Test design techniques & BVA

How to choose a few smart test cases instead of many random ones. The interviewer asked a boundary value question about a snowflake at 5 °C — this chapter gives you a perfect, complete answer and lets you play with it.

Why we need techniques

Equivalence partitioning (EP)

More car examples: fan speed 1–7 (valid) vs 0 and 8 (invalid); gear P/R/N/D (each gear is its own partition); phone types iPhone / Android; language groups: left-to-right vs right-to-left scripts.

Boundary value analysis (BVA)

2-value and 3-value BVA (ISTQB v4.0)

The "step size" (resolution) decides the neighbours

The neighbour is one step away, and the step is the smallest change the signal can make:

• Integer values (volume, fan level): step = 1 → 39, 40, 41.
• Temperature signal with 0.1 °C resolution: step = 0.1 → 4.9, 5.0, 5.1.
• Temperature signal with 0.5 °C resolution (factor 0.5 in the DBC!): step = 0.5 → 4.5, 5.0, 5.5. You can't even send 4.9!

"Below 5" vs "5 and below": open and closed boundaries

Only the value exactly on the boundary changes between the two wordings. That's why 5.0 is the single most important test value.

The snowflake question: the full answer

This is the exact scenario from your interview:

Step 1: Spot the problem in the requirement (this impresses!)

Look closely: it appears below 5 but disappears above 4. What happens at 4.5 °C? Going down, 4.5 is below 5 → ON. Going up, 4.5 is above 4 → OFF. The requirement contradicts itself between 4 and 5 °C.

Real cars solve this with hysteresis: two different thresholds, one for switching ON and one for switching OFF, with a zone in the middle where the icon keeps its previous state. It stops the icon from flickering when the temperature hovers around one value (4.9, 5.0, 4.9, 5.0…).

Step 2: Choose the test values

Assuming the clarified rule above and a resolution of 0.1 °C (3-value BVA on both boundaries):

If instead the rule is the simple "ON below 5 °C, OFF at 5 °C or above" (the version in your notes), the core values are 4.9 → ON, 5.0 → OFF, 5.1 → OFF. The notes' table (3, 4, 4.9, 5, 5.1, 6 °C) is the same idea: normal values on each side plus the values at and around the boundary.

Step 3: Test both directions and the transitions

• Falling 6 → 5.1 → 5.0 → 4.9 → 4.1 → 4.0 → 3.9: note exactly when it switches ON.
• Rising 3 → 3.9 → 4.0 → 4.1 → 4.9 → 5.0 → 5.1: note exactly when it switches OFF.
• Oscillation: 4.9 ↔ 5.0 ↔ 4.9 many times → no flicker.
• Hold time: keep each value for a few seconds (or the time in the spec) before judging; check the reaction time.
• How: step the signal in CANoe (panel slider or a CAPL script), record the screen and the trace.

Step 4: Extra cases a strong tester adds

• Signal minimum and maximum (e.g. −40 °C and 85 °C).
• Invalid / SNA value and timeout (sensor failure) → no snowflake, dashes shown (per spec).
• Ignition cycle while the icon is ON → correct state after restart.
• °F setting: 41 °F = 5 °C, 39.2 °F = 4 °C; the logic must still use the same thresholds.
• Display rounding: signal 4.96 might be displayed as "5.0" — is the icon logic based on the raw value or the rounded one?
• Related: an ice warning pop-up/chime may also appear — test it together.

Step 5: If it fails, report it

Interviewer: "The snowflake appears at 5.0 °C, but the requirement says only below 5 °C. What do you do?"

1. Check the CAN trace: was the signal really 5.0 °C (not 4.96 displayed as 5.0)?
2. Repeat 3 times: is it reproducible? Test both directions.
3. Check the requirement version and whether this is already reported (search Jira).
4. Raise the ticket:

Summary     : [Cluster] Ice warning (snowflake) shown at exactly 5.0 °C; REQ-HMI-TEMP-012 requires < 5.0 °C
Environment : SW 24.3.1, bench B-03, HW C-sample, variant EU, language EN, units °C
Precondition: Ignition ON, cluster started, Outside_Temp = 6.0 °C for 3 s
Steps       : 1. CANoe: set Outside_Temp = 5.0 °C
              2. Wait 3 s and observe the cluster
Expected    : No snowflake at 5.0 °C (REQ-HMI-TEMP-012)
Actual      : Snowflake shown after 0.4 s
Repro rate  : 3/3 (also when rising from 4.0 °C)
Evidence    : video, CAN trace (.blf, 10:15:32 = raw 0x5A = 5.0 °C), DLT log
Analysis    : Correct value on the bus; cluster logic likely uses <= instead of <
Impact      : False ice warning at exactly 5.0 °C; all variants using this logic
Severity    : Minor/Major per project definitions   Links: TC-TEMP-003, REQ-HMI-TEMP-012

Follow-up questions (Q62–69) with short answers

Decision tables

State transition testing

Experience-based techniques

Error guessing: 25 things that often break in IVI

Exploratory testing

Testing freely, but with a goal written as a charter, in a time box (e.g. 45 min), taking notes. Example charter: "Explore the media player with a 10,000-song USB stick to find performance and display problems." At the end, report what you tested, what you found, and ideas for new test cases.

Checklist-based testing

Use a list of things to check (like the UI checklist or the 8-point scan in the eye tests).

Practice exercises (try, then open the answer)