[deadlykitten]

hi beefheart, i was going to tell you to check the backlash, at least in the zone where you cut the part

however, your errors were pretty normal :
... post 1 : width varies about 0.04 mm
... post 3 : varied up and down by some hundreds

with errors so small, like hundreds, i wouldn't have believed that there is a baklash problem; so far, i was blaming your tool actually, i was also thinking about an issue with your dial indicator

here, during long term setup, dimensions shift with <0.03 after a few hours, because machine gets warmer, so i would expect a few more hundreds from your machine

however, from few hundreds, to 0.2mm backlash, is a long way good luck


ps : now, it does not matter anymore, but :
... if you wish to check if the turret is clamping slowly, like underthetire said, you may need to use 2 dial indicators : one on the turret ( revolver ), and another on the saddle if turret is clamping slowly, then the dial that is on it, will shows some deviation
... a lot of pecking will damage the insert faster, so try to keep it @ a minimum; for your part, i would recomand :

    G97 S2000 M03 M08 M42
    G00 X22.5 Z-75
    G01 X4.5 F0.06789 10 11 whatever
    G01 X-2*0.4 S250 M03 M63 ( rpm is changed during feeding, X axis is not reversed )
    G00 X22.5

[Beefheart]

... a lot of pecking will damage the insert faster, so try to keep it @ a minimum; for your part, i would recomand :

    G97 S2000 M03 M08 M42
    G00 X22.5 Z-75
    G01 X4.5 F0.06789 10 11 whatever
    G01 X-2*0.4 S250 M03 M63 ( rpm is changed during feeding, X axis is not reversed )
    G00 X22.5

Thanks for the code, new way to cut a part for me. I have a part time shop, working 3 days a week (and weekends) I use Fusion 360 to my Haas wmc and tried to use it for my lathe but I don't like the lathe module so I'm going to program the Okuma manually. The other two days I'm working for a bigger company programming a Mori Seiki NTX using Esprit cad/cam so my code typing is a little bit rusty.

/B

[deadlykitten]

hy pontiac

teahole is a rare presence, but when it shows up, it hits i am glad that there was an easy fix for your 98 okuma

you have to run some maintenance, inspections on the machine; like this, you would know the numbers for your machine : look behind the covers, check the screws, guides, clean the chuck, check the lube flow, at least 1 / year; check backlash at each 20-30mm, check turret tir, etc

i have a geometry record for each machine, and like this, i know when something is out of alignment ... if the nut was loose, then you should observe 0.2backlash all over the Z axis, from the spindle, to the tailstock

Thanks for the code, new way to cut a part for me.

it may not work, because of that M63

I have a part time shop, working 3 days a week (and weekends)

so you work for you 3+2=5 days ... thus, your ' main - esprit - job ', is during the week-end your have a shifted calendar

i will share few tricks, so to be able to increase precision class on older models

ok, go and unscrew that nut, put the backlash back to 0.2, and let's deliver nice parts

during parting off, when you said " I made another shaft after that and it varied in length about 0.07mm. ", it was because the cutting load may have moved your turret, towards the backlash; there is no fix for this

during grooving, when you said " width varies about 0.04 mm ", you could have used a gentle approach, like G01 Z_at_least_1/2_screwball_revolution F_gentle, so to keep the backlash behind the nut, in a repetitive manner also, a small doc + gentle feeding, should generate lower cutting forces, keeping the turret steady ... i use such tricks, when i have to deliver tight tolerances

if you wish to deliver tight tolerances, you have to keep the cutting load under a zone that does not generate dif, and to keep the backlash only on one side of the nut / kindly