Stage 01
Measurement and test planning
A noise problem is decided before the first run: whether the data can carry the answer at all.
Most campaigns that fail do not fail in the analysis. They fail because the runs cannot distinguish the candidates — no angle reference on the shaft that matters, a speed range that never crosses the suspected order, or a load point that was held for four seconds. Planning the campaign against the questions it has to answer is the cheapest stage of the whole investigation.
How it runs
- 01
Collect the questions
What has to be decided at the end, and which components are seriously suspected. A campaign without that list measures everything and separates nothing.
- 02
Derive the orders
From the gear train: which orders each candidate produces, where they coincide, and which operating point pulls them apart.
- 03
Design the runs
The minimum set of ramps, load steps and repeats that makes those orders separable, plus the repeatability runs that prove the data is trustworthy.
What it covers
- Run matrices for speed ramps, load steps and coast-down, derived from the orders that have to be separated
- Sensor and tacho layout, so shaft angle is available for every channel that will be analysed
- Sampling rate, block length and ramp rate checked against the lowest order of interest
- Repeatability and linearity checks before the campaign scales up to dozens of runs
What you get
- A run matrix you can hand to the test bench, with the purpose of each run written next to it
- A channel and tacho plan, including which shaft carries the angle reference and why
- A short note on which questions the campaign can answer — and which it cannot
What I need from you
- Gearbox drawing or gear data: tooth counts, shaft arrangement, bearing types
- What the test bench can actually do — speed range, load capability, available channels
- The noise complaint in the words of whoever raised it