Circular Interpolation out of round, help with trouble shooting.

Bill in PA · Mar 29, 2015

Hello,

NASA III knee mill, Fagor 8020 control, analog drives, rotary encoders.

Interpolated inside diameters are on size on the x axis, +.007" on the y axis.

Backlash checked with .0005" indicator at the center of the table where the vise is: x axis .000", y axis .000".

Following error is .001" per inch of feed rate. Checked at the following feed rates: 5, 10, 20, 30, 40, 50, 60.

I observe no marks in the machined surface in any of the circle quadrants upon the axis direction change (like one would see when there is a backlash issue).

This problem surfaced suddenly within the last week. I took some time to look at it today. I interpolated a 2" inside diameter and the error would repeat with every cut; with or without cutter comp.

I will find out if the y axis is locating correctly when I put up the next job which requires a bolt circle.

About 2 weeks ago I experience a y axis following error during program execution. It just happened the one time (of course scrapping the part).

I am looking for any ideas of what I can check next.

Thanks much,

Bill

specfab · Mar 29, 2015

I can't offer much in the way of specifics, but I would suggest doing some sort of test with a reference feature at center, like a hole, so you can then see whether the error is symmetric about center (or not). Also look at whether feed direction makes any difference. It sounds as if there is some sort of drift with the encoder (maybe loose encoder disk?) if there is no step artifact at axis direction. Sheer speculation...

Bill in PA · Mar 29, 2015

specfab said:
I can't offer much in the way of specifics, but I would suggest doing some sort of test with a reference feature at center, like a hole, so you can then see whether the error is symmetric about center (or not). Also look at whether feed direction makes any difference. It sounds as if there is some sort of drift with the encoder (maybe loose encoder disk?) if there is no step artifact at axis direction. Sheer speculation...

I thought about a dirty encoder but I expected that would be coupled with a following error. I did not think about some sort of drift however. The servo is inside the knee below the y axis ball screw. I won't have time to pull it until Wednesday afternoon.

I will be running a part tonight that has a bore inside a circular pocket. I can check the pocket in relation to the bore and the bore location in relation to the sides to see where the error lies.

Thanks.

Bill

mike_kilroy · Mar 29, 2015

You say . Your pos gain is 1in/min/mil, but not how close it is really adjusted on the 2 circ inter axes...... if cutting circle at 10ipm are both really perfect .010in fe? If not identical on both axes that will cause your egg shape.

Bill in PA · Mar 29, 2015

mike_kilroy said:
You say . Your pos gain is 1in/min/mil, but not how close it is really adjusted on the 2 circ inter axes...... if cutting circle at 10ipm are both really perfect .010in fe? If not identical on both axes that will cause your egg shape.

At 20 ipm:

x axis following error is .0200"-.0202"

y axis following error is .0200"-.0203"

Bill in PA · Mar 30, 2015

I am not sure what is happening now. I started the next job. I bored the center hole and cut the 2" circular pocket. Roundness of the pocket according to the CMM: .00075" - .00083". Bore and pocket are concentric and located at the right spot.

I have not tried rapid approaches from different directions to see if anything changes. I am not going to right now since I am getting a good part.

I still welcome any thoughts to this issue. I cannot assume any part coming off is correct.

Thanks,

Bill

Tonytn36 · Mar 30, 2015

Is it timing belt drive from motor to screw? (I'm assuming so if the motor is tucked under the knee.) If so, check for crap in the sprockets and the condition of teeth on the belt. Pay close attention to the belt, they can sheer and it changes the pitch of the teeth slightly but you have to really look good at it or you won't notice it. Also check that the sprockets are not slipping slightly on the shafts. Scribe a sharp line across the end of the shafts and the sprockets while you have them exposed.

DMF_TomB · Mar 30, 2015

hole not round

Bill in PA said:
Hello,

NASA III knee mill, Fagor 8020 control, analog drives, rotary encoders.

Interpolated inside diameters are on size on the x axis, +.007" on the y axis.

Backlash checked with .0005" indicator at the center of the table where the vise is: x axis .000", y axis .000".

Following error is .001" per inch of feed rate. Checked at the following feed rates: 5, 10, 20, 30, 40, 50, 60.

I observe no marks in the machined surface in any of the circle quadrants upon the axis direction change (like one would see when there is a backlash issue).

This problem surfaced suddenly within the last week. I took some time to look at it today. I interpolated a 2" inside diameter and the error would repeat with every cut; with or without cutter comp.

I will find out if the y axis is locating correctly when I put up the next job which requires a bolt circle.

About 2 weeks ago I experience a y axis following error during program execution. It just happened the one time (of course scrapping the part).

I am looking for any ideas of what I can check next.

Thanks much,

Bill

.
i have had cnc where y was at the travel limit and could not move farther, if machine takes 5 seconds to get servo overload error but you are cutting a circle i have seen y stop moving but x is still moving and eventually y starts to move other way. basically cutting a circle with a flat spot where y was not moving
.
i just had to have part not at travel limit and then i got the normal circle being circular within about .001" or less

palr717 · Mar 30, 2015

Would like to know the solution to this problem if it's ever figured out, thanks.

Bill in PA · Mar 30, 2015

DMF_TomB said:
.
i have had cnc where y was at the travel limit and could not move farther, if machine takes 5 seconds to get servo overload error but you are cutting a circle i have seen y stop moving but x is still moving and eventually y starts to move other way. basically cutting a circle with a flat spot where y was not moving
.
i just had to have part not at travel limit and then i got the normal circle being circular within about .001" or less

Been running small parts in the middle of the table. Not near the limits.

Thanks,

Bill

Bill in PA · Mar 30, 2015

palr717 said:
Would like to know the solution to this problem if it's ever figured out, thanks.

I will post the solution when found.

Had an all-nighter to finish a job, and I will probably have another for a job that is due tomorrow afternoon.

I will carefully inspect the belt. I will also check the encoder if the belt looks OK. I have been in that encoder before, so that is a possibility.

Making runs on Wednesday, so Wednesday night or Thursday, I will spend some time with the y axis.

Thanks,

Bill

Bruce Griffing · Mar 30, 2015

If you are able to reproduce the problem, I suggest you follow the problem interpolation run with another exactly the same, but in the other direction.

Bill in PA · Mar 31, 2015

I checked the belt today. It had an aluminum chip on it, but otherwise it appears OK. I made a small adjustment to the pulley alignment. The servo mounting plate was not as tight as it should be. There was plenty of chips lying in the bottom of the belt cover. I used some foam rubber to plug up the gaps in the top of the cover.

I also loosened the y axis gib a little. Appeared to be a slight stick-slip once-in-a-while? I did have some difficulty getting the gib to slide once I loosened the screw.

The mill is running very well today. I am currently milling a circular pocket with a circular boss in the middle. I am very happy with size, concentricity, roundness, and location.

I am not planning to pull the servo unless the problem resurfaces.

Thank you for all the input.

Bill

Atomkinder · Mar 31, 2015

You said this is a knee mill, yes? How's your lubrication system functioning?

trevj · Mar 31, 2015

The Milltronics I used to run would sometimes work loose the nut that held the thrust bearing on the X axis.

If it got loose enough, then running rapids down to the other end sometimes dragged it back in and seated it, so that it would run loose again later. General PITA. Easy fix once we learned what to look for. Eventually I waited until the boss wasn't around to say no, and drilled the nut and lockwired it (Military, they thought that fixing your own stuff as 'not our problem', even though the guys that used to do it had long since been downsized away).

Frustrating as heck.

Any possibility that it is something in the code being turned on or off in the way of compensations? Finger coded or CAM produced? I got pretty close to zero experience hand coding, but have found a lot, if not nearly all, the ways to make my own life miserable in GibbsCam.

Cheers
Trev

Bill in PA · Mar 31, 2015

Atomkinder said:
You said this is a knee mill, yes? How's your lubrication system functioning?

Very well. It pumps oil every 15 mins. of spindle on time. It makes quite a mess.

Bill in PA · Mar 31, 2015

trevj said:
The Milltronics I used to run would sometimes work loose the nut that held the thrust bearing on the X axis.

If it got loose enough, then running rapids down to the other end sometimes dragged it back in and seated it, so that it would run loose again later. General PITA. Easy fix once we learned what to look for. Eventually I waited until the boss wasn't around to say no, and drilled the nut and lockwired it (Military, they thought that fixing your own stuff as 'not our problem', even though the guys that used to do it had long since been downsized away).

Frustrating as heck.

Any possibility that it is something in the code being turned on or off in the way of compensations? Finger coded or CAM produced? I got pretty close to zero experience hand coding, but have found a lot, if not nearly all, the ways to make my own life miserable in GibbsCam.

Cheers
Trev

When I was having the problem on Sunday, I tried both CAM and hand coding. Same result.

Thanks.

Bill in PA · Apr 3, 2015

Y axis had another hiccup this morning. You'll see in the photo that the first outside roughing pass was offset about .040" in the positive Y direction. All subsequent passes were correct.

I will be checking the screw and servo tomorrow unless I cannot get three good parts off today. The first was scrap, the next OK.

Bill

SteveEx30 · Apr 3, 2015

Is it possible to switch the X & Y encoders to determine if that is your problem?

Bill in PA · Apr 3, 2015

SteveEx30 said:
Is it possible to switch the X & Y encoders to determine if that is your problem?

I could, but the problem is intermittent. I don't know how long until the error may occur again.

I don't have any spare encoders. I would prefer to leave the x axis alone.

Thanks.

Circular Interpolation out of round, help with trouble shooting.

Hot Rolled

Titanium

Hot Rolled

Stainless

Hot Rolled

Hot Rolled

Diamond

Diamond

Aluminum

Hot Rolled

Hot Rolled

Titanium

Hot Rolled

Titanium

Titanium

Hot Rolled

Hot Rolled

Hot Rolled

Stainless

Hot Rolled

Similar threads