The present invention relates in general to transmission line terminators, and more particularly, to a digital filter circuit and method for blocking transmission line reflection signals arising from an incomplete termination impedance being coupled to an end of the transmission line.
In order to properly transmit data in a network, reflection signals on the transmission lines of the network need to be managed. Reflection signals are typically avoided by terminating a transmission line at both its input and output with the line's characteristic impedance. One disadvantage of using termination of both transmission line ends, however, is the amount of power that is consumed by the termination resistors.
The shortcomings of the prior art are overcome and additional advantages are provided through the provision of a blocking filter for a device receiving signals from a transmission line. The blocking filter includes a pulse generator for generating a masking pulse timed and of sufficient duration to block at the device a reflection signal within a received signal from the transmission line, wherein the reflection signal results from incomplete termination of the transmission line. The filter further includes logic for combining the masking pulse with the received signal, such that the reflection signal is blocked by the masking pulse at the blocking filter.
In another aspect, a network is provided which includes a bus system having a transmission line which is incompletely terminated, wherein a reflection signal arises with transmission of a signal across the transmission line. The network further includes a device connected to the transmission line for receiving the signal. The device includes a blocking filter having a pulse generator for generating a masking pulse timed and of sufficient duration to block at the device the reflection signal, and logic for combining the masking pulse with the signal received from the transmission line, wherein the reflection signal is blocked by the masking pulse.
In a further aspect, a method of filtering received signals from a transmission line is provided. The method includes: generating a masking pulse timed and of sufficient duration to block at a device connected to the transmission line a reflection signal within the received signal from the transmission line, the reflection signal resulting from incomplete termination of the transmission line; and combining the masking pulse with the received signal, wherein the reflection signal is blocked by the masking pulse.
Additional features and advantages are realized through the techniques of the present invention. Other embodiments and aspects of the invention are described in detail herein and are considered a part of the claimed invention.
The subject matter which is regarded as the invention is particularly pointed out and distinctly claimed in the claims at the conclusion of the specification. The foregoing and other objects, features, and advantages of the invention are apparent from the following detailed description taken in conjunction with the accompanying drawings in which:
FIG. 1 is an example of a complex network topology within which transmission line reflection signal filtering can be employed, in accordance with an aspect of the present invention;
FIG. 2 is a schematic of an approach for operating a transmission line with termination resistors at the input end and output end of the line;
FIG. 3 is an alternate embodiment for operating a transmission line with the termination resistor at the output end of the line removed and only capacitive loading remaining;
FIG. 4 graphically represents a signal at the input end of the transmission line of FIG. 3 for different values of input termination resistance as signal voltage falls from voltage V to ground;
FIG. 5 graphically represents the signal at the output end of the transmission line of FIG. 3 for different values of input termination resistance as signal voltage falls from voltage V to ground;
FIG. 6 depicts another embodiment of a transmission line operated with a termination resistance at the input end and capacitive loading at the output end, and having a capacitance at the input end modeling the capacitance of a common mode filter at the input of the line;
FIG. 7 graphically depicts voltage input and voltage output signals for the transmission line depicted in FIG. 6, with a signal state change from voltage V to ground occurring;
FIG. 8 is a high level, partial schematic of a receiver device with a blocking filter for handling a transmission line reflection signal, in accordance with an aspect of the present invention;
FIG. 9 is a schematic of one embodiment of a blocking filter for the receiver of FIG. 8, in accordance with an aspect of the present invention;
FIG. 10 graphically illustrates a received voltage signal (A), a desired output voltage signal (B) and a masking voltage signal (C) occurring within the blocking filter of FIG. 9, in accordance with an aspect of the present invention;
FIG. 11 depicts another embodiment of a blocking filter showing one embodiment of the pulse generator of FIG. 9, in accordance with an aspect of the present invention;
FIG. 12 graphically illustrates a received voltage signal (A), a desired voltage out signal (B), a masking signal (C), and a delayed reset signal (D) occurring within the blocking filter of FIG. 11, in accordance with an aspect of the present invention;
FIG. 13 depicts another embodiment of a blocking filter, which allows part of the reflection signal to be outside of the masking signal while still blocking the reflection signal, in accordance with an aspect of the present invention;
FIG. 14 graphically illustrates a received voltage signal (A), a desired voltage out signal (13), a masking signal (C), and an extended delayed reset signal (D) for the circuit of FIG. 13, in accordance with an aspect of the present invention;
FIG. 15 depicts another embodiment of a blocking filter, which again allows part of the reflection signal to be outside of the masking signal, and which has fast reset of the delay circuit, in accordance with an aspect of the present invention;
FIG. 16 graphically illustrates voltage signals using the circuit of FIG. 15, including a received voltage signal (A), a desired voltage out signal (B), a masking signal (C), a delayed reset signal (D), and an extended masking signal (E), in accordance with an aspect of the present invention;
FIG. 17 is a schematic of one embodiment of a blocking filter for blocking both falling and rising edge reflection signals, in accordance with an aspect of the present invention;
FIG. 18 is a schematic of one detailed embodiment of the blocking filter of FIG. 15, in accordance with the aspect of the present invention; and
FIG. 19 is a schematic of one detailed embodiment of the blocking filter of FIG. 17, in accordance with an aspect of the present invention.
One problem in applying termination resistors to the ends of network transmission lines arises in complex networks such as depicted in FIG. 1. In FIG. 1 a master device 10, such as a computer, may drive signals across transmission lines 12 to a plurality of slaves 14, in this example, slave 1 through slave 12. Each slave in the network topology of FIG. 1 comprises, for example, a sensor or actuator. In view of the complexity of the network topology, it may be difficult to correctly apply termination at the ends of the transmission lines because a particular line may have multiple ends, and certain slave devices should terminate the line, while most slave devices should not.
The conventional technique for driving and terminating a transmission line is shown schematically in FIG. 2. In this example, the transmission line TL is driven by a source signal VS and the line is terminated at the input by resistor RIN as well at the output by resistor RL. If the value of both resistors is equal to the characteristic impedance ZO of the transmission line, no reflections occur and high speed data communications is possible. As noted, however, a problem with this approach is the power that is dissipated in the termination resistors. This problem can be solved, in part, by connecting load resistor RL to a latch instead of to ground connection so that no static power is dissipated. This solution, however, does not address the problem of how to insert termination resistors in a complex system such as depicted in FIG. 1.
When resistive termination at the output end of a transmission line is omitted, or cannot be used, the transmission line can be represented by the schematic depicted in FIG. 3. In this example, transmission line TL is driven by voltage VS and input termination resistor RT. The capacitive load CL at the output represents the loading of the slave devices. When the input voltage falls from voltage V down to zero, the waveforms shown in FIG. 4 can be found at the input end of the transmission line and the waveforms plotted in FIG. 5 can be found at the output end of the transmission line. These waveforms are shown for different values of termination resistance RT. For the case where the termination resistance is equal to the characteristic impedance ZO of the transmission line, the output voltage displays an exponential decay determined by RT and CL after an initial delay caused by the time the signal needs to go from the input to the output of the line. At the input end of the line, the voltage immediately drops to one-half V because of the voltage division between RT and the characteristic impedance of the transmission line. After a delay equal to two times the time needed for a signal to travel the length of the transmission line, a reflection caused by the capacitor load at the output end of the transmission line returns to the input. This reflection causes the input line voltage to spike (e.g., return to voltage V), before dissipating into the input end termination resistor. Thereafter, the input voltage decays exponentially. For lower values of RT more reflections occur, but the signal settles much faster to the final value. Therefore, it may be advantageous to apply a termination resistor at the input end that is lower than the characteristic impedance ZO.
FIG. 6 depicts another model of a transmission line or bus system. In this schematic, capacitor CCM is added to model the capacitance of a common-mode filter connected between the output of a transmitting device and the input of the transmission line (see FIG. 1). Exemplary falling edge induced signals at the input end and output end of the transmission line are plotted in FIG. 7. These signals are similar to the signals depicted in FIGS. 4 & 5, but the sharp reflections of FIG. 4 are reduced by capacitor CCM. The result is a reflection bump at the input end of the transmission line. In the middle of the line, a signal occurs that has a shape that is in between the shape of the input and output transmission line signals of FIG. 7, yielding a smaller reflection signal. Thus, in a complex system such as depicted in FIG. 1, the slave nodes near the input end of a transmission line may see an extra digital pulse, which could affect the device's performance. For example, if a device is using the received signal to drive a clock, an additional clock pulse might be incorrectly sensed due to the reflection signal. A goal of the present invention is to substantially block these reflection signals or bumps, to the extent received at a device (e.g., slaves connected to the transmission line).
FIG. 8 depicts one embodiment of a device or receiver 100 which includes a blocking filter 120 for removing a received reflection signal. Receiver 100 includes a comparator 110, which compares the received bus voltage (VIN) to a reference voltage (VREF). The output of comparator 110 changes when the bus voltage passes the reference voltage. Responsive to the reflection signal or bump, an additional pulse may be generated at the output of comparator 110 as shown in FIG. 10 (See Signal (A)) with falling of the received signal to ground. When the reference voltage is close to the voltage of the reflection signal, this additional pulse may corrupt the logic of the receiving circuit, particularly when the receiver employs the received signal to extract a clock signal. Therefore, blocking filter 120 is presented to suppress any additional pulse which may be produced from a reflection signal on either a falling or rising edge of the received signal.
One possibility for suppressing the reflection signal would be to use a filter such as an RC filter. The drawback of such a filter, however, is that delay is introduced. This delay can be avoided by the use of a filter (such as described herein) which allows the initial signal edge transition to pass, but blocks the additional reflection pulse. One embodiment of such a circuit is depicted in FIG. 9, with signals at various nodes of the circuit being shown in FIG. 10 responsive to the received voltage (VIN) falling from voltage V to ground.
The filter circuit of FIG. 9 includes an AND gate 130 which receives as input the received voltage signal (VIN) from the transmission line, and outputs therefrom a voltage (Vout) to be employed by the receiver (e.g., master or slave device). A second input to AND gate 130 is received from an inverter 150, which is driven by output from a pulse generator 140. Generator 140 is driven by the output of AND gate 130. Initially, the output of pulse generator 140 is low so that inverter 150 presents a high signal to AND gate 130. Therefore, the output signal Vout is identical to the input signal VIN and the initial falling edge of the input signal is immediately present at the output. This falling edge at the output starts the pulse generator. The pulse created by the pulse generator is inverted by inverter 150 so that a low signal is presented to AND gate 130. Consequently, the output voltage Vout remains low responsive to the masking pulse of the pulse generator, and the reflection pulse presented at the input VIN is blocked. This is shown graphically in FIG. 10, wherein an initial falling edge in received signal (VIN) is depicted in graph (A), the output signal (Vout) is depicted in graph (13), and the masking signal is depicted in graph (C).
FIG. 11 illustrates one embodiment of pulse generator 140 for the blocking filter, which includes the same AND gate 130 and inverter 150 as depicted in FIG. 9. Pulse generator 140 is implemented (in this embodiment) by a memory element 142 (i.e., flip flop FF1) and a delay circuit 144. Memory element 142 is set by the falling edge of output signal Vout (see graph (B) of FIG. 12), and output Q of element 142 goes high (see graph (C) of FIG. 12) and drives inverter 150 and AND gate 130 in such a way that a reflection pulse derived from the received signal (VIN) (see graph (A) of FIG. 12) is blocked. The output of element 142 also drives a delay circuit 144. After a certain delay, the output of delay circuit 144 (see graph (D) of FIG. 12) goes high and memory element 142 is reset. Consequently, the output of element 142 is low and the received signal (VIN) controls the output signal (Vout) again. Also, the delay circuit 144 is reset so that its output goes low and the circuit is again ready for blocking a reflection after a next falling edge.
When the reflection pulse at the received signal (VIN) does not completely fall within the time delay set by the pulse generator, such as depicted in FIG. 14 (compare graph (A) of FIG. 12 and graph (A) of FIG. 14), the filter circuit of FIG. 11 might still generate an additional reflection pulse at the output Vout (which has the undesired effect of making Vout look like VIN). This possibility can be avoided by increasing the duration of the blocking pulse, but this also increases the risk of missing the next rising edge of the received input signal. Another solution, employing logic for lengthening the masking pulse, is depicted in FIG. 13. In this circuit, the reset pulse generated by delay circuit 144 is blocked as long as the input signal is high using a second inverter 12 155 and AND gate A2 160 (see graph (D) of FIG. 14). Thus, the reset is delayed or extended to as long as the additional reflection pulse lasts. A slight disadvantage of this approach is that the reset of the delay circuit is delayed which might be a problem when the delay circuit is used for blocking rising-edge pulses as well as for blocking falling-edge pulses (described further below).
FIG. 15 depicts still another filter circuit embodiment in accordance with an aspect of the present invention. This circuit again uses the same pulse generator as the circuit shown in FIG. 11 so that resetting of the delay circuit is not delayed. A second memory element, latch L1 170, is used to stretch or extend the masking pulse (see graph (E) of FIG. 16). Latch L1 170 is set when the masking pulse and the received input voltage (VIN) are high. Latch L1 170 is reset when both the masking pulse and the received voltage signal are low. The output of the pulse generator (i.e., output Q from memory element 142) and the output of the latch 170 are combined by an OR gate 01175. Thus, the output of OR gate 175 goes high on the falling edge of the received voltage signal (VIN) and stays high until both the masking pulse and the received voltage signal have returned to a low state. FIG. 16 graphically depicts this. Graph (A) of FIG. 16 represents the received voltage signal (VIN), which as shown has a second wide reflection signal on the falling edge of the actual voltage signal. Graph (B) of FIG. 16 represents the desired output voltage (Vout), while graph (C) represents the initial masking pulse output by the pulse generator. Graph (D) of FIG. 16 represents the delayed reset pulse for FF1 generated by the delay circuit 144 of the pulse generator, and graph (E) represents the extended masking performed by latch 170 and OR gate 175. As shown, the reflection signal in the received signal (A) is fully masked by the extended masking pulse (E).
The circuit of FIG. 15 can be extended so that reflection pulses on both the rising edge and the falling edge of the received signal can be blocked. One example of such a circuit, generally denoted 200, is depicted in FIG. 17. Initially, the lower inputs to NAND gates 230 and 240 are high so that the output voltage Vout is identical to the input voltage. At the falling edge of the output voltage, memory element 242 is set while the rising edge of the output voltage sets memory element 243. The output of memory elements 242 & 243 are combined by OR gate 234, which starts the delay circuit 244. Thus, on both the falling and the rising edge, the same delay circuit is started. After a certain time, the delay circuit resets memory elements 242 and 243. The result is a masking pulse at the output of memory element 242 after each falling edge and a masking pulse at the output of element 243 after each rising edge.
The output of memory element 242 is extended by latch L1 262 and NOR gate NO1 272 until the input image goes low. The NOR gate drives NAND gate 230 to keep the output low during the extended masking pulse. The output of memory element 243 is extended by latch L2 263 and NOR gate NO2 273 until the actual received voltage input signal goes high. The NOR gate drives NAND gate 240 to keep the output high during the extended masking pulse.
FIG. 18 illustrates one detailed implementation of the circuitry of FIG. 15. In this example, the AND gate A1 shown in FIG. 15 is replaced by NAND gate NA1 and an inverter I1. Flip flop FF1 in FIG. 15 is implemented by D-flip flop FF1 in FIG. 18. The delay circuit can be implemented using a digital counter, however, in this case an analog circuit consisting of transistors M0-M9 and capacitor C1 is used. The latch L1 shown in FIG. 15 is created using NAND gates NA2-NA5 and inverters I3 and I4. Initially, the upper input of NAND gate NA1 is high and the input voltage is inverted by NA1 and inverter I1 so that the output voltage is identical to the input voltage. At the falling edge of the input voltage, flip flop FF1 is clocked so that its Q output becomes high responsive to the rising edge of the input clock. Consequently, the input of inverter I2 is high, the output of inverter I2 is low and transistor MO is turned off so that current source M7 starts to charge capacitor C1. After a certain time, the voltage across capacitor C1 is larger than the reference voltage VREF and the output of the comparator (comprising differential stage M1, M2, current mirror M3, M4 and output stage M5) goes low. As a result, flip flop FF1 is reset and its output Q is low. Therefore, transistor M0 is turned on by inverter I2 and the capacitor is discharged so that the delay circuit is ready for the next pulse. When the output of flip flop FF1 is high, the output of NOR gate NO1 is low and NAND gate NA1 blocks the input voltage. The blocking pulse at the output of flip flop FF1 also sets the latch when the input voltage returns to the high state: if both the input voltage VIN and the output voltage of flip flop FF1 are high, the output of NA2 is low, while the inverters I3 and I4 drive the inputs of NA3 with a low signal so that the output of NA3 is high. This situation sets the latching structure NA4, NA5, yielding a high voltage at the output of NA4 and at the lower input of NO1, so that the blocking pulse is extended as long as the latch is set. When the blocking pulse generated by FF1 is low and also the input voltage is low, the output of NA2 is high and the output of NA3 is low resulting in resetting of the latching structure, driving the output of NA4 low. As a result, the output of NO1 goes high and the output voltage is again identical to the input voltage.
A detailed implementation of the pulse-blocking filter for rising and falling edges is depicted in FIG. 19. This implementation is based on the principle presented in FIG. 17. Initially, the upper inputs of NAND gates NA1 and NA10 are high so that the output voltage Vout is identical to the input voltage VIN. Flip flop FF1 is clocked by the falling edge of the output signal Vout via inverter I1. Flip flop FF2 is clocked by the rising edge of the output signal Vout. The output of both flip flops is combined by NOR gate NO4 in order to trigger the delay circuit on the rising edge as well as on the falling edge of Vout. The output of the delay circuit drives the reset of the flip flops via inverter I2 and NOR gate NO3. The latch consisting of NAND gates NA2-NA5 and inverters 13 and I4 together with NOR gate NO1 extends the falling-edge blocking pulse until the input signal is low. The latch comprising NAND gates NA6-NA9 and inverters I4 and I6 together with NOR gate NO2 extends the rising-edge blocking pulse until the input signal is high. The input signal RESET resets the flip flops FF1 and FF2 via NOR gate NO3 and it also resets the latches via inverter I5 using additional inputs on NAND gates NA5 and NA9.
To summarize, conventional matching impedance termination at the output of transmission lines may be impractical in complex bus systems. Without full termination, however, a reflection signal or pulse can occur on the bus creating additional pulses at the input of the receiving devices. Reflection blocking filters are presented herein to substantially suppress these additional reflection pulses without adding delay to the received signal.
Although preferred embodiments have been depicted and described in detail herein, it will be apparent to those skilled in the relevant art that various modifications, additions, substitutions and the like can be made without departing from the spirit of the invention and these are therefore considered to be within the scope of the invention as defined in the following claims.