 provides a regulated voltage VOUT to the row control circuit 420 and column control circuitry 410 for various operations such as reading memory elements. The sense amplifiers 504 are used to read the state in which memory elements are programmed.
Integrated circuits incorporating a memory array usually subdivide the array into a sometimes large number of sub-arrays or blocks. Blocks can be further grouped together into bays that contain, for example, 16, 32, or a different number of blocks. As frequently used, a sub-array is a contiguous group of memory elements having contiguous word and bit lines generally unbroken by decoders, drivers, sense amplifiers, and input/output circuits. This is done for any of a variety of reasons. For example, the signal delays traversing down word lines and bit lines which arise from the resistance and the capacitance of such lines (i.e., the RC delays) may be very significant in a large array. These RC delays may be reduced by subdividing a larger array into a group of smaller sub-arrays so that the length of each word line and/or each bit line is reduced. As another example, the power associated with accessing a group of memory elements may dictate an upper limit to the number of memory elements which may be accessed simultaneously during a given memory cycle. Consequently, a large memory array is frequently subdivided into smaller sub-arrays to decrease the number of memory elements which are simultaneously accessed. Nonetheless, for ease of description, an array may also be used synonymously with sub-array to refer to a contiguous group of memory elements having contiguous word and bit lines generally unbroken by decoders, drivers, sense amplifiers, and input/output circuits. An integrated circuit may include one or more than one memory array.
In some embodiments, memory element operation is based on a bi-stable resistance change in the material that forms the state change element 204. The state change element 204 can be made to change state by the application of a high bias voltage (e.g., 4 V). Current through the memory element 200 is a function of the resistance of the state change element 204. The memory elements 200 are read at a lower voltage than the program voltage such that reading will not change the resistance of the state change element 204.
For example, if the memory element 200 has been programmed to state “1”, the current “IMEM” is measurably larger than if the memory element 200 has been programmed to state “0”. In some implementations, IMEM may be an order of magnitude larger in one state than the other state. The state change element 204 is depicted as a resistor in
The memory sense circuit 500 includes a voltage regulator 100 that supplies a voltage “VOUT”. Basic operation of the voltage regulator 100 is as follows. The voltage converting circuit 102 converts an input voltage “VWR” into an output voltage “VOUT,” and supplies an output current “ILOAD” in accordance with requirements of a load. In this example, the load includes the memory element 200. The memory element 200 has a steering element 202 (e.g., a diode) and a resistive state change element 204. Note that it is not desirable to keep the memory element 200 charged to VOUT at all times. Rather, the memory element 200 is charged to VOUT just prior to reading the memory element 200.
A resistive voltage divider formed of series-connected resistors “R1” and “R2” in the voltage regulator 100 generates a feedback voltage “VOUTDIV” representative of the output voltage VOUT. By comparing the feedback voltage VOUTDIV and a predetermined reference voltage “VBG,” amplifier 514 generates and applies an error voltage to a gate of output transistor 516. The drain of output transistor 516 is connected to the input voltage VWR and the source of output transistor 516 is connected to R1 at the node that provides the output voltage VOUT. The voltage converting circuit 102 has a compensation capacitor CCOMP with one plate connected to the output of the amplifier 514 and the other plate connected to the drain of transistor 533. Transistor 533 provides a compensation current ICOMP. This compensation current ICOMP is part of the normal operation of the voltage converting circuit 102 and is not a part of generating a current that is proportional to the maximum load current ILOADMAX.
The voltage regulator 100 also has a maximum load current determination circuit 104 and a proportional current generation circuit 106. In this embodiment, the maximum load current determination circuit 104 is connected to the output of amplifier 514 and the node that supplies VOUT. The proportional current generation circuit 106 is connected to the node that supplies VOUT (and to the maximum load current determination circuit 104). Basic operation of these circuits 104, 106 has been discussed with respect to
A sense amplifier (or comparator) 504 is used to determine the state to which the memory element 200 is programmed by comparing a voltage “VSENSE” at node A with a reference voltage “VREF
Prior to reading the state of a memory element 200 the voltage of the memory element 200 is raised to VOUT. At other times, the memory element 200 may be kept at ground. Thus, prior to reading the memory element 200, the memory element 200 is connected to the voltage regulator 100, which causes an abrupt change to the voltage regulator load. In general, when the load on a linear voltage regulator makes a significant transition (e.g., from heavy loading to light loading or from light loading to heavy loading) the output voltage is disturbed. In the circuit 500, this disturb of the output voltage charges (or discharges) the parasitic capacitance “CPSELB” at node B (the anode of the steering element 202 in the memory element 200). Typically, the voltage regulator output voltage VOUT overshoots in an attempt to re-establish the target voltage. More specifically, VOUT is typically a ripple that dies down over time.
While SA_ENABLE is still low, a signal “precharge” goes high (
After a peak current interval that follows the precharge signal going low, signals BLP (
When SA_ENABLE goes high, switch 520 opens such that the direct output of the voltage regulator 100 is no longer connected to the memory cell 200. However, when SA_ENABLE goes high, amplifier 526 charges the memory element 200 to VOUT. Specifically, amplifier 526 receives VOUT at the non-inverting input. The inverting input is coupled to the source of transistor 528. The output of amplifier 526 is the voltage SFGATE, which is provided to the gate of transistors 528 and 530. Transistor 530 provides a current to bias the memory element 200 to VOUT.
After BLP goes to 0, the voltage at node A starts to decrease based on the state of the memory element 200 being sensed. That is, the (parasitic) capacitance CPSENSE at node A is discharged. FIG. 6E(i)-6(E)(iii) depicts several example cases that depict the discharge of CPSENSE based on the relative magnitude of the memory element current Imem to the reference current IREF generated by the current mirror. FIG. 6(E)(iv) depicts the reference voltage VREF
Referring to FIG. 6(E)(i), if IMEM is less than IREF, then VSENSE drops very slowly. VSENSE will not drop sufficiently to cause the output (SA_DATA) of the sense amplifier 504 to switch. Thus, referring to FIG. 6(F)(i), SA_DATA remains low.
Referring to FIG. 6(E)(ii), if IMEM is somewhat less than IREF, then VSENSE drops at a rate that depends on how much greater IMEM is than IREF. When VSENSE drops below VREF
Referring to FIG. 6(E)(iii), if IMEM is substantially greater than IREF, then VSENSE drops rapidly. When VSENSE drops below VREF
The memory sense circuit 500 of
It may take some time for the voltage ripple to settle down. The recovery period is depicted as the oscillation of VOUT, which in this example occurs during the memory element sensing phase. That is, the oscillation of VOUT occurs when SA_ENABLE is high. Unfortunately, this oscillation interferes with the accurate sensing of the memory elements 200. For example, the voltage VSENSE may fall at a much more rapid rate than it should, which can prevent an accurate measurement.
If the voltage ripple extends beyond the point when BLP goes low (
One possible solution to this problem is to delay the sensing of the memory element 200 until after the ripple on VOUT settles down. For example, the period at which BLP is high might be extended to delay sensing the memory element current. However, this solution may not be desirable as it extends the amount of time needed to read the memory element 200.
Another possible solution is to add an amplifier that can absorb the extra current due to the load change. Specifically, an amplifier might provide a precharge current just prior to the signals BLP and SA_ENABLE going high. In other words, this solution would add a fast amplifier that is used to absorb all (or at least most) of the current needed to precharge the memory elements 200 before SA_ENABLE is set to 1. However, the voltage regulator 100 now has to provide the input voltage to the fast amplifier, along with its other duties. In this possible solution, ILOAD would not have an abrupt load current variation before SA_ENABLE. Thus, VOUT would remain at the target level. Therefore, the detection of VSENSE at node A would not be affected by the parasitic elements (e.g., CPSELB).
A possible draw-back of this potential solution is the added amplifier circuit should be fast to be able to recover from the abrupt change to the load current and consequentially would sink substantial current from the power supply. Moreover, this additional amplifier might have a very low efficiency since it should be designed to be able to sink (or source) a very large current.
Both implementations have circuits that sense a peak output of a voltage regulator 100 (e.g., a peak voltage regulator load current ILOADMAX. The circuit of each embodiment adds a current that is proportional to ILOADMAX. The proportional current is added during the memory element sensing phase (i.e., when SA_ENABLE=1). Note that it is not necessary for the compensation current to be as large as the peak current. It can be beneficial for the compensation current to be substantially less than the peak load current such that the additional circuits sink a very small current. Therefore, these solutions do not negatively impact the overall power efficiency of the voltage regulator 100.
The following is a description of the operation of voltage regulator 100a of
Operation of the voltage regulator 100a of
During this sensing phase (VOUT
After sensing the peak load current, the voltage VCTRL is provided to the gate of transistor MNAD to generate a current that is proportional to the peak load current, as the following will illustrate. When SA_ENABLE goes high (
The following describes details of selecting circuit parameters to arrive at a desired proportionality of the compensation current to the peak load current. Some of the circuit parameters that can be varied include, but are not limited to, the sizes of capacitor C1 and capacitor C2, and the sizes/widths of transistors MNDRV and MNAD. In one embodiment, the ratio of C1/C2 and (W/L)MNDRV/(W/L)MNAD determines the amount of current added to the output of the voltage regulator 100a to improve the settling time of the output voltage VOUT.
For example, if C1=C2, then VCTRL=0.5*VMAX. Continuing this example, if transistors MNDRV and MNAD are identical, then the compensation current will be half the peak current. Note that it is not required to use capacitor C2. In the case in which C2=0, then VCTRL=VMAX.
The proportionality can be further tuned by selection of the widths and lengths of transistors MNDRV and MNAD. Note that the voltage that is being sensed by capacitor C1 is the gate to source voltage of transistor MNDRV. Further note that the voltage VCTRL is applied between the gate and source of transistor MNAD. Thus, by appropriate selection of the ratio of (W/L)MNDRV/(W/L)MNAD the proportionality of the current of MNAD to MNDRV can be achieved.
It may be desirable to have a linear relationship between ILOADMAX and the compensation current supplied by MNAD. In one implementation, to guarantee a linear relationship between ILOADMAX and the compensation current supplied by transistor MNAD, the capacitances of capacitors C1 and C2 should be independent of the voltages applied to their plates. In other words, the capacitors C1 and C2 are linear capacitors. Metal-metal capacitors can be fabricated such that the capacitance is independent of voltage. Hence, in one embodiment, capacitors C1 and C2 are metal-metal capacitors. Linear capacitors other than metal-metal capacitors might also be fabricated.
In the option depicted in
Specifically, transistor MNDSM is a scaled down replica of driver transistor MNDRV 516. The scale factor will be referred to as “m” and in this example is much higher than 1. Therefore, the voltage regulator output current (ILOAD) is supplied almost entirely by the driver transistor MNDRV. Note that the gate of these two transistors are coupled together and also the sources are coupled together. Therefore, the current in transistor MNDSM tracks that of the driver transistor MNDRV, but is much smaller due to its smaller scale. Transistor MNDSM is thus biased with the current ILOAD/m.
Transistors MP1 and MP2 form a current mirror in order to mirror the current ILOAD/m. Specifically, transistor MP1 is coupled to transistor MNDSM such that it is biased with a current that is a fraction “m” of the load current ILOAD. Transistors MP1 and MP2 may be identical such that transistor MP2 provides the current ILOAD/m to transistor MNR. However, transistors MP1 and MP2 could be sized differently such that a current that is smaller or larger than current ILOAD/m is provided to transistor MNR.
Capacitor C3 tracks the gate-source voltage of transistor MNR when SA_ENABLE is low. That is, switch 1003 is closed when SA_ENABLE is low. When SA_ENABLE is low, switch 1004 is closed to connect the gate of transistor MNAD to ground. Therefore, transistor MNAD is shut off when SA_ENABLE is low.
When SA_ENABLE is high, capacitor C3 is disconnected from transistor MNR. However, capacitor C3 maintains the charge that was stored when SA_ENABLE was low. At this time, amplifier AMPAD buffers the voltage that is across capacitor C3 and biases transistor MNAD, which delivers to the output of the voltage regulator 100b a current that is proportional to ILOADMAX. Therefore, the voltage regulator 100b delivers a compensation current that is proportional to the maximum load current ILOADMAX. Note that by appropriate selection of the ratio of (W/L)MNDRV/(W/L)MNAD the relationship between the compensation current and the load current can be further tuned.
Note that capacitor C3 is not required to be a linear capacitor C3 in order to assure a linear relationship between the maximum load current and the compensation current. Thus, capacitor C3 is not required to be a metal-metal capacitor. Therefore, capacitor C3 can easily be fabricated. A reason why capacitor C3 is not required to be a linear capacitor is that a single capacitor C3 both stores the signal that is based and the peak load current and supplies that stored signal to transistor MNAD. Thus, even if the capacitance of capacitor C3 depends on the voltage across its plates, a linear relationship between the peak load current and the compensation current provided by transistor MNAD is maintained.
While example embodiments of using a voltage regulator within a memory array have been provided, the present invention is not limited to use in a memory array. Moreover, while examples in which the memory elements are state change elements have been provided, a voltage regulator in accordance with embodiments of the present invention may be used with memory arrays have other types of memory elements. An example of a different type of memory array is a NAND memory array. Relevant examples of NAND type memories and their operation are provided in the following U.S. Patents/Patent Applications, all of which are incorporated herein by reference: U.S. Pat. No. 5,570,315; U.S. Pat. No. 5,774,397; U.S. Pat. No. 6,046,935; U.S. Pat. No. 6,456,528; and U.S. Pat. Publication No. US2003/0002348. The discussion herein can also apply to other types of flash memory in addition to NAND, as well as other types of non-volatile memory.
The foregoing detailed description of the invention has been presented for purposes of illustration and description. It is not intended to be exhaustive or to limit the invention to the precise form disclosed. Many modifications and variations are possible in light of the above teaching. The described embodiments were chosen in order to best explain the principles of the invention and its practical application to thereby enable others skilled in the art to best utilize the invention in various embodiments and with various modifications as are suited to the particular use contemplated. It is intended that the scope of the invention be defined by the claims appended hereto.