System and method of determining the noise sensitivity characterization for an unknown circuit

Abstract

The present invention includes a system for and a method of determining noise characteristics of a circuit of an integrated circuit. The circuit is classified based on its topology and measured circuit parameters. Noise characteristics are retrieved using the circuit classification and circuit parameters to calculate a noise response. Classification and characterization may be performed on each individual input.

Description

TECHNICAL FIELD

[0002] This invention generally relates to microprocessor and integrated circuit design techniques and specifically relates to the characterization of the noise sensitivity of integrated circuits.

BACKGROUND

[0003] Signal noise, line loading, and line coupling all contribute to signal degradation and cause much concern in the design and production of high speed Very Large Scale Integrated (VLSI) circuits. The effects of these phenomena are especially troublesome in the design of circuitry comprising numerous closely packed signal lines and densely packed logic circuits. To achieve minimum size and maximum design of integrated circuits involves a lengthy phase of design optimization followed by multiple design iteration. During design validation, a database representing the proposed integrated circuit is used to model the proposed silicon integrated circuit as closely as possible before a prototype or production integrated circuit is manufactured. The model of the integrated circuit is used to test and verify performance of the design and to identify and avoid potential and actual problems and the issues which are expected to occur once the integrated circuit is manufactured.

[0004] In accordance with Moore's law, miniaturization of the integrated circuits and the wires which connect the various transistors on an integrated circuit must support the doubling of circuit density every 18 months. State of the art processors utilizing 0.18 micron and smaller feature sizes, cramming tens-of-millions of transistors on a single die, have between six and eight layers of metal used for wires interconnecting the underlying logic and transistors. In an effort to reduce the resistance associated with these wires, the width of the wires has been decreased to maintain compatibility with ever decreasing feature sizes, while the individual wires have grown in height. This reduced resistance allows an increase in connecting speed between transistors. However, these modifications in the width and height of the wires also affect interferences associated with neighboring wires.

[0005] In densely packed 0.18 micron and smaller integrated circuits, the capacitance effects between neighboring wires is difficult to accurately predict. Capacitance effects between wires can be impacted by a change in the direction of current in neighboring wires, the amount of current in neighboring wires, and the accumulative effects of neighboring wires within a specific distance. This interference is the result of the parasitic capacitance which is a product of the electric field between wires. This static electric field is caused by a voltage potential between two conductors when an insulator is located between the two conductors, resulting in capacitor coupling of the wires.

[0006] Capacitance loading can also have an adverse effect on signals carried on wires which were intended to maintain a constant voltage. The capacitance loading will effectively superimpose an additional current on the wire that is required to maintain the constant voltage and may result in a voltage spike which must be absorbed by the components (e.g., signal driver or signal receiver) connected to this wire. These voltage spikes may result in circuit failures. A wire in a VLSI circuit is usually viewed as having a driving circuit (a driver), the wire, and one or more receiving circuit (a receiver). If these voltage spikes reach the receivers connected to the effected or victim wire, which are required to maintain a constant voltage, the receiver may switch states within the dense wire environment of an integrated circuit resulting in a circuit failure.

[0007] If these loading effects are identified early, they may be prevented by spacing the wires further apart, increasing the driver strength (e.g., current driving capacity) or by making the receiver less sensitive to a voltage spike. Sensitivity to noise or noise problems are typically identified by the receiver's response to the voltage spike. If the receiver ignores the voltage spike the noise problem may be ignored, because it does not cause a malfunction in the integrated circuit. However, if the voltage spike causes an adverse reaction in the receiver, the interference is at such a level that it must be resolved.

[0008] Systems and methods for investigating electrical characteristics of a multi-level interstructure are known. Such systems use complicated algorithms for simulating a circuit. The process is complicated by the fact that adjacent interconnects, or wires, sometimes follow non-parallel paths. Another complexity is that there is a non-linear relationship between some process variables and electrical characteristics. For example, the relationship between capacitance and the space between adjacent interconnects, or wires, is non-linear. Consequently, the computation and simulation procedures are complex and are typically performed using Electrical Design Automation (EDA) tools. For example, HIVE is a software package that performs 2D numerical field simulations for interconnects having given geometries to arrive at the closest-fit analytical functions. As another example, Simulation Program with Integrated Circuit Emphasis (SPICE) is a software package that is commercially available for simulating inter alia electrical performance of complex Very Large Scale Integrated (VLSI) chips. SPICE requires inputs in the form of a SPICE subcircuit datafile, known in the art as a “SPICE deck” which numerically characterizes and describes the value and type of every conductor and component of the VLSI chip.

[0009] Many common computer simulators are also variations of the simulator tool SPICE. These programs typically operate by accepting circuit frequency response parameters, either directly from a Computer Aided Design (CAD) package, a simulator (using discrete frequencies to directly measure frequency response of a circuit prototype) or other means. Based upon these parameters, the simulator is then typically used to, simulate special signal conditions for the circuit which are usually not discrete frequencies, i.e., to predict transit responses in an integrated circuit. The computer based simulator typically use numbers which represent test input signals, e.g., initial voltages, currents and frequencies. The simulators are then usually used to conduct a time based analysis of responses to the input signal conditions of the different measurement points of the circuit. These tools can be used to determine the noise characteristics and sensitivity of circuit designs.

[0010] While numerous SPICE simulations can accurately simulate the operation of the integrated circuit, the required number of SPICE simulations is expensive both financially and computationally. The SPICE simulator accurately simulates wires and circuits with respect to their voltage and current behavior. However, in order to identify potential noise problems on an integrated circuit, numerous SPICE simulations must be performed. In these numerous SPICE simulations, variables such as the switching speed of neighboring wires, the distance between the aggressor wire and the victim wire, the length of the victim wire which runs parallel to the aggressor wire, the strength of the driver, and the sensitivity of the receiver must all be known and possibly varied to adequately investigate the potential noise problem. To accurately determine the extent of the noise problem in a typical LSI or VLSI circuit literally millions of SPICE iterations would need to be performed. Normally, simplifying assumptions are made to reduce the number of SPICE iterations which must be performed. These simplifying assumptions may impose restrictions upon the design itself. For example, restrictions may be imposed on wire spacing, the number of stages of inverters used to reject voltage spikes in receivers, the shielding required to reduce or eliminate line capacitance, and other such integrated circuit design limitations may be imposed. These limitations may result in a less than optimal design.

SUMMARY OF THE INVENTION

[0011] A need exists for a system and methodology that will analyze integrated circuits with “SPICE-like” accuracy at a fraction of the cost and computational intensity required by SPICE or similar full-featured circuit simulator. A further need exists for the identification of interference determining parameters or criteria that can be used to classify receivers within an integrated circuit as either being susceptible to noise interference or predicted to be operating within their design limitations.

[0012] This need and other features and technical advantages are achieved by a method of determining noise characteristics of a circuit of an integrated circuit which includes the steps of classifying the circuit from the topology of the circuit to provide a circuit classification and measuring circuit parameters of the circuit to provide circuit parameter measurements. The method of determining noise characteristics further includes retrieving prestored noise characteristics using the circuit classification and the circuit parameter measurements and calculating a noise response of the circuit using the prestored noise characteristics. The step of classifying the circuit may be performed for each input terminal of the circuit and may further comprises the step of identifying each input as one of an invertor, a NAND gate, a NOR gate or a dynamic circuit. The step of measuring circuit parameters may include measuring a bad capacitance, determining a pn ratio, and counting a number of input terminals. The method may also include the steps of precalculating default schematics for common circuit types, identifying circuit parameters for each of the default schematics, sweeping the circuit parameters through a set of discrete values, calculating a noise response for each of the noise characterizations, and storing default schematics, circuit parameters and the noise responses for later use.

[0013] In another embodiment, the invention includes a tool for identifying noise responses for circuits within an integrated circuit which includes a circuit identifier which identifies a target circuit, and a circuit parameter identifier which identifies parameters for the target circuit. The tool further comprises a controller configured to lookup in a prestored table noise characteristics for the target circuit and the parameters, and a noise determiner which displays the noise characteristics for the target circuit with the parameters. The circuit parameter identifier may operate on each input terminal of the circuit. The circuit identifier may identify the target circuit as one of an inverter, a NAND gate, a NOR gate or a dynamic circuit. The circuit parameter identifier may identify a bad capacitance of each of the input terminals of the target circuit, a pn ratio of each of the input terminals of the circuit, and/or the number of input terminals of said target circuit.

[0014] Another embodiment of the current invention includes a tool for determining noise characteristics of an integrated circuit which includes means for classifying the circuit from the topology of the circuit to provide a circuit classification and means for measuring circuit parameters of the circuit to provide circuit parameter measurements. The tool includes the means for retrieving prestored noise characteristics using the circuit classification and the circuit parameter measurements and means for calculating a noise response of the circuit using the prestored noise characteristics. The means of classifying the circuit may operate for each input terminal of the circuit, and may identify each input as one of an invertor, a NAND gate, a NOR gate or a dynamic circuit. The means for measuring circuit parameters may include a means for measuring a bad capacitance, a means for determining a pn ratio, and/or a means for counting a number of input terminals. The tool may further include a means for precalculating default schematics for common circuit types, a means for identifying circuit parameters for each of the default schematics, a means for sweeping the circuit parameters through a set of discrete values, a means for calculating a noise response for each of the noise characterizations, and a means for storing default schematics, circuit parameters and the noise responses for later use.

[0015] It should be appreciated by those skilled in the art that the conception and specific embodiment disclosed may be readily utilized as a basis for modifying or designing other structures for carrying out the same purposes of the present invention. It should also be realized by those skilled in the art that such equivalent constructions do not depart from the spirit and scope of the invention as set forth in the appended claims.

BRIEF DESCRIPTION OF THE DRAWING

[0016] For a more complete understanding of the present invention, reference is now made to the following descriptions taken in conjunction with the accompanying drawing, in which:

[0017] FIG. 1 is a diagram illustrating a simplified example of a number of wires which are affected through capacitive coupling from nearby aggressor wires;

[0018] FIG. 2 is a circuit diagram used to model a victim wire for determining noise characteristics in the presence of aggressor wires;

[0019] FIG. 3A is a SPICE simulation plot;

[0020] FIG. 3B is a plot of characterization results;

[0021] FIG. 3C is a flow diagram showing how noise characteristics are precalculated;

[0022] FIG. 4 is a flow diagram showing how precalculated noise characteristics are used;

[0023] FIG. 5 is a block diagram outlining the creation of the noise characteristic library for known cells;

[0024] FIG. 6 is dynamic circuit included in the preferred embodiment of the library of the present invention;

[0025] FIG. 7 is an inverter circuit included in the preferred embodiment of the library of the present invention;

[0026] FIG. 8 is a two input nand circuit included in the preferred embodiment of the library of the present invention;

[0027] FIG. 9 is a three input nand circuit included in the preferred embodiment of the library of the present invention;

[0028] FIG. 10 is a four input nand circuit included in the preferred embodiment of the library of the present invention;

[0029] FIG. 11 is a two input nor circuit included in the preferred embodiment of the library of the present invention;

[0030] FIG. 12 is a three input nor circuit included in the preferred embodiment of the library of the present invention;

[0031] FIG. 13 is a four input nor circuit included in the preferred embodiment of the library of the present invention;

[0032] FIG. 14 is a diagram of the steps used to determine noise characteristics;

[0033] FIG. 15 is a detailed diagram of a traversal algorithm to identify a node; and

[0034] FIG. 16 shows a circuit used to implement the current invention.

DETAILED DESCRIPTION

[0035] FIG. 1 is a diagram depicting a simplified example of a number of wires and their components 100 in which wire 101 may be affected through capacitive coupling or loading by other, adjacent wires and signals. As previously described, a wire in a VLSI design can be classified as having a driver circuit (driver) providing a signal, the interconnect or wire communicating the signal, and one or more receiving circuits (receiver) detecting the transmitted signal. Wire 101 may therefore be referenced as a “victim” wire 101 when subject to analysis. Thus, victim wire 101 is connected to a victim driver 102 and a victim receiver 103.

[0036] FIG. 1 also shows additional wires in the vicinity of victim wire 101 which may result in capacitance loading or coupling with wire 101. The wires or components which cause capacitance loading or coupling are designated as aggressor wires or components. FIG. 1 shows aggressor wire 104 immediately adjacent to one side of victim wire 101 which is connected to aggressor driver 105 and aggressor receiver 106. Second aggressor wire 107 located on the other side of wire 101 is connected to aggressor driver 108 and aggressor receiver 109. Capacitance coupling affects on victim wire 101 are not limited to affects from wires 104 and 107. As also shown in FIG. 1, aggressor wire 110, connected to aggressor driver 111 and aggressor receiver 112 and aggressor wire 113, connected to aggressor driver 114 and aggressor receiver 115 may also have an affect on victim wire 101.

[0037] The affects from aggressor wires 110 and 113 as diagramed may reduce the capacitive loading affects of aggressor wires 104 and 107 because of the orientation of the associated drivers and receivers. Additionally, one of ordinary skill in the art would understand that the affects of aggressor wire 113 will be reduced as compared to the affect of wire 110 because of the shorter span of aggressor wire 113 which runs parallel to victim wire 101 (again as compared to the longer length of aggressor wire 110 which runs parallel to victim wire 101). These comparisons assume that all other capacitance coupling factors are identical between aggressor wire 110, aggressor driver 111 and aggressor receiver 112 with aggressor wire 113, aggressor driver 114 and aggressor receiver 115, and that the differences are limited to the different lengths of the two aggressor wires which run parallel to victim wire 101.

[0038] If victim receiver 103 requires a constant voltage signal to be supplied from victim driver, via victim wire, capacitance loading bleeding part of the signal or capacitance coupling adding to the signal level may cause victim wire 101 to vary the signal voltage supplied to victim receiver 103. The capacitance loading or coupling affecting signal levels on victim wire 101 may be the result of a combination of the effects of signals on aggressor wire 104, aggressor wire 107, aggressor wire 110, aggressor wire 113, or any other components, interconnects, or wires in the vicinity of victim wire 101. One of ordinary skill in the art would understand the parasitic capacitance in the form of capacitance loading or capacitance coupling is not limited to aggressor wire 104 and aggressor wire 107.

[0039] Prior art tools such as SPICE simulate circuits and their signals including an analysis of capacitive coupling and other noise. This analysis is performed for each discrete signal frequency or continuous ranges of frequencies within a specified frequency range of interest and a noise value is calculated which corresponds to the spectral density of the circuit variables when viewed as a stationary gaussian stochastic process. Once the spectral densities are calculated, a noise analysis integrates these values over the specified frequency range to arrive at the total noise voltage/current.

[0040] In order to perform a noise analysis, SPICE requires the circuit under analysis to be described as a set of element lines, where the element lines define the circuit topology and the element values, and a set of control lines which define the model parameters and the run control. The element lines contain the element name, the circuit nodes to which the element is connected, and the values of the parameters that are used to determine the electrical characteristics of the element. Most simple circuit elements typically only require a few parameter values, while semiconductor devices typically require many parameter values. For example to simulate a Bi-polar Junction Transistor (BJT), SPICE uses a model which is an adaptation of the integral charge control model of Gummel-Poon. When certain parameters are not provided to SPICE, the model simplifies to the Ebers-Moll model BJT. Table 1 shows the parameters used by SPICE, and supplied by the user for a BJT transistor for the modified Gummel-Poon model.

TABLE 1
Parameters Used By SPICE (supplied by the user) For
A BJT Transistor For The Modified Gummel-Poon
Model
include transport saturation current
ideal maximum forward beta
forward current current emission coefficient
forward Early voltage
corner for forward beta high current roll-off
B-E leakage saturation current
B-E leakage emission coefficient
ideal maximum reverse beta
reverse current emission coefficient
reverse Early voltage
corner for reverse beta high current roll-off
B-C leakage saturation current
B-C leakage emission coefficient
zero bias base resistance
current wear base resistance falls halfway to its minimum
value
minimum base resistant at high currents
emitter resistance
collector resistance
B-E zero-bias depletion capacitance
B-E built-in potential
B-E junction exponential factor
ideal forward transit time
coefficient for bias dependence of TF
voltage describing VBC dependence of TF
high-current parameter for effect on TF
excess phase at frequency = 1.0/(TF*2PI)Hz
B-C zero-bias depletion capacitance
B-C built-in potential
B-C junction exponential factor
fraction of B-C depletion capacitance connected to
internal base node
ideal reverse transit time
zero-bias collector-substrate capacitance
substrate junction built-in potential
substrate junction exponential factor
forward and reverse beta temperature exponent
energy gap for temperature effect on IS
temperature exponent for effect in IS
flicker-noise coefficient
flicker-noise exponent
coefficient for forward-biased depletion capacitance
formula
parameter measurement temperature

[0041] In a similar manner, 15 parameters are required for a diode, 14 parameters are required for a JFET model, 42 parameters are needed for a MOSFET model, 13 parameters are required for a MESFET model and so on. Additional SPICE information is available in the SPICE user's manual, particularly in Appendix B entitled “Model And Device Parameters.”

[0042] Line-to-line coupling noise which may distort signals supplied to victim receivers is of primary concern in predicting electrical failures in modern circuits. Line-to-line coupling noise can be characterized from four key elements including: the percentage of capacitance effects on a signal that come from aggressor signal lines that switch when this victim signal is being held at a constant voltage level; the output impedance of the driving circuit; the length of the victim line; and the nature of the receiver. The first element, percentage of capacitance on the signal that comes from aggressor signals that switch when this victim signal is attempting to hold a constant voltage, hereinafter called “bad capacitance,” is greatly effected by wire spacing and proximity to supply shields. Another effect of “bad capacitance” is an increase in coupling noise percentage resulting in an increased noise margin. With respect to the second element, output impedance of the driving circuit, one of ordinary skill in the art would understand that the lower the impedance of the victim driving circuit, the better the victim driver will tolerate and mitigate any signal coupled to the victim wire and thereby minimize any adverse effect on the victim receiver. With respect to the third element, as the length of the victim line increases so does the resistance to the charge received from the aggressor wires which serves to increase the magnitude of the voltage applied to the receiver. Finally, the fourth element, the nature of the receiver, includes the trip point (i.e., domino gates trip at NFET Vt etc.), multi-input behavior (i.e. all inputs to a NOR could bias the NOR in such a way which greatly increases its sensitivity to spurious signals), output load, and number of stages (i.e., a buffer provides more noise immunity than an inverter).

[0043] Noise characteristics for a specific receiver can therefore be determined as a function of three factors: percentage of coupling capacitance (bad cap), output impedance of the driving circuit (strength of driver), and length of victim line (route length). Using these three factors, a three dimensional matrix of a circuit's response to stray capacitance can be produced. In a preferred embodiment, this matrix is further dependent on two assumptions. First, it is assumed that the wires are made of M3. Since M3 has the worse coupling behavior as a function of length, higher layers of metal (those with better coupling behavior as a function of length) will report maximum noise values for shorter length wires than is actually observed. The second assumption is that the driver strength minimum is set at 8 μm/pf for NFETs and 21 μm/pf for PFETs.

[0044] FIG. 2 is a diagram of a circuit used for the characterization of noise sensitivity for a victim wire 201. Victim wire 201 is connected to victim driver 202 of strength “X” and victim receiver 203. As shown, victim line 201 is formed between two aggressor lines 204 and 207. Aggressor line 204 is connected to aggressor driver 205 of magnitude 4X and aggressor receiver 206. Aggressor line 207 is connected to aggressor driver 208 also of magnitude 4X and aggressor receiver 209. For example, if the victim driver is a NFET, its strength would be 8 μm/pf and aggressor driver 205 and 208 would each be 32 μm/pF. Alternatively, if victim driver is a PFET, victim driver strength “X” would be 21 μm/pF and both aggressor driver 205 and aggressor driver 208 would each have a magnitude of 84 μm/pF.

[0045] The noise characteristics are created by applying a transition on the aggressor wire while keeping the victim wire at a constant value. The noise response is measured at the output of the victim receiver and is in the form of an excursion, or variance, from the quiescent, or steady state voltage. The magnitude of the voltage excursion is stored for each value of % bad capacitance, length, and driver size. After the transition on the aggressor wire has been terminated, the voltage on the victim output may or may not return or recover to the initial quiescent value.

[0046] A number of separate simulations are performed which typically encompass the following cases: aggressor transitioning from low to high with the victim attempting to hold the voltage high; aggressor transitioning from low to high with the victim attempting to hold the voltage low; aggressor transitioning from high to low with the victim attempting to hold the voltage high; and aggressor transitioning from high to low with the victim attempting to hold the voltage low. These simulations are performed, by way of example, in a computer sized for the application as desired.

[0047] FIG. 3A is a plot of SPICE simulation showing a transient case. In FIG. 3A aggressor 301 is transitioning from low to high while victim receiver input signal 302 is configured to maintain a low voltage. The effect on voltage on victim receiver input 302 propagates through the victim receiver and the noise can be observed on victim receiver output 303. When victim receiver output 303 droops a specified voltage below the supply rail, or VDD, a failure occurs.

[0048] FIG. 3B is a plot of the noise characterization results for a particular receiver. The “x” axis represents the wire capacitance and the “y” axis is a fraction of bad capacitance. Curve 304 shows the maximum bad capacitance when the driver is equal to {fraction (21/8)} μm/pf. Curve 305 shows the maximum bad capacitance percentage when the driver is equal to {fraction (37/14)} μm/pf and curve 306 shows the maximum bad capacitance percentage when the driver is equal to {fraction (52/20)} μm/pf. Values of bad capacitance below the curve for a given driver size are passing conditions.

[0049] SPICE calculations are then performed for this configuration while varying the wire lengths, the percentage of coupling capacitance (or bad cap) as well as the victim driver strength from the minimum values previously discussed (8 μm/pf for NFET and 21 μm/pf for PFETs) to twice the minimum values. As the values of victim driver strength are changed so are the corresponding aggressor driver strengths. Once these SPICE calculations are performed, the noise characteristics for a specific receiver can be applied as a function of the route length, the percentage of coupling capacitance and the strength of the driver without having to recalculate values using a SPICE simulator. Thereafter the precalculated noise characteristics for receivers as a function of route length, bad capacitance and driver strength allows the evaluation of the noise characteristics of an overall circuit, and the identification of required circuit updates to reduce noise characteristics, in a shorter amount of time and effort.

[0050] As shown in FIG. 3C, flow 307 shows how noise characteristics for receivers are precalculated. Flow 307 consists of identifying assumptions block 308, identify applicable noise susceptibility factors block 309, run SPICE programs for receiver noise characteristics block 310, modify noise susceptibility factors block 311, modify receiver characteristics block 312 and store 3-dimensional matrix for receiver block 313. In identify assumptions block 308 reasonable assumptions are identified which will ease the calculations, without substantially reducing the accuracy of the results, and a decision is made as to whether or not these assumptions should be incorporated. The assumptions which have been identified and incorporated in block 308 include an assumption that the wires are made of M3 material which will give the worst coupling behavior as a function of length. This assumption ensures that the worst case noise characteristics possible are identified. Additionally, a second assumption as to the driver signal minimum strength for NFET and PFET have been identified. As previously described, the minimum driver strength for NFET is 8 μm/pf and the minimum driver strength for PFET is 21 μm/pf.

[0051] Identify applicable noise susceptibility factors block 309 is an optional block in which the user of the system can make a determination beforehand of which factors that affect the noise sensitivity are the most critical and which will allow ease of use of the resulting matrix. For example, while the SPICE modeling technique uses 41 parameters per BJT device, the effects of some of these parameters are less important than others. This step is optional in that numerous SPICE iterations can be run and all 41 parameters for BJT can be analyzed to determine which of those parameters have the most effect on the noise characteristics of the BJT. However, if the user can precharacterize those parameters which they believe may be the applicable noise susceptibility factors in block 309 this analysis may be simplified. Feedback between identify applicable noise susceptibility factors 309 and identifying assumptions block 308 through link 315 is available for the modifications of assumptions based on the perceived applicable noise susceptibility factors.

[0052] In run SPICE for receiver noise characteristics block 310 the SPICE program is run to determine the noise characteristics for this specific configuration.

[0053] In modify noise susceptibility factor block 311, the user changes one of their perceived noise susceptibility factors to determine the effects on the noise characteristics calculated in block 310. If the user has not identified what they believe are applicable noise susceptibility factors in block 309, each of the parameters included in the SPICE model for the specific receiver can be modified to determine which parameters actually affect the noise characteristics.

[0054] Once the noise susceptibility factor is modified connector 315 returns to the run SPICE program for receiver noise characteristics block 310 to recalculate the noise characteristics for the updated parameters for the receiver. This interaction between run SPICE program for receiver noise characteristics block 310 and modified noise susceptibility factor 311 is repeated until each of the applicable parameters are thoroughly tested to determine the sensitivity of the noise characteristics on changes in the parameters. Once these sensitivities are determined, modify receiver characteristics block 312 is encountered which modifies the receiver characteristics to determine the new noise susceptibility factors for the updated receiver. Once the receiver characteristics are changed, link 316 returns to run SPICE program for the receiver noise characteristics 310. Once again, run SPICE program for receiver noise characteristics block 310 and modify noise susceptibility factor block 311 through link 315 together fully analyze the noise characteristics of the updated receiver for changes in the noise susceptibility factors. This process is continued until this type of receiver is thoroughly analyzed. Once the receiver is thoroughly analyzed, the results are stored in a 3-dimensional matrix for the receiver in block 313.

[0055] Once noise characteristics as a function of noise susceptibility factors are determined and stored, FIG. 4 allows these results to be used to identify the noise characteristics for a specific receiver. Flow 400 consists of identify receiver 405, identify receiver characteristics 410, access 3-dimensional table with bad capacitance total length and drive impedance 415, obtain noise characteristics 420, a decision step in which the question asked is if the noise sensitivity is within the limits block 425 and blocks for if the noise sensitivity is within the limits when the design is accepted in block 435 and a block for when the noise sensitivity is not within the limits which requires a modification in the circuit to reduce the noise sensitivity block 430. As previously described with respect to FIGS. 3A-3C, numerous 3-dimensional matrices are generated for various receivers with various characteristics. In order to determine the correct 3-dimensional matrix to access for this specific receiver, the receiver must be identified in block 405. Once the receiver itself is identified the receiver characteristics must be identified to determine which specific 3-dimensional matrix for this type of receiver must be accessed to determine the noise characteristics associated with this specific instance of the receiver. Once the receiver is identified in block 405 and receiver characteristics are identified in block 410, a specific 3-dimensional matrix for that receiver with those characteristics is accessed in block 415 with the bad capacitance, the total length of the victim wire susceptible to capacitance loading and the driver impedance. These three factors allow specific noise characteristics to be obtained from the 3-dimensional matrix in block 420. At this point, the noise characteristics could be displayed to the user, or a next step can be taken to determine whether the noise sensitivity is within acceptable limits in Step 425. If the noise sensitivity is within the limits, the design can be accepted as in block 435. If the noise sensitivity is not within the limits the circuit must be modified to reduce the noise characteristics of the circuit in block 430. These modifications may include shielding, movement of a victim wire, movement of an aggressor wire, or some other change in the circuit design to reduce the capacitance coupling felt on the victim wire.

[0056] More specifically, when accessing the three dimensional characterization matrix for a receiver, the maximum transition allowable at the output of the victim receiver is calculated. The maximum transition allowable should be less than a FET Threshold Voltage to ensure the transition, when it occurs, does not propagate the transition to elements connected to the victim receiver output. The total victim wire length and the driver impedance are used to calculate the maximum bad cap percentage which can be present on the victim wire such that the victim output transition amplitude is less than the maximum amplitude variation allowed. If the actual bad cap percentage is less than the maximum bad capacitance percentage, the receiver successfully passes the noise check. If, however, the actual bad capacitance percentage is greater than the maximum bad cap percentage the receiver fails this test. The extent of the failure can also be measured in that the greater the actual bad capacitance percentage is over the maximum bad capacitance, the greater the failure.

[0057] Alternatively, the characterization matrix may also be accessed by a determination of the actual transition amplitude at the victim receiver output. This parameter may be calculated by using actual bad capacitance, the total victim wire length and the victim driver impedance. The receiver passes the sensitivity test as long as the actual transition amplitude is less than the maximum allowed transition amplitude.

[0058] One of ordinary skill in the art would further understand that the characterization matrix of the present invention is not limited to three dimensions. For instance, the characterization matrix may be expanded to a fourth dimension through the addition of the actual capacitive loading on the victim receiver output. One of ordinary skill in the art would understand that the actual capacitive loading on the victim receiver output will affect the response of the receiver to the noise event. As the loading increases, a smaller response is expected and the receiver would be able to tolerate a higher bad capacitance on the victim wire.

[0059] The maximum allowable transition amplitude specification may be relaxed or reduced for certain circuit configurations. If the victim receiver is non-recoverable, meaning it contains a latch node or pre-charge node, no relaxation of the maximum allowable transition amplitude specification is permitted Alternatively, if the victim receiver is recoverable, meaning the static logic gate will always return to the original logic value after the noise event has subsided, the fanout of the victim receiver output may be traversed to identify the shortest path to a non-recoverable node. The maximum allowable transition amplitude may be increased, as a function of the number of recoverable stages which are present between the victim receiver and the non-recoverable node. This increase is because the intermediate stages will attenuate the noise event before it reaches the non-recoverable node.

[0060] Default cell characterization is used when a static noise check is performed on a circuit or cell input (a single input of the circuit or an input terminal) for which noise characterization data is not available. As a first step a library of noise characteristics for known cells is generated. This noise characteristic library, or library, is then used to determine the noise response of uncharacterized circuits or input terminals by matching the attributes (circuit type and parameters) of the uncharacterized input terminal with a cell from the library. Once a cell match is determined, the characterization data for the library cell, an individual entry within the library, can be used to accurately estimate the noise response of the uncharacterized input terminal.

[0061] The matching or selection process between the uncharacterized input terminal and the library cell involves identifying the library cell which includes a circuit type and circuit parameters identical or similar to the uncharacterized input terminal. Multiple input terminals of an uncharacterized cell may be considered independently and different library cells may be selected and used for different input terminals. The library includes, but is not limited to, common circuit types including inverters, nand circuits, nor circuits and dynamic circuits. As one of ordinary skill in the art would understand, the circuit parameters for the different circuit types vary and may include input capacitance for each of the common circuit types, the number of inputs to the circuit for the nand and nor circuits, the staging ratio and the forward inverter pn ratio for the dynamic circuits.

[0062] FIG. 5 is a block diagram outlining the creation of a database including a noise characteristic library for known cells. In Step 501, default schematics for common circuit types are created wherein each of the parameters of interest is adjustable. In Step 502, the schematics for each of the circuit types is replicated and the adjustable parameters are varied through a set of discrete values and a separate cell for each combination of the parameters is created within the library. Once each of the parameter combinations has been identified, Step 503 is used to calculate the noise characterization of each cell. In Step 504 each of the calculated noise characterizations, along with the circuit types and the values of the discrete parameter values is save within the library. The number of cells and therefore the size of the library is dependent on the number of types of circuits defined, the number of parameters, and the number of parameter values used. For example, in one embodiment of the current invention the library contains 1631 cells which includes 231 cells for dynamic circuits, 260 cells for inverters, and 190 cells each for 2, 3, and 4 input nand and 2, 3, and 4, input nor gate circuits. In this embodiment of the invention the parameters of the common circuit types are shown in the following table.

Dynamic
circuits	Inverter circuit	nand circuit	nor circuit
input capa-	input capacitance:	input capacitance:	input capacitance:
citance: 2.6 to	3 to	3 to	3 to
20 femtofarad	80 femtofarad	80 femtofarad	80 femtofarad
staging ratio	pn ratio	pn ratio	pn ratio
(width_dyn—	(width—	((width—	(width_p/
inv_n/w	p/width_n):	p*number	(width_n*
width_dyn—	0.5 to 6.0	of inputs)/	number of
n): 0.14 to 0.16		width_n):	inputs)):
		0.5 to 6.0	0.5 to 6.0
forward invert-
er pn ratio
(width_dyn—
inv_p/w
idth_dyn—
inv_n):
2.0 to 6.0

[0063] One of ordinary skill in the art would understand that the number of cells contained within the library is variable and that the foregoing describes only an example of the current invention; the invention covers libraries with different numbers of cells, different parameters and different formats.

[0064] FIGS. 6-13 are circuit diagrams of the various circuits included in the preferred embodiment of the library of the current invention including a dynamic circuit (FIG. 6), inverter (FIG. 7), 2, 3 and 4 input nand gate circuits (FIGS. 8-10 respectively) and 2, 3 and 4 input nor gate circuits (FIGS. 11-13 respectively).

[0065] FIG. 14 is diagram which includes the steps used by the present invention to determine the noise characteristics of an uncharacterized circuit or input terminal using the present invention. In Step 1401 an uncharacterized circuit is encountered during the static noise checking. In Step 1402 the topology of the uncharacterized circuit is identified and the number of inputs present in the circuit are determined. Step 1402 begins with an application of a pattern matching algorithm which compares the topology of the uncharacterized cell with the known topologies of circuits included in the library as further defined below. In a preferred embodiment the topology is compared to dynamic, inverter, nand and nor circuits (FIGS. 6-13).

[0066] In Step 1403 the circuit parameters of the uncharacterized circuit are measured to provide information for matching the input terminal to a cell within the library. In Step 1404 the library of noise characteristics, or the library, is searched to identify the entry or cell which is included in the library which most closely matches the topology and measured circuit parameters of the uncharacterized cell. In Step 1405 the expected noise response of the uncharacterized cell is calculated using the characterization results obtained from the appropriate library cell. Details of determining the circuit topology, Steps 1402 and determining the circuit parameters, Step 1403, of a preferred embodiment are described below.

[0067] The pattern matching algorithm to determine the circuit type is as follows:

[0068] 1) The input is defined to connect to a NOR circuit if all of the following conditions are met:

[0069] a) Input is connected to a n-fet gate, and

[0070] b) the source of that n-fet is connected to Ground, and

[0071] c) the drain of that n-fet is connected to the stage output, and

[0072] d) there are two or more n-fet's in parallel with their drains connected to the stage output, their sources connected to Ground and their gates connected to different signals, one of which being the input, and

[0073] e) the input is connected to a p-fet gate, and

[0074] f) there are two or more p-fet's in series between VDD and the stage output

[0075] A node is defined as the stage output (see condition (c)) if it is connected to both a p-fet drain and an n-fet drain. This node is found using a recursive traversal beginning with a fet whose gate is connected to the input. FIG. 15 shows the traversal algorithm, which illustrates the steps used to identify a node.

[0076] 2) The input is connected to a NAND circuit if all of the following conditions are met:

[0077] a) Input is connected to a p-fet gate, and

[0078] b) the source of that p-fet is connected to VDD, and

[0079] c) the drain of that p-fet is connected to the stage output, and

[0080] d) there are two or more p-fet's in parallel with their drains connected to the stage output, their sources connected to VDD and their gates connected to different signals, one of which being the input, and

[0081] e) the input is connected to an n-fet gate, and

[0082] f) there are two or more n-fet's in series between GND and the stage output.

[0083] 3) The input is defined to connect to an INVERTER circuit if all of the following conditions are met:

[0084] a) Input is connected to a p-fet gate, and

[0085] b) the source of that p-fet is connected to VDD, and

[0086] c) the drain of that p-fet is connected to the stage output, and

[0087] d) the input is connected to an n-fet gate, and

[0088] e) the source of that n-fet is connected to Ground, and

[0089] f) the drain of that n-fet is connected to the stage output.

[0090] 4) The input is defined to connect to a DYNAMIC circuit if all of the following conditions are met:

[0091] a) Input is connected to an n-fet gate, and

[0092] b) input is not connected to a p-fet gate, and

[0093] c) stage output is a precharge node

[0094] A node is defined as a precharge node if, and only if:

[0095] a) The node is connected to the input of an inverter (nfet gate and pfet gate). (This is referred to as the forward inverter); and

[0096] b) The node is connected to the drain of a feedback p-fet, whose source is connected to VDD, and whose gate is connected to the output of the forward inverter; and

[0097] c) The node is connected to VDD, and gate is connected to a clock signal.

[0098] Once the circuit type is determined, the parameters are calculated as follows:

[0099] a) Variable definitions:

[0100] input_pfet_gate_width=total width of pfets whose gates are connected to the input

[0101] input_nfet_gate_width=total width of nfets whose gates are connected to the input

[0102] parallel_nfet_width=total width of nfets in parallel with the input nfet

[0103] parallel_pfet_width=total width of pfets in parallel with the input pfet

[0104] fwd_inv_pfet_width=width of the pfet of the forward inverter

[0105] fwd_inv_nfet_width=width of the nfet of the forward inverter

[0106] precharge_nfet_drain_width=total width of nfets whose drains are connected to the precharge node

[0107] feedback_pfet_drain_width=width of the feedback pfet

[0108] b) Process specific constants:

[0109] Cgate=fet gate capacitance per unit width

[0110] K=staging ratio constant

[0111] INVERTER: 1$\mathrm{PN}\mathrm{ratio}=\frac{\mathrm{input\_pfet}\mathrm{\_gate}\mathrm{\_width}}{\mathrm{input\_nfet}\mathrm{\_gate}\mathrm{\_width}}$

[0112] Input cap=(input_pfet_gate_width+input_nfet_gate_width) * Cgate

[0113] NAND: 2$\mathrm{Number}\mathrm{of}\mathrm{inputs}=\mathrm{integer}\frac{\mathrm{input\_nfet}\mathrm{\_gate}\mathrm{\_width}+\mathrm{parallel\_nfet}\mathrm{\_width}}{\mathrm{input\_nfet}\mathrm{\_gate}\mathrm{\_width}}$$\mathrm{PN}\mathrm{ratio}=\frac{\mathrm{input\_pfet}\mathrm{\_gate}\mathrm{\_width}*\mathrm{number}\mathrm{of}\mathrm{inputs}}{\mathrm{input\_nfet}\mathrm{\_gate}\mathrm{\_width}}$

[0114] Input cap=(input_pfet_gate_width+input_nfet_gate_width) * Cgate

[0115] NOR: 3$\mathrm{Number}\mathrm{of}\mathrm{inputs}=\mathrm{integer}\frac{\mathrm{input\_pfet}\mathrm{\_gate}\mathrm{\_width}+\mathrm{parallel\_pfet}\mathrm{\_width}}{\mathrm{input\_pfet}\mathrm{\_gate}\mathrm{\_width}}$$\mathrm{PN}\mathrm{ratio}=\frac{\mathrm{input\_pfet}\mathrm{\_gate}\mathrm{\_width}}{\mathrm{input\_nfet}\mathrm{\_gate}\mathrm{\_width}*\mathrm{number}\mathrm{of}\mathrm{inputs}}$

[0116] Input cap=(input_pfet_gate_width+input_nfet_gate_width) * Cgate

[0117] DYNAMIC:

[0118] Input cap=input_nfet_gate_width * Cgate 4$\mathrm{Forward}\mathrm{inverter}\mathrm{pn}\mathrm{ratio}=\frac{\mathrm{fwd\_inv}\mathrm{\_pfet}\mathrm{\_width}}{\mathrm{fwd\_inv}\mathrm{\_nfet}\mathrm{\_width}}$$\mathrm{Staging}\mathrm{ratio}=\frac{\mathrm{fwd\_inv}\mathrm{\_nfet}\mathrm{\_width}}{\mathrm{precharge\_nfet}\mathrm{\_drain}\mathrm{\_width}*\frac{\mathrm{precharge\_nfet}\mathrm{\_drain}\mathrm{\_width}}{K*\mathrm{feedback\_pfet}\mathrm{\_drain}\mathrm{\_width}}}$

[0119] When the circuit type is identified and all parameters are calculated, the cell from the default library whose parameters most closely match those calculated for the uncharacterized cell input is selected. The characterization for that default cell is then used to determine the noise response of the uncharacterized cell input.

[0120] FIG. 16 is a block diagram of a system which implements the present invention. CPU 1601 runs the aforementioned programs from memory 1603. The library also resides in memory 1603. Circuit identifier and parameter identification functionalities are provided by software stored in memory 1603 and executed by CPU 1601. Input devices 1602 and output device 1604 are used by the user in interfacing with CPU 1601. CPU 1601 may be a general purpose microprocessor capable of supplying controller functionality by the execution of operating system and application software stored in memory 1603 and processing data and information also stored in memory 1603.

[0121] As one of ordinary skill in the art will readily appreciate from the disclosure of the present invention, processes, machines, manufacture, compositions of matter, means, methods, or steps, presently existing or later to be developed that perform substantially the same function or achieve substantially the same result as the corresponding embodiments described herein may be utilized according to the present invention. Accordingly, the appended claims are intended to include within their scope such processes, machines, manufacture, compositions of matter, means, methods, or steps.

Claims

1. A method of determining noise characteristics of a circuit of an integrated circuit comprising the steps of: classifying the circuit from the topology of said circuit to provide a circuit classification; measuring circuit parameters of said circuit to provide circuit parameter measurements; retrieving prestored noise characteristics using said circuit classification and said circuit parameter measurements; and calculating a noise response of the circuit using said prestored noise characteristics.

2. The method of claim 1 wherein said step of classifying said circuit is performed for each input terminal of said circuit.

3. The method of claim 2 wherein said step of classifying said circuit further comprises the step of identifying each input as one of an invertor, a NAND gate, a NOR gate or a dynamic circuit.

4. The method of claim 1 wherein said step of measuring circuit parameters further comprises measuring a bad capacitance.

5. The method of claim 1 wherein said step of measuring circuit parameters further comprises determining a pn ratio.

6. The method of claim 1 wherein said step of measuring circuit parameters further comprises counting a number of input terminals.

7. The method of claim 1 further comprising the steps of: precalculating default schematics for common circuit types; identifying circuit parameters for each of said default schematics; sweeping said circuit parameters through a set of discrete values; calculating a noise response for each of said noise characterizations; and storing default schematics, circuit parameters and said noise responses for later use.

8. A tool for identifying noise responses for circuits within an integrated circuit comprising: a circuit identifier which identifies a target circuit; a circuit parameter identifier which identifies parameters for said target circuit; a controller configured to lookup in a prestored table noise characteristics for said target circuit and said parameters; and a noise determiner which displays the noise characteristics for said target circuit with said parameters.

9. The tool of claim 8 wherein the circuit parameter identifier operates on each input terminal of said circuit.

10. The tool of claim 8 wherein the circuit identifier identifies the target circuit as one of an inverter, a NAND gate, a NOR gate or a dynamic circuit.

11. The tool of claim 9 wherein said circuit parameter identifier identifies a bad capacitance of each of said input terminals of said target circuit.

12. The tool of claim 9 wherein said circuit parameter identifier identifies a pn ratio of each of said input terminals of said circuit.

13. The tool of claim 8 wherein said circuit parameter identifier identifies the number of input terminals of said target circuit.

14. A tool for determining noise characteristics of an integrated circuit comprising: means for classifying the circuit from the topology of said circuit to provide a circuit classification; means for measuring circuit parameters of said circuit to provide circuit parameter measurements; means for retrieving prestored noise characteristics using said circuit classification and said circuit parameter measurements; and means for calculating a noise response of the circuit using said prestored noise characteristics.

15. A tool according to claim 14 wherein said means of classifying said circuit operates for each input terminal of said circuit.

16. A tool according to claim 14 wherein said means of classifying said circuit further comprises the means for identifying each input as one of an invertor, a NAND gate, a NOR gate or a dynamic circuit.

17. A tool according to claim 14 wherein said means for measuring circuit parameters further comprises a means for measuring a bad capacitance.

18. A tool according to claim 14 wherein said means of measuring circuit parameters further comprises a means for determining a pn ratio.

19. A tool according to claim 14 wherein said means of measuring circuit parameters further comprises a means for counting a number of input terminals.

20. A tool according to claim 14 further comprising: means for precalculating default schematics for common circuit types; means for identifying circuit parameters for each of said default schematics; means for sweeping said circuit parameters through a set of discrete values; means for calculating a noise response for each of said noise characterizations; and means for storing default schematics, circuit parameters and said noise responses for later use.