Defining "low-range" of a data set

Hi all!

So here is the scenario,
I have a linear non-weighted calibration curve. I run a bunch of tests and gather data from which I can derive a control chart. This gives me "acceptance criteria" for a passing test, from a statistical perspective. Let's say +-20% from a known value. But this covers the whole range of data -low to high. Now, the scattering is tighter at higher values and looser at lower ones. I want to define acceptance criteria for the "edges" of the curve, if it makes sense, specifically the lower end.
The problem is, I don't know how to do that. I don't know how to say in a "non-arbitrary" way "well, anything below X can be considered low end". How do I define X? I don't want to apply weights and change the curve.
Can anyone help me with that?

Thanks so much,