CNC contouring prblem - surface finish

"ADkeit" wrote in news:1108633177.375406.183380 @o13g2000cwo.googlegroups.com:

Aaron, You need to check the following: Set up a base and indicator on a solid piece of the machine in line with the x axis. Using incremental mode, step the x axis toward the indicator a few steps. Reverse direction and step away from the indicator, taking count of the distance required before the indicator moves. It _should_ move exactly one step backwards when you change direction. The distance it does not, is your backlash. If it is less than about 0.25 mm, you should be able to compensate for this in the control parameter for backlash. Check Y axis using the same method. If the backlash is more than about 0.25 mm, the control will most likely not be able to compensate for it, and you must find the cause of the excessive motion. The cause could be several things. a) worn ballscrew/nut, b) loose or worn ballscrew thrust bearing, c) problems in the drive tuning. Problems a) and b) will most likely be the culprit, as if it were c) you would have other problems besides those you describe.

Hey Aaron,

Does the feed rate make any difference, or the cutter size? If you do a mirror or duplicate, do you get EXACTLY the same thing?

Take care.

Brian Laws>I have a Supermax CNC mill with an anilam Crusader-M control, made in

Loose gibs will do this, too. Power feeds should have a little friction against which to work. The tables' gibs should be adjusted tight enough that a given bit at a given feed depth and rate won't "self-feed" into the work. Of course, don't over-tighten them, or you'll overwork your drives.

DROs give an accurate output regardless of play in the system, but if the table cannot hold its position as a bit bites into the work, then the table can overshoot the desired feed point, because of lost motion in the drive. You can minimize this, but it can never be completely eliminated. Adequate friction in the gibs is the only solution, shy of _true_ backlash-free drives.

LLoyd

LLoyd

I'm not familiar with this particular machine, but in the late 80's - early 90's I did CNC service and virtually all the larger machines I worked on did not use a "DRO" with scales on the tables. The majority used the 2,000 line encoder on the back of the Fanuc servo and pitch error mapping for the ballscrew. Some of these machines had axis travels of 4' x 4' x 8' so a linear scale on an 8' axis would be difficult at best. In this case the control is not aware of any play/backlash/wear in the system.

Pete C.

When was the last time you tuned the servos? Sounds like they are out of balance and one has more following error than the others. Do a full servo amp tune up and see what it does.

Gunner

Rule #35 "That which does not kill you, has made a huge tactical error"

I had a similar problem with contouring aluminum tooling plate. We were using a 1" diameter ball-end cutter with a PVD insert and a spindle speed of 5400 RPM. The surface was step-profiled previously with a regular end mill to approximate the contour, so the final cuts with the ball mill were very light -- 0.005 - 0.05 per pass. The feed in these passes was 20-40 IPM.

The finish was "hairy" with roughness much worse than you'd expect from the cutter geometry alone. I checked with Valenite to see why their tooling was doing such a poor job, and the answer surprised me. We needed to up our feed rate by a factor of 5-10 to allow the cutter to eject chips instead of rubbing the aluminum surface and rewelding them.

Try a big > I have a Supermax CNC mill with an anilam Crusader-M control, made in

OK, this helps narrow it down. If you are having a 3-5 thousandths error in the circular arc, that is likely to be backlash in one or both axes (X and Y). With that much, you should be able to SEE the errors at the points where the axis turns around. I have backlash of .001 - .0015 on my machine, and I can see the slight bumps where this happens. (My backlash is mostly flex in the leadscrew end bearing mounts, not true slack in the ballscrew.)

Anyway, I never mill out a bearing nest. I may mill it to leave .010" of material, and then finish with a boring head, so i know it is truly round.

For the other problems, what is the encoder increment (or step size, if stepper driven)? With the profile milling, is this all G01 moves, or does it have G02/G03 moves, too? Are the G02/G03 moves smoother than the G01? If so, then the resolution of the CAM file is suspect. In other words, if the G-code says to move in rough linear steps, then the machine is only doing what it is told. If the G02/G03 are bumpy, too, then it probably indicates it is a motion problem.

Are you climb milling? I try to do everything in the climb mill direction, if possible, as you get better tool life and better surface finish, most especially on aluminum.

Jon

Thanks for all the replies so far... here is some more information: The mill uses glass encoder scales, so checking for backlash with an indicator won't work because if I tell it to move one step of .001", it does so.

The specific cutter or size doesn't seem to make much of a difference, and yes, the problem occurs exactly the same way if I cut a mirror or duplicate part. I'm not sure if feed rate makes a difference or not, I'll do some tests later today and get back to you. I imagine if it is a problem with servo tuning it would be much easier for the control to keep an accurate position with very low feed rates. (The cuts I was making recently were I lowered the feed rate to 15 ipm with a finish cut of .010", 3/8" deep with a 3/8" diameter cutter. Although I did do some only 1/8" deep and is didn't make much any difference.)

friction

OK...I think the gibs have something to do with holding the table against the ways??? Not sure how to go about adjusting them or what they look like... The machine is a knee mill, about Bridgeport size, probably a bit heavier.

Well, I've had the machine for about 3 years, and I haven't tuned the servos. Is this something I would have to hire a service tech for? I know there are a few potentiometers on the servo amps (made by Glentek or Westamp I think). Or is there more to it than that? Are their parameters inside the Anilam control that affect servo tuning? Perhaps I'll have to find a service manual...

The surface texture or not rough at all, I believe that the cutter is doing its job. Like I said, the surface on moves parallel to the X or Y axes is fine. One the arc or angled cuts the surface looks overly "wavy". If you run you finger along it is not rough, just a little bumpy. If you run your finger in the direction of the Z axis it feels smooth. The waves all run parallel to each other, kind of like driving over an endless series of tiny parallel speed bumps (...or inverted speed bumps).

The occurs with G02/G03 moves, and with simple straight line moves that are not parallel to an axes, so its not the programming. However, your note about backlash reminds me of something that may be related. I have noticed that there is excess noise (like a grindings or vibrating noise) eminating from one end of the table during X-axis moves. It appears to be coming from the pillow block that holds the ballscrew. I have noticed this for a while and was planning on looking into it when I have the time, and don't need to use the machine for a while. The noise (much louder than noise coming from a Y axis move) shows up mostly at feed rates, and is curiously quieter during Rapid moves. Either way, I figure I have a bearing that is going out. Could this contribute to the problem?

Thanks much for all your help. Any more is greatly appreciated.

Aaron Keit

We had a Monarch VMC 150 at work that did this (Y Axis). They have a piece of plastic like material that epoxy's to the adjuster (I believe it was), and when it started coming loose, it provided the exact same finish OP speaks of. Its not that the machine doesn't know where it is, encoders were working fine. The problem is the table wobbles on the gibs. When drilling holes and such, it wasn't so much of an issue because the backlash comp. routine always loaded the table in the same direction, but when you profile mill, you don't have that luxury. OP should put an indicator on your base and chk. runout in the cross slide as you try to manually shove it around on the gibs from each end. If that is your problem you likely see lots of indicator movement instead of just a few thousands. If not, you've narrowed it down. Easy chk.