Method for generating test access procedures
AssigneeMitsubishi Electric Research Laboratories, Inc.
InventorMichael Bershteyn
Life of this patent
Estimating the term…
Abstract
A method is presented for generating test access procedures for digital circuits embedded within larger circuits. The method of the present invention utilizes logical implication techniques to restrict the set of initial possible signal combinations at the inputs of the circuit. The implications allow values to be calculated from input assignments already made. The implication step also allows the exclusion of other values as a result of other assignments already made. In addition, the method of the present invention includes the performance of a netlist transformation technique to model sequential circuits with iterative arrays for purposes of developing proper input signals and calculating expected output values in accordance with standard test synchronizing procedures.
Claims14
A method for generating test access sequences for an embedded block in a digital circuit, said digital circuit having a set of circuit signals including primary inputs for receiving a circuit test sequence, a set of embedded block terminals including inputs and output for the embedded block, primary outputs, and internal signals associated with access circuitry providing a T-path from the primary inputs to the inputs for the embedded block and from the outputs for the embedded block to the primary outputs, said method comprising the steps:
- loading a netlist representation of the digital circuit including the embedded block;
- identifying a first set and a second set of terminals from the set of embedded block terminals;
- loading a set of test targets for the embedded block to create a set of terminal signal values for testing the first set and the second set of terminals, wherein said test target loading step comprises defining a set of initial test targets, which includes the steps of:
- specifying in the set of test targets a first test sequence for the first set of terminals, said first test sequences comprising a first set of signal values and a first set of assigned time intervals for applying the first set of signal values;
- and
- specifying in the set of test targets a second test sequence for the second set of terminals, said second test sequence comprising a second set of signal values and a second set of assigned time intervals for applying the second set of signal values;
- and
- calculating a set of primary input signal values for the circuit test sequence in accordance with the set of test targets.
A method for generating test patterns for a digital circuit having sequential elements, wherein the test synchronization scheme for the sequential circuits comprises applying a set of data values to data inputs to the digital circuit, latching the set of data values by applying clock signals to clock inputs of the digital circuit, and subsequently reading a set of output values of the digital circuit, said method comprising:
- receiving an initial netlist representation of the digital circuit;
- transforming the initial netlist representation of the digital circuit to create a model of the digital circuit represented in the form of an iterative array providing model signal values equal to the values expected during testing of the digital circuit under the test synchronization scheme, wherein said transforming step comprises the steps of:
- identifying a portion of the netlist representation of the digital circuit having a signal which is both combinationally observable and sequentially observable;
- and
- duplicating the portion into a first copy having only a combinationally observable signal and a second copy having only a sequentially observable signal;
- and
- applying proper model equations for the combinationally observable signal and the sequentially observable signal.
A method for generating test patterns for a digital circuit having sequential elements, wherein the test synchronization scheme for the sequential circuits comprises applying a set of data values to data inputs of the digital circuit, latching the set of data values by applying clock signals to clock inputs of the digital circuit, and subsequently reading a set of output values of the digital circuit, said method comprising:
- receiving an initial netlist representation of the digital circuit;
- transforming the initial netlist representation of the digital circuit to create a model of the digital circuit represented in the form of an iterative array providing model signal values equal to the values expected during testing of the digital circuit under the test synchronization scheme, wherein said transforming step comprises:
- the steps of:
- first marking all signals of the digital circuit passing exclusively through combinational circuit elements to any primary circuit output;
- first storing for each circuit signal in the digital circuit a set of all synchronizing clock primary inputs for the circuit signal passing exclusively through combinational elements;
- first determining for every data input of every circuit element, a set of flip flops from which the data input can be reached exclusively through combinational elements;
- second storing for each data input of every circuit element the union of the sets of primary input clock signals for each of the sets of flip flops identified in said first determining step;
- second marking the data input for each circuit element if the intersection of the clock signals from said first storing step with the clock signals from said second storing step is non-empty;
- third marking every circuit signal having a path from the signal to any other signal that is identified in said second marking step which passes exclusively through combinational elements;
- duplicating every element that has an output identified in said first marking step and either of said second or third marking step into one circuit element that is only combinationally observable and another equivalent circuit element that is only sequentially observable, creating fanout at its input, and if a duplicated element's output is a fanout stem with every branch being either only combinationally observable or only sequentially observable, splitting the fanout by connecting its branches to the corresponding copy of the duplicated element;
- assigning for every sequential element whose output is sequentially observable but not combinationally observable the internal model for the element where the output for the element changes one time unit after application of the clock signal;
- and
- assigning for every sequential element whose output is combinationally observable but not sequentially observable the internal model for the element where the output for the element changes during the same time frame as the clock signal.
Dependencies are inferred from the claim text.