Post on 30-Jan-2016
description
CMP238: Projeto e Teste de Sistemas VLSI
Marcelo Lubaszewski
Aula 1 - Teste
PPGC - UFRGS
2005/I
Lecture 2 - Fault Modeling• Defects, Errors, and Faults• Why model faults?• Some real defects in VLSI and PCB• Common fault models• Stuck-at faults
• Single stuck-at faults• Fault equivalence• Fault dominance and checkpoint theorem• Classes of stuck-at faults and multiple faults
• Other Common Faults• Faults in FPGAs
Defects, Faults, and Errors• Defect: unintendent difference between the implemented
HW and its intendent design– May or may not cause a system failure
• Fault: representation of a defect at the abstracted function level
Defect X Fault
Imperfections in the HW Imperfections in the function
Defects, Faults, and Errors• Error: Manifestation of a fault that results in
incorrect circuit (system) outputs or states– Caused by faults
• Failure: Deviation of a circuit or system from its specified behavior– Fails to do what it should do– Caused by an error
• Defect --> Fault ---> Error ---> Failure
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 defects Dielectric breakdown Electromigration . . .
Packaging defects Contact degradation Seal leaks. . .
Ref.: M. J. Howes and D. V. Morgan, Reliability and Degradation - Semiconductor Devices and Circuits, Wiley, 1981.
Example
• Defect: a short to ground
• Fault: signal b stuck at logic 0
• Error: z has the wrong value if a = b = 1
a
b
1
1z
0 1
0
Example
• Defect: a short to ground
• Fault: signal b stuck at logic 0
• Error: z has the wrong value if a = b = 1
• But, if a = 0, fault exists, but no error!
a
b
0
1z
1 1
0
Why Model Faults?• Real defects (often mechanical) too numerous and
often not analyzable
• A fault model identifies targets for testing– Model faults most likely to occur
• Fault model limits the scope of test generation – Create tests only for the modeled faults
• A fault model makes analysis possible– Associate specific defects with specific test patterns
• Effectiveness measurable by experiments– Fault coverage can be computed for specific test patterns
to reflect its effectiveness
Common Fault Models• Single stuck-at faults
• Transistor open and short faults
• Memory faults
• PLA faults (stuck-at, cross-point, bridging)
• Functional faults (processors)
• Delay faults (transition, path)
• Analog faults
Single Stuck-at Fault• Three properties define a single stuck-at fault
• Only one line is faulty• The faulty line is permanently set to 0 or 1• The fault can be at an input or output of a gate
Single Stuck-at Fault• Three properties define a single stuck-at fault
• Only one line is faulty• The faulty line is permanently set to 0 or 1• The fault can be at an input or output of a gate
• Example: NAND gate has 3 fault sites ( ) and 6 single stuck-at faults
a
b
1
1z
s-a-0 fault, s-a-1 fault
Single Stuck-at Fault• Three properties define a single stuck-at fault
• Only one line is faulty• The faulty line is permanently set to 0 or 1• The fault can be at an input or output of a gate
• Example: NAND gate has 3 fault sites ( ) and 6 single stuck-at faults
a
b
1
1z
1 (0)
1
Faulty circuit valueGood circuit value
s-a-0
Single Stuck-at Fault• Three properties define a single stuck-at fault
• Only one line is faulty• The faulty line is permanently set to 0 or 1• The fault can be at an input or output of a gate
• Example: NAND gate has 3 fault sites ( ) and 6 single stuck-at faults
a
b
1
1z
1 (0)
1
Test vector for a s-a-0 fault
Faulty circuit valueGood circuit value
s-a-0
Single Stuck-at Fault
a
b
c
d
e
f
1
0
g h i 1
j
k
z
0
1
1
Good circuit value
Example: XOR circuit has 12 fault sites and 24 single stuck-at faults
Single Stuck-at Fault
a
b
c
d
e
f
1
0
g h i 1
s-a-0j
k
z
0(1)
1(0)
1
Test vector for h s-a-0 fault
Good circuit valueFaulty circuit value
Example: XOR circuit has 12 fault sites and 24 single stuck-at faults
Fault Equivalence
• Fault equivalence: Two faults f1 and f2 are equivalent if all tests that detect f1 also detect f2.
• If faults f1 and f2 are equivalent then the corresponding faulty functions are identical.
• Fault collapsing: All single faults of a logic circuit can be divided into disjoint equivalence subsets, where all faults in a subset are mutually equivalent. A collapsed fault set contains one fault from each equivalence subset.
Equivalence Rules
sa0 sa1
sa0 sa1
sa0 sa1
sa0 sa1
sa0 sa1
sa0 sa1
sa0 sa1
sa0 sa1
sa0 sa1
sa0 sa1
sa0 sa1
sa0 sa1
sa0
sa1
sa0
sa0sa1
sa1
sa0sa1
AND
NAND
OR
NOR
WIRE
NOT
FANOUT
Equivalence Rules
sa0 sa1
sa0 sa1
sa0 sa1
sa0 sa1
sa0 sa1
sa0 sa1
sa0 sa1
sa0 sa1
sa0 sa1
sa0 sa1
sa0 sa1
sa0 sa1
sa0
sa1
sa0
sa1
sa0
sa0sa1
sa1
sa0
sa0
sa0sa1
sa1
sa1
AND
NAND
OR
NOR
WIRE
NOT
FANOUT
Equivalence Example
sa0 sa1sa0 sa1
sa0 sa1
sa0 sa1
sa0 sa1
sa0 sa1
sa0 sa1
sa0 sa1
sa0 sa1
sa0 sa1
sa0 sa1
sa0 sa1
sa0 sa1sa0 sa1
sa0 sa1
Equivalence Example
sa0 sa1
sa0 sa1
sa0 sa1
sa0 sa1sa0 sa1
sa0 sa1
sa0 sa1sa0 sa1
sa0 sa1
sa0 sa1sa0 sa1
sa0 sa1
sa0 sa1sa0 sa1
sa0 sa1 16Collapse ratio = ----- = 0.533 30
Equivalence Example
sa0 sa1sa0 sa1
sa0 sa1
sa0 sa1
sa0 sa1
sa0 sa1
sa0 sa1
sa0 sa1
sa0 sa1
sa0 sa1
sa0 sa1
sa0 sa1
sa0 sa1
sa0 sa1
sa0 sa1
sa0 sa1
Equivalence Example
sa0 sa1sa0 sa1
sa0 sa1
sa0 sa1
sa0 sa1
sa0 sa1
sa0 sa1
sa0 sa1
sa0 sa1
sa0 sa1
sa0 sa1
sa0 sa1
sa0 sa1
sa0 sa1
sa0 sa1
sa0 sa1
Faults in redremoved byequivalencecollapsing
20Collapse ratio = ----- = 0.625 32
Fault Dominance• If all tests of some fault F1 detect another fault F2,
then F2 is said to dominate F1.
• Dominance fault collapsing: If fault F2 dominates F1, then F2 is removed from the fault list.
• When dominance fault collapsing is used, it is sufficient to consider only the input faults of Boolean gates. See the next example.
Dominance Example
s-a-1F1
s-a-1F2
Dominance Example
s-a-1F1
s-a-1F2 001
110 010 000101 100
011
All tests of F2
Only test of F1
Dominance Example
s-a-1F1
s-a-1F2 001
110 010 000101 100
011
All tests of F2
Only test of F1s-a-1
s-a-1
s-a-1 s-a-0
A dominance collapsed fault set
Dominance Example
sa1
sa1
sa0 sa1
sa0sa1
sa1
sa0sa1
sa1
sa0sa1
sa1
sa0sa1
sa1
Dominance Example
sa1
sa1
sa0 sa1
sa0sa1
sa1
sa0sa1
sa1
sa0sa1
sa1
sa0sa1
sa1
Dominance Example
sa0sa1
sa0 sa1
sa1
sa1
sa1
sa0 sa1
sa1
sa1
sa0 sa1
sa0
sa0
sa0
sa1
sa1
sa0 sa1
Dominance Example
sa0sa1
sa0 sa1
sa1
sa1
sa1
sa0 sa1
sa1
sa1
sa0 sa1
sa0
sa0
sa0
sa1
sa1
sa0 sa1
Faults in redremoved byequivalencecollapsing
Checkpoints• Primary inputs and fanout branches of a combinational
circuit are called checkpoints.
Total fault sites = 16
Checkpoints• Primary inputs and fanout branches of a combinational
circuit are called checkpoints.
Total fault sites = 16
Checkpoints ( ) = 10
Checkpoints• Primary inputs and fanout branches of a combinational
circuit are called checkpoints.
Total fault sites = 16
Checkpoints ( ) = 10
Checkpoint theorem: A test set that detects all single (multiple) stuck-at faults on all checkpoints of a
combinational circuit, also detects all single (multiple) stuck-at faults in that circuit.
Why Fault Collapsing?
• Memory & CPU- Time saving– To ease the burden for test generation and fault
simulation in testing
Multiple Stuck-at Faults
• A multiple stuck-at fault means that any set of lines is stuck-at some combination of (0,1) values.
• The total number of single and multiple stuck-at faults in a circuit with k single fault sites is 3k-1.
• A single fault test can fail to detect the target fault if another fault is also present, however, such masking of one fault by another is rare.
• Statistically, single fault tests cover a very large number of multiple faults.
Why Single Stuck- At Fault Model?• Complexity is greatly reduced.
– Many different physical defects may be modeled by the same logical single stuck- at fault.
• Single stuck- at fault is technology independent– Can be applied to TTL, ECL, CMOS, etc.
• Single stuck- at fault is design style independent– Gate Arrays, Standard Cell, Custom VLSI
• Even when single stuck- at fault does not accurately model some physical defects, the tests derived for logic faults are still valid for most defects.
• Single stuck- at tests cover a large percentage of multiple stuck- at faults.
Bridging Faults• Two or more normally distinct points (lines) are
shorted together– Logic effect depends on technology– Wired- AND for TTL
– Wired- OR for ECL
– CMOS?
Transistor (Switch) Faults
• MOS transistor is considered an ideal switch and two types of faults are modeled:
• Stuck-open -- a single transistor is permanently stuck in the open state.
• Stuck-short -- a single transistor is permanently shorted irrespective of its gate voltage.
• Detection of a stuck-open fault requires two vectors.
• Detection of a stuck-short fault requires the measurement of quiescent current (IDDQ).
CMOS Transistor Stuck- Short
• Transistor stuck- on may cause ambiguous logic level– depends on the relative impedances of the pull- up & pull-
down networks
• When input is low, both P and N transistors are conducting causing increased quiescent current, called IDDQ fault.
CMOS Transistor Stuck- OPEN
• Transistor stuck- open may cause output floating.
Functional Faults
• Fault effects modeled at a higher level than logic for function modules, such as– Decoders – Multiplexers – Adders – Counters – RAMs – ROMs
Functional Faults of Decoder
• f( L i /L j ): Instead of line L i , Line L j is selected
• f( L i /L i +L j ): In addition to L i , L j is selected
• f( L i /0): None of the lines are selected
Memory Faults
• Parametric Faults – Output Levels– Power Consumption– Noise Margin– Data Retention Time
• Functional Faults – Stuck Faults in Address Register, Data Register,
and Address Decoder– Cell Stuck Faults– Adjacent Cell Coupling Faults– Pattern- Sensitive Faults
Memory Faults
• Pattern- sensitive faults: the presence of a faulty signal depends on the signal values of the nearby points – Most common in DRAMs
• Adjacent cell coupling faults – Pattern sensitivity between a pair of cells
PLA Faults
• Stuck Faults
• Crosspoint Faults– Extra/ Missing Transistors
• Bridging Faults
• Break Faults
Missing Crosspoint Faults in PLA
• Missing crosspoint in AND- array – Growth fault
• Missing crosspoint in OR- array – Disappearance fault
Extra Crosspoint Faults in PLA
• Extra crosspoint in AND- array – Shrinkage or disappearance fault
• Extra crosspoint in OR- array – Appearance fault
Gate- Delay- Fault
• Slow to rise, slow to fall – x is slow to rise when channel resistance R1 is
abnormally high
Gate- Delay- Fault
• Disadvantage:– Delay faults resulting from the sum of several
small incremental delay defects may not be detected.
Path- Delay- Fault
• Propagation delay of the path exceeds the clock interval.
• The number of paths grows exponentially with the number of gates.
State Transition Graph
• Each state transition is associated with a 4- tuple:– source state, input, output, destination state
Single State Transition Fault Model
• A fault causes a single state transition to a wrong destination state.
Faults in FPGAs
ffF1F2F3F4
Configuration Memory CellM
M
M M M M M MLUT
BlockRAM
SEU(Bit flip)
clk
clk
E1E2E3
E1E2
E1E3
E2E3
Permanent faults: same ASIC models apply
But for transients ...
Virtex (Xilinx)
FPGA building blocks:
Effect of Transients in SRAM-based FPGAs
ffF1F2F3F4
Configuration Memory CellM
M
M M M M M MLUT
BlockRAM
SEU(Bit flip)
clk
clk
E1E2E3
E1E2
E1E3
E2E3
Possible Bit flip Transient effect Corrected at the next load
Virtex (Xilinx)
CLB Comb. Logic:~0.5 % of the FPGA sensitive area
Effect of Transients in SRAM-based FPGAs
ffF1F2F3F4
Configuration Memory CellM
M
M M M M M MLUT
BlockRAM
SEU(Bit flip)
clk
clk
E1E2E3
E1E2
E1E3
E2E3
Bit flip Transient effect Corrected at the next load
Virtex (Xilinx)
CLB Flip-flops:~0.5 % of the FPGA sensitive area
Effect of Transients in SRAM-based FPGAs
ffF1F2F3F4
Configuration Memory CellM
M
M M M M M MLUT
BlockRAM
SEU(Bit flip)
clk
clk
E1E2E3
E1E2
E1E3
E2E3
Bit flip Permanent effect Corrected by reconfiguration
Virtex (Xilinx)
CLB LUTs:~8% of the FPGA sensitive area
Effect of Transients in SRAM-based FPGAs
ffF1F2F3F4
Configuration Memory CellM
M
M M M M M MLUT
BlockRAM
SEU(Bit flip)
clk
clk
E1E2E3
E1E2
E1E3
E2E3
Virtex (Xilinx)
Short or open circuit Corrected by reconfiguration
Routing and CLB customization:~91.0 % of the FPGA sensitive area
Summary
• Fault models are analyzable approximations of defects and are essential for a test methodology.
• For digital logic single stuck-at fault model offers best advantage of tools and experience.
• Many other faults (bridging, stuck-open and multiple stuck-at) are largely covered by stuck-at fault tests.
• Stuck-short and delay faults and technology-dependent faults require special tests.
• Memory and analog circuits need other specialized fault models and tests.
• Transient faults may have permanent effects in FPGAs