If stable descent can't be maintained at "one dot low" then the incorrect VDA has been used to design the approach.
Yes and no.
Non-precision CDFA is also a relatively new concept. There are literally
thousands of non-precision approaches in the US alone -- developed long before the CDFA concept -- which have obstructions in the 34:1 and 20:1 protection surfaces.
These old approaches were never designed for continuous descent. What mattered in the past was meeting the required clearance at the MDA floor after a series of step-downs.
So the VDAs for these approaches are constrained by the existing designs. The VDA isn't "selected". By rule, the VDA is simply the angle from
the existing FAF to the TCH.
I.e., for these thousands of approaches, the VDA is calculated "after the fact".
This "after the fact" calculation is ok if the VDA is only used to the MDA. Below the MDA, however, there's no obstacle clearance guarantee if the 34:1 surface is penetrated.
Question for everyone. How does removing the VDA make this situation better or safer?
FAA did some tests. With the computed VDA, predictably every approach even when flown perfectly got too close "to the house" (~190 ft) and triggered EGPWS warnings. This is an accident waiting to happen.
Without the VDA, these old approaches get flown "as they were designed" (dive-and-drive). Meaning, pilots would "correctly" level off at the MDA and pass well clear of the house at a protected altitude, before making a visual landing.
So until the approach can be redesigned, we have to do a risk analysis: Which option is worse? a) using CDFA but getting perilously close to obstacles; b) reverting back to dive-and-drive but remaining well clear of obstacles.
FAA picked option b).