TP20 Calibration

not likely, I have had a lot of experience with TP20s and those sound like its time to bin and replace. In my experience, they really don't last more than a couple years, but then again, I used to work for a facility that allowed EVERYBODY to run the CMMs, YMMV

Calibrating a probe will tell the machine where the probe's spherical centroid is and will work great even if bent a bit... Until you try to measure with the probe deep into a part.

The machine can't discern or compensate for when you try to measure something deep into a bore, and the probe's shank/shaft/extension is contacting the part before the sphere does (aka a "Shank hit").  This will produce terrible dimensional results in-error. 

I've dealt with this for TP200's.  Some new, directly out of the box from Renishaw had >1.5° projected axial error.  It doesn't sound like much, but it's enough to shank-hit a 60mm deep 3mm sphere with a 2mm diameter shank.