Trunnion Troubles --- Ep. 74 

Yep, tooling balls are magical for locating parts in space! I've designed machining fixtures for 5 axis parts and we always add a precision hole with tooling ball. Then everything is always based off it. 💪

I’ve used 2 5 axis machines (3 if you count PocketNC). One of them provided a tooling ball and a script to calibrate with it. That trunnion had pretty poor referencing so if I wanted any accuracy I had to calibrate it every power-on, but at least I had the power to run it myself. With the other machine they did during install but didn’t want to empower me to do it myself, so I’m always wondering if it needs calibration. I’ll have to look into the cube thing. It’d be super useful to document what you’ve learned, and the code would obviously be useful too if you can talk the tech into letting you share it.

It seems kinda silly that they hold onto these small scripts with a tight grip. If I had a HAAS I couldn’t deal with that and would have to write my own. Is MRZP translational offsets? It doesn’t seem like a crazy amount of work. FYI I have never used any 5-axis so I could be way off with what the problem is. But here's how I would go at it: Find the center coord transformation of the ball on the XY plane through the euler rotations of the trunnion. Inverse transform back to the machine origin should get you the offsets of the 2 trunnion axis. You could get really advanced and run multisampling through the full range of the trunnion’s rotational limits. This would account for possible wear spots on the ball or if you don’t trust how round it is. Would give you a higher degree of accuracy than could be measured from a single sample. With lots of samples of the offset from lots of angles you get the true offset center coord in 3d space by the median of the sampled points and I think also how spherical the ball is based on how close together the subsamples are. People that know this stuff really well are game and VFX programmers. I think the whole thing could be simulated and verified in a game engine such as Unreal Engine with blueprints with a virtual haas machine rigged up. It’s far easier to find the solution to problems such as this when you can run through it on a computer and then convert it back down into simple gcode logic.

Keep up the great work .👍👍

I have a umc750 and hold +/-.0005 and I haven had any trouble holding tolerances ..

when you sweep the platter at A90 did you rotate to A180 and see if it matches? You need to split the diff if they are not the same ya?

The Harvey 4 makes the pidcast! (8fl rougher) I ran those in stainless for months (no better tool)

AJ, if you need any FEA work done on those flextures let me know!

With Fanuc it depends on the control. A 0IMF is 3+2 or 4+1, up to 4 axis simul. A 31iB5 is full simultaneous 5 axis. That's the difference.

This business of Haas NOT installing my trunnion is pretty dam disturbing. I have a new VF4SS fully locked and loaded including a TRT210 being installed next Friday. I HAVE ZERO 5th axis experience, and this would be the first time I've heard that they may not install the TRT. I'm getting more pissed off by the moment. Swear that if this is the case Ill sell the machine and purchase a DN Puma.