I'm not sure what a PCD is, but it seems to me that the live tool positioning is about as accurate as the cutter positioning given that the tools are held in the same tool turret as the cutting tools. I'm easily holding sub 0.001" tolerances on the static tools so I'm guessing that the live tool positioning in the Z-Y directions has to be that good plus whatever radial play you've got in the live tool bearings. My live tools are pretty worn now (they were bought used) and have developed about 0.001" radial TIR just off the collet face when I hold a ground shaft and push on them. That being said, I think they'll end up drilling centred when they spin.
C-axis interpolation seems to be very good on my lathe. I'm doing some circular interpolation to finish out a hole that's not on the spindle axis. My gauge pins only go to 0.0005" increments, but I haven't been able to measure any variations in my hole diameter so I don't think the interpolated holes are out of round by more than 0.0005" or out of diameter. I've looked at the holes under a microscope and I can't see any interpolation ribs or other funny obround features around the pin. Surface quality of slots or polygons seems to be quite nice if the part doesn't flex too much.
0.002" is a pretty big distance for a machine that is in good adjustment, but I note that my turning diameter can vary by nearly 0.001" from cold start to warmup so if you want 0.002" positioning repeatability, you'll probably need to do a warmup cycle to get 0.002" positioning over a big 800mm part. My machine travel is much shorter at sub 300mm. I would expect more cool-warm variations for a much bigger machine. Still, thermal expansion issues seem to be be fairly repeatable if you have a steady part to part production cycle. If I was really picky, I'd run a different program at the beginning of each day with a compensated program after the first couple hours.
Everything kind of goes out of whack after a turret bump. I find that the entire turret may either shift in a single axis or it can twist out of axis after a crash (depends on the geometry and energy of the hit). I pretty much touch every tool in and check turret parallelism after a hit. I don't think there's any sense of accuracy after a crash if you don't reset all the offsets and check the turret parallelism. I've got a lot of operations with long boring bars. Everything goes to chattering sh*t after a crash if I don't confirm that they're cutting across spindle centre height.
My Daewoo is significantly smaller than the one you're considering. I've got a Puma 200, but I've found it to be a pretty decent design from a crash recovery perspective. The toolchanger is fairly easy to realign and so is the spindle. Plenty of space to access the adjustment and fixturing screws in this design and my machine is a pretty old '97 build.  |