What's new
What's new

5 Axis Centres Thermal Movement

That sounds to me like your centrelines are out if one side is perfect and the other is out. Not sure they they are set on a mazak, but on a roeders there is a probing cycle that probes a tooling ball at all different c and a angles and updates the centreline coordinates.
I've done it twice now manually with clocks, and the engineers done it a bunch of times with the renishaw. We got the two methods to *mostly* agree for the first time a couple of hours ago. Just ran the test part and still got one side up about 14 microns, much closer than 40, but I've had it to 5 before with clocks, the next day it was all out again due to the thermals by the sound of it.
 
I have used various, from a Roeders that would move less than 0.002mm week to week. That was the nuts. It had full cooling circuit running through the casting, and also on the drives it's self along with spindle growth comp. I've also used stuff that's moved the same as your seeing. The first 5x I worked on was a Mikron UPC600, when we first got it we found the same as yourself if not more. We had major issues trying to blend toolpaths in from different tool vectors. We initially thought it to be a machine problem, then the supplier paid to have it put in a temperature controlled tent for 2 weeks as a trial, and that helped the issue straightaway. It still wasn't perfect but it helped massively. If you're mould making and trying to blend 2 tool paths in on an impression it's always going to be a battle. I also found that varying tool diameters caused blending issues as well.

Its a difficult one as it's not really feasible to reset your spin centres numerous times a day.
Out of interest, how much does a Roeders with a 500mm-ish table cost if you know?
Hopefully out of budget to save me some buyers remorse.
 
glass scales are absolutely necessary for low single digit micron accuracy.
The exception to this being machines that have fine pitch ball screws, good climate control, and all of the other "tricks" to get high accuracy without scales.

A really good machine can sometimes ball-bar test with better accuracy with the scales turned off. The battle between encoder and scale can lead to goofy movement at an extremely high resolution. This is why some of the "old school" high end machines don't come with scales as standard.

That is almost certainly not the case on a budget Mazak though!
 
The exception to this being machines that have fine pitch ball screws, good climate control, and all of the other "tricks" to get high accuracy without scales.

A really good machine can sometimes ball-bar test with better accuracy with the scales turned off. The battle between encoder and scale can lead to goofy movement at an extremely high resolution. This is why some of the "old school" high end machines don't come with scales as standard.

That is almost certainly not the case on a budget Mazak though!
Gotchya, I can see that being a servo tuning nightmare.
Would you expect scales to help with control of the centres?
The machines been pretty damn tight until it's asked to translate the work coordinate system. It's 1 or 2 micron repeatable and probably accurate within 5 if I was doing 3 axis work. I'm just about to do some circular interpolation so I will gauge it at multiple orientations.
 
Gotchya, I can see that being a servo tuning nightmare.
Would you expect scales to help with control of the centres?
The machines been pretty damn tight until it's asked to translate the work coordinate system. It's 1 or 2 micron repeatable and probably accurate within 5 if I was doing 3 axis work. I'm just about to do some circular interpolation so I will gauge it at multiple orientations.

This is the type of job that really kicks 5 axis machining up to the next level. There are a TON of dynamics involved here. Hard to make blanket statements.

The reason scales are critical for most applications is because of tolerance stackup. By the time you factor center-of-rotation issues, tool growth, part growth, rotational error, etc... There are a lot of small things adding up.

In order of importance, the three things I see most frequently causing grief during five axis machining are.

1 - center of rotation not correct
2 - tool length not correct
3 - climate control and growth issues
*bonus* machines with poor rotary accuracy and parts projected a mile off of the center of the rotaries

None of those things are really that important for 3 axis work, but they are super critical for getting consistent accuracy when working in 5 axis. I've been in multiple shops where they are successfully making 5 axis parts with tight tolerances for YEARS, but they have to dick around with feature and plane shifting constantly at every new orientation. When things are tightly controlled and set correctly, this is rarely necessary.
 
Understandable, the best answers always start with "it depends" lol.
Yeah quickly I'm getting the impression just how much stackup can get you in trouble.
I think if I can get everything to stay put thermally I can get this machine to do what I'm asking, but the temp is gonna be the big issue.
Keep everything controlled and the machine should take care of itself: I only get 1 go normally so I'm gonna have to get some of these variables nailed down.
 
We had a production job that had 2 off 8mm dowel holes over 300mm in an ali part with true position of Ø0.05. Holding that size was ok during proving off the part. Once in production they were struggling to hold it. Turned out they hadn't spec'd a coolant chiller with the machine and over the day the coolant was warming up, which was transferring in to the part in machining. The part would then shrink in room temperature and the holes would fall out of spec.
I recently cut a medical mold where the cl distance was extremely important, I cut it in the Hermle and it was 72 degrees in the cabinet, by the time I took the block out and got it into the 68 degree inspection room, the cl distance shrunk by .00015" on each side of the cl. This was a deal breaker, and had to rethink our process.
 
All excellent points that I will do my best to execute on.
Thankfully (I think) all my work is one and twos, the machine is normally cold so perhaps getting it stabilized and keeping it at 10% rapid to avoid heating the screws up is the way to control that. Getting the blend perfect I'm happy comping the toolpath out in CAM and sneaking up on it, but I'm getting a missmatch side to side of 40microns. One side a perfect blend, the other 40 too deep. They've thrown kit at it, we've probed the centres a bunch and got things pretty stable so just about to run the test program for the real part. It'll probably be all out again tomorrow when the weather changes.
This sounds like a kinemetics issue more than anything. When the machine is at operating temp, calibrate laser, then calibrate probe, then run kenimetics cycle, in that order, if you don't your chasing your tail. I'm not familiar with the Mazak 5axis, but this holds true for just about every other I know.
 
Last edited:
My experience is that scales help most as the machine ages. That said I have also seen non scale equipped machines retain excellent accuracy and repeatability when maintained very well and regularly pitch comp adjusted via laser test results.
 
One of the engineers that's been (we've had about 7 different guys come look at this point for various issues (definitely built this one on a friday afternoon)) wanted it lasered and ball barred since they had to slacken the x and z rails off and retighten to get it run straight after delivery. I'm inclined to agree with him. I'm also wondering doing some reading if there's an issue with the thermal comp system. It should be able to adjust for most of it surely, it's covered in thermocouples even if the screws aren't cooled.
 
he did say that he'd got it better than what they work too in the factory for straightness and perpendicularity so that's in the win category.
 
I’d point out “ambient stability” is a derivative of the number of measurement points and airflow. The larger the thermal mass absorbing a smaller temperature gradient the greater the the stability.

There are generally always gradients of differential heat transfer. More data points better than fewer. Which data points being actually relevant a completely different and generally unique question
 
Ballbar was mentioned several times above as one of the aims to trace the 5 axis machine performance.
Unfortunately, conventional use of the Ballbar (simultaneous movement of 2 linear axes) is not helpful in 5 axis tasks. In the past I described here the routine which establishes the position of centers of rotation of rotary axes in machine coordinates, but this is based on idle positions of the axes.
The ISO standard 10791-6 describes kinematic cycles, which involve simultaneous movement of 2 linear and one rotary axes (AK1 and AK2) or 3 linear and 2 rotary axes (AK3 and AK4) and specifies the expected results and tolerances. Renishaw's Ballbar Trace software is one of the aims to perform these tests. This (
) shows how it looks in action.

Stefan
Cogito Ergo Sum
 
I think Mazak have got all that. They seem to have all the kit and software in the Renishaw catalogue. The challenge will be getting it out of the factory and on site by the sound of it.
 
To the Op, have you found a state where the machine has reached thermal equilibrium?
It would be interesting to run a probing cycle on a reference part through the day while the machine is doing a warm up cycle.

I also think high rapid speeds contribute to accuracy loss. Back and forth on one part of a screw is creating heat.

Interesting thread, not many on this forum are worrying about this level of accuracy.
 








 
Back
Top