Testing every board from design A before every board from design B can confound the design comparison with time. Instrument drift, board warmup, operator practice, or ambient conditions may change during the session. Plan the run order before measuring the alternatives.
NIST’s completely randomized design discussion describes random assignment of experimental run sequence. Randomization is a design tool, not a cure for every uncontrolled condition; practical constraints and known influential factors still need explicit treatment.
Identify time-dependent behavior
List conditions that could change while the comparison is running: equipment warmup, fixture wear, battery state, thermal equilibrium, software state, or operator learning. Determine which can be controlled and which should be measured or incorporated into the experimental plan.
Define how each board is prepared before a run. If one alternative is measured immediately after power-on and another after an extended setup period, the apparent difference may be preparation rather than design.
Create and preserve the planned sequence
| Record | Purpose |
|---|---|
| Board and configuration identity | Connect the result to the intended alternative |
| Planned and actual run order | Expose departures from the experiment |
| Preparation and elapsed time | Explain state-dependent behavior |
| Station and environmental observations | Support investigation of drift or interruptions |
Use a randomization or balancing method suitable for the experiment and retain how it was generated. Do not rearrange the sequence after seeing early results to make a preferred alternative appear more consistent.
Handle interruptions transparently
- Record the pause and any equipment or setup change.
- Assess whether the remaining runs need a revised block or session identity.
- Restore the defined preparation conditions.
- Keep failed or interrupted runs identifiable.
- Analyze results with their actual order available.
If a test permanently changes a board, repeated measurements may not be interchangeable trials. Plan separate units or an appropriate repeated-measures approach rather than assuming that random order removes the physical history.
Plot the measured response against run order as well as grouping it by design alternative. A trend across the session can reveal an issue that a pair of averages hides. Investigate that trend before assigning the entire difference to the PCB change. Record actual instrument settings at each run rather than relying on the planned sequence when operators can change conditions manually.
Close the comparison with the effect being estimated, the actual sequence, and the conditions it represents. The blocked-comparison guide addresses experiments where known session, lot, or station differences should be represented directly in the design.
Move from reading to a build-ready project
Use the fabrication and assembly checklists to prepare drawings, production files and test requirements.
Open the DFM checklist