SAFETY MANUAL - Tuv-fs.com
SAFETY MANUAL - Tuv-fs.com
SAFETY MANUAL - Tuv-fs.com
- No tags were found...
You also want an ePaper? Increase the reach of your titles
YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.
<strong>SAFETY</strong> <strong>MANUAL</strong><br />
3.11.3.4 Main and Alternative Comparison Pass/Fail Flag<br />
The results of the alternative implementation shall be <strong>com</strong>pared with the results of<br />
the function under test; discrepancies shall cause a “main and alternative<br />
<strong>com</strong>parison fail flag” to be set.<br />
3.11.3.5 Test Results Register<br />
Each harness shall include registers that record the functionality of the function<br />
block. This registration should be as <strong>com</strong>prehensive as possible and should utilise<br />
as many predictable features as possible.<br />
For example, a 2 input logical “Or Gate” stimulated by the two lower bits of a 16-bit<br />
counter will record 32768 logical high states if the counter is allowed to make one<br />
<strong>com</strong>plete up count from 0 to 65536. The results register would count these states<br />
and present a number to the human operator. In this case the results register<br />
should also record that no two consecutive states of the counter caused a logical<br />
“1” at the output of the Gate.<br />
3.11.3.6 Test Coverage<br />
Where possible, all <strong>com</strong>binations of input shall be simulated.<br />
For certain functions, such as adders and <strong>com</strong>parators, this is not practical. In<br />
these cases, the test harness shall utilise a significant number of test cases to<br />
prove the functions operation. The use equivalence class, boundary cases and<br />
random numbers shall be used as the preferred method of generating these cases.<br />
Functions containing <strong>com</strong>plex algorithms or with extensive retained state or value<br />
dependence require an extensive number of test cases, and are therefore<br />
considered impractical to achieve a sufficient level of test coverage and shall be<br />
used in non-safety programs only.<br />
3.11.3.7 Recording and Filing of Results<br />
The tests shall utilise formally approved test procedures and the test results shall<br />
be formally recorded. The test harness, details of the test environment and test<br />
result shall be retained.<br />
Any deviation between the results and expected results shall be examined; where<br />
this results from deficiencies in the test harness these shall be corrected and the<br />
test repeated. Should any function fail it shall be:<br />
• Not used within safety related applications, or<br />
• The conditions that result in erroneous operation shall be explicitly<br />
recorded and published. If the function is used, other function(s) shall be<br />
added to the application to specifically detect the conditions leading to<br />
erroneous operation and take a fail-safe action.<br />
To maintain system certification, any test harness used to prove a function block<br />
should be archived as part of the test record so that the tests can be repeated at<br />
later date and if required, reviewed by TÜV.<br />
Doc Number P8094<br />
Issue 14 September 2003 Page 61 of 67