Abstract: Computer-implementable recursive summation algorithms are disclosed that are useful for efficiently performing recursive convolution, such as is often required in Statistical Signal Analysis (SSA) techniques. The disclosed recursive summation algorithms can be more computationally-efficient from both a speed and memory perspective than other recursive convolution techniques known in the prior art, such as the techniques relying on Fast Fourier Transforms (FFTs).

Abstract: The computer-implementable method allows for the fast creation of a multi-unit interval data signal suitable for simulation. The created signal represents the output of an otherwise ideal Discrete Time Filter (DTF) circuit, and the quick creation of the signal merely requires a designer to input the number of taps and their weights without the need of laying out or considering the circuitry of the DTF. A matrix is created based on a given data stream, and the number of taps and weights, which matrix is processed to create the multi-unit-interval data signal. Noise and jitter can be added to the created signal such that it now realistically reflects non-idealities common to actual systems. The signal can then be simulated using standard computer-based simulation techniques.

Abstract: Disclosed herein are circuitry and methods for transmitting data across a parallel bus using both high common mode and low common mode signaling. The transmitter stages are configured to work with two of three possible power supply voltages: a high Vddq voltage, a low Vssq voltage, and an intermediate Vx voltage. In one embodiment, the odd numbered transmitter stages, that drive the odd numbered outputs to the bus, use the Vddq and Vx supplies, such that the odd numbered outputs comprise high common mode signals. The even numbered transmitter stages, that drive the even numbered outputs to the bus, use the Vx and Vssq supplies, such that the even numbered outputs comprise low common mode signals. With the transmitter and power supplies so configured, no one of the three power supplies must source or sink current to or from more than half of the transmitters at any given time, which reduces power supply loading and minimizes switching noise.

Abstract: Decision feedback equalization (DFE) circuits are disclosed for use with fractional-rate clocks of lesser frequency than the data signal. For example, a one-half-rate clocked DFE circuit utilizes two input data paths, which are respectively activated on rising and falling edges of an associated half-rate clock. Each of the input data paths has a pair of comparators with differing reference voltage levels. The comparators in each input data path output to a multiplexer, which picks between the two comparator outputs depending on the logic level of the previously received bit. The output of each input data path is sent as a control input to the multiplexer of the other data path. Thus, the results from previously-detected bits affect which comparator's output is passed to the output of the circuit, even though the synchronizing clock is half the frequency of the data. A quarter-rate DFE circuit is also disclosed which operates similarly.

Abstract: Methods implementable in a computer system for simulating the transmission of signals across a plurality of data channels (bus) are disclosed. The disclosed techniques simulate the effects of Intersymbol Interference (ISI), cross talk, and Simultaneous Switching Output (SSO) noise by generating Probability Distribution Functions (PDFs) for each. The resulting PDFs are convolved to arrive at a total PDF indicative of the reception of data subject to each of these non-idealities. The total PDF, and its underlying terms, can be indexed to particular channels of the bus as well as to particular logic states. Use of the disclosed technique allows bit error rates and sensing margins to be determined with minimal computation and simulation.

Abstract: Methods implementable in a computer system for simulating the transmission of signals are disclosed. The disclosed techniques simulate the effect of the transmitter as well as the channel on a positive and negative pulse, which assures that asymmetry in the transmitter is captured. The resulting positive and negative pulse responses are then used to generate two separate PDFs: one indicative of received logic ‘1’s and another indicative of received logic ‘0’s at a point in time. Generating a plurality of such PDFs at different times allows the reliability of data reception to be assessed, and appropriate sensing margins to be set at a receiver, without the need to simulate the transmission of a very long random stream of data bits.

Abstract: Apparatus are disclosed, such as those involving a transmitter circuit that is configured to generate multi-level signals based on a plurality of data digits. One such transmitter circuit includes a signal output and an encoder configured to provide control signals based at least partially on the plurality of data digits. The transmitter circuit also includes a first set of switches configured to receive one or more of the control signals, and to selectively conduct a first or second voltage reference to the signal output. The transmitter circuit further includes first and second voltage drop circuits that provide third and fourth voltage references, respectively. The third and fourth voltage references have voltage levels between those of the first and second voltage references. The transmitter circuit also includes a second set of switches configured to receive one or more of the control signals, and selectively conduct the third or fourth voltage reference to the signal output.

Abstract: Methods and apparatuses for calculating the location of an optimal sampling point for a receiver system are disclosed. In brief, a first method comprises determining a maximum voltage margin and a maximum timing margin of a received signal, and from these margins, determining an optimal sampling point, which includes a reference voltage level (Vref) and a relative sample phase. The location of the optimal sampling point is based on the locations of the sampling point of the maximum voltage margin and the sampling point of the maximum timing margin. A second method comprises establishing an initial sampling point, and then successively refining each of the voltage and timing components of the sampling point until an optimal sampling point is reached.

Abstract: Methods for generating realistic waveform vectors with controllable amplitude noise and timing jitter, simulatable in a computer-based simulation environment are disclosed. In one implementation, a transition vector is created from a sequence of bits having a rise time and a fall time, in which the transition vector comprises voltage values at timings corresponding to midpoints of transitions in the bit sequence. A jittered transition vector is created from the transition vector, in which the timing of the transitions in the jittered transition vector include timing jitter. An upscaled jittered transition vector is then formed having additional points, in which at least some of the additional points comprise corners of the sequence of bits. The voltages of the additional points are set by the sequence of bits, and the timing of the corners are set in accordance with the rise time and the fall time.

Abstract: Apparatus are disclosed, such as those involving a transmitter circuit that is configured to generate multi-level signals based on a plurality of data digits. One such transmitter circuit includes a signal output and an encoder configured to provide control signals based at least partially on the plurality of data digits. The transmitter circuit also includes a first set of switches configured to receive one or more of the control signals, and to selectively conduct a first or second voltage reference to the signal output. The transmitter circuit further includes first and second voltage drop circuits that provide third and fourth voltage references, respectively. The third and fourth voltage references have voltage levels between those of the first and second voltage references. The transmitter circuit also includes a second set of switches configured to receive one or more of the control signals, and selectively conduct the third or fourth voltage reference to the signal output.

Abstract: The computer-implementable method allows for the fast creation of a multi-unit interval data signal suitable for simulation. The created signal represents the output of an otherwise ideal Discrete Time Filter (DTF) circuit, and the quick creation of the signal merely requires a designer to input the number of taps and their weights without the need of laying out or considering the circuitry of the DTF. A matrix is created based on a given data stream, and the number of taps and weights, which matrix is processed to create the multi-unit-interval data signal. Noise and jitter can be added to the created signal such that it now realistically reflects non-idealities common to actual systems. The signal can then be simulated using standard computer-based simulation techniques.

Abstract: An improved data transmission system is disclosed in which data encoding such as Data Bus Inversion (DBI) in a transmitting device is matched to the termination scheme being used in a receiving device. In the improved system, the transmitting device is able to automatically discover the termination scheme being used in the receiving device, and is thereby able to automatically implement a data-encoding algorithm to best match the termination scheme being used. In one example, Information concerning the termination scheme can be communicated to the transmitting device via a control channel, or another channel otherwise dedicated to data encoding such as a DBI channel. In another example, the transmitting device can infer the termination scheme being used via measurements, or by understanding how the receiving device will modify its termination scheme given current data transmission conditions.

Abstract: A method and apparatus for balancing an output load using data bus inversion is disclosed. In brief, one such technique comprises measuring the “balance” of data bits across a data bus (e.g., the number of zero values compared to the number of one values in a set of parallel data bits). If the data bits are unbalanced by a specified amount, a portion of the bits on the data bus are inverted, and the data bits, including the inverted portion, are transmitted. Also, a data bus inversion bit is set to a particular value and transmitted with the data bits to indicate that data bus inversion was used. If the data signal is not unbalanced (i.e., the bits on the data bus do not comprise an unbalanced number of logic values), then the bits on the data bus are transmitted as they are detected, and the data bus inversion bit is set to another particular value to indicate that data bus inversion was not used.

Abstract: An improved reference voltage (Vref) generator for a single-ended receiver in a communication system is disclosed. The Vref generator in one example comprises a cascoded current source for providing a current, I, to a resistor, Rb, to produce the Vref voltage (I*Rb). Because the current source isolates Vref from a first of two power supplies, Vref will vary only with the second power supply coupled to Rb. As such, the improved Vref generator is useful in systems employing signaling referenced to that second supply but having decoupled first supplies. For example, in a communication system in which the second supply (E.g. Vssq) is common to both devices, but the first supply (Vddq) is not, the disclosed Vref generator produces a value for Vref that tracks Vssq but not the first supply. This improves the sensing of Vssq-referenced signals in such a system.

Abstract: Methods and apparatuses for calculating the location of an optimal sampling point for a receiver system are disclosed. In brief, a first method comprises determining a maximum voltage margin and a maximum timing margin of a received signal, and from these margins, determining an optimal sampling point, which includes a reference voltage level (Vref) and a relative sample phase. The location of the optimal sampling point is based on the locations of the sampling point of the maximum voltage margin and the sampling point of the maximum timing margin. A second method comprises establishing an initial sampling point, and then successively refining each of the voltage and timing components of the sampling point until an optimal sampling point is reached.

Abstract: The computer-implementable method allows for the fast creation of a multi-unit interval data signal suitable for simulation. The created signal represents the output of an otherwise ideal Discrete Time Filter (DTF) circuit, and the quick creation of the signal merely requires a designer to input the number of taps and their weights without the need of laying out or considering the circuitry of the DTF. A matrix is created based on a given data stream, and the number of taps and weights, which matrix is processed to create the multi-unit-interval data signal. Noise and jitter can be added to the created signal such that it now realistically reflects non-idealities common to actual systems. The signal can then be simulated using standard computer-based simulation techniques.

Abstract: Decision feedback equalization (DFE) circuits are disclosed for use with fractional-rate clocks of lesser frequency than the data signal. For example, a one-half-rate clocked DFE circuit utilizes two input data paths, which are respectively activated on rising and falling edges of an associated half-rate clock. Each of the input data paths has a pair of comparators with differing reference voltage levels. The comparators in each input data path output to a multiplexer, which picks between the two comparator outputs depending on the logic level of the previously received bit. The output of each input data path is sent as a control input to the multiplexer of the other data path. Thus, the results from previously-detected bits affect which comparator's output is passed to the output of the circuit, even though the synchronizing clock is half the frequency of the data. A quarter-rate DFE circuit is also disclosed which operates similarly.

Abstract: Decision feedback equalization (DFE) circuits are disclosed for use with fractional-rate clocks of lesser frequency than the data signal. For example, a one-half-rate clocked DFE circuit utilizes two input data paths, which are respectively activated on rising and falling edges of an associated half-rate clock. Each of the input data paths has a pair of comparators with differing reference voltage levels. The comparators in each input data path output to a multiplexer, which picks between the two comparator outputs depending on the logic level of the previously received bit. The output of each input data path is sent as a control input to the multiplexer of the other data path. Thus, the results from previously-detected bits affect which comparator's output is passed to the output of the circuit, even though the synchronizing clock is half the frequency of the data. A quarter-rate DFE circuit is also disclosed which operates similarly.

Abstract: Methods for generating simulation vectors incorporating periodic jitter, or phase-shifted periodic jitter are disclosed. Periodic jitter, such as sinusoidal jitter, is preferably represented by a mathematical equation which defines the amount of jitter experienced at each cycle of a clock or data signal. The calculated periodic jitter for each cycle is used to form a new multi-cycle vector incorporating the jitter. If a particular signal to be simulated additionally needs to travel a particular distance such that it would experience a time delay, that time delay may also be incorporated into the jitter equation as a phase shift. So incorporating the time delay into the jitter equation allows for the easy simulation of circuits receiving the vectors without the need to actually design or “lay out” the circuits that imposing the time delay.

Abstract: A computer-implementable method comprises a matrix-based approach to generating in parallel a plurality of realistic simulatable signal vectors, which vectors include the addition of amplitude noise and/or timing jitter and encoding. Each channel in a parallel bus can be populated in a matrix, with each row comprising ideal voltage values for the channel, and the columns comprising bits of the sequence of voltage values for that channel. Encoding can be employed to modify the data in the matrix. Amplitude noise and/or timing jitter can then be applied to each channel (row) in the matrix. This modifies the time basis from a bit basis as used in the matrix to a time-step basis. With such modification accomplished, each row in the matrix can be transformed into simulatable vector, which vectors can then be simulated in parallel to test, the robustness of the parallel bus of which the channels are part.