I ended up upgrading to 1.4.7 from 1.3.8 too once I realised that the pressure sensor calibration didn’t provide reproducable enough values between prints. It felt like the variance was something around 0.05 mm, which results in the annoying “sticks well half of the time, not that well sometimes and not at all still far too often”.
In 1.4.7, internal_endstop_offset is by default set to -0.20 (not the -0.25 you’ve mentioned). Testing a one layer PLA print with that value made it clear that’s not the correct value at least for my printer either. Next test with -0.15 looked far more promisintg with only some tiny holes here and there if viewed against a light. Then with -0.10 I finally got the mythical “perfect first layer”. Remains to be seen if the reproducability is there too, but that will require more prints over a longer period of time.
However, 1.4.7 does the nozzle cleanup far worse than 1.3.8. The new logic appears to be to heat the bed first, then bring the nozzle to 200C (at which PLA starts slowly oozing) and then moving the nozzle to touch the bed at X30 Y30. The calibration is done using the eddy sensor even if it’s a touch calibration and I think this part is only for the nozzle cleanup with the side brush. This touching however has the side effect of leaving a PLA blob at that point if there was any oozing. Next comes the brush moves which have been changed to do more moves that previously. Lastly comes the issue that the nozzle is moved back to X30 Y30 for another touch and the cooldown back to 130C. The issue is that the cleaned nozzle has every time during testing collected back the blob that was dropped here before the cleanup so that cleanup end up being for nothing and that blob will move to be part of the print or priming line. Probably all this could be avoided by using a different point instead of the same twice. Going too much closer to X0 Y0 however isn’t likely to be a good idea since that may cause the eddy sensor to be outside the bed plate and not stop the nozzle lowering as a result.
@Fabio - What’s your movement speed for the first layer? If there’s acceleration that can happen then that picture could suggest your pressure advance isn’t optimal as the wrinkles are more visible near the edges.