Method and apparatus for programmable active termination of input/output devices

1. Technical Field of the Present Invention

The present invention generally relates to programmable active terminations, and more specifically to programmable active terminations of Input/Output devices.

2. Background of the Present Invention

The electronic industry is in a state of evolution spurred by the recent changes in technology which have allowed greater functionality in smaller devices. This has resulted in the explosion of new found uses for such small devices (e.g. medical, monitoring etc.), as well as greater functionality in increasingly smaller electronic devices.

The evolution has caused electronic devices to become an inseparable part of our society. Consumers are now buying and demanding electronic devices which are smaller, more powerful, and faster at unprecedented rates. These demands are constantly driving the electronic industry to exceed limitations which were previously considered unsurpassable.

One such area ripe for improvement is how various electronic devices communicate with one another. For example, the communication between the memory and cpu of a computer. This communication is accomplished using a system data bus. The primary concerns of this type of communication are signal propagation delay and capacitive loading. Propagation delay is minimized by placing the memory as close to the cpu as possible. Capacitive loading is minimized by using point-to-point networks. As the speed of the system data bus approaches one Giga Hertz and beyond, the data transfers must be point-to-point terminated. The point-to-point termination is accomplished by terminating each signal trace to a reference voltage via the use of resistors (“terminating resistors”).

The implementation of the point-to-point termination while minimizing signal propagation delay and capacitive loading has resulted in requiring the terminating resistors to reside on the chip itself (“on-chip”). Since the termination resistors are critical to both the timing and integrity of transmitted signals, they must be highly accurate (i.e. resistance values will not vary when exposed to changes in process and/or temperature). Unfortunately, the terminating resistors also require a considerable amount of space and consume a significant amount of Direct Current (DC) power.

It would therefore be a distinct advantage to have a method and apparatus for providing on-chip termination resistance that would be highly accurate, occupy less space, and consume less DC current than current terminating resistors. The present invention provides such a method and apparatus.

The present invention is a method and apparatus for providing programmable active termination of transmission lines.

The present invention will be better understood and its numerous objects and advantages will become more apparent to those skilled in the art by reference to the following drawings, in conjunction with the accompanying specification, in which:

FIG. 1 is a diagram of a data processing system in which the present invention can be practiced;

FIG. 2 is a diagram of a computer card that may reside in the processor of FIG. 1 according to the teachings of the present invention;

FIG. 3 is a schematic diagram illustrating an example of how a conventional unterminated point-to-point network configuration for communication between the CPU and memory of FIG. 2 can be implemented;

FIGS. 4A-B are simulations of how the distortions of the voltage wave form at both the far and near ends, respectively of the transmission line of FIG. 3 may appear during high frequency operation;

FIG. 5 is a schematic diagram illustrating an example of how an ideal terminated point-to-point network configuration for communication between the cpu and memory of FIG. 2 can be implemented;

FIGS. 6A-B are simulations of the how the voltage wave form at both the near and far ends, respectively of the transmission line of FIG. 5 may appear during high frequency operation;

FIG. 7 is a circuit diagram of how an active terminated point-to-point network configuration for communication between the cpu and the memory of FIG. 2 can be implemented according to the teachings of the present invention;

FIGS. 8A-B are simulations of the voltage wave form at both the near and far ends, respectively of the transmission line of FIG. 7 according to the teachings of the present invention;

FIG. 9A is a timing diagram illustrating the voltage waveforms of FIGS. 6B (ideal termination) and 8 B (active termination) superimposed for comparison upon one another according to the teachings of the present invention;

FIG. 9B is a timing diagram illustrating the current waveforms for both the active termination network of FIG. 7 and the ideal termination network of FIG. 5 superimposed upon one another in accordance with the teachings of the present invention;

FIG. 10 is a circuit diagram of a preferred embodiment for active parallel termination (APT) that has been incorporated into an off-chip driver circuit according to the teachings of the present invention;

FIG. 11 is a circuit diagram of a programmable quad-state buffer constructed according to the teachings of the present invention; and

FIG. 12 is a circuit diagram of a active terminator that can be used in uni-directional networks according to a preferred embodiment of the present invention.

In the following description, numerous specific details are set forth, however, it will be obvious to those of ordinary skill in the art that the present invention can be practiced with different details. In other instances, well-know circuits have been shown in block diagram form in order not to obscure the present invention in unnecessary detail. For the most part, details concerning timing considerations and the like have been omitted inasmuch as such details are not necessary to obtain a complete understanding of the present invention, and are within the skills of persons of ordinary skill in the relevant art.

FIG. 1 is a diagram of a data processing system 20 in which the present invention can be practiced. The data processing system 20 includes processor 22 , keyboard 82 , and display 96 . Keyboard 82 is coupled to processor 22 by a cable 28 . Display 96 includes display screen 30 , which may be implemented using a cathode ray tube (CRT) a liquid crystal display (LCD) an electrode luminescent panel or the like. The data processing system 20 also includes pointing device 84 , which may be implemented using a track ball, a joy stick, touch sensitive tablet or screen, track path, or as illustrated a mouse. The pointing device 84 may be used to move a pointer or cursor on display screen 30 . Processor 22 may also be coupled to one or more peripheral devices such as modem 92 , CD-ROM 78 , network adapter 90 , and floppy disk drive 40 , each of which may be internal or external to the enclosure or processor 22 . An output device such as printer 100 may also be coupled with processor 22 .

FIG. 2 is a diagram of a computer card ( 3 ) that may reside in the processor 22 of FIG. 1 according to the teachings of the present invention. Computer card ( 3 ) includes a Central Processing Unit (CPU) ( 1 ) and a memory chip ( 2 ). Communication between the CPU ( 1 ) and the memory chip ( 2 ) is accomplished using data bus wires ( 4 ).

The present invention provides a novel method and apparatus for using active CMOS device structures, or their equivalents, located on chip for providing termination of buses. In describing the present invention, the computer card ( 3 ) of FIG. 2 is used as a convenient and non-obfuscating example, and by no means is to be considered a limitation to the numerous applicability's to which the present invention can be used.

FIG. 3 is a schematic diagram illustrating an example of how a conventional unterminated point-to-point network configuration 300 for communication between the cpu ( 1 ) and memory ( 2 ) of FIG. 2 can be implemented. The network configuration 300 includes an ideal voltage source E 0 302 , an internal resistance of the ideal voltage source R 0 (typically 50 Ohms), a TX 0 306 lossy transmission line, and a capacitance load C 0 308 . The TX 0 306 represents one of the card traces ( 4 ) connecting the cpu ( 1 ) to the memory ( 2 ) and is typically about 50 Ohms with a delay of 70 ps/cm and a length of 8 cm. C 0 308 is typically about 5 Pico Farads (pf) and represents the capacitance load at the far end of TX 0 306 . The operation of the network configuration 300 is described below in connection with low and high frequency parameters.

During low frequency operation, E 0 302 drives a single voltage level for a period of time sufficient so as to allow both the near end and the far end of the transmission line TX 0 306 to settle to a predetermined DC voltage value (either positive or negative power supply level of the driver circuit) absent any DC current flow. This settling typically occurs after one round-trip delay along the signal connection for impedance-matched networks.

During high frequency operation, E 0 302 switches its drive level before the entire (far and near ends) transmission line TX 0 306 has settled to a predetermined DC Level. The premature switching results in a reduced amplitude signal swing and “history” effects (superimposed forward and backward traveling waves) which can cause signal distortion. While the reduced amplitude and signal distortion which occurs may be negligible during continuous switching, significant ringing occurs at the far end of the transmission line TX 0 306 when the near-end driver goes into a high impedance state after switching (R 0 304 of FIG. 3 becomes infinite).

FIGS. 4A-B are simulations of how the distortions of the voltage wave form at both the far and near ends, respectively of the transmission line TX 0 306 of FIG. 3 may appear during high frequency operation. The timing scale (x-axis) is nanoseconds, the voltage scale (y-axis) is volts. The illustrated wave form was produced by switching E 0 302 as a one Giga Hertz sine wave for 2 Nano Seconds, and then causing R 0 304 to increase by eight orders of magnitude, reproducing the effect of placing the driver in the high impedance mode. As illustrated, the transmission line TX 0 306 fails to settle even after 6 Nano seconds of ringing. The amplitude of the ringing is sufficient to cause the receiver at the far end to switch and/or timing skews if the driver at the far end becomes active (e.g. a bi-directional signal network).

Another disadvantage of an unterminated point-to-point network is that the driving circuit at the near end must absorb energy reflected from the far end of the transmission line every time a signal is driven. This near end absorption can cause significant on-chip power supply noise if all ones or zeros are driven. This condition of driving all ones or all zeros is commonly referred to as “simultaneous switching.” Simultaneous switching causes all of the absorbed energy to have a path to the same power supply. The noise generated during simultaneous switching results from the real world constraints of a non-ideal power connection to the driver circuit. Depending upon the particular implementation, the package connection between the driver source node and the stable card-level supply can have an inductance as little as a fraction of a Nano Henry or as much as several Nano Henries.

At very high frequencies, the absorption of these reflections and the resulting noise can be superimposed with the noise resulting from switching the drivers. The magnitude of the total resulting noise can exceed one Volt. This magnitude of noise poses a drastic performance limitation in systems which operate with power supplies of 1.5 Volts or less. This power supply noise will cause a quiet line to have an unacceptable transient voltage when all other signal lines are switched simultaneously (i.e. noise on the power supply is transmitted to the quiet line through the quiet line driver).

Termination techniques have been developed which eliminate the superposition of these two power supply noise sources by causing forward traveling waves to be completely absorbed at the far end of the transmission line via the termination. Thus, in an ideal environment, the near end of the transmission line will not absorb reflected energy.

FIG. 5 is a schematic diagram illustrating an example of how an ideal terminated point-to-point network configuration 500 for communication between the cpu ( 1 ) and memory ( 2 ) of FIG. 2 can be implemented. In similarity to the signal network 300 of FIG. 2, the network configuration 500 includes an ideal voltage source E 1 502 for driving the network 500 , a resistor R 1 504 which represents the internal resistance of the voltage source E 1 502 (typically 50 Ohms), and a TX 1 506 lossy transmission line (typically 50 Ohms with a delay of 70 ps/cm and a length of 8 cm) which represents the card trace ( 4 ) connecting the cpu ( 1 ) to the memory ( 2 ). In addition, Rterm 1 508 and Rterm 2 510 represent ideal passive terminating resistors (each typically 100 Ohms). It should be noted that the signal network 500 represents the best possible network for signal integrity, since both the far-end capacitance and any inductance in the connection from the far end of the transmission line TX 1 506 to the terminating resistors Rterm 1 508 and Rterm 510 have been excluded.

In general, the far end of the terminated network 500 will have a DC voltage bias equal to the power source divided by two (Vdd/2). In addition, the far end of the terminated network will usually have a Thevinen resistance substantially equivalent to that of the TX 1 506 transmission line.

FIGS. 6A-B are simulations of the how the voltage wave form at both the near and far ends, respectively of the transmission line TX 1 506 of FIG. 5 may appear during high frequency operation. The timing scale (x-axis) is nanoseconds, the voltage scale (y-axis) is volts. The illustrated wave form was produced by switching E 1 502 as a one Giga Hertz sine wave for 2 Nano Seconds, and then causing R 1 504 to increase by eight orders of magnitude, reproducing the effect of placing the driver in the high impedance mode. In contrast to the wave forms of FIGS. 4A-B, the wave forms of FIGS 6 A-B settle almost immediately once the driver has been placed in the high impedance mode. As shown, the signal experiences no deformations, and the transmission line TX 1 506 is able to reach a DC voltage level after only one propagation delay (half of the round-trip delay) along the signal trace (TX 1 506 ). The most significant disadvantages of using a resistive termination such as that shown in FIG. 5 are: 1) reduced signal amplitude; 2) DC currents which remain even after the signal network has settled; 3) significant space requirement for placing on a computer card; and 4) the wire connection between the data bus and the termination resistor can be long enough to substantially reduce the effectiveness of the termination.

FIG. 7 is a circuit diagram of how an active terminated point-to-point network configuration 700 for communication between the cpu ( 1 ) and the memory ( 2 ) of FIG. 2 can be implemented according to the teachings of the present invention. In similarity to the signal network 500 of FIG. 5, Network 700 includes an ideal voltage source E 2 702 for driving the network 700 , a resistor R 2 704 which represents the internal resistance of the voltage source E 2 702 (typically 50 Ohms), a TX 2 706 lossy transmission line (typically 50 Ohms with a delay of 70 ps/cm and a length of 8 cm) which represents the card trace connecting the cpu ( 1 ) to the memory ( 2 ). In addition PFET P 0 708 and NFET N 0 710 are included to provide the active termination components. These active termination components 708 and 710 can be located on a cpu chip, memory chip, or any other high speed peripheral device. It should be noted that the design of network 700 has the added benefit of occupying a smaller area in immediate proximity to the on-chip far end terminal of the transmission line TX 2 706 .

In general, the active termination components 708 and 710 have desirable AC characteristics such as their self biasing CMOS structure (i.e. the “turn on” characteristics of one will compensate for the “turn off” characteristics of the other).

FIGS. 8A-B are simulations of the voltage wave form at both the near and far ends, respectively of the transmission line TX 2 706 of FIG. 7 according to the teachings of the present invention. The timing scale (x-axis) is nanoseconds, the voltage scale (y-axis) is volts. The illustrated wave form was produced by switching E 2 702 as a one Giga Hertz sine wave for 2 Nano Seconds, and then causing R 2 704 to increase by eight orders of magnitude, reproducing the effect of placing the driver in the high impedance mode. Since the impedance matching of network 700 is not ideal, a small signal reflection (less than 50 milli Volts) occurs at the far end after the driver at the near end has gone into a high impedance mode. The DC biasing for the network 700 is correct for a parallel-terminated signal network (i.e. 0.75 V for a VDDQ output supply of 1.5V), and the network only requires approximately a single cycle (1 nano second) to settle completely.

FIG. 9A is a timing diagram illustrating the voltage waveforms of FIGS. 6B (ideal termination) and 8 B (active termination) superimposed for comparison upon one another according to the teachings of the present invention. As shown, the data signals of the active termination are nearly identical in amplitude and timing to the data signals of the ideal termination.

FIG. 9B is a timing diagram illustrating the current waveforms for both the active termination network 700 of FIG. 7 and the ideal termination network 500 of FIG. 5 superimposed upon one another in accordance with the teachings of the present invention. As shown, the active termination network 700 of FIG. 7 consumes less DC power without compromising AC performance. More specifically, the ideal termination network 500 of FIG. 5 allows 7.54 milli Amps to flow from VDDQ to ground when the driver is in the high impedance state. In contrast, the active termination network 700 of FIG. 7 only allows 3.57 milli Amps to flow from VDDQ (output supply) to ground when the driver is in high impedance state. The active termination network 700 has a decreased power consumption of over 50 percent in comparison to the ideal network 500 . The decreased power consumption is ideal for allowing on-chip termination without exceeding the thermal limitations of the chip's package.

Decreasing the power consumption without affecting the overall performance is made possible by DC biasing of the active devices (P 0 708 and N 0 710 ) such that their AC performance very nearly reproduces that of 100 Ohm passive resistors.

FIG. 10 is a circuit diagram of a preferred embodiment for active parallel termination (APT) that has been incorporated into an off-chip driver circuit 1000 according to the teachings of the present invention. Normally, an off-chip driver has three states, driving a low (ZERO)—P 13 off, N 23 on, driving a high (ONE)—P 13 on, N 23 off, and high impedance (HIZ)—P 13 off, N 23 off. Incorporating active parallel termination into an off-chip driver circuit introduces a fourth state—P 13 partially on, N 23 partially on. This new circuit structure is best referred to as a “quad-state buffer”. Each of these four modes of operation are listed in Table 1 and described in more detail below.

In the first mode (ZERO), OE and DATA are both high. OE high causes the output of NOR gate NR 0 , node AT, to be low and the output of inverter I 0 , node ATN, to be high. In this condition, zero-threshold NFET devices N 3 and P 3 are on and zero-threshold NFET devices Z 13 and Z 23 are off. Device N 3 couples node N_PASS electrically to node NDRIVE, causing node N_PASS to have a voltage level very nearly equal to that of node NDRIVE as long as zero-threshold device Z 23 remains off. Similarly, device P 3 couples node P_PASS electrically to node PDRIVE, causing node P_PASS to have a voltage level very nearly equal to that of node PDRIVE as long as zero-threshold device Z 13 remains off. The high state on DATA is inverted by inverter 12 which causes the output of NAND gate ND 0 to be high. This high potential on node PDRIVE is driven through zero-threshold device P 3 onto node P_PASS resulting in device P 13 being off. The low on node DATAN and node OEN (OE inverted) causes the output of NOR gate NR 1 to be high. This high potential on node NDRIVE is driven through zero-threshold device N 3 onto node N_PASS which results in device N 23 being on thus causing node OUTPUT_SIGNAL to be at a low state.

In the second mode (ONE), OE is high and DATA is low. OE high has the same affect as described above. DATA low causes the output of NOR gate NR 1 to be low. This low potential on node NDRIVE is driven through zero-threshold device N 3 onto node N_PASS which results in device N 23 being off. Both OE and DATAN being high causes the output of NAND gate ND 0 to be low. This low potential on node PDRIVE is driven through zero-threshold device P 3 onto node P_PASS resulting in device P 13 being on thus causing node OUTPUT_SIGNAL to be at a high state.

In the third mode (HIZ), OE is low and TERMMODE is low. TERMMODE low causes the output of NOR gate NR 0 to be low and the output of inverter I 0 to be high. In this condition, zero-threshold devices N 3 and P 3 are on and zero-threshold devices Z 13 and Z 23 are off. OE low causes the output of NAND gate ND 0 to be high thus keeping device P 13 off, and also causes the output of NOR gate NR 1 to be low thus keeping device N 23 off.

In the fourth mode (APT), OE is low and TERMMODE is high. This causes the output of NOR gate NR 0 to be high which results in zero-threshold devices N 3 and P 3 being off and zero-threshold devices Z 13 and Z 23 being on. Device Z 13 couples node P_PASS electrically to node OUTPUT_SIGNAL, causing node P_PASS to have a voltage level very nearly equal to that of node OUTPUT_SIGNAL as long as zero-threshold device P 3 remains off. Similarly, device Z 23 couples node N_PASS electrically to node OUTPUT_SIGNAL, causing node N_PASS to have a voltage level very nearly equal to that of node OUTPUT_SIGNAL as long as zero-threshold device N 3 remains off. With their gates essentially connected to their drains, devices P 13 and N 23 are only partially on (gate to source voltage of approximately VDD/2) compared to the ONE or ZERO mode of operation, where the on device has a gate to source voltage of VDD. To summarize, signal TERMMODE is used to alter the typical function of the output drivers output enable signal (OE). Instead of placing the driver into a high impedance state (HIZ), when the chip is transitioning from the drive mode to the receive mode, TERMMODE high will allow a low OE to partially keep both drive transistors on, thereby providing an active means of termination.

In order to better facilitate the performance of the active devices, their width must be adjusted to give the correct terminating resistance independent of variations in process, temperature, and power supply voltage. It would be further advantageous if the termination impedance was programmable (i.e. programmable termination device width to match a desired value).

FIG. 11 is a circuit diagram of a programmable quad-state buffer 1100 constructed according to the teachings of the present invention. Each of these four modes of operation are listed in Table 1 and described in more detail below.

In the first mode (ZERO), OE and DATA are both high. OE high causes the output of NOR gate NR 31 , node AT, to be low and the output of inverter I 0 , node ATN, to be high. In this condition, zero-threshold devices Z 19 and Z 20 are on and zero-threshold devices Z 13 and Z 23 are off. Device Z 20 on couples node N_PASS electrically to node N_OFFFSET, causing node N_PASS to have a voltage level very nearly equal to that of node N_OFFSET as long as zero-threshold device Z 23 remains off. Similarly, device Z 19 couples node P_PASS electrically to node P_OFFSET, causing node P_PASS to have a voltage level very nearly equal to that of node P_OFFSET as long as zero-threshold device Z 13 remains off. Node OE high also forces the outputs of NOR gates NRO 1 , NR 11 , NR 21 , NR 31 , NR 41 , NR 51 , NR 61 , and NR 71 low, causing zero-threshold devices Z 10 , Z 11 , Z 12 , Z 13 , Z 22 , Z 23 , Z 24 , and Z 25 to be off and zero-threshold devices Z 14 , Z 16 , Z 18 , Z 21 , Z 26 , and Z 27 to be on. The high state on DATA is inverted by inverter I 2 , node DATAN, which causes the output of NAND gates ND 00 , ND 10 , ND 20 , and ND 30 to be high which in turn causes PFETs P 10 through P 13 to be off. The low on node DATAN and node OEN (OE inverted) causes the output of NOR gate NR 41 to be high. This high potential on node N_OFFSET is driven through zero-threshold device Z 20 onto node N_PASS which results in device N 23 being on. Some combination of nodes N_SMALL, N_MEDIUM, and N_LARGE may be high as well, corresponding to a combination of nodes N, 2 N, and 4 N being high. Which of nodes N, 2 N, and 4 N are high depends on the combination of devices N 20 through N 22 which, when turned on, results in an output impedance of the quad-state buffer 1100 that matches a desired output impedance. This desired output impedance is programmed through the impedance evaluation circuitry 1102 . Nodes P, 2 P, and 4 P are don't cares (x) in this mode.

In the second mode (ONE), OE is high and DATA is low. OE high has the same affect as described above. DATA low causes the output of inverter I 2 , DATAN, to be high. This high potential on node DATAN causes the outputs of NOR gates NR 41 , NR 51 , NR 61 , and NR 71 to be low thus placing NFETs Z 22 through Z 25 in an off state. OE high and node DATAN high also cause the output of NAND gate ND 30 , P_OFFSET, to be low. This low potential on node P_OFFSET is driven through zero-threshold device Z 19 onto node P_PASS resulting in device P 13 being on. Some combination of nodes P_SMALL, P_MEDIUM, and P_LARGE may be low as well, corresponding to a combination of nodes P, 2 P and 4 P being low. Which of nodes P, 2 P, and 4 P are low depends on the combination of devices P 10 through P 12 which, when turned on, results in an output impedance of the quad-state buffer that matches a desired output impedance. This desired output impedance is programmed through the impedance evaluation circuitry 1102 . Nodes N, 2 N, and 4 N are don't cares (x) in this mode.

In the third mode (HIZ), OE is low, TERMMODE is low, and DATA is a don't care. TERMMODE low causes the output of NOR gate NR 31 to be low and the output of inverter I 0 to be high. In this condition, zero-threshold devices Z 19 and Z 20 are on and zero-threshold devices Z 13 and Z 23 are off. OE low causes the output of NAND gates ND 00 , ND 10 , ND 20 , and ND 30 to be high thus causing devices P 10 through P 13 to be off. The low state of OE is also inverted by inverter I 1 causing the output of NOR gates NR 41 , NR 50 , NR 60 and NR 70 to be low thus placing devices N 20 through N 23 in an off state.

In the fourth mode (APT), OE is low and TERMMODE is high. This causes the output of NOR gate NR 31 to be high which results in zero-threshold devices Z 19 and Z 20 being off and zero-threshold devices Z 13 and Z 23 being on. Device Z 13 couples node P_PASS electrically to node OUTPUT_SIGNAL, causing node P_PASS to have a voltage level very nearly equal to that of node OUTPUT_SIGNAL as long as zero-threshold device Z 19 remains off. Similarly, device Z 23 couples node N_PASS electrically to node OUTPUT_SIGNAL, causing node N_PASS to have a voltage level very nearly equal to that of node OUTPUT_SIGNAL as long as zero-threshold device Z 20 remains off. In a similar fashion, OE low and TERMMODE high will couple the drains and gates together of the remaining programmable output devices (P 10 , 11 , 12 ,N 20 , 21 , 22 ), depending on whether the respective programmable inputs (P, 2 P, 4 P,N, 2 N, 4 N) are active. For example, if 2 P is active and 4 P is inactive during the termination mode, then the output of nor gate NR 11 will be high, coupling the gate of P 11 to the output signal and thereby allowing P 11 to contribute to the termination; nor gate NRO 1 will provide a low on node TP_Large, preventing device P 10 from providing current for the termination. This allows the termination to be precisely controlled by either the impedance evaluation circuitry, or possibly by direct user control (test circuit inputs for example, not shown here but common in the art).

It should be apparent to those of ordinary skill in the relevant art that the concepts behind the present invention can be extended to provide different values of termination by simply activating various combinations of the programmable impedance inputs. Further, by simply providing different termination enable signals for the pull-up and pull-down devices, the type of termination provided can be altered from the common mid-level termination scheme (discussed above) to either a simple pull-up or pull-down termination scheme.

FIG. 12 is a circuit diagram of a active terminator 1600 that can be used in uni-directional networks according to a preferred embodiment of the present invention. The active terminator 1600 includes NFETs N 0 -N 1 , PFETs P 0 -P 1 , zero-threshold NFETs Z 0 -Z 1 , and inverter I 0 . When node Enable is high, NFETs Z 0 -Z 1 are active and PFET P 1 and NFET N 1 are off. This results in the shorting of the gates of PFET P 0 and NFET N 0 to their common drains, thus forming a structure electrically equivalent to the active termination network 700 of FIG. 7 . When node Enable is low, PFET P 1 and NFET N 1 are active while devices NFETs Z 0 -Z 1 are off, causing the active terminator 1600 to be in a high impedance state. Obviously, the active terminator 1600 could be modified in manner similar to that illustrated in FIG. 11 in order to be programmable (i.e. feedback controlled).

It is thus believed that the operation and construction of the present invention will be apparent from the foregoing description. While the method and system shown and described has been characterized as being preferred, it will be readily apparent that various changes and/or modifications could be made wherein without departing from the spirit and scope of the present invention as defined in the following claims.