Page 1
Predictive versus Experimental
Approach for the Evaluation of Global
Robustness of Digital Apparatus
Piero Belforte, Paolo Fogliati
Luca Giacomello, Flavio Maggioni CSELT
Luigi Bragagnini Hewlett Packard
Carla Giachino, Emmanuel Leroux High Design Technology
Flavio Canavero Polytechnic of Turin
Page 2
Outline
• Hardware robustness issues
• Tool requirements for testing hardware robustness
• Validation requirements for predictive tools
• The THRIS environment
• Validation examples
• THRIS application examples
Page 3
Achieving Hardware Robustness: Key Issues
Hardware design phase
- signal integrity optimization (ringing, crosstalk, ground bounce, power supply distribution analysis), timing problem control
- filtering/shielding optimization (EMI simulations)
- reliability evaluation (components, architectural conceps, thermal analysis)
Hardware/Software Integration phase
- fault insertion, noise injection, experimental EMI precompliance tests
EMC qualification phase
- emission/susceptibility conformity tests
Overall Fault Tolerance Verification phase
- fault insertion
Installation phase
- environment interaction evaluation (grounding, shielding, EMI tests)
Page 4
Signal Integrity
• Most critical to manage:
– modeling for high speed/EMI
– simultaneous switching noise
• Less critical to manage:
– ringing/reflection control
– propagation delay evaluation
– crosstalk control
References:
A High-Performance Environment for Modelling and Simulation of Digital Systems - 1993 HP
High Speed Digital Systems Design & Test Symposium
Advanced Simulation and Modeling for Telecom System Hardware Design - 1994 HP
ATM/Broadband Design Symposium
Page 5
Fault Insertion / Noise Injection
– Typical procedure for physical fault insertion:
• put the system in normal operation
• introduce the fault: verify the system behavior (alarms, reconfiguration, etc)
• remove the fault: verify the system behavior (manual/automatic recovery)
Fault Insertion: is a procedure to verify the behavior of an apparatus
in case of hardware faults
Noise Injection: is a procedure to verify noise margins within the
apparatus
Page 6
Hardware Reliability Prediction
Prediction methodology applicable to existing
products, when complete data about electronic
devices used in the system are available in the CAD
environment
A “parts count” methodology applicable at early stage
of the project, for reliability fast predictions“Early” evaluation
Accurate prediction
Page 7
Requirements for a Reliable SI/EMI
Simulation
• Accurate modelling techniques
• Real exploitability of the simulation tool:
• powerful simulation engine (speed and complexity
management)
• integrability in the design flow
• customizability
• Experimental validation
Page 8
Accurate Modelling
A good prediction requires accurate models of components
Data supported by
IBIS standard
Enough for low-speed
(< 100MHz) applications
IBIS
BTM*
+ Necessary for high-speed
( > 100MHz), EMI
* BTM= Behavioral Time Modelling
Page 9
Why Accurate Models are
Required for EMI Prediction?
Driver waveform spectrum Radiated spectrum
Page 10
Requirements of the Simulation Tool
• Simulation speed
• Taking into account all the effects at the same time
• Customizability and customer support
• Transparency on the applied methods
Page 11
Requirements for EMI aspects
• Adequate validation environment
• Knowledge of the physical basis of the predictive algorythm
adopted inside the software
• Knowledge of the occurring EM phenomena
Page 12
Considerations on the Choice of the
EMI Prediction Method
complex PCB configurations
can be idealized and analyzed
with efficient methods in the
spectral domain
SIMPLIFIED
models leads to
approximations
simple PCB configurations
can be fully analyzed with
rigorous 3D full-wave
methods as MoM, FDTD,
finite elements method
RIGOROUS models require
long computational times
and are difficult to extend
to complex situations
BUT
BUT
Page 13
A unique method is
not available yet
Numerical method ( PEEC)
where necessary
Analytical formulation
where enough Good trade-off
between accuracy
and simulation time
Solution adopted;
HYBRID METHOD
Considerations on the Choice of the
Method: Possible Solution
Page 14
Knowledge of the Occurring Phenomena When
Comparing Simulated and Measured Radiated Field
knowledge of the
simulation algorithms limitations
evaluation of the systematic errors of the simulation
Definition of the best
simulation setupsexperience of
the designer +
Page 15
CSELT’s EMC Laboratory Layout
Page 16
Objective of an EMC Laboratory
ACCURATEACCURATERIPETIBLERIPETIBLE
TEST ENVIRONMENTTEST ENVIRONMENT
TESTTEST
RESULTSRESULTS REPEATABLE
Page 17
NSA: Basic Principles
Ideal test site
Semi Anechoic
Chamber (SAC)
Page 18
NSA: Volumetric Measurement Method
4 m
1 mh1
h2
d
D
Ar
At
Page 19
30
25
20
15
10
5
0
- 5
- 10
- 15
- 20
NSA
30 40 50 70 90 120 160 200 300 500 700 900
Frequency ( MHz )
5
4
3
2
1
0
-1-2
-3
-4
-5
30 100 200 1000
FREQUENCY [MHz]
DEVIATI
ON [dB
]
THEORETICAL NSA OF AN IDEAL SITE SAC MEASURED NSA
SAC PerformanceMeasured NSA
Page 20
Commonality Requirements
The tool must implement a common test
methodology for both manufacturers and their
clients
- design and integration tool for manufacturers
- incoming qualification and quality monitoring tool for clients
Page 22
THRIS
• Hardware/Software test environment
• Hardware: standard workstation, standard instrumentation.
Partner: HP
• Software: integration/customization of commercial software for both
predictive and experimental issues.
• Custom Front-end (CSELT patented)
• THRIS validation is performed in CSELT laboratories
• New specifications from THRIS User Group (TUG)
• During „97: New functionalities added (reliability,thermal,EMC)
Page 23
THRIS Functionalities (today)
• Modeling environment (including TDR)
• Signal integrity prediction
• EMI prediction (radiated/conducted emission,conducted susceptibility)
• EMI performance optimization (What-If analysis)
• Reliability evaluation (RAPSODIA, METRICA)
• Fault injection (pin forcing technique)
• Noise injection
• Electrical/thermal monitoring
• EMI precompliance analysis (near-field, common mode currents,
conducted susceptibility)
Page 24
Some THRIS Validation Examples
- Radiated emissions (differential/common modes)
- Conducted susceptibility
- Fault insertion
- Noise injection
Page 25
Radiated Emission Tool Validation(Test vehicle #1)
Top view
Bottom view
Copper tape
Termination
Shielded oscillator
Battery
pack
Oscillator PCB
50 ohm coax cableConnector
Line length = 20 cm
Line imped. = 50 ohm
pcb thickness=1.6mm
Er = 4.5
PCB size = 20x30 cm
Ground plane
Page 26
ECL Oscillator Model
GND
Package
VoSTF DTF
S_param
Output stage of the
ECL driver
V
I
Logic “1”
Logic “0”
Constant R
zone
Varying R
zone
Static VI characteristic
-1
-0.85
Dynamic characteristic
Equivalent Thevenin
circuit of the driver
Page 27
Validation of ECL Oscillator Model
Comparison between the spectra of
measured and simulated output
waveform V(2) of the ECL driver
with a 50ohm termination
50
1 2
V(2)Vo
S param.
Electrical model
of the ECL driver
0
20
40
60
80
100
120
68
,8
20
6,4
34
4
48
1,6
61
9,2
75
6,8
82
5,6
frequency [MHz]
dB
uV
/m
simulation
measure
Page 28
10 m
h
H
shielded
oscillator
Typical Measurement Set-up
Page 29
H=1m
h=1-4m with a 0.5m step
ECL shielded oscillator
Microstrip load = open
Frequency [MHz] Measurements [dBuV/m] Simulation [dBuV/m]
68,8 22,47 21,67
206,4 37,38 35,87
344 41,73 42,22
481,6 38,61 38,40
619,2 31,88 35,60
756,8 32,61 30,05
894,4 N.A.* 26.82
Comparison Between
Measurements and Simulations
H
h
frequenza misura simulazione mis+4 mis-4 misura
68,8 22,47 21,67 22,47
206,4 37,38 35,87 41,38 33,38 37,38
344 41,73 42,22 45,73 37,73 41,73
481,6 38,61 38,40 42,61 34,61 38,61
619,2 31,88 35,60 35,88 27,88 31,88
756,8 32,61 30,05 36,61 28,61 32,61
894,4 26,7 2,4
Esempio in cui:
0
10
20
30
40
50
68,8
206
,4
344
481
,6
619
,2
756
,8
894
,4
Freq(MHz)
dB
uV
/m
simulation
measure
* Measure under noise level
Page 30
Common Mode Case Study: PCB alone
H=1.36m
h=1-4m with a 0.5m step
CMOS 10MHz shielded oscillator
Microstrip load = open
Antenna polarization = horizontal
Hh
0,00
5,00
10,00
15,00
20,00
25,00
30,00
35,00
50
60
70
80
90
10
0
11
0
12
0
13
0
14
0
15
0
16
0
17
0
18
0
frequency [MHz]
dB
uV
/m
simulationmeasure
Page 31
frequency (MHz) Simulation Measurement
50 10,30 12,23
70 18,64 20,01
90 21,52 21,82
100 22,83 23,76
120 27,15 31,71
140 35,92 36,27
160 34,29 29,98
180 34,04 28,24
H=136cm
h=1-4m with a 0.5m step
CMOS 10MHz shielded oscillator
microstrip load = open
Cable length = 60cm
Cable termination = open
Antenna polarization = horizontal
0,00
5,00
10,00
15,00
20,00
25,00
30,00
35,00
40,00
50 70 90 110
130
150
170
frequency [MHz]
dB
uV
/m
Hh
simulation
measure
Common Mode Case Study: PCB + attached cable
Page 32
Susceptibility Analysis Example
+
-
P1_8
P3_2
P3_8
P3_1
U1
Distributed ground
5
6
N
Page 33
Measurement Setup
Power
splitterHP8648Cadded ground plane
TR3 2 1
HP54750 DSO
TD
R
flexible coax. cable
PCB under test
Page 34
Measurement/Simulation Comparison
25.00 25.50 26.00 26.50 27.00 27.50 28.00 28.50 29.00 29.50 30.00 TIME[nS]-200.00
-150.00
-100.00
-50.00
0.00
50.00
100.00
150.00
200.00V [mV]U1_6 meas.
U1_6 simul.
U1_5 simul. U1_5 meas.
900MHz residual noise on comparator inputs
Page 35
Pin Forcing: Why Simulation is Needed
Driver Receivers
3
12
1 cm
Actual signal at location 1
and 2 in normal conditions
0TIME[nS]
-2 V-1 V0 V
1 V
2 V3 V
4 V
5 V
6 V
7 V
10 20 30 40 50 60
0TIME[nS]
-1 V
0 V1 V
2 V
3 V
4 V
5 V
6 V
7 V
10 20 30 40 50 60
Residual noise measured at
location 1 and 2 when a THRIS
stuck_at_0 is applied in 1Actual signal at location 1 and 2
after one inverter stage
location 1
location 2
location 1
location 2
10 20 30 40 50 600-1 V0 V1 V2 V3 V4 V5 V6 V7 V
TIME[nS]
Page 36
Pin Forcing: High Speed Systems
0.00 0.25 0.50 0.75 1.00 1.25 1.50 1.75 2.00
TIME[uS]
-1.30 V
-1.30 V
-1.30 V
-2.00 V
0.00 V
-2.00 V
0.00 V
-2.00 V
0.00 V
-2.00 V
0.00 V
FAULT CONTROL
ON OFFOFF ON
STM1 INPUT STREAM (155 Mbit/s)
FAULTY NODE (FIRST STAGE OUTPUT)
FAULTY OUTPUT STREAM
32 155Mbit/s
output streams
32 155Mbit/s
input streams
Fault
Stage A Stage B Stage C
Description:
3 Boards + backplane
32 155Mbit/s streams
6 16x16 switching matrices
THRIS simulation model:
50000 circuit elements
35000 nodes
20000 timepoints
20‟ simulation time on typ. ws.
Page 37
Noise Injection: Noise Margin Evaluation
0.0 2.0 4.0 6.0 8.0 10.0 12.0
TIME[nS]
-1.9 V
-1.7 V
-1.5 V
-1.3 V
-1.1 V
-0.9 V
-0.7 V
-0.5 V
0.0 2.0 4.0 6.0 8.0 10.0 12.0
TIME[nS]
-1.9 V
-1.7 V
-1.5 V
-1.3 V
-1.1 V
-0.9 V
-0.7 V
-0.5 V
0.0 2.0 4.0 6.0 8.0 10.0 12.0
TIME[nS]
-1.9 V
-1.7 V
-1.5 V
-1.3 V
-1.1 V
-0.9 V
-0.7 V
-0.5 V
Normal operation
5mA current injection
12mA current injection
ecl interconnection
N
output testpoint
Page 38
Pre-compliance Near-Field Testing
HP 8590EM EMC Analyzer
PreamplifierHP-IB cable
to THRIS
ROM card
Near field probe
DUT
Page 39
Pre-compliance Common Mode
Current Testing
HP8590EM EMC Analyzer
PreamplifierHP-IB cable
to THRIS
ROM card
CABLE
UNDER
TEST
HP 11967A
Page 40
Thermal Monitoring
Page 41
Conclusions
• EMI behavior of electronic equipment must be optimized in early design
stage. Simulation is the key.
• The simulation tool should be accurately validated versus measurements in
suitable site (e.g. accreditated EMC labs).
• Reliable EMI simulations require accurate SI models based on wide -band
measurements on components (e.g.: TDR, network analyzer, ecc).
• Measurements are still mandatory for the certification phase. If some
problems are detected, corrective actions can be experimented by means of
simulation (e.g. What-if analysis)
• Specific EMI measurements executed after installation can identify
problems due to the environment impact, even if the device under test has
been already certified.
Page 42
THRIS Solutions
THRIS_SI
THRIS_IG
THRIS_EMC
Designers, R&D
engineers
Hardware/software
system integration
engineers
Designers, R&D
engineers, EMC
specialists
Optimized design,
redesign cycle reduction
Increased tests
coverage,
increased system level
robustness
EMI problems found and
fixed at design stage
WHO BENEFITS
Page 43
THRIS Resources
THRIS_SI THRIS_IG THRIS_EMC
Workstation x x x
VEE x x
VXI, Front_end x
EMC instruments x
THRIS manager x x x
Fault insertion test suite x
Post-layout analysis package x x x
Crosstalk, bouncing analysis x
EMI radiated emissions analysis x
What-IF analysis opt. opt. opt.
Modeling development suite opt. opt.
Page 44
THRIS Application Example
Design of a high speed board (ATM switch)
Page 45
THRIS Application Example
Fault injection probes
Page 46
THRIS Application Example
Qualification trials on a digital exchange
Page 47
THRIS Application Example
Design verification of a telecom auxiliary apparatus
Page 48
THRIS Application Example
Fault insertion trials on a SDH apparatus
Page 49
THRIS Application Example
Signal monitoring inside a public telecom terminal