50mm sounds coarse, and monitoring pressure for penetration sounds backwards.
I would use a moving/scanning pivot-arm approach, to get a continual surface record, and set a nominally fixed pressure, and measure the deflection as the pressure point and axis is moved.

Something like a nylon wheel on the probe end, would help avoid side-ways sticking effects, and simplify operator control to the level of orbit position.
Orbit position could be set with a spin-handle on a gear+track, like
http://www.sdp-si.com/web/html/newprdgears3.htm
and/or a stepper motor.

With an arm design, a rotary encoder+maths can give end-point location, but it is quite distance compliant, and safe.

Another track-following quadrature encoder would record where on the orbit-circle the axle was, and a third one could check the spring deflection (or the operator could just manually keep a fixed pressure setting )

So the problem has morphed into 2 or 3 encoders, which are easy to scale and choose, and pressure is nominally fixed per scan.
Multiple scans would build up a contour style plot.

You can get very fine scan resolution, sub-mm.

-jg