I think you did an excellent job here. This is coming from a proud owner of a deckel fp1 myself that can basically only cut straight lines, I have to say I'm pretty jealous! Anyone who hasn't operated a 1950s something manual mill and has done everything all by hand and grew up in the cnc world has no clue what you have done. Great job!!!

What do you think about the outcome? I'm a bit worried about the 50.3 dimension of the part but the rest seems to be quite accurate. Let me know your thoughts! Thanks for watching!

good video GBWM

I'd map each measurement to the X/Y/Z axis, see if you have a consistent error with one of them (appears you might). Big hulking machine like that should have very good results (< 40um) as it'll have the rigidity to achieve it, so any errors can (and should) be able to be ironed out of it. > Throw an indicator on your endmill - what's the runout there? Should be < 10um. > Perform backlash tests for each axis - testing parallel to the axis. This can be mitigated for in some CAM packages. > Perform linear test - test forces perpendicular to the axis (ie. worn ways/dovetails). This will require gibs to be tightened, maybe even re-grind/scraping.

Backlash compassion in the software will resolve the problem but if you want to resolve this problem for ever the hand scraping is the best solution

You should probably check your hysteresis, you seem to have bigger tolerances on one axis, so check the steps for the motor, and tweek it, for positive-direction and return, it seem it's lacking, or maybe you just havr sligthly bigger tolerances in the machine itself, plays in the parts...

i love your videos! thanks for sharing! i am facing a similar issue with my cnc machine and i noted that my problem was backlash on one axis. As soon as i improved the backlash in that axis (adjusting gibs and increasing acceleration to avoid stiching) the results start to be much more accurate ;) let us know how would you solve this problem, thankssssss

Как для самодельного станка, очень хорошо!

Taki pomiar dokładności mało wnosi. Powinieneś zrobić pomiar pierwszej sztuki później zrobić poprawki w programie i zobaczyć czy wtedy wyjdzie to co miało wyjść 🙂 Błędy wymiaru na pierwszej sztuce często się zdarzają. Najważniejsze jeśli maszyna jest powtarzalna i trzyma wymiar po winiesieniu poprawek.

Good job. How can you make it 5 axis cnc mill?

what software do you use to design and CAM bro ?

Very nice accuracy test. Please bear in mind I'm a complete amateur, but like others have said it seems clear that one axis has a little slop in it. I can't tell but I think the 10.08 dimension on the square boss is probably the same axis as the 50.31 on the outside of the part - they both seem to be about 0.6-0.8% over what they should be, or like 7 extra microns per mm. I just wonder if this is actual slop (somewhere between the motors and the ball screw nuts) or if it's a linear error with your ball screws that you can just dial out by recalibrating. Assuming your gibs are tightened well, I'm not sure how much scraping would help with this. Unless your dovetails are getting really tight and it's stopping them from moving properly, none of them should affect your horizontal dimensions to that degree. 0.3mm of wear on a dovetail should be a showstopper, so despite the mill's age I don't think it's that bad. I would say scraping should be a last resort, especially since it's the most time-consuming and risky change you could make. As for surface finish, yeah it's not bad but far from stellar. In mild steel it's kinda expected, though I have to wonder about how sharp that endmill is. Try a brand new one if you can? As before I would check the head gibs and double-check runout on the spindle (on the shank of the tool you're using - you never know if your ER collet or chuck could be causing the problem). Otherwise maybe it's rigidity issues which *could* be fixed by scraping, but again I would check and double-check everything else first. It's really not that bad imho, just needs a bit of fiddling to get it cutting accurately.

Hi, can I ask you for some info on using the software? Which do you use to create the gcode? In practice you use a stl as file format?

Buenos videos como siempre, lástima q no tenga mucha precisión. Las máquinas con colas de Milano es lo q tienen. Hace tiempo pase una Perrin u1( parecida a la deckel) a cnc. Deje los husillos originales y le puse reglas Digitales. La mejor manera de lograr algo de precisión con esas máquinas es usando unas placas 5i25+7i76 y emplear las reglas para corregir el error. Hay varios post en foros linuxcnc. Esperando nuevos videos, un saludo.

one of the axis is pretty good, like +- 0,01mm but the other one is much worse, like +-0,1mm, the surface finish is about what you would expect from mild steel

L I K E 👍👍👍👍👍💥💯💥💯💥😻😻😻😻😻

looks terrible. Might be the controller or backlashes, most evident on corners