Slew rate control of output drivers using PVT controlled edge rates and delays

[0002] The present invention relates generally to integrated circuit pad circuits, and more particularly to controlling the slew rate of output drivers using external resistance and programmed delays.

[0003] Integrated circuits communicate with one another using digital signals. In the digital world, a digital signal may be in one of a plurality of predefined quantized states. Because digital signals are transmitted using an analog signal along a transmission line, the predefined quantized states of the digital signal are represented by different ranges of voltages within the total voltage range of the signal. For example, a typical digital integrated circuit (IC) based on a binary system will communicate using two states—zero (“0”) and one (“1”). The digital state of “0” is represented by the range of voltages between a minimum voltage V_MIN(e.g., 0 volts) of the potential voltage range of the signal and a voltage V_LOWthat is low relative to the total range of voltage, whereas the digital state of “1” is represented by the range of voltages between a voltage V_HIGHthat is high relative to the total range of voltages and a maximum voltage V_MAX(e.g., 1.5 volts) of the potential voltage range of the signal. In binary system example, the state of the digital signal is unknown when the voltage level of the signal is between V_LOWand V_HIGH, which typically only occurs during transitions of the signal from either the “0” state to the “1” state or vice versa.

[0004] Because the transmission signal is actually analog, the transition between digital states does not occur instantaneously, but instead occurs over a period of time T_TRANSITIONthat is dependent on the physical conditions present on the transmission line. It is well known that signal transitions over a transmission line will suffer a delay known as a propagation delay due to the parasitic resistance, inductance, and capacitance of the line. This delay increases with the length of the line. In addition, it is also well-known that unless the impedance of the transmission line matches that of the load it drives, the signal will degrade because the mismatch in impedance leads to reflections from the load that are passed back to the driver circuit, which may then be re-reflected causing further signal degradation.

[0005] Furthermore, when the driver circuit drives multiple loads with differing impedances, the transmission line requires multiple stubs to properly match each of the loads during realtime operation. However, the use of multiple stubs then generates multiple reflections. One way of ensuring proper detection of signal states is to control the edge rates of the signal.

[0006] However, this competes with the trend towards ever increasing signal frequencies, which results in higher edge rates. Accordingly, a need exists for a technique for controlling the slew rate of signal edge transitions without sacrificing the signal frequency.

[0007] The present invention is a method and circuit for controlling the slew rate of integrated circuit output drivers by controlling the resistance of a pre-driver circuit that generates the drive signal.

[0008] In particular, the present invention allows the ability to vary the slew rate of the signal on the output pad by controlling the current flow through a set of pre-driver FETs that driver the output stage FETs. In a preferred embodiment, this is accomplished using a programmable resistance pre-driver circuit to drive the output stage of the output driver. The slope of the pre-driver signal driving the output stage FETs is controllable by varying the source resistance of the pre-driver FETs.

[0009] In addition to controlling the slew rate of the output signal, the use of the programmable resistance pre-driver circuit may also be advantageous to overcome chip-to-chip parameter differences due to variations in voltage, temperature, and manufacturing process.

[0010] For even slower slew rate requirements, the invention may also implement a staged turn-on of the output driver legs. This gives a slower possible output slew rate than possible with edge rate control of a single output driver leg alone.

[0011] The invention will be better understood from a reading of the following detailed description taken in conjunction with the drawing in which like reference designators are used to designate like elements, and in which:

[0012] FIG. 1 is a block diagram of a slew rate controlled output driver circuit in accordance with the present invention;

[0013] FIG. 2 is an operational flowchart of a method in accordance with the invention;

[0014] FIG. 3 is a schematic diagram of a slew-rate controlled output driver circuit implemented in accordance with the invention;

[0015] FIG. 4A is a gate-voltage-vs.-timing diagram illustrating a pre-drive signal due to operation of the invention of the first embodiment of the invention shown in FIG. 3;

[0016] FIG. 4B is a gate-voltage-vs.-timing diagram illustrating the output signal corresponding to the pre-drive signal of FIG. 4A using the first embodiment of the invention shown in FIG. 3;

[0017] FIG. 5 an alternative embodiment of a slew rate controlled output driver circuit implemented in accordance with the invention;

[0018] FIG. 6A is a gate-voltage-vs.-timing diagram illustrating a pre-drive signal due to operation of the invention of the second embodiment of the invention shown in FIG. 5; and

[0019] FIG. 6B is a gate-voltage-vs.-timing diagram illustrating the output signal corresponding to the pre-drive signal of FIG. 4A using the second embodiment of the invention shown in FIG. 5.

[0020] A novel method and circuits for controlling the slew rate of an output signal by an output driver is described in detail hereinafter. Although the invention is described in terms of specific illustrative embodiments, such as specific output driver designs, it is to be understood that the embodiments described herein are by way of example only and that the scope of the invention is not intended to be limited thereby but is intended to extend to any embodiment that controls the output signal edge rate by adjusting the resistance in the pre-driver circuit to control the slope of the pre-drive signal.

[0021] Turning now to a general preferred embodiment, FIG. 1 depicts a slew rate controlled output driver circuit 10 implemented in accordance with the present invention. As will be described in detail hereinafter, the slew-rate controlled output driver circuit 10 provides the functionality for controlling the slew rate of the signal driven onto the output pad by controlling the source resistance of the pre-drive devices.

[0022] As known in the art, a typical output driver will include a pull-up circuit S1 and a pull-down circuit S2. Each circuit S1 and S2 will typically include a pre-driver stage 28 (typically implemented with an inverter 12, 22 comprising a pair of complementary CMOS devices 14,15 and 24, 25 respectively) having an input coupled to receive a data signal DATA 11. The pre-driver stage 28 produces pull-up and pull-down pre-drive signals 13, 23 used to control output driver pull-up and pull-down devices 16 and 26 in an output stage 29 which drives the output pad 18 to a high voltage level, a low voltage level, or possibly a tri-state voltage level. Typically the output stage 29 employs a pair of complementary CMOS devices 16 and 26 each having a gate connected to receive a respective pre-drive signal 13, 23, a drain connected to the output pad 18, and a source connected to alternate ones of either a high voltage source (e.g., V_DD) or a low voltage source (e.g., V_SSor ground), as shown. In accordance with the invention, the pre-driver stage 28 is responsive to a pre-driver resistance control circuit 20 which adjusts the source resistance of the pre-driver stage devices 14, 15, and 24, 25 to speed up or slow down the rate of current flow through the pre-driver stage devices 14, 15, and 24, 25, respectively, in order to adjust the slew rate of the respective pull-up and pull-down pre-drive signals 13 and 23. This in turn directly affects the slew rate of the signal on the output pad 18.

[0023] FIG. 2 is an operational flowchart of the general method of the invention. In accordance with the method, shown generally at 30, in a step 32, the source resistance of the pre-driver device(s) is varied to adjust the rate of current flowing through the pre-drive device(s) to achieve a desired slew rate of the edge(s) of the pre-drive signal(s). The slew-rate-controlled pre-drive signal(s) are then used to drive 34 the output driver device(s).

[0024] Turning now to a specific embodiment, FIG. 3 depicts an output driver 100 implemented in accordance with the principles of the present invention. As illustrated, the output driver 100 includes three stages: a programmable current source 131, a pre-driver stage 132, and an output stage 133.

[0025] Output stage 133 includes output drive devices 127 and 128 connected respectively between a high voltage source V_DDand low voltage source (ground) and an output pad 118. The output drive devices 127 and 128 are controllable via the pre-driver stage 132 to drive output pad 118 which is connected to a load 117 having a characteristic impedance of Z_O. Programmable current source 131 determines a composite source impedance for the pre-driver stage 132. The composite source impedance of the pre-drive devices can be separated into a value R_SC(the source resistance while charging) and a value R_SD(the source resistance while discharging). Generally speaking, it is desirable that R_SCand R_SDbe equal to each other, although one can imagine that there might be special circumstances that would require them to be different. R_SCand R_SDmay be varied, as described in detail hereinafter, to alter the slope of the output on the pre-driver stage 132. If R_SC/R_SDis increased, the slope of the output 116, 126 of the predriver stage 132 decreases and therefore the transition time of the pre-drive signal 116, 126 increases. Conversely, if R_SC/R_SDis decreased, the slope of the output 116,126 of the predriver stage 132 increases and therefore the transition time of the pre-drive signal 116,126 decreases.

[0026] In the pull-up portion of the circuit, the pre-driver circuit 132 comprises four CMOS devices 112-115 in series. Devices 113 and 114 act as switches to respectively pull up (charge to V_DD) and pull down (discharge to ground) the pre-drive signal on line 116 that drives the pull-up device 127 of the output stage. It will be understood that switching devices 113 and 114 are driven on and off in suitable alternation in accordance with the desired output waveform (which represents the bit pattern of the data being output), and that although both devices 113 and 114 may be off to tri-state pre-drive signal on line 116, both devices will never be on at the same time. Device 112 acts as a resistance of programmable value to combine with the very low on-resistance of device 113 to produce R_SC. Similarly, device 115 acts as a resistance of programmable value to combine with the relatively low on-resistance of device 114 to produce R_SD. The resistance of device 112, having generally equal transconductance as device 115, is controlled by the value of the voltage PGATE 120, while in similar fashion the resistance of device 115 is determined by the value of the voltage NGATE 119.

[0027] Turning now to the programmable current source 131, a voltage V_REF102 is derived from V_DDby a voltage divider including two resistive devices 103 and 104, that are connected in series between V_DDand GND. The geometry of these two devices is chosen to produce, for a V_DDof say, 3.3 V, a V_REFof 1.8 V.

[0028] An external programming resistor R_PROG107 is connected between an external source of V_DDand a terminal 109 of the chip, characterized by voltage V_PROG. The voltage V_PROGis produced by a feedback controlled voltage divider formed by the external programming resistor R_PROG107 and an N-type device 108 having a drain connected to terminal 109 and a source connected to ground. V_PROGand V_REFare applied to an error amplifier 106 (an operational amplifier of suitable gain) whose output is the signal NGATE 119. NGATE is applied to the gate of n-type device 108.

[0029] In operation, V_PROGequals V_REF, within the error limits of the feedback loop. A gain of forty in the error amplifier 106 is a reasonable gain and will keep V_PROG109 within, say, 50 mv of V_REF. Second, the characteristics of device 108 are included in the feedback loop. This means that the gate voltage V_GSN(which is also NGATE 119) varies as needed to advantageously null variations in V_PROGthat are due parameter shifts in device 108 arising from temperature and process variations. Thus, NGATE varies in a way that can be used to supply compensation to other devices that experience generally identical parameter shifts for those same process and temperature excursions.

[0030] So, for example, if device 108 is considered “fast” (i.e., the current through the device is relatively large for a given V_GSN) compared to a hypothetical design center device, the voltage V_PROGwill tend to be lower than it would otherwise be (which is set at V_REFby the feedback loop). (Presumably, devices 112 and 115 will also be “fast”, which causes them to exhibit decreased values for R_SCand R_SD, which is undesirable.) However, if V_PROGdecreases below V_REF, the error amplifier will decrease the value of NGATE and raise the resistance of device 108 to increase V_PROGback to near V_REF. As will be seen, decreasing the value of NGATE increases the resistance of devices 112 and 115. This is what is wanted, since they are also “fast”, having been fabricated in the same process, and would otherwise then presumably operate with a resistance lower than desired. Similar examples are obtained for “slow” devices, as well as for shifts produced by temperature excursions.

[0031] Accordingly, by including device 108 in the feedback loop for V_REF, variations in NGATE are produced that can be used for compensation of deviation away from a programmed value of source impedance.

[0032] Returning now to the novel aspects of the invention, R_PROGmay be varied to adjust the source resistances R_SCand R_SDaffecting the rate of current flow by pre-drive drive devices 113 and 114. The slew rate of the pre-drive signal 116 can be adjusted by programming the source resistance R_SCand R_SDof the pre-drive devices 113, 114 which affects the rate of charge/discharge of the pre-drive signal 116. By slowing down the charge/discharge rate of the pre-drive signal 116, it takes longer to charge/discharge the gates of the output signal drive device 127, and therefore increases the amount of time over which the output signal drive device 127 conducts in the linear region before reaching saturation/pinch-off. As known in the art, the range of the gate-to-source voltage V_GSdefining the linear region of a FET device is small, and the drain current increases linearly with the drain-to-source voltage V_DSup to a saturation voltage V_DS(sat)at which point the FET becomes a current source. The slope in the linear region, I_D/V_DS, is proportional to V_GS−V_T. Accordingly, the longer the drive device 127, 128 remains in the linear region, the slower the charge/discharge rates will be on the output pad, and therefore the slower the edge rates of the output signal.

[0033] The pull-down portion of the circuit 100 is similar to the pull-up portion of the circuit, including predriver devices 122-125 connected in series between high-voltage source V_DDand ground and generating a pre-drive signal 126 which drives the gate of drive device 128. The pull-down portion of the circuit operates similarly to the pull-up portion of the circuit, except that the drive device 128 discharges the output pad to ground.

[0034] FIGS. 4A and 4B are timing diagrams illustrating the effect of increasing the source resistance of the pre-drive devices on the edge rate of the output signal. As illustrated, without slew rate control, the pre-drive signal (shown by the dashed line in FIG. 4A) switches quickly, resulting in sharp edge transitions. This leads to sharp edge transitions in the output signal (shown by the dashed line in FIG. 4B). In contrast, with the invention's slew rate control of the pre-drive signal (indicated by the solid line in FIG. 4A), the output signal (indicated by the solid line in FIG. 4B) transitions more slowly.

[0035] Returning again to FIG. 3, the programmable current source 131 is implemented such that devices 108 and 110 comprise a 1:1 current mirror. Device 110 is operated in a region where it tends to behave as a constant current source, where the value of the current is a function of V_GSN(i.e., of NGATE). That is, the current through device 110 (and 111, too) will be I_PROG, but as adjusted (for compensation) by any movement in V_GSNproduced by the error amplifier 6 as it servos V_PROGto track V_REF. Device 111 also operates in a constant current region, and owing to symmetry of construction, it will have the same magnitude gate voltage at a given current as does device 110. Since devices 110 and 111 are connected in series, as constant current sources they produce and share exactly the same current. Thus, the current through device 110 produces, or is accompanied by, gate voltage V_GSP(PGATE) for device 111 that, when referenced to V_DD, corresponds in magnitude and direction of change to V_GSNreferenced to DGND. In other words, devices 110 and 111 operate as a gate voltage mirror. The results are signals NGATE 119 and PGATE 120 whose values are determined in a major fashion according to the value selected for R_PROGand that vary in a minor fashion according to variations in process and temperature.

[0036] The signal NGATE 119 drives the gate of the n-channel FET 115, while the signal PGATE 120 drives the gate of the p-channel FET 112. Devices 108 and 115 also constitute a current mirror with a current ratio proportional to the ratio of the geometries of the devices. For example, suppose that the geometries selected for FET 115 generates a 1:30 mirror. The current that flows through FET 115 (when allowed by device 114 being on) is thirty times the amount of current flowing through device 108 (I_PROG). In this example, the geometries of devices 108 and 112 are chosen to also constitute a 1:30 current mirror. Hence, R_PROGsets I_PROG, which in turn programs and also compensates the values of R_SCfor device 112 and R_SDfor device 115.

[0037] In some applications, for example buses that have a very slow slew rate requirement and/or have multiple loads on the bus, additional measures for slowing the slew rates is needed. FIG. 5 is an alternative embodiment of a slew rate controlled output driver circuit 200 in accordance with the invention that employs multiple staged-turn-on/off output driver drive devices 227, 227a, and 228, 228ato pull up or pull down the output pad 218 of the driver. In this embodiment, output driver 200 includes a programmable current source 231, a pre-driver stage 232, and an output stage 233. The output stage 233 includes a plurality of pull-up devices 227, 227a, connected between a high voltage source V_DDand the output pad 218, and a plurality of pull-down devices 228, 228a, connected between a low voltage source (ground) and output pad 118. The output drive devices 227, 227a, 228, 228a, are each controllable via respective corresponding pre-drive circuits which belong to the pre-driver stage 232 whose respective source impedances R_SCand R_SDare programmable via the programmable current source 231 according to the principles discussed above with respect to the embodiment of FIG. 3. As in the embodiment of FIG. 3, by setting up known ratios between the current mirrored devices (i.e, the ratios between devices 208 and 210, 208 and 211, 208 and 213, and 208 and 212, a single programmable resistor R_PROGmay be used to control the slew rate of each of the pre-drive signals 216, 216a, 226, 226asuch that output drive devices 227, 227a, and 228, 228aturn on/off in a staged manner.

[0038] It will be appreciated that one programmable current source 231 may serve to set, and also maintain through compensation, the drive level (source impedance) of an arbitrary plurality of output driver stages; e.g., for an entire bus. It will further be appreciated that there could easily be multiple arbitrary pluralities of output driver stages, with each such multiple having a source impedance that is independently controlled by an associated separate programmable current source.

[0039] FIGS. 6A and 6B are timing diagrams illustrating the effect of increasing the source resistance of the multi-staged pre-drive devices on the edge rate of the output signal. As illustrated, without slew rate control, the pre-drive signal (shown by the dashed line in FIG. 6A) switches quickly, resulting in sharp edge transitions, leading to sharp edge transitions in the output signal (shown by the dashed line in FIG. 6B). In contrast, with the invention's slew rate control of the pre-drive signal with staged turn-on (indicated by the solid line in FIG. 6A), the output signal (indicated by the solid line in FIG. 6B) transitions more slowly.

[0040] While illustrative and presently preferred embodiments of the invention have been described in detail herein, it is to be understood that the inventive concepts may be otherwise variously embodied and employed and that the appended claims are intended to be construed to include such variations except insofar as limited by the prior art.