Zeroing load cell data

In what way? The ASCII graph had no values associated with it so the actual nonlinearity % is unknown. Generally a load cell plot shows an error from a straight line. Y-Axis being the error as a % of fullscale and X-Axis being the force. If the graph shown was simply load vs output then I agree the error would appear to be rather large.

The original OP's question appears to have never been answered. Normally when a load cell is calibrated the results are normalized with the initial zero reading subtracted such that zero load = zero output. A load cells zero reading is arbitrary since mounting methods etc. can shift the initial zero considerably. A load cell calibration therefore typically shows span.

I work for Interface Inc. who manufactures load cells and have written the calibration software they use and sell.

Miles,

You're jumping into the middle of an abandoned thread. The graph was represented as output, not error. The foldback it shows is the unusual behavior I had in mind when I said that functioning load cells don't do that.

The curve turns out to have been invented for the purpose of illustrating an ill-formed linearization question and is unrelated to any reality. The rest of the thread is a fine illustration of a sci-fi writer's dilemma.

Jerry