I'm not fully on top this issue but; I've heard that fourier series can be implemented into CMM raw data of a spline inspection to eliminate some errors which helps to reduce the rejection of parts.
Anyone got an idea on how to do it?
The question may not be %100 correct, but I think if you know about this issue you can understand what I'm asking for
Welcome to EE!
Most CMM software I am aware of has the ability to mathematically smooth or fit as-measured data.
“Fourier Series” maybe, though I think most software uses some sort of least square fit polynomial. I do think that before one manipulates the as-measured data points to smooth the data readings a full review of the CMM program or measurement plan setup should be undertaken.
This would include:
1. Verify the CMM has a robust or recent self-calibration.
2. Verify that the stylus probes have been properly calibrated.
3. Review the components zero or datum setup. Be sure you are getting sufficient measurement points and if not possible utilize true datum simulators to ensure a proper zero has been established.
4. Ensure that when measuring the spline or other complex geometry that the proper measurement data point increments are being taken. The fewer the data points taken along the geometry the less the error is possible. This is where I think most inspection folks fall short as they generally want to take the fewest number of data points possible and then trust the software to smooth the data mathematically. The tighter the curve and the assigned tolerance the more measurement points one needs. Tolerance values and boundaries are most-correctly absolute and maneuvering so that data point measurements get an acceptable part tolerance can be defeating the purpose of a quality inspection.