0.5mm tip + TP20

0.5x20 TP20 sees STDEV of 0.003-0.004mm on my 20 year old B&S ONE machine with no vibration isolation. My Global S Blue with same type of probe with the X1 with vibration isolation sees 0.0002-0.0003mm STDEV. 

Build of the machine, vibration isolation, probing system, etc... all plays into this. 

Back before I moved into Gantry Land, we just used a TP-20 low force with around 1/2 the touch speed of the standard force for the .5mm probe. Never had any issues with accuracy or calibration on our 7.10.7SF with a HH-A-T5

Accuracy is not the issue. Repeatability is. Do you remember the values of std you were getting? For me Low force or std TP20 modules seem to not play any role. Still shitty results as opposed to 1 plus mm tips

Yah, that tiny lil shank on there isn't the best. I don't remember the numbers, only that it was deemed "acceptable" 
It was always worse than more rigid probes though. 

It's not just the force of the module that will affect accuracy but also the touch speed being used.  Ø0.5 tips have a very thin stem that is quite flexible, probing at low touch speeds (~1 mm/second) with a low force module should help.  Also, remember that you should always calibrate your probes at the same touch speed that you use in your routine when measuring.

Also remember there is an inherent tri-lobing error with touch trigger probes due to the trigger mechanism that you will not be able to eliminate, regardless of which force module or touch speed you use.  Those errors are usually in the sub-micron range but are one of the reasons why you can achieve more accurate results with an analogue system (like the LSP-X1-S / HPS-X1S).  

It's not just the force of the module that will affect accuracy but also the touch speed being used.  Ø0.5 tips have a very thin stem that is quite flexible, probing at low touch speeds (~1 mm/second) with a low force module should help.  Also, remember that you should always calibrate your probes at the same touch speed that you use in your routine when measuring.

Also remember there is an inherent tri-lobing error with touch trigger probes due to the trigger mechanism that you will not be able to eliminate, regardless of which force module or touch speed you use.  Those errors are usually in the sub-micron range but are one of the reasons why you can achieve more accurate results with an analogue system (like the LSP-X1-S / HPS-X1S).  

Touch speed 0.5mm/s I believe it to be ok. I also measure with the same parameters as calibrated. I ordered new fresh TP20 modules to repeat my tests but until then for me all 0.5 tips are getting around 5 micros of std in calibration results. This does translate indeed in repeatability in my measurements especially in TP callouts where any deviation is doubled.