I would think it would be more of a mechanical problem, if the Z height changes over time but the readout does not, I would suspect a stepper coupler or other mechanical issue.
I was running the same part for several days.
When I compared the first part to the last part I ran I noticed considerable tolerance changes.
I think my 1100 must slowly change (head drop a little) over time.
This has led me to re-reference my parts every 2-4 hours run time.
Anyone else notice this happening?
Similar Threads:
I would think it would be more of a mechanical problem, if the Z height changes over time but the readout does not, I would suspect a stepper coupler or other mechanical issue.
mike sr
The Z axis coupler is visible by taking the cover off the Z stepper adapter, put some marks on the coupler and shafts, use the machine and see if there is any movement between the couplers and shafts.
Most other things would be taken care of by the weight of the head as far as slack is concerned. I have seen posts in the past about couplers slipping etc.
Good luck, no one likes a machine problem!
mike sr
I am curious as to how much a difference you found, and how? Were you doing different operations to the same parts, days apart? Generally speaking, if you are facing both sides of the part in a short time, it wouldn't matter if the head lost a thou or two over a couple of days.
I've seen tools move around in holders too, so check their length as well if you're hunting for error sources.
If the head was lower by .002 and you face both sides of the part then the thickness ends up smaller by .004. Depending on the acceptable tolerances it can matter alot.
I have had issues with the XY drifting when left on for days at a time. A while back I was making a part and ran out of time on the weekend. Roughing and pre-finish passes done with just the finish pass left to go. I didn't get back to it until the following weekend. Machine was left on and waiting for a tool change (no ATC). Loaded the next tool and hit go. No clean up on the left side of the features and excessive cleanup on the right. Hmm... checked a hole location and X had drifted around .020. I don't know how, perhaps the 20-30 degree temperature change every day, maybe a power fluctuation not enough to restart the control but enough to cause the stepper to lose position?. Can't say why. But now I always re-probe the part if it has been static for any significant period of time. Takes a minute.
Just for reference, I did a test on this today. I set z height on one unused offset. Then run code, loaded material, set and reset normal offsets for a number of different parts for about 4-5 machine hours over a day. Went back and checked the offset I set a beginning of day and found a 0.0008 difference. This is near the tolerance of my Hamier. . The measurements were made by sending the probe to z 0 height at 15 ipm from full height. So stiction should be limited.
Edit:
My bad, forgot I shut machine off for lunch. Will have to try this again.
Last edited by mountaindew; 03-17-2018 at 08:57 AM.
This sort of thing can happen sometimes when you have the Step signal referenced wrongly. That is, its rest state might be HI when it should be LO.
What happens is that a Dir change can interfere with a single Step pulse under these conditions. That means that over time the Z axis can creep, a single step at a time.
Yeah, very technical - but it happens, and it matters.
Cheers
Roger
Do the ball screws grow as the machine warms up and do they then, shrink back down when sitting idle for a while?
Yes they do, steel expands at around 6 parts per million per degree F. So over 14 inches the change in length due to a 30 degree F temperature change is: 14x.000006x30 = .00252
Enough to take into consideration when setting up on a cold morning and running all day. But the next cold morning it should come back to the original position, the issue observed is that it does not.
It's interesting that only the expansion of the ball screws ever seems to get a mention on this forum but never the expansion of the entire machine. Cast Iron also expands at around 6 parts per million per degree F so most of the difference will cancel out if the temperature distribution is uniform (which it won't be). Consider a machine made of cast iron, with a cast iron vice and using ball screws with the same temperature coefficient. Move the spindle down until it just touches the vice and then warm up the machine uniformly. Everything will expand the same amount. The spindle will still be just touching the vice. It's like enlarging a photograph of the mill.
The only parameters that concern us in reality are the DIFFERENCES in thermal expansion coefficients and the DIFFERENCES in temperature between or along the components.
Step
After cleaning my machine and performing a tune up 2 weeks ago. I then replaced an 16 year old agp memory sharing, pci computer used as a controller with a brand new one from Tormach on Monday. "took all of 15 minutes including update of 2.02 release and restoring my backup of PP settings".
I also carefully recalibrated my Hamier probe in all directions including resetting zero "being very careful as the tormach video stresses". Then carefully measuring new height and set in tool table. I run a number of base line tests all week and found I get 0.0008 to 0.0005 difference in z axis measurement after using machine for 1 hour or 1 day or shutting the machine off and homing it the next day. Overall I see lots of zeros on my x and y axis measurements, because the Hamier had not been checked and adjusted in over a year. Homing precision almost doubled to 0.0009 it was around 0.0018. I bet because my probe has much lower run out and better height tolerance from proper adjustment that was over due. The new controller also provides real time dro numbers and even smoother operation, something that was lacking with the old iron and overall very happy I replaced it.
Anyway this system is just getting better with age, upgrades, experience and proper maintenance
Still would like to try and use a tool setter system to see if I can improve on the height gauge to tool table entry system. I can see tiny differences in those numbers when compared to the tool table from day to day. Maybe a higher quality height gauge and or take more time setting tool heights would help also.
Note: adjusting zero on the older Hamier probes is different then shown in Tormach video. The screws are on top and bottom left. Carefully loosen left bottom first then left top. then adjust bottom to zero then top screw is lightly tightened. I set run-out as per directions included with probe replacement tips
I have something similar if not worse. Using a test block i probe and set origin to zero. Moving the axis in one direction only (say x) back and forth using the jog wheel i re-probe and get a new DRI reading off say 0.001mm. Running the same simulation using code it can be out as much as 0.15mm or more! I have checked every from Gibbs, Couplings, Controllers and even tried a different PC. The axis driver for Z wasnt showing the issue so i swapped that over to X and ran the simulation only for it to still show an offset of 0.1 or greater (sometimes less). X and Y are the issue were Z is fine. The big question here is why using the job when i get great accuracy but as soon as i use code it screws up. Tool compensation is off and G40 is active. Its driving me nuts!
Sounds mechanical to me.
What happens if you slow both speed and acceleration for the axes?
Cheers
Roger
Unfortunately i get the same results..Im trying to narrow it down and all i can say at this stage is its looking like a communication problem. I can move the table around with the jog wheel in all directions at full speed and have it come back to origin within 0.005mm. Running code its still 0.1mm or more
For what it is worth ... using the jog wheel uses the same communications hardware and software as running code. I have never known any of these problems to be in either Mach or the communications, but I guess there is always a first time.
Cheers
Roger