FIND THE NEEDLE IN THE HAYSTACK February 2017 CHALLENGES OF SYSTEM - - PowerPoint PPT Presentation

FIND THE NEEDLE IN THE HAYSTACK

Automated Trace Analysis for Testing of CANopen Devices

by Andrew Ayre & Olaf Pfeiffer

CHALLENGES OF SYSTEM INTEGRATION

 CANopen Device Testing

has its limits

 Official CANopen

conformance test is very basic

 No stress test  No interaction with multiple

devices

 Limited timing tests

February 2017 Automated Trace Analysis

 Highly dynamic systems

can not be fully tested with a single DUT

 Dynamic node IDs  Multiple SDO clients

CHALLENGES OF CIA 447

 Wake up and sleep  Dynamic Node IDs  The 16 devices are both

 SDO servers (16)  SDO clients (1-15)  Lots of scanning

 Many timings and timeouts

 When to start SDO server

 Impossible to test all

combinations

February 2017 Automated Trace Analysis

1 4 5 6 3 2

February 2017

SPREADSHEET APPROACH

Trace analysis

Automated Trace Analysis

TRACE ANALYSIS BY SPREADSHEET

 Base is a CAN trace with all

messages and timestamp

 Who is who?

 LSS node ID assignements

 Look for the obvious

 Emergency messages  Unexpected bootups  Sleep objections

 Verify Heartbeat timing

 Use spreadsheet formula

 Verify SDO timeouts

 Ok for single, global client  Challenging for 16*15 clients

 Size limit

 More than 1,000,000 lines?

February 2017 Automated Trace Analysis

February 2017

AUTOMATED ANALYSIS APPROACH

Trace analysis

Automated Trace Analysis

RELEVANT INFORMATION

• Input can be large
• Millions of lines
• Identify relevant portions

Requirement

• Generate analysis of trace log
• Filter to only show relevant events
• Linked to original trace

Solution

February 2017 Automated Trace Analysis

RELEVANT EVENTS - EXAMPLE

February 2017 Automated Trace Analysis

WHO IS WHO?

• LSS Node ID assignment
• Which device has which node ID

Requirement

• Node ID assignment is an event
• Mark with LSS ID (= 1018h Object)
• Monitor sequence, event on error

Solution

February 2017 Automated Trace Analysis

BOOTUPS AND HEARTBEATS

• NMT Master messages (reset?)
• Unexpected bootup?
• Heartbeat timeouts

Requirement

• NMT Reset marked as event
• Unexpected bootups marked as event
• Heartbeat timeouts marked as event

Solution

February 2017 Automated Trace Analysis

SDO COMMUNICATION

• Does a device always respond?
• In which time?
• Does a client swamp the network?

Requirement

• Analyze all SDO channels
• Analyze all timings, event on timeout
• Analyze all repetitive client requests for load
• Analyze sequence, event on error

Solution

February 2017 Automated Trace Analysis

PDO COMMUNICATION

• Verify event time
• Verify inhibit time
• Verify SYNC reaction

Requirement

• Analyze all PDOs
• Analyze all timings, SYNC responses
• Mark as event if timings are wrong

Solution

February 2017 Automated Trace Analysis

WHERE DO TIMEOUT VALUES COME FROM?

 Timeout for

 Heartbeat  SDO response

 Allowed timing variation

 Percentage that a cyclic

timing may vary

 PDO +/- 10%?  Heartbeat +/- 30%?

 When to mark something as

warning, when as error?

 Application Profile  Custom?

February 2017 Automated Trace Analysis

February 2017

SYSTEM AND DEVICE SUMMARY

Trace analysis

Automated Trace Analysis

REPORTS IN ADDITION TO EVENT LIST

Global System Summary

 Maximum message rates  Maximum burst length  Number of nodes  Number of system (re-)starts

Per Device

 Unexpected bootups  Min/max heartbeat  Max SDO response time  CAN message rate generated by

a device

 SDO server responses count

towards client requesting it

February 2017 Automated Trace Analysis

CURRENT IMPLEMENTATIONS

 Hardware

 CANopen Diag hand-held

diagnostic tool

 Live analysis

 System summary display  Analyzer event list

February 2017 Automated Trace Analysis

 Software

 Logxaminer  Input

 Trace format „Vector“  Trace format „PEAK“  Trace format „ESAcademy“

 Output

 Display  PDF report