Connect each symptom to your hypothesis and a check that passes the required design and rejects the fault you proposed.
What to notice. The matrix consumes all three earlier stages; clean and failing observations are distinct evidence; every symptom needs a row.
The clean-data branch must meet the requirement and the proposed faulty branch must show the failure the test is intended to detect.
Create columns for symptom ID, requirement, hypothesis, test, observed result and conclusion. Fill result columns from your own runs. Find and inject the fault you think explains each symptom. When tests share a dependency, explain the additional failures rather than assuming each test must identify one fault alone.
All ten symptoms in the symptom table.
The artefacts from stages 1 to 3: map and link tables, clock, crossing and reset tables, interrupt map and driver trace, boot record.
One clean design record to run every check against.
Deliverable. An evidence matrix with ten rows and the columns symptom, requirement, hypothesis, test, observed result and conclusion.
Pass criterion. Every test passes on the clean record and fails on the faulty variant you propose for its row; where one fault trips several tests, the matrix says which test is primary and why the others also fail.
The ten-row matrix, filled from your own runs.
For each row, the clean result and the faulty result side by side.
A written explanation of every test that fails for more than one row.
Keep your diagnosis in your own report. After the stage checks, compare it with Solutions and run the provided injection checks.