This invention relates to signal receivers, and, more particularly, to a method and system that generates a reference voltage to which an input signal is compared in a signal receiver in a manner that minimizes the timing skew of the input signal.
Digital signals are commonly coupled to and from electronic devices, such as memory devices, at a high rate of speed. The Digital signals are normally coupled to an input buffer or receiver, which generates a digital signal corresponding to the Digital signal applied to the input of the receiver. The timing at which the signal at the output of the receiver changes state is often critically important for timing the relationships within the electronic device. In particular, it is important that the transition of the digital signal not become skewed relative to other digital signals in the electronic device. The difficulty of avoiding signal skew is increased when the digital signals applied to input receivers switch between two voltages that are relatively close to each other.
One technique for preventing the timing of digital signals becoming skewed is to use differential signals, which tend to avoid skewing because of their inherent symmetry even where the voltage between which the signals transition is relatively small. However, some memory bus signaling protocols do not couple differential signals to memory devices. In these devices, the input signal must be compared to a reference voltage to determine the trip point of the receivers. More specifically, when the magnitude of the input signal is greater than the reference voltage, the output of the receiver has a first binary value. When the magnitude of the input signal is less than a reference voltage, the output of the receiver has a second binary value. Ideally, the reference voltage is centered between the two voltages between which the input signal transitions. However, because the input signal may lack symmetry, the ideal reference voltage may be at some other level. In either case, the use of a reference voltage can allow the input receivers to accurately convert digital signals switching within a relatively small voltage range to a receiver output signal that switches within a substantially larger voltage range.
Reference voltages for input receivers are traditionally generated at a single source, and then distributed to each of the input receivers. Distributing the reference voltage in this manner can create a number of problems, all of which can skew the signals generated at the output of the input receivers. For example, noise signals can be coupled to input signal lines from various sources, such as power supply noise and switching noise, may alter the voltages between which the input signal switches. As a result, the reference voltage may no longer be centered between the voltages between which the input signal switches. Furthermore, a lack of symmetry in the input signal can alter the ideal value of the reference voltage. As a result, the signal at the output of the receiver may be skewed, which can lead to degraded timing margins within the memory device.
Not only is it possible for timing skew to result from noise signals to the input signal lines, but timing skew can also result from noise signals coupled to reference voltage distribution lines. The susceptibility of noise signals being coupled to reference voltage distribution lines is exacerbated by the widespread routing of such lines to sometimes hundreds of input receivers at various locations throughout the memory device. Noise can be coupled to these distribution lines through various mechanisms, such as power supply coupling through decoupling capacitors connected to the voltage reference distribution lines. The noise signals can significantly alter the reference voltage at various input receivers thereby altering the voltage of the input signal at which the output of the input receiver switches. The result is a skewing of the output signal, which, as mentioned above, can adversely affect timing relationships within the memory device. Under the circumstances, it may not be possible for the memory device to function properly at high operating speeds.
There is therefore a need for a method and system for generating a reference voltage having an optimum magnitude and for maintaining the reference voltage at that value at the input of each of many input receivers.
A reference voltage generator and method generates a reference voltage for use by a digital signal receiver. A test signal is applied to an input of the receiver during a calibration mode while a reference voltage is applied to another input of the receiver. The magnitude of the reference voltage is varied while an output of the receiver is examined to determine if the test signal has been correctly coupled through the receiver. A reference voltage in the middle of the range of reference voltages that allowed the test signal to be coupled through the receiver is then used as a reference voltage during normal operation of the receiver.
The data capture circuit 4 is coupled to calibration logic 8, which controls the operation of the data capture circuit 4 and selects a final reference voltage VR based on a range of reference voltages that allow the receiver 6 to couple the received digital pattern to its output. As described in greater detail below, the calibration logic 8 causes the data capture circuit 4 to incrementally vary the reference voltage VR over a substantial range, and to determine if the receiver 6 is able to couple the digital pattern to its output at each of the reference voltages. The calibration logic 8 then calculates a final reference voltage in substantially the middle of the range of reference voltages that allow the receiver 6 to pass the digital pattern.
With reference to
The data capture circuit 10 operates in either of two modes: a calibration mode and a normal operation mode. In the calibration mode, a digital signal generator 58 successively applies a data test pattern to the data pad 16 as the magnitude of the reference voltage VR is varied. The data test pattern may be generated by suitable means, such as by a conventional digital signal generator 38. In the embodiment illustrated in
The deserializer 48 stores data bits captured by the flip-flop 42 8-bits at a time. The deserializer 48 converts the 8 serial data bits to 8 bits are parallel data, and applies the 8 parallel data bits to pattern recognition logic 50 of conventional design. The pattern recognition logic determines whether or not each 8-bit group of captured data pattern matched the data test pattern applied to the data pad 16, and generates a corresponding pass/fail* signal. The deserializer 48 also outputs a data latch pulse each time 8-bits are coupled to the pattern recognition logic 50. Therefore, the deserializer 48 will normally generate 16 data latch pulses for the 128 bits of data applied to the data pad 16.
The pass/fail* signal and the data latch signal are coupled to the calibration logic 8 (
With reference to
As explained above, 128 bits of a data test pattern are then applied to the data pad 16, coupled through the data receiver 12, captured by the flip-flop 42 and output from the deserializer 48 to the pattern recognition logic 50, which generates a pass/fail* signal. The pass/fail* signal is coupled to pass/fail logic 74. Assuming that, as would be expected, the pass/fail* signal is initially low indicating a capture failure, the pass the logic 74 couples a clock signal to the register 68, which causes the register 68 to capture the count from the counter 64. As will be explained below, the count stored in the register 68 is underwritten with subsequent counts until four active high pass/fail* signals have been received by the pass/fail logic 74.
As explained above, a data latch pulse is generated by the deserializer 48 for each 8 bits of data captured by the flip-flop 42. Therefore, after 128 bits of the data test pattern have been captured, 16 data latch pulses have been applied to a counter 80. The counter 80 then outputs a 4-bit binary count “1111” indicative of the number “15,” which is connected by test enable logic 84. The test enable logic 84 then drives an active low test enable TestEn* signal high, which increments the counter 64, thereby incrementally increasing the magnitude of the reference voltage VR. The high test enable TestEn* signal disables the calibration circuitry for the next 16 data latch pulses, which are generated responsive to 128 bits of the clock signal. These 16 data latch pulses allow the reference voltage VR time to settle before a check is made to determine if the receiver 12 can pass data at the new level of the reference voltage VR. After 128 additional clock pulses have been received, the resulting 16 data latch pulses again increment the counter 80 to 16, which is detected by the test enable logic 84. The test enable logic 84 then generates an active low TestEn* signal, which enables the pass/fail logic 74 to again process a pass/fail* signal.
The above-described procedure is repeated until a high pass/fail* signal is received by the pass fail logic 74. The pass fail logic 74 then refrains from applying a clock signal to the register 68 so that a count corresponding to the last failing reference voltage is retained in the register 68. The procedure is repeated 3 additional times and, if the pass/fail* signal remains high, the count is retained in the register 68 as the count corresponding to a low calibration reference voltage. If, one or more additional low pass/fail* signals are received by the pass/fail logic 74, the count corresponding to the last failing reference voltage VR is retained in the register 68 as the low calibration voltage.
After the low calibration voltage has been found and stored in the register 68, the above-described procedure is repeated until a low pass/fail* signal is again received by the pass/fail logic 74. The pass/fail logic 74 then outputs a clock signal to a register 90, which then stores the count from the counter 64 corresponding to the current reference voltage VR as the high calibration voltage. At this time, the register 90 will retain a count corresponding to the high calibration voltage, and the control logic 62 outputs an active high “Done” signal.
The count from the register 68 corresponding to the low calibration voltage and the count from the register 90 corresponding to the high calibration voltage are applied to an adder/divider 94, which sums the count and divides the result by two to provide a calibration count corresponding to a reference voltage VR that will be used in normal operation. The calibration count is coupled to the multiplexer 70. The active high “Done” signal causes the multiplexer 70 to couple the calibration count from the adder/divider 94 to the digital-to-analog converter 34. The digital-to-analog converter 34 then generates a reference voltage VR that is used in normal operation. Significantly, the above procedure adaptively causes a magnitude for the reference voltage VR to be selected that best allows the data receiver 12 to couple data to the capture flip-flop 42.
The operation of the SDRAM 100 is controlled by a command decoder 102 responsive to high-level command signals received on a control bus 104. These high level command signals, which are typically generated by a memory controller (not shown in
The SDRAM 100 includes an address register 112 that receives row addresses and column addresses through an address bus 114. The address bus 114 is generally coupled to a memory controller (not shown in
After the row address has been applied to the address register 112 and stored in one of the row address latches 126, a column address is applied to the address register 112. The address register 112 couples the column address to a column address latch 140. Depending on the operating mode of the SDRAM 100, the column address is either coupled through a burst counter 142 to a column address buffer 144, or to the burst counter 142 which applies a sequence of column addresses to the column address buffer 144 starting at the column address output by the address register 112. In either case, the column address buffer 144 applies a column address to a column decoder 148, which applies various column signals to corresponding sense amplifiers and associated column circuitry 150, 152 for one of the respective arrays 120, 122.
Data to be read from one of the arrays 120, 122 is coupled to the column circuitry 150, 152 for one of the arrays 120, 122, respectively. The data is then coupled to a data output register 154, which includes a set of read data receivers collectively shown as receiver 156. The data output register 156 applies bit of the data to respective conductors of a data bus 158 through the read data receivers 156. Data to be written to one of the arrays 120, 122 are coupled from the data bus 158 through a data input register 160. The data input register 160 includes a plurality of write data receivers 162 that couple a respective bit of write data from the data bus 158 to internal circuitry in the data input register 160. A reference voltage VR from the reference voltage generator 108 is coupled to each of the read data receivers 156 and the write data receivers 162. Alternatively, a separate reference voltage generator (not shown) may be used to supply a reference voltage VR to all of the read data receivers 156 and the write data receivers 162, or a separate reference voltage generator (not shown) may be used to supply a reference voltage VR to each of the read data receivers 156 and the write data receivers 162. In any case, the write data from the write data receivers 162 are coupled to the column circuitry 150, 152 where they are transferred to one of the arrays 120, 122, respectively. A mask register 164 responds to a data mask DM signal to selectively alter the flow of data into and out of the column circuitry 150, 152, such as by selectively masking data to be read from the arrays 120, 122.
From the foregoing it will be appreciated that, although specific embodiments of the invention have been described herein for purposes of illustration, it will be understood by one skilled in the art that various modifications may be made without deviating from the spirit and scope of the invention. For example, although the embodiments of the invention are primarily shown as generating reference voltages only for data signal receivers, it will be understood that they may also be used for generating reference voltages for other signals, such as command signals and address signals, as shown in
This application is a continuation of pending U.S. patent application No. 10/930,543, filed Aug. 30, 2004.
Number | Name | Date | Kind |
---|---|---|---|
4060834 | Mounts et al. | Nov 1977 | A |
4433298 | Palm | Feb 1984 | A |
4857725 | Goodnough et al. | Aug 1989 | A |
4894656 | Hwang et al. | Jan 1990 | A |
5402666 | Chalpin | Apr 1995 | A |
5553034 | Jones et al. | Sep 1996 | A |
5898912 | Heck et al. | Apr 1999 | A |
6046632 | Straw | Apr 2000 | A |
6052026 | Tiller et al. | Apr 2000 | A |
6160851 | Brown et al. | Dec 2000 | A |
6240100 | Riordan et al. | May 2001 | B1 |
6294949 | Kojima et al. | Sep 2001 | B1 |
6347287 | Beckett et al. | Feb 2002 | B1 |
6356487 | Merritt | Mar 2002 | B1 |
6546343 | Batra et al. | Apr 2003 | B1 |
6586964 | Kent et al. | Jul 2003 | B1 |
6762617 | Iwase et al. | Jul 2004 | B2 |
20030001667 | Casper et al. | Jan 2003 | A1 |
20040085085 | Muhtaroglu et al. | May 2004 | A1 |
20050078530 | Han | Apr 2005 | A1 |
20050177771 | Miller | Aug 2005 | A1 |
Number | Date | Country | |
---|---|---|---|
20060203938 A1 | Sep 2006 | US |
Number | Date | Country | |
---|---|---|---|
Parent | 10930543 | Aug 2004 | US |
Child | 11433322 | US |