Pilou -

I zoomed in to the microscopic level on my model and indeed it's not tangent. Then I tried the Jfs (PAQUICINNO) method in his video (which I think is the better solution), and it also was not tangent. Of course these are visual checks.

I don't know the mathematical solution, or even if displayed curves are 100% visually accurate when zoomed in that far.

Practically speaking, the error I saw when zoomed in that far is below the resolution on my micrometer :)

But it would be nice to verify what the precise solution is.

Update - I see Michael has linked to a solution above. Who wants to test it? :)

Ed Ferguson