Integrated circuit with current limited charge pump and method

[0002] This application claims priority under 35 USC §119(e)(1) of provisional application No. 60/166,655 filed Nov. 18, 1999.

[0003] This patent application is related to co-pending U.S. application entitled INTEGRATED CIRCUIT WITH CHARGE PUMP AND METHOD, Serial No. 60/166,655, filed Nov. 18, 1999.

[0004] This invention relates generally to the field of integrated circuits and more particularly to integrated circuits having a current limited charge pump.

[0005] Integrated circuits often employ circuitry such as charge pumps to increase or decrease voltage levels within the integrated circuit. Conventional charge pump designs typically include diodes to couple multiple stages. Such charge pumps typically utilize several stages to achieve a voltage step-up. Current losses within each of these stages may reduce efficiency and may result in poor predictability of the current flow out of the pump, thus complicating designs for current limiting. Adding stages may decrease the efficiency of such a pump due to forward drops in voltage across each diode.

[0006] Existing charge pumps sometimes employ current limiting circuitry at the output. One prior approach includes a current mirror coupled to the output of the charge pump. This approach harms power efficiency due to the loss of the current drawn down the reference limb of the current mirror. The size of the devices within the charge pump must also be increased to handle the additional current drawn by the current mirror. Since many charge pumps used in integrated circuits produce very small currents (1 uA or less), the increase in device size may be considerable.

[0007] One aspect of the invention is an integrated circuit comprising a current source coupled to an input element and an output load. The integrated circuit further comprises a first stage charge pump coupled to the current source at a first node and to the output load at a second node and a recirculation circuit coupled to the first node and the second node. The recirculation circuit is operable to limit to a known value the current that flows between the second node and the output load.

[0008] The invention provides several important technical advantages. The invention improves current efficiency while providing a current-limited output. The invention may utilize any charge pump. It achieves these advantages with simple circuitry that consumes little integrated circuit area.

[0009] For a more complete understanding of the present invention and the advantages thereof, reference is now made to the following descriptions taken in connection with the accompanying drawings in which:

[0010] FIG. 1 illustrates a schematic diagram of one embodiment of a charge pump constructed in accordance with the present invention;

[0011] FIG. 2 illustrates a cross section of one embodiment of an n-channel transistor used in one embodiment of the charge pump of FIG. 1;

[0012] FIG. 3 illustrates a cross section of one embodiment of a p-channel transistor used in one embodiment of the charge pump of FIG. 1;

[0013] FIG. 4 illustrates a schematic diagram of another embodiment of a charge, pump constructed in accordance with the present invention;

[0014] FIG. 5 illustrates a block diagram of a multi-stage charge pump constructed in accordance with the present invention;

[0015] FIG. 6 illustrates a schematic diagram of a current-limited charge pump constructed in accordance with the present invention;

[0016] FIG. 7 illustrates an exemplary recirculation circuit for a step-up charge, pump utilizing the teachings of the present invention; and

[0017] FIG. 8 illustrates another exemplary recirculation circuit utilizing the teachings of the present invention.

[0018] The present invention and its advantages are best understood by referring the FIGS. 1 through 8 of the drawings, like numerals being used for like and corresponding parts of the various drawings.

[0019] FIG. 1 illustrates a schematic diagram of one embodiment of a charge pump. Charge pump 10 comprises two n-channel field effect transistors MN1 and MN2, two p-channel field effect transistors MP1 and MP2, and two capacitors C1 and C2 operable to couple to two signal sources φ₁and φ₂. These components couple first node 11 of charge pump 10 to second node 12 of charge pump 10.

[0020] Charge pump 10 comprises two generally symmetric half cells A and B. Each half cell comprises one n-channel transistor, one p-channel transistor, and a capacitor. In each half cell, the gate and drain of the n-channel transistor are coupled to the gate and drain of the p-channel transistor, respectively. The drains are coupled to one capacitor while the gates are coupled to another capacitor from the other half-cell. Each capacitor is coupled to its associated signal source. The source of the n-channel transistor is coupled to first node 11, and the source of the p-channel transistor is coupled to second node 12. In this embodiment, half-cell A comprises MN1, MP1, and C2, and half-cell B comprises MN2, MP2, and C1.

[0021] Any suitable field effect transistors, such as MOSFETS, may be used. To minimize the effective-on-resistances of transistors MP1, MP2, MN1 and MN2, it is desirable that the backgate of each transistor connect to its source. The backgates of transistors MP1 and MP2 should be isolated from the backgates of transistors MN1 and MN2. Any suitable method of isolation can be used, such as the junction isolation schemes discussed in further detail in conjunction with FIGS. 2 and 3. Other isolation techniques may also be employed in order to accommodate manufacturing needs.

[0022] Capacitors C1 and C2 may be any suitably-sized capacitors. In cases where capacitors C1 or C2 are created by a process wherein one electrode of the capacitor has a larger parasitic coupling capacitance than the other electrode, it is desirable to connect the electrode having the larger parasitic capacitance to the associated signal source, rather than coupling it to the associated transistors. Such a configuration reduces charge injection from internal nodes of the charge pump into the substrate, thus improving the overall energy efficiency of charge pump 10.

[0023] Signal sources φ₁and φ₂may be any suitable signal sources that are approximately 180 degrees out of phase. In one embodiment, signal sources φ₁and φ₂are two square wave signals having equal peak-to-peak voltages V_φ and duty cycles of approximately 50%. Other signal sources may be also used within the scope of the invention.

[0024] When operated without a significant load current being drawn from second node 12, charge pump 10 generates a voltage at second node 12 that is equal to the sum of the peak-to-peak voltage V_Ø of φ₁and the voltage (V_in) applied to first node 11. Charge pump 10 thus operates to step up the voltage V_inapplied to first node 11 by a voltage V_φ provided by signal sources φ₁and φ₂. The charge pump may also be operated to step down the voltage V_inby a voltage V_φ by connecting voltage V_into second node 12 and by connecting the load to first node 11. For example, if voltage V_inequals zero volts (ground potential), then the voltage generated at first node 11 will be negative, and charge pump 10 acts to invert voltage V_in.

[0025] When signal sources φ₁and φ₂are disabled, internal nodes N1 and N2 charge through the first and second back gates of transistors MN1 and MN2 respectively, to approximately equal input voltage V_inapplied to first node 11. Similarly, second node 12 charges through the back gates of transistors MP1 and MP2 to approximately equal voltage V_in. When clock φ₁rises from zero volts to V_Ø while the other clock φ₂remains at zero volts, the voltage at node N1 rises from V_into a summed voltage V_Ø+V_in. Because half cell B is driven by first clock φ₁, the gate-source voltage, V_gs, of transistor MN2 increases from zero volts to V_φ, causing transistor MN2 to conduct. Additionally, V_gsfor transistor MP2 increases from zero volts to V_φ and thus transistor MP2 is not conducting. Node N2 is thus connected to first node 11 and may charge to approximately equal voltage V_in. First node 11 also transfers charge to capacitor C2.

[0026] When clock φ₁falls from V_into zero volts, approximately simultaneously clock φ₂rises from zero volts to V_in. The voltage on node N2 rises from V_into V_in+V_φ . Approximately simultaneously, the voltage on node N1 falls from V_in+V_φ to V_in. The gate-to-source voltage V_gsacross transistor MN2 falls from V_φ to −V_φ, and transistor MN2 ceases to conduct. The gate-to-source voltage V_gsacross transistor MP2 approximately simultaneously falls from V_φ to −V_φ, and transistor MP2 begins to, conduct. Thus, the voltage V_in+V_φ on node N2 is coupled to second node 12 through transistor MP2, and charge is transferred from capacitor C2 to the load coupled to second node 12. When clocks φ₁and φ₂change phase, current flow through half cells A and B is reversed. For example, when clock φ₁is high and clock φ₂is low, capacitor C2 is charged from first node 11 through transistor MN2, and capacitor C1 discharges to second node 12 through transistor MP1. When clock φ₁is low and clock φ₂is high, capacitor C1 is charged from first node 11 through transistor MN1, and capacitor C2 discharges to second node 12 through transistor MP2. Thus, second node 12 is constantly charged by one of the capacitors C1 or C2, and if the charge pump is not heavily loaded and the circuit has had time to reach steady state, second node 12 is not subjected to large voltage swings. Charge pump 10 may also operate similarly in reverse to step down the voltage from second node 12 to first node 11, in which case charge is transferred from first node 11 to second node 12 in the manner above described.

[0027] Typically, the amount of current that may be provided to second node 12 for any given output voltage between V_inand V_in+V_φ is proportional to the frequency of signal sources Ø₁and Ø₂. This output current is also proportional to the sum of the capacitance of capacitors C1 and C2. Capacitors C1 and C2 may be equal in value. In some embodiments, it may be desirable for one capacitor to be significantly larger in value than the other. Charge pump 10 may also operate at a higher frequency in order to compensate for lower capacitance values C1 and C2.

[0028] FIG. 2 illustrates a cross section of one embodiment of an n-channel transistor that may be used in the charge pump of FIG. 1. In this embodiment, an n-channel transistor, such as MN1, comprises a junction isolated NMOS transistor such as may be formed in an N-well BiCMOS process. NMOS transistor MN1 may be formed by two shallow heavily doped N-type diffusions (N+ diffusions) 101, 102 in a P-type epitaxial region (P-epi) 125. In this embodiment, source 20 of transistor MN1 comprises N+ diffusion 102, and drain 24 comprises N+ diffusion 101.

[0029] In this embodiment, the back gate of transistor MN1 comprises a P-epi region 125, which may be contacted by a shallow heavily doped P-type diffusion 103 abutting source 102. A PN-junction formed between regions 102 and 103 may create excessive current leakage. To eliminate this leakage, source 102 and back gate 103 may be coupled together electrically, as shown in FIG. 4. The PN-junction between back gate P-epi region 125 and drain N+ diffusion 101 forms a parasitic back gate diode which may be biased into conduction if the source voltage exceeds the drain voltage by more than a few tenths of a volt. Forward-biasing the drain-backgate junction injects minority carriers (electrons) into the P-epi, some of which traverse to the isolation. This current flow represents a loss of charge from the charge pump circuit and reduces the energy efficiency of the charge pump.

[0030] Effective isolation techniques may be used to improve the efficiency of charge pump 10 by eliminating back gate modulation which would otherwise occur if the back gates of NMOS transistors MN1 and MN2 were connected to the substrate.

[0031] The isolated NMOS configuration illustrated in FIG. 2 may also use an additional combination of N-buried layer 111 (or NBL) and deep N+ sinker 112 (or alternatively, an N-well) as an isolation “ring” to isolate the section of P-epi forming the back gate of the transistor. The NBL severs the isolated P-epi tank 125 from the P-substrate 110 beneath, and the deep N+ sinker ring isolates it from adjacent P-epi region 113. In order to create effective isolation, such a ring contains no gaps and the NBL overlaps it sufficiently to allow for misalignment. Such an isolation ring is generally connected to a voltage equal to, and generally not greater than, that applied to the isolated P-epi tank to avoid “punch through” from the drain N+ diffusion through the lightly doped epi. This isolation technique may improve the voltage efficiency of charge pump 10 by minimizing voltage drops across the MOS transistors (due to the diminution of backgate modulation effects and the consequent reduction in effective gate drive voltage), and may improve energy efficiency by blocking minority carrier flow to the substrate, thereby minimizing parasitic transistor action. N-channel transistors MN1 and MN2 may also be oxide-isolated. Many manufacturing methods for oxide or dielectric isolation are known to the industry.

[0032] FIG. 3 illustrates a cross section of one embodiment of a p-channel; transistor, such as MP1, that may be used in the charge pump of FIG. 1. A p-channel transistor comprises an isolated PMOS transistor such as may be formed in an N-well BiCMOS process. PMOS transistor MP1 may be formed by two shallow, heavily doped P-type diffusions or P+ diffusions 201, 202 in a deep, lightly doped N-well well diffusion. In this embodiment, source 36 of transistor MP1 comprises P+ diffusion 201, and drain 32 comprises P+ diffusion 202. P-channel transistors MP1 and MP2 may also include oxide or dielectric isolation without departing from the scope of the invention.

[0033] In this embodiment, the back gate of transistor MP1 comprises an N-well diffusion 203, which is terminated upon its bottom surface by N-buried layer (NBL) 204, and upon all sides by a ring of deep-N+ sinker diffusion 205. The backgate is coupled via a shallow N+ diffusion 206 residing within, and electrically coupled to, the deep-N+ sinker 205. The NBL layer 204 resides at the interface between a lightly doped P-type epitaxial layer (P-epi) 208 deposited upon a lightly doped P-type substrate 207. The back gate of transistor MP1, comprising N-type regions 203, 204, 205, and 206, is electrically isolated from the P-epi 208 and P-substrate 207 by a reverse-biased PN-junction.

[0034] The PN-junction between drain diffusion 201 and N-well 203 may be biased into conduction when the back gate voltage exceeds the drain voltage by more than a few tenths of a volt. For example, in FIG. 4, the source and backgate are electrically common, and the PN-junction between the drain and backgate of transistor MP1 as illustrated in FIG. 4 will become forward-biased if the source voltage exceeds the drain voltage by more than a few tenths of a volt.

[0035] Deep-N+ sinker 205 and NBL 204 may be used to minimize current loss to the substrate through parasitic substrate PNP transistor action. For example, deep-N+ sinker 205 and NBL 204 may repel minority carriers back into the N-well (due to the built-in potential and consequent electric field present at the N+/N− interface). Any minority carriers that surmount the electrical potential across the N+/N− interface may recombine in the heavily doped deepN+ sinker or NBL before they can reach the P-epi or P-substrate. The deep-N+ sinker and NBL may reduce the escape of minority carriers from the N-well, and thereby reduce parasitic transistor action and concomitant loss of current to the substrate.

[0036] FIG. 4 illustrates a schematic diagram of another embodiment of a charge, pump constructed in accordance with the present invention. This embodiment of charge pump 10 comprises resistors R1 and R2, in addition to the elements comprising the embodiment as shown in FIG. 1. Resistor R1 couples source 36 of transistor MP1 to second node 12. Similarly, resistor R2 couples source 42 of transistor MP2 to second node 12.

[0037] Such resistors typically limit the voltage drop encountered across the parasitic back gate diodes of each of the transistors, and prevent such diodes from conducting. The use of such resistors thus may improve the voltage and energy efficiency of a charge pump by limiting losses due to parasitic transistor action and loss of current to the substrate. One example for such an implementation includes utilizing standard BiCMOS processes with junction isolation, with or without the additional isolation rings as shown and discussed in conjunction with FIG. 3.

[0038] Resistors R1 and R2 are in series with each p-channel transistor MP1 and MP2. Other means for providing equivalent resistance are known to those skilled in the art. For example, resistors R1 and R2 may be manufactured within each transistor MP1 and MP2. Additional resistors R3 and R4 (not explicitly shown) may also similarly couple the source of transistor MN1 and the source of transistor MN2 to first node 11 without departing from the scope of the invention.

[0039] FIG. 5 illustrates a block diagram of a multi-stage charge pump constructed in accordance with the present invention. Charge pump 500 is connected between a first node N4, fed by a power source 515, and a second node N5, which is coupled to additional circuitry 520. Cascading a plurality of charge pumps 510, 511 enables higher voltages to be achieved than in a single stage charge pump. Charge pump 500 may utilize alternative schemes to couple charge pumps 510 and 511 to step up the input voltage at node N4 to produce a larger voltage at node N5, or to step down the input voltage to produce a smaller (or more negative) output voltage. As an input, power source 515 may comprise any suitable source of electrical energy, such as a voltage source or a battery. Additional circuitry 520 may be any type of circuitry whether analog or digital.

[0040] Charge pumps 510, 511 may be of the type illustrated in FIG. 1, and provide a step-up in voltage between power source 515 and node N5. In that case, first charge pump 510 provides an output voltage on node N3 of V_in+V_Ø, and on output node N5, a voltage of V_in+V_Ø+V_Ø. Any number of charge pump stages may be included without departing from the scope of the invention.

[0041] Each charge pump 510, 511 may be configured with capacitor values C1 and C2 as discussed in conjunction with FIG. 1. Each charge pump 510, 511 may operate independently. For example, independent signal sources φ₁and φ₂may comprise peak-to-peak voltages, duty cycles, phase relationships and/or frequencies of operation that may vary from one charge pump to the other. Furthermore, charge pumps 510, 511 may or may not be configured identically.

[0042] Transistors MN1, MN2, MP1, and MP2 as used in each charge pump 510, 511 may be isolated as described in conjunction with FIGS. 2 and 3. The back gates used in transistors MN1, MN2, MP1 and MP2 are isolated independently between charge pumps 510, 511.

[0043] FIG. 6 illustrates a schematic diagram of a current-limited charge pump 613 constructed in accordance with the present invention. Current-limited charge pump 613 comprises a current source 611 coupled between voltage source 610 and a first node 612 that limits a current Is drawn from voltage source 610 into node 612. Current-limited charge pump 613 further comprises charge pump 600 coupled to and drawing current I₁from first node 612, and coupled to and providing current I₂to second node 620. Recirculation circuit 650 is also coupled to and provides a current I₃to second node 620 and is also coupled to and draws a current I₄from first node 612. Current-limited charge pump 613 delivers a current I_Lto load 635. Charge pump 600 may provide a voltage V₂to second node 620 that is less than or equal to a voltage V₁at first node 612, or vice-versa. Currents I₁-I₄flow from a higher voltage potential to a lower voltage potential, although illustrated in FIG. 6 as flowing from left-to-right. Any charge pump 600, including those discussed in conjunction with FIGS. 1-4, may be used without departing from the scope of the invention. As will be discussed in further detail below, charge pump 600, current source 611 and recirculation circuit 650 may combine to limit the current provided to load 635 to a known value. For example, current may be limited to no more than a maximum, current provided by current source 611, and to no less than a minimum current provided by charge pump 600.

[0044] In general, a charge pump 600 provides an output current I₂limited by the values of its internal capacitors, the frequencies at which its switches are operated, and the magnitudes of the voltage (and current) losses within these switches. Furthermore, current 12 may vary drastically depending upon the voltage differential across the charge pump. On the other hand, charge pump 613, which comprises current source 611 and recirculation circuit 650, may minimize variations in load current I_Lcaused by variations in charge pump output current I₂.

[0045] To illustrate the operation of charge pump 613, consider a charge pump 600 that receives a voltage V₁from first node 612 and provides a higher voltage V₂to second node 620, and where the current I₁drawn by charge pump 600 from first node 612 approximately equals the current I₂delivered by charge pump 600 to second node 620. Where voltage V₁is less than V₂, which is less than the voltage V_Sdelivered by voltage source 610, current source 611 limits current I_Sto first node 612. By Kirchoff's current law, I_S=I₁+I₄, and I_L=I₂+I₃. In the embodiment currently under discussion for charge pump 600 and recirculation circuit 650, I₂approximately equals I₁, while I₄approximately equals I₃, and thus (I₂+I₃) approximately equals (I₁+I₄), and I_Lapproximately equals I_S. In this embodiment, charge pump 613 limits the current I_Lprovided to load 635 to approximately I_S.

[0046] As charge pump 600 draws current I₁from first node 612 and delivers current I₂to second node 620, current I₁may exceed the current I_Sprovided by current source 611, and voltage V₁on first node 612 will begin dropping. Without recirculation circuit 650, voltage V₁may drop low enough to impair proper operation of charge pump 600 and current source 611, by (for example) forward-biasing isolation junctions. Recirculation circuit 650 may desirably avoid this situation by conducting current from second node 620 to first node 612 when voltage V₁at first node 612 drops below a threshold value. Because (I₂+I₃) is approximately equal to (I₁+I₄), load current I_Lshould approximately equal current I_S.

[0047] As voltage V₂further increases, the voltage differential (V₂−V₁) across charge pump 600 increases and the amount of current I₁that it draws diminishes, eventually to a value below I_S. Voltage V₁on first node 612 rises until it approaches voltage V_Sdelivered by voltage source 610. The voltage differential (V_S−V₁) across current source 611 decreases sufficiently that current source 611 cannot deliver its full rated current, and I_Sdecreases to match I₁.

[0048] In embodiments where charge pump 600 draws a current I₁unequal to current I₂, recirculation circuit 650 should operate to provide a ratio I₃/I₄that is approximately equal to the ratio I₂/I₁, or N. In this embodiment, (I₂+I₃) is approximately N*(I₁+I₄), and I_Lis approximately N*I_S, regardless of the relative magnitudes of I₁and I₄.

[0049] For step-up charge pumps whose unloaded output voltage V₂asymptotically approaches a value larger than their input voltage V₁, it may be desirable for recirculation circuit 650 to provide the following relationships:

[0050] (1) when V₂>V₁and V₁>V_thres, then I₄approximately=I₃approximately=0;

[0051] (2) when V₂>V₁and V₁<V_thres, then I₄approximately=I₃/N; and

[0052] (3) when V₂<=V₁, then I₄approximately=I₃/N, or I₄approximately=0.

[0053] In these equations, V_thresrepresents a lower bound on voltage V₁chosen to ensure proper operation of current limited charge pump 613, and N approximately equals the ratio of charge pump output current I₂to charge pump input current I₁.

[0054] Similarly, for step-down charge pumps whose unloaded output voltage V₂asymptotically approaches a value smaller than their input voltage V₁, it may be desirable for recirculation circuit 650 to provide the following relationships:

[0055] (1) when V₂<V₁and V₁<V_thres, then I₄approximately=I₃approximately=0;

[0056] (2) when V₂<V₁and V₁>V_thres, then I₄approximately=I₃/N; and

[0057] (3) when V₂>=V₁, then I₄approximately=I₃/N or I₄approximately=0.

[0058] In these equations, V_thresrepresents an upper bound on voltage V₁chosen to ensure proper operation of current limited charge pump 613, and N approximately equals the ratio of charge pump output current I₂to charge pump input current I₁.

[0059] FIG. 7 illustrates an exemplary recirculation circuit 650 for a step-up charge pump 600, utilizing the teachings of the present invention. In this implementation, recirculation circuit 650 comprises an NMOS transistor 800 whose source couples to first node 612, whose drain couples to second node 620 and whosle gate couples to the output of voltage source 610 at node 614. When V₂<=V₁, V₁is also less than V_S, and NMOS transistor 800 conducts current from first node 612 to second node 620. Because the ratio of drain current to source current will approximately equal one, recirculation circuit 650 may be used with charge pumps whose output current approximately equals their input current.

[0060] When V₂>V₁, transistor 800 conducts current from second node 620 to first node 612 when voltage V₁at first node 612 is less than (V_s−V_t) where V_tis the threshold voltage of NMOS transistor 800. An NPN transistor or any form of N-channel field effect transistor may be substituted for NMOS transistor 800. It is also within the scope of the invention for a step-down charge pump 600 to utilize a PMOS transistor, a PNP transistor or any form of P-channel field effect transistor instead of NMOS transistor 800.

[0061] FIG. 8 illustrates another exemplary recirculation circuit 650 utilizing the teachings of the present invention. In this implementation, recirculation circuit 650 comprises an avalanche diode 801 whose anode couples to first node 612 and whose cathode couples to second node 620. This configuration may be used with step-up charge pumps 600 where output current I₂is approximately equal to input current I₁.

[0062] When V₂<V₁−V_f, where V_fis a forward voltage of the diode, diode 801 conducts current from first node 612 to second node 620. When V₂>V₁, diode 801 conducts current when V₂>V₁+V_r, where V_ris the reverse breakdown voltage of diode 801. When the output of the charge pump is initially started from a voltage approximately equal to V_s, a diode 801 whose V_r<V_smay prevent voltage V₁from falling below zeros which may be desirable when charge pump 600 and current source 611 are constructed using junction isolation. As another example, for step-down charge pumps, the cathode of diode 801 may couple to first node 612 and the anode of the diode may couple to second node 620.

[0063] Use of cascaded charge pumps 600A, 600B (not explicitly shown) as a multi-stage charge pump may also be used to enable higher voltages to be obtained, as discussed in conjunction with FIG. 5. Recirculation circuit 650 may couple to nodes N4 and N5 as illustrated in FIG. 5, and cascaded charge pumps 600A, 600B may provide an output voltage on second node N5 of V_in+V_Ø+V_Ø. Any number of charge pump stages may be used without departing from the scope of the invention.

[0064] While the invention has been particularly shown and described by the foregoing detailed description, it will be understood by those skilled in the art that various other changes in form and detail may be made without departing from the spirit and scope of the invention.