I agree that the chunk does not follow the spline fit on your lower right. The spline does shows that the CTQ_DCBN increases for increasing CTQ_WRT_WDTH, although the increase in CTQ_WDTH does not
explain all of it. I think that if you add EST_BP to the model that would explain much of the remaining difference.
We need to keep in mind that the CTQ_DCBN on the pi lot was 6.871. Later when we built the child lots, the CTQ_DCBN came in at 12.327. The chunk was the same.
So what is the difference? The child lots were overtilted. That's why they came in wide. The shorter EST_BP also increased the CTQ_DCBN, in addition to the CTQ_WRT_WDTH.
So the risk is the standard writer ATI risk that we have with all wide writers with short break points. The CTQ_DCBN is a test artifact. The reader did not change from tilting. The written track that the
reader sees is obviously much different since the patterns are larger.
I appreciate the plot of drive yield vs CTQ_DCBN. Does the drive yield roll of with wider CTQ_WRT_WDT? Can you contour CTQ_WRT_WDT and EST_BP? Maybe yield rolls off on wide heads with
short breakpoints, but is okay or wide heads with nominal breakpoints.