Overview
Two questions, deliberately kept separate
Verification checks the product against its specification. Validation checks the specification against the actual need. A product can pass every verification test and still be wrong, which is why we keep the two activities distinct.
Across hardware, firmware and the integration between them, we build a coverage picture: what has been verified, by what method, with what result, and what remains an accepted risk. Gaps are stated rather than left implicit.
That coverage picture is what makes a release decision defensible. It also becomes the baseline against which future changes are re-verified, so a small modification does not quietly invalidate earlier testing.
Engineering principles
- Verification: does it meet the specification
- Validation: was the specification right
- Coverage stated, gaps named
- Baseline retained for future change
Validation scope
Where validation effort is spent
Each area answers a different class of question about the product.
Hardware validation
Rails, signals, timing, thermal behaviour and margin measured against the design intent across the operating range.
Firmware verification
Behaviour verified against specification including error paths, recovery and worst-case timing rather than typical operation.
Integration testing
Hardware and software exercised together, particularly at startup, shutdown, fault and concurrent-load conditions.
Reliability assessment
Extended and cyclic operation to surface failures that only appear over time or repeated use.
Design verification
A traceable mapping from each requirement to the evidence that it has been met, including the ones that have not.
Evidence and reporting
Recorded results, coverage summaries and a clear statement of residual risk to support the release decision.
Coverage model
What gets verified, and how
| Area | Typical method | Evidence produced |
|---|---|---|
| Power and supply behaviour | Measurement across load, temperature and start-up conditions | Rail measurements, sequencing traces, margin data |
| Signal behaviour | Instrumented measurement of critical nets and interfaces | Waveform captures against expected characteristics |
| Firmware functional behaviour | Specification-based testing on target hardware | Test results per requirement, including error paths |
| Timing | Worst-case measurement under representative load | Response and jitter measurements against budgets |
| Integration | Combined hardware and software exercises | Results for startup, fault, recovery and concurrent load |
| Reliability | Extended and cyclic operation | Run duration, failure records and analysis |
Validation process
Building a defensible verification picture
Structured so that a release decision can be made on evidence rather than on how the last demo went.
-
Requirement review
Requirements examined for testability; anything that cannot be measured is rewritten until it can be.
-
Verification planning
Method chosen per requirement: inspection, analysis, measurement or test, with the rationale recorded.
-
Test development
Procedures, fixtures and automation built where repetition or precision justifies the effort.
-
Execution
Testing carried out on representative units with conditions and results captured as they occur.
-
Failure analysis
Failures driven to root cause, with the distinction between design, build and test faults made explicit.
-
Coverage reporting
A traceable view of what is verified, what is outstanding and what risk is being accepted.
-
Regression baseline
The verification set retained so future changes can be re-checked against a known reference.
Applications
When formal validation is worth the effort
Validation effort should scale with consequence, not with project size.
- Application 01
- Products approaching a production release decision
- Application 02
- Designs where field failure carries high cost or risk
- Application 03
- Equipment operating in demanding environments
- Application 04
- Products with regulatory or customer evidence requirements
- Application 05
- Designs undergoing significant change or modernisation
- Application 06
- Projects inherited from another team needing an independent picture
FAQ
Testing and validation questions
How validation work is usually structured and delivered.
Verification asks whether the product meets its specification. Validation asks whether the specification was right for the need. We keep them separate because a product can pass all verification and still fail the user.
Yes, and it is a common request before a production decision. We build the coverage picture from the existing requirements and evidence, then test the gaps.
By consequence. High-consequence functions get deeper coverage; low-consequence ones get proportionate checks. What matters is that the decision is deliberate and written down rather than accidental.
We prepare designs and evidence for compliance activity and support the process. Accredited testing is performed by qualified laboratories, and our specific compliance scope is to be confirmed by the company.
Next step
Approaching a release decision without a clear coverage picture?
We can build the verification map, run the gaps and tell you what risk you are actually carrying.