analysis of single event transient effects in analogue...
TRANSCRIPT
Universidad de Sevilla (AICIA)
Centro Nacional de Microelectrónica (IMB-CNM)
ANALYSIS OF SINGLE EVENT TRANSIENT EFFECTS IN ANALOGUE
TOPOLOGIES Fourth International Workshop on Analogue and Mixed Signal Integrated
Circuits For Space Applications (AMICSA) August 27th 2012
ESA/ ESTEC Noordwijk
F. Márquez, F. Muñoz, M. A. Aguirre, F. R. Palomo (Universidad de Sevilla-AICIA)
M. Ullán (Instituto de Microelectrónica de Barcelona)
Universidad de Sevilla (AICIA)
Centro Nacional de Microelectrónica (IMB-CNM)
Outline Introduction Motivation Description of the implemented tool SET emulation models Single Event Simulation Analyzer methodology Examples of tool performance
Conclusions and future work
27/08/2012 AMICSA’12 2
Universidad de Sevilla (AICIA)
Centro Nacional de Microelectrónica (IMB-CNM)
Introduction
High-energy particle impacts become critical as technology shrinks.
Single Event Effects (SEE) are a major concern not only in digital domain but also in analogue cells.
Dealing with the analysis of analogue
circuits is a challenging tread as the number of transistors increases.
27/08/2012 AMICSA’12 3
Universidad de Sevilla (AICIA)
Centro Nacional de Microelectrónica (IMB-CNM)
Motivation: SET sensitivity evaluation
The work presented has been carried out in colaboration with IMB-CNM and Arquimea. Need for a SET sensitivity evaluation in a reconfigurable System-
on-chip for radiation-hardening purposes.
A tool to evaluate the SET sensitivity at transistor level in complex circuits. Provide a useful information about critical nodes of the circuit
under test to the analog designer. Easy to use by the circuit designer.
Development of the tool will be continued under FT-Unshades (digital SEU sensitivity evaluation)
27/08/2012 AMICSA’12 4
Universidad de Sevilla (AICIA)
Centro Nacional de Microelectrónica (IMB-CNM)
INPUT NETLIST
AutomaticTool
Simulationparameters(nodes, Qc,
impact times,heuristics)
OceanScript
Cadencesoftware
Output File(Extracted
data)
Comparisonwith non-irradiated
performance
AUTOMATIC TOOL
InstrumentedNetlist
Single Event Simulation Analyzer
27/08/2012 AMICSA’12 5
The circuit designer has to define the test-bench of the circuit under test: A netlist that describes the circuit is
required as an input.
The tool automatically creates an instrumented netlist. SET models for impacts emulation are
added to every possible target. The user provides configuration
parameters for simulation and analysis of the circuit.
A simulation script is automatically
created. Compares an ideal (non-irradiated)
output and the signal affected by injected SETs.
Results of the analysis of circuit’s vulnerabilities are extracted.
Universidad de Sevilla (AICIA)
Centro Nacional de Microelectrónica (IMB-CNM)
INPUT NETLIST
AutomaticTool
Simulationparameters(nodes, Qc,
impact times,heuristics)
OceanScript
Cadencesoftware
Output File(Extracted
data)
Comparisonwith non-irradiated
performance
AUTOMATIC TOOL
InstrumentedNetlist
Single Event Simulation Analyzer
27/08/2012 AMICSA’12 6
The circuit designer has to define the test-bench of the circuit under test: A netlist that describes the circuit is
required as an input.
The tool automatically creates an instrumented netlist. SET models for impacts emulation are
added to every possible target. The user provides configuration
parameters for simulation and analysis of the circuit.
A simulation script is automatically
created. Compares an ideal (non-irradiated)
output and the signal affected by injected SETs.
Results of the analysis of circuit’s vulnerabilities are extracted.
Universidad de Sevilla (AICIA)
Centro Nacional de Microelectrónica (IMB-CNM)
INPUT NETLIST
AutomaticTool
Simulationparameters(nodes, Qc,
impact times,heuristics)
OceanScript
Cadencesoftware
Output File(Extracted
data)
Comparisonwith non-irradiated
performance
AUTOMATIC TOOL
InstrumentedNetlist
Single Event Simulation Analyzer The circuit designer has to define the
test-bench of the circuit under test: A netlist that describes the circuit is
required as an input.
The tool automatically creates an instrumented netlist. SET models for impacts emulation are
added to every possible target. The user provides configuration
parameters for simulation and analysis of the circuit.
A simulation script is automatically
created. Compares an ideal (non-irradiated)
output and the signal affected by injected SETs.
Results of the analysis of circuit’s vulnerabilities are extracted.
27/08/2012 AMICSA’12 7
Universidad de Sevilla (AICIA)
Centro Nacional de Microelectrónica (IMB-CNM)
SET emulation
27/08/2012 AMICSA’12 8
SET emulation based on charge injection models: Current sources with double exponential dynamics* Use of configurable parameters (AHDL implementation)
*REF: G. Messenger, “Collection of Charge on junction nodes from ion tracks”, IEEE Transactions on nuclear science, vol.29, nº 6, Dec. 1982
Universidad de Sevilla (AICIA)
Centro Nacional de Microelectrónica (IMB-CNM)
Single Event Simulation Analyzer Starting point: Input netlist
27/08/2012 AMICSA’12 9
To extract the Spectre netlist, the tool requires a test-bench for the analog scheme under test.
INPUT NETLIST
AutomaticTool
Simulationparameters(nodes, Qc,
impact times,heuristics)
OceanScript
Cadencesoftware
Output File(Extracted
data)
Comparisonwith non-irradiated
performance
AUTOMATIC TOOL
InstrumentedNetlist
Universidad de Sevilla (AICIA)
Centro Nacional de Microelectrónica (IMB-CNM)
Single Event Simulation Analyzer Starting point: circuit test-bench
27/08/2012 AMICSA’12 10
This test bench is the same used by the designer to test the performance of the analog circuits
Universidad de Sevilla (AICIA)
Centro Nacional de Microelectrónica (IMB-CNM)
Single Event Simulation Analyzer Starting point: circuit test-bench
27/08/2012 AMICSA’12 11
This test bench is the same used by the designer to test the performance of the analog circuits
Universidad de Sevilla (AICIA)
Centro Nacional de Microelectrónica (IMB-CNM)
Single Event Simulation Analyzer Starting point: circuit test-bench
27/08/2012 AMICSA’12 12
The only extra effort required is to place the SET models (provided to the user)
Universidad de Sevilla (AICIA)
Centro Nacional de Microelectrónica (IMB-CNM)
Single Event Simulation Analyzer From the input netlist, the tool:
• Step 1: generates an instrumented netlist • Step2 : creates a simulation script that automatically:
• analyzes the SET sensitivity of the analog circuit . • Extracts critical information and provides an output file with
the results.
27/08/2012 AMICSA’12 13
Universidad de Sevilla (AICIA)
Centro Nacional de Microelectrónica (IMB-CNM)
Single Event Simulation Analyzer Step 1: Netlist instrumentation: Technology information is
provided to the designer. SET emulation models are
added to generate a new netlist.
27/08/2012 AMICSA’12 14
INPUT NETLIST
AutomaticTool
Simulationparameters(nodes, Qc,
impact times,heuristics)
OceanScript
Cadencesoftware
Output File(Extracted
data)
Comparisonwith non-irradiated
performance
AUTOMATIC TOOL
InstrumentedNetlist
Universidad de Sevilla (AICIA)
Centro Nacional de Microelectrónica (IMB-CNM)
Define the technology Welcome to SET analyzer >> loadtech ST130nm.txt Successful
27/08/2012 AMICSA‘12 15
A technology file is loaded (provided to the designer) with the required information for SET models placement, making this process technology independent.
Example: SET Sensitivity for a 130 nm analog cell
Universidad de Sevilla (AICIA)
Centro Nacional de Microelectrónica (IMB-CNM)
Instrumented netlist Welcome to SET analyzer >> loadtech ST130nm.txt Successful >> source netlistCM130nm Source netlist netlistCM130nm successfully opened Successful
27/08/2012 AMICSA‘12 16
A Spectre netlist is loaded as an input for the tool, which creates an instrumented version adding the SET emulation models
A technology file is loaded (provided to the designer) with the required information for SET models placement and particle impacts, making this process technology independent.
Example: SET Sensitivity for a 130 nm analog cell
Universidad de Sevilla (AICIA)
Centro Nacional de Microelectrónica (IMB-CNM)
Script generation Step 2: Automatic script
generation
27/08/2012 AMICSA’12 17
• Configuration parameters should be defined by the user.
• Example: Impact nodes.
INPUT NETLIST
AutomaticTool
Simulationparameters(nodes, Qc,
impact times,heuristics)
OceanScript
Cadencesoftware
Output File(Extracted
data)
Comparisonwith non-irradiated
performance
AUTOMATIC TOOL
InstrumentedNetlist
Universidad de Sevilla (AICIA)
Centro Nacional de Microelectrónica (IMB-CNM)
Impact nodes selection >> listAlltargets Possible targets: SET_I0_M13 SET_I0_M12 SET_I0_M11 SET_I0_M10 SET_I0_M9 SET_I0_M8 SET_I0_M7 SET_I0_M6 SET_I0_M5 SET_I0_M4 SET_I0_M3 SET_I0_M2 SET_I0_M1 SET_I0_M0 Successful
27/08/2012 AMICSA’12 18
The tool provides a list of the possible targets which have to be properly configured for SET emulation
Example: SET Sensitivity for a 130 nm analog cell
Universidad de Sevilla (AICIA)
Centro Nacional de Microelectrónica (IMB-CNM)
Impact nodes selection >> listAlltargets Possible targets: SET_I0_M13 SET_I0_M12 SET_I0_M11 SET_I0_M10 SET_I0_M9 SET_I0_M8 SET_I0_M7 SET_I0_M6 SET_I0_M5 SET_I0_M4 SET_I0_M3 SET_I0_M2 SET_I0_M1 SET_I0_M0 Successful >> Addtarget SET_I0_M5 Successful >> Addtarget SET_I0_M7 Successful >> Addtarget SET_I0_M10 Successful
27/08/2012 AMICSA’12 19
We can add all the targets for a massive campaign or only a group of them to perform a more selective analysis at a sub-block level
The tool provides a list of the possible targets which have to be properly configured for SET emulation
Example: SET Sensitivity for a 130 nm analog cell
Universidad de Sevilla (AICIA)
Centro Nacional de Microelectrónica (IMB-CNM)
Script generation Step 2: Automatic script generation Configuration parameters should be defined:
• Impact nodes. • SET model parameters:
– Charge injected – Impact times
27/08/2012 AMICSA’12 20
Universidad de Sevilla (AICIA)
Centro Nacional de Microelectrónica (IMB-CNM)
SET model parameters >> listtargets Possible impacts in: Source Q T SET_I0_M7 (0.5p) (10n 20n) SET_I0_M5 (0.5p) (10n 20n) SET_I0_M10 (0.5p) (10n 20n) Successful
27/08/2012 AMICSA’12
21
Critical charge to be injected to emulate a particle impact (defined in the SEE models).
Example: SET Sensitivity for a 130 nm analog cell
Universidad de Sevilla (AICIA)
Centro Nacional de Microelectrónica (IMB-CNM)
SET model parameters >> listtargets Possible impacts in: Source Q T SET_I0_M7 (0.5p) (10n 20n) SET_I0_M5 (0.5p) (10n 20n) SET_I0_M10 (0.5p) (10n 20n) Successful
27/08/2012 AMICSA’12 22
Impact times. A generated SET can be more or less critical depending on the impact time selected.
Critical charge to be injected to emulate a particle impact (defined in the SEE models).
Example: SET Sensitivity for a 130 nm analog cell
Universidad de Sevilla (AICIA)
Centro Nacional de Microelectrónica (IMB-CNM)
SET model parameters >> listtargets Possible impacts in: Source Q T SET_I0_M7 (0.5p) (10n 20n) SET_I0_M5 (0.5p) (10n 20n) SET_I0_M10 (0.5p) (10n 20n) Successful >> setT SET_I0_M7 (13n 26n 31n ) Successful >> setQ SET_I0_M10 (0.25p) Successful
27/08/2012 AMICSA’12 23
Impact times. A generated SET can be more or less critical depending on the impact time selected.
Charge to be injected to emulate a particle impact (defined in the SET models).
Definition of impact times and charge
Example: SET Sensitivity for a 130 nm analog cell
Universidad de Sevilla (AICIA)
Centro Nacional de Microelectrónica (IMB-CNM)
SET model parameters >> listtargets Possible impacts in: Source Q T SET_I0_M7 (0.5p) (10n 20n) SET_I0_M5 (0.5p) (10n 20n) SET_I0_M10 (0.5p) (10n 20n) Successful >> setT SET_I0_M7 (13n 26n 31n ) Successful >> setQ SET_I0_M10 (0.25p) Successful >> listtargets Possible impacts in: Source Q T SET_I0_M7 (0.5p) (13n 26n 31n ) SET_I0_M5 (0.5p) (10n 20n) SET_I0_M10 (0.25p) (10n 20n) Successful
27/08/2012 AMICSA’12 24
Impact times. A generated SET can be more or less critical depending on the impact time selected.
Charge to be injected to emulate a particle impact (defined in the SET models).
Example: SET Sensitivity for a 130 nm analog cell
Universidad de Sevilla (AICIA)
Centro Nacional de Microelectrónica (IMB-CNM)
Script generation Step 2: Automatic script generation Configuration parameters should be defined:
• Impact nodes. • SEE model parameters:
– Critical charge – Impact times
• Outputs selected.
27/08/2012 AMICSA’12 25
Universidad de Sevilla (AICIA)
Centro Nacional de Microelectrónica (IMB-CNM)
Outputs selection >> listallnodes Possible output nodes: /net19 /vcn /vcp /vdd /net10 /Vout /net11 /vss /0 /I0/net12 /I0/net16 /I0/net32 /I0/net33 /I0/net24 /I0/net45 /I0/Vout0 /I0/net42 Successful
27/08/2012 AMICSA’12
26
A list of available nodes to be selected for analysis is provided, including internal nodes of different sub-blocks.
Example: SET Sensitivity for a 130 nm analog cell
Universidad de Sevilla (AICIA)
Centro Nacional de Microelectrónica (IMB-CNM)
Outputs selection >> listallnodes Possible output nodes: /net19 /vcn /vcp /vdd /net10 /Vout /net11 /vss /0 /I0/net12 /I0/net16 /I0/net32 /I0/net33 /I0/net24 /I0/net45 /I0/Vout0 /I0/net42 Successful >> addOutput /Vout Successful >> addOutput /I0/net32 Successful
27/08/2012 AMICSA’12 27
A list of available nodes to be selected for analysis is provided, including internal nodes of different sub-blocks.
- The tool will only consider for analysis the nodes selected by the designer
- Only the information of this nodes will be saved during simulations
Example: SET Sensitivity for a 130 nm analog cell
Universidad de Sevilla (AICIA)
Centro Nacional de Microelectrónica (IMB-CNM)
Outputs comparison
27/08/2012 AMICSA’12 28
Two main transient parameters have been considered: • Recovery time •Voltage deviation
Example: SET Sensitivity for a 130 nm analog cell
Universidad de Sevilla (AICIA)
Centro Nacional de Microelectrónica (IMB-CNM)
Transient simulations are performed
27/08/2012 AMICSA’12 29
Example: SET Sensitivity for a 130 nm analog cell
Universidad de Sevilla (AICIA)
Centro Nacional de Microelectrónica (IMB-CNM)
Analysis of the results Output file Results are automatically
extracted, analyzed and saved into an output file.
The process is transparent to the user.
27/08/2012 AMICSA’12 30
INPUT NETLIST
AutomaticTool
Simulationparameters(nodes, Qc,
impact times,heuristics)
OceanScript
Cadencesoftware
Output File(Extracted
data)
Comparisonwith non-irradiated
performance
AUTOMATIC TOOL
InstrumentedNetlist
Universidad de Sevilla (AICIA)
Centro Nacional de Microelectrónica (IMB-CNM)
Database is automatically generated
27/08/2012 AMICSA’12 31
Output Impact Qc Timp Trec Vmax /I0/net32 I0_M5 5,00E-13 1,00E-08 5.100.000 0.109233 /Vout I0_M5 5,00E-13 1,00E-08 6.400.000 0.225535 /I0/net32 I0_M5 5,00E-13 2,00E-08 5.100.000 0.109186 /Vout I0_M5 5,00E-13 2,00E-08 6.400.000 0.225493 /I0/net32 I0_M7 5,00E-13 1.3e-08 2.300.000 0.426629 /Vout I0_M7 5,00E-13 1.3e-08 3.600.000 0.183520 /I0/net32 I0_M7 5,00E-13 2.6e-08 2.300.000 0.370461 /Vout I0_M7 5,00E-13 2.6e-08 3.000.000 0.147083 /I0/net32 I0_M10 2,50E-13 1,00E-08 2.800.000 0.205726 /Vout I0_M10 2,50E-13 1,00E-08 3.900.000 0.186725 /I0/net32 I0_M10 2,50E-13 2.0e-08 3.300.000 0.143872 /Vout I0_M10 2,50E-13 2.0e-08 4.600.000 0.231184 . . .
Example: SET Sensitivity for a 130 nm analog cell
Universidad de Sevilla (AICIA)
Centro Nacional de Microelectrónica (IMB-CNM)
Database is automatically generated
27/08/2012 AMICSA’12 32
Example: SET Sensitivity for a 130 nm analog cell
Output Impact Qc Timp Trec Vmax /I0/net32 I0_M5 5,00E-13 1,00E-08 5.100.000 0.109233 /Vout I0_M5 5,00E-13 1,00E-08 6.400.000 0.225535 /I0/net32 I0_M5 5,00E-13 2,00E-08 5.100.000 0.109186 /Vout I0_M5 5,00E-13 2,00E-08 6.400.000 0.225493 /I0/net32 I0_M7 5,00E-13 1.3e-08 2.300.000 0.426629 /Vout I0_M7 5,00E-13 1.3e-08 3.600.000 0.183520 /I0/net32 I0_M7 5,00E-13 2.6e-08 2.300.000 0.370461 /Vout I0_M7 5,00E-13 2.6e-08 3.000.000 0.147083 /I0/net32 I0_M10 2,50E-13 1,00E-08 2.800.000 0.205726 /Vout I0_M10 2,50E-13 1,00E-08 3.900.000 0.186725 /I0/net32 I0_M10 2,50E-13 2.0e-08 3.300.000 0.143872 /Vout I0_M10 2,50E-13 2.0e-08 4.600.000 0.231184 . . .
Universidad de Sevilla (AICIA)
Centro Nacional de Microelectrónica (IMB-CNM)
Tool performance The analysis of several
analog topologies has been performed: 130nm technology Amplifier topologies:
a) Two stage Miller OpAmp
b) Telescopic OpAmp c) Fully cascoded two
stage OpAmp d) Current mirror
OpAmp (OTA)
27/08/2012 AMICSA’12 33
V
Vin
in
+
+
V
Vout
out
I
I
B
B
M
M
1
1
M
M
2
2M
M
3
3
MM4 4M M5 5
M
M
7
7
M M6 6 M
M
7
7
I
I
B
B
V
V
in
in
-
-
V
V
out
out
V
Vcp
cp
V
V
cn
cn
V
V
in
in
+
+
M
M
1
1
M
M
2
2M
M
3
3
MM
44
M M5 5M
M6
6
M
M
8
8
M
M
9
9
VV
in
in
+
+
VV
in
in
-
-
I
I
B
B
M
M1
1M
M2
2
M
M
3
3
M M4 4 M M5 5 M M6 6
MM
77
MM
88
M
M
9
9
MM1010
MM
1111
MM
1212
M
M
13
13
II
B
B
VVout
out
V Vcp cp
VV
cncn
22
VV
in
in
+
+
VV
cn
cnMM1
1M
M2
2
MM44 M M5 5
MM66 M M7 7
MM
33
MM
99
MM88
M M10 10
M M11 11
VV
in
in
-
-
MM
1212
MM
1313
(a) (b)
(c) (d)
Vin-
Vout
Example: SET Sensitivity for 130 nm analog cells
Universidad de Sevilla (AICIA)
Centro Nacional de Microelectrónica (IMB-CNM)
Tool performance The analysis of several
analog topologies has been performed: Test-bench:
Input signal: • A= 400mV • Fin= 200MHz (T=5ns)
27/08/2012 AMICSA’12 34
V
Vin
in
+
+
V
Vout
out
I
I
B
B
M
M
1
1
M
M
2
2M
M
3
3
MM4 4M M5 5
M
M
7
7
M M6 6 M
M
7
7
I
I
B
B
V
V
in
in
-
-
V
V
out
out
V
Vcp
cp
V
V
cn
cn
V
V
in
in
+
+
M
M
1
1
M
M
2
2M
M
3
3
MM
44
M M5 5M
M6
6
M
M
8
8
M
M
9
9
VV
in
in
+
+
VV
in
in
-
-
I
I
B
B
M
M1
1M
M2
2
M
M
3
3
M M4 4 M M5 5 M M6 6
MM
77
MM
88
M
M
9
9
MM1010
MM
1111
MM
1212
M
M
13
13
II
B
B
VVout
out
V Vcp cp
VV
cncn
22
VV
in
in
+
+
VV
cn
cnMM1
1M
M2
2
MM44 M M5 5
MM66 M M7 7
MM
33
MM
99
MM88
M M10 10
M M11 11
VV
in
in
-
-
MM
1212
MM
1313
(a) (b)
(c) (d)
Vin-
Vout
Example: SET Sensitivity for 130 nm analog cells
Universidad de Sevilla (AICIA)
Centro Nacional de Microelectrónica (IMB-CNM)
Example: SET Sensitivity for 130 nm analog cells
Tool performance
27/08/2012 AMICSA’12 35
V
Vin
in
+
+
V
Vout
out
I
I
B
B
M
M
1
1
M
M
2
2M
M
3
3
MM4 4M M5 5
M
M
7
7
M M6 6 M
M
7
7
I
I
B
B
V
V
in
in
-
-
V
V
out
out
V
Vcp
cp
V
V
cn
cn
V
V
in
in
+
+
M
M
1
1
M
M
2
2M
M
3
3
MM
44
M M5 5M
M6
6
M
M
8
8
M
M
9
9
VV
in
in
+
+
VV
in
in
-
-
I
I
B
B
M
M1
1M
M2
2
M
M
3
3
M M4 4 M M5 5 M M6 6
MM
77
MM
88
M
M
9
9
MM1010
MM
1111
MM
1212
M
M
13
13
II
B
B
VVout
out
V Vcp cp
VV
cncn
22
VV
in
in
+
+
VV
cn
cnMM1
1M
M2
2
MM44 M M5 5
MM66 M M7 7
MM
33
MM
99
MM88
M M10 10
M M11 11
VV
in
in
-
-
MM
1212
MM
1313
(a) (b)
(c) (d)
Vin-
Vout
Longest Trec= 4.4ns
Longest Trec= 6.4ns Longest Trec= 6.3ns
Largest Vmax = 290mV
The analysis of several analog topologies has been performed: Test-bench:
Input signal: • A= 400mV • Fin= 200MHz (T=5ns)
Longest Trec= 5.1ns Largest Vmax = 394mV
Largest Vmax = 390mV Largest Vmax = 390mV
Universidad de Sevilla (AICIA)
Centro Nacional de Microelectrónica (IMB-CNM)
Conclusions and future work A first version of the tool for SET sensitivity analysis in
analog schemes has been designed Allows a rapid SET sensitivity analysis of critical nodes at
schematic level. Technology independent.
Tool posibilities and future trends Implementation of alternative SET models. Extension of the fault injection to layout simulation
(charge sharing) Unify this tool with FT-Unshades to obtain a mixed signal
SET sensitivity analizer.
27/08/2012 AMICSA’12 36
Universidad de Sevilla (AICIA)
Centro Nacional de Microelectrónica (IMB-CNM)
Thanks for your attention
27/08/2012 AMICSA’12 37
Contact:
[email protected] AICIA-Universidad de Sevilla
Camino de los descubrimientos S/N 41092 Sevilla (Spain)