BIST (Built in self test)
In the era of submicron technology memories may develop failure during the operation within their expected lifetime. The circuit needs to put on test mode and validates that whether it is free of fault or not, with the assumption that they would not damage within the expected life span, as the technology is shrinking assumption does not hold for the modern day. In the virtue of submicron technology to conquer such problem redundant circuitry kept on chips which replace the faulty part. It tests circuit every time before they start up. Module sensitizes the fault and propagates the effect to the output of CUT, It generates many patterns for 0 to 2^n if there are n flip-flops. The BIST a redundant circuit kept on the chip which replaces the erroneous circuit with the error-free part. Reduce power consumption is the main focus point of the project. mainly describe the equipment scan be test edit itself in the circuit. In this additional software and hardware component are integrated on the board itself and there is no need of using the additional equipment to check the functionality and performance of the equipment, so by integrating the hardware and software will reduce the complexity of using external component of testing. the circuit will itself observe the performance BIST is used to test the complex circuit there are no external connections. It is the cheapest technique to implement the self test. Logic BIST implements most ATE function on chip so that the test cost can be reduced through less time less tester memory requirement or cheaper tester. Logic BIST applies large number of test patterns so that more defects can be detected. As LFSR and BIST are essential part of testing are executed at system clock modified LFSR is used which target to reduce power consumption, without affecting fault coverage and reduce switching transition we are able to reduce power consumption. Low correlation between test vector and switching activity leads to power dissipation. Hence modified LFSR help to reduce power leakage.
#. Fault Modeling
• Real defects (often mechanical) too numerous and often not analyzable
• A fault model identifies targets for testing
• A fault model makes analysis possible
• Effectiveness measurable by experiments
Some Real Defects in Chips
• Processing defects
• Missing contact windows
• Parasitic transistors
• Oxide breakdown
• Material defects
• Bulk defects (cracks, crystal imperfections)
• Surface impurities (ion migration)
• Time-dependent failures
• Dielectric breakdown
• Electro migration
• Packaging failures
• Contact degradation
• Seal leaks
When any of this part is affected may cause failure, so it becomes essential to test and remove the fault. Among them, a few are explained here.
1.... Stuck at Fault
A type of fault due to which cell does not change its state, it stuck at the previous state and become faulty, e.g. If any cell
Stuck at zero, it does not change its value from 0 to 1 it remains at 0. It stays in its state and this cause failure, due to this it doesn’t allow to fetch the accurate address or data value.
2..... Transition Faults
Transition fault is a special type of stuck at fault in which at write operation, it fails to write 1 and stay at its low state, hence it is called up transition fault, and if it unable to write 0 and stuck at 1 then this type of fat called us down transition fault. According to van de Goor [4] each cell must sustain an up transition (cell value goes from low to high) and a down transition (cell value goes from high to low) and be read after each transition before undergoing any further tra
3......Coupling Faults
Coupling faults are of three types [1].
• Inverse coupling fault: due to some reason changes in the value of one cell unexpectedly inverts the content of another cell. [13]
• Idempotent coupling fault: a change in one cell forces an affix logic value of one cell into another cell. [13]
• State coupling fault: a cell is forced to a stay in its state only if the coupling cell/line is in a given state (pattern sensitivity fault (PSF)). [13]
4..... Bridging fault
The Bridging fault is a type of fat which is determined by its logical level instead of transition write operation because it is caused due to the short circuit between two or more cells. We have found three types of bridging faults [13]
....AND Bridging fault:
This is called so because the logical value of two faulty cells is AND of the two shortened cells acronym as ABF
.....OR Bridging fault:
This is called so because the logical value of two faulty cells is OR of the two shortened cells acronym as OBF.
6..... State Coupling fault
SCF is also a type of logical fault because a coupled cell is forced to certain value only because coupling cell is in that state
7...... Retention Faults (RF)
A cell fails to regain its logic value after some time. This fault is caused by a broken pull-up resistor
8...... Neighborhood Pattern Sensitive
The NPSF is a type of pattern sensitive fault in which cells are influenced by the neighboring cells; contents of cells are changed automatically by the transition of the neighboring cell from 0 to 1 or 1 to 0. Three types of NPSF faults are found: Active, Passive, and Static.
9..... Address Decoder Faults
During the read-write operation (AFs) address decoder faults are the fault due to which decoders are not able to fetch accurate address to fetch data, to simplify such problems we consider a different system in which each cell are tested is called bit oriented memory. The Functional fault within the address decoder can be classified into four AFs For bit-oriented memories, because each cell is linked to a dedicated address, none of the faults listed above is individual. For example, when fault 1 occurs, then either fault 2 or fault 3 will occur as well. [4]
10..... Linked Faults
Linked faults are more complex fault than the other faults explained in this paper in this type of faults many faults are linked with each other i.e. One fault occurs due to another one
11...... Data Retention Fault
Memory loses its content spontaneously not caused by read-write.
12.....Recovery Fault
When some part of memory may not recover fast enough from its previous state, Sense amplifier recovery: after reading, writing long string of 0s and 1s, write recovery: write followed by read or write in a different location resulting in reading or writing at the same location
No comments:
Post a Comment