One side washboard?

[StillLearning]

OK, so I finally have a little time to dial in a NM-135 and I am noticing a difference in the cutting finish between travel direction in the Y axis. As the bed moves away from the column, the finish is obviously a washboard. When the bed moves toward the column, the finish is smoother. This is side milling in Delrin so it should be leaving a pretty nice finish.

I have recently adjusted the gibs (they were waaay loose) and found that the mounting screws for the Y axis ball screw nut needed to be tightened (fixed those).

Any ideas? I appreciate any and all help. I am, after all, StillLearning.

[JoeBean]

You say you're taking a side cut - are cuttingthe same way in both directions (ie. climb cuts both ways or conventional both ways)? Climb and conventional milling will leave different side finishes.

How are you holding the delrin? Can you post pics of your setup & finish?

If you haven't already you may also want to read this.

[StillLearning]

JoeBean,

Thanks for responding and that Link - that will take some time to digest. Both of the cuts are conventional milling and the feed rate is the same. The cutter is a Cleveland center cutting 4 flute single end mill 1/2" (12.7mm) diameter.

I am holding the delrin in a vise supported on parallels (gripping only 0.18" or 4.5mm) and protruding about 5/8" (15.875 mm) on both sides of the vise. Delrin is 3/4" (19.05 mm) thick.

I am making a profile around this part, but both ends (those protruding outside of the vise) are straight travels and should be leaving a nice finish as observed on the sides. I am stepping down the profile at about 0.078" (2mm) per travel around and removing around 0.05" (1.27mm) on the periphery.

This does not seem to be chatter (usually one can hear that). The thing is, the washboard seems to be present on both sides of the piece, but only visible (not able to be felt) on one side, on the other it is significant enough to be felt. I am a little suspicious that the frequency of the washboard almost matches the ball screw pitch.

I don't have access to a camera at the moment (next week...) but will post when I do.

[cteufert]

When conventional milling there is a tendency that the material being milled is pushed away (and the end mill too) at the start of the cut. This is due to only a tiny sliver being all that is cut. The tool rides up on the material as if it were a ramp. As it continues up the ramp the deflection force increases. When this force increases beyond a certain point the cutter will suddenly cut in to the material, scooping out a chunk. The ramp / scooping will continue and create the surface finish you are seeing.

Climb milling does not have this effect, since the cutter is entering the material being cut at a steeper angle, taking more material so a ramp never develops unless high speed / low feed is used.

As for why it is more pronounced on one edge than the other could be caused buy a bit of backlash or looseness in the ways. Traveling past where the side cut is to begin then coming back the other way may work around this as the ball screw / ball screw nut will have all the slack taken up.

Charlie.

[JoeBean]

So if I'm understanding correctly the object is rectangular in shape, correct?

Without seeing pictures it's hard to say the cause for sure (even then it's not necessarily obvious). But the way you're holding it seems like it might be part of the problem. Delrin likes to vibrate, and you won't need to be able to hear it to have finish problems. Since you're holding it in a vise what I'd suggest doing first, assuming you have some material to play with, is to rotate the vise 90 degrees and mount it that way, then redo the cuts. If you get the marks on the other 2 sides then the problem is likely related to the mill - maybe the ballscrew as you suggested - but if you still get the marks on the ends it's the likely the way it's held.

There's a good thread over here on PM that gives some of the best info I've found on machining Delrin. That might help you out a bit.

As well, what are your speeds & feeds? I haven't done a huge amount of Delrin but when I do I use aluminum EMs, usually 2 flute, with high helix angle (as recommend on the PM thread) and run it fairly fast. I use coolant as mentioned on PM but only because of the off-gassing issue (which I've never experienced, but I'm very sensitive to chemicals so I don't take the chance if possible). The high feedrate & EM choice helped a lot in finish compared to my first few jobs with carbide 4 flute EMs.

Also, have you verified that both ways are getting sufficient lubricant? I've got an NM-200, so the setup's a bit different, but I had a number of problems with the lubrication system that I had to fix, and other have as well. In some cases that can mean that you won't get any lubricant to your ways. If you move the table back and lift any covers you should be able to see if they're dry or lubricated. One (or both) ways being dry can cause chatter.

Finally, one way to check play in an axis is to mount an accurate dial indicator or a DTI to a fixed point off the table (eg. the spindle) and set it up to have it measure off something fixed on the table (eg. the vise). Then run the axis in one direction for a few inches, stop, set the indicator on the object & zero it, zero the axis in the DRO in mach3, then slowly (ie. by steps) move the axis in the opposite direction until the indicator moves. Subtract the amount shown on the indicator (compensating for cosine error if you use a DTI and can't get it in range of the correct angle) from the amount shown in the DRO (which should be a fairly accurate calculated value of what travel should have been). That's fairly accurately what your slop in that axis is.

Hopefully that'll be of some help to you. Take care!

[StillLearning]

Jobean and Charlie, Thanks for the responses.

The attached images should help clarify the issue.

Image 001s is the final surface when the bed is moving away from the column to the right side of the vise (in image 003s).

Image 002s is the final surface when the bed is moving toward the column while cutting to the left side of the vise (in image 003s).

The finish in 002s is acceptable, but I would like to improve the finish on 001s.

This was cut using a 4 fl. 1/2" Cleveland Twist single end mill, and then followed using a 1/8" ball end mill, but not all the way down to the ledge (the light step is visible in 001s, about half way down the machined surface). The CAM was generated using BobCAD assuming feed rates similar to aluminum. What stumps me is that it is OK on one side and goofy on the other.



Thanks for any and all help.

[cteufert]

Just to eliminate some crazy possibilities:

Could the CAM program be creating G-Code that actually creates this? Is the G-Code anything other than a simple straight line milling operation?

Is backlash compensation turned on in MACH3?

Is the outside of the part milled before or after the inside pocket? Could the remaining material thickness be less on one side allowing more deflection?

If you can place a scrap piece of material in the vise try milling just the end of it and see what happens. Try it again reversing the direction to use climb milling.

I think the solution may be switching to climb milling to get a better finish.

Charlie.

[StillLearning]

Thanks for the response Charlie,

So in order:

I checked the G-code and it is just straight line milling - I can post it if you think it necessary.

Backlash compensation is NOT turned on for this system. I wanted to do the adjustments before I resorted to the software compensation.

The outside of the part is milled prior to the inside. There was a little flexibility in the inner dimensions, so it makes sense to machine the outside first.

I placed some stock in the machine and ran some simple code so that I could do a conventional and a climb cut on either side of the vise. The finish was washboard in both types of milling on the right side of the vise. The finish for the left side was much reduced washboard for conventional milling and was ideal for the climb milling. So the climb was actually an improvement - I will try to use that in the future.

As I have been working on this, I am getting the impression that the gib adjustment may still be an issue. My rationale is that 1) the most significant washboard occurs as the x-axis drive motor is positioned farther away from the saddle 2) that it seems to be movement in the x-axis itself that is making the washboard either in response to the pressure of the cutter touching the stock or in response to the y-axis travel and 3) that it seems like the tension on the gib screws are different when checked as the bed is at either extreme of travel. Does this make sense?

What is the correct way to adjust the gib on a NM-135 CNC? I really don't want to lock up the bed and munch something up...

Thanks for the help!

[JoeBean]

Hi,
I think he asked about b/l comp because turning it on will often cause finish problems.

I think problems with the gib adjustment would be reflected on both sides. But you might have wear in the ballscrew at one point. Another possibility is uneven pressure from the vise between the 2 sides. I had an old vise once that had that problem. Squaring the jaws wouldn't fix it either as it pulled more on one side, I'm guessing due to a fault in the casting.

I still suggest the simplest way to narrow down the issue is by experimental process of elimination. Otherwise you're just guessing and introducing more variables. Why not either:
1. Move the vise a several inches over to the side so that the problem side is near where the good side currently is. Problem fixed: It's likely the ballscrew, or maybe the gib. If the problem's still there the issue is likely due to workholding.
Or,
2. Rotate the vise 90 degrees, with one long side in roughly the same place as the problem is now. Long side where the problem was cuts good and/or the problem side is still a problem: The problem was workholding. If it's still there it's something with the machine, like the ballscrew.

You pretty much know now that the problem has to be either in the X axis or workholding, as you had finish problems in conventional and climbing on the one side but not the other. I keep harping on workholding because it's far and away the #1 cause of finish problems. But in any case, a systematic approach I think would be best to find the culprit.

As for gib adjustment google bridgeport gib adjustment, it's similar (at least for the NM-200).

There's no published specs for torque but generally I use the following procedure:
Loosen the lock nuts significantly,loosen gib bolts, loosen all gib screws a couple turns (completely removing & oiling any screws that are sticky), move the opposite axis by hand or stepper a tiny amount so as to ensure the tension is fully off the gib, hand tighten (this was first trial and error, then you get a feel for it - it doesn't take much) the screws evenly, then move the axis by hand with the grip on the end of the ballscrew (ensure the stepper is not powered or it won't move) to make sure that it's free. Tighten nuts & bolts. Check for slop.

I found this to work better than the procedure in the manual as I like the screws as loose as possible to reduce friction & wear. But for thoroughness sake, the manual's procedure is:
Make sure the slide is lubricated.
Loosen the gib bolts.
Tighten the set screws all the way so there's no table movement.
Loosen all the setscrews one at a time a quarter of a turn and check the motion of the table.
Adjust setscrews until motion is even.
Tighten hex head bolts (max 100in-lb).

Hope that helps!

[cteufert]

Certainly a strange problem. The gib adjustments are backwards to what you might expect. The Y gib prevents the table from moving left and right by tightening up the clearance along the Y axis. The X gib will tighten up front to back movement. Being that you are seeing left to right problems it is the gib on the Y axis you want to adjust.

More things to consider:
The pressure on the Y axis ways changes from side to side depending on what side the cutting force is on. Perhaps there is a bit of crud along one side of the Y axis way.

Set up to conventional mill an edge of scrap. Then without changing anything or moving any axis other than moving X to do a second pass to remove some additional material (whatever typical amount you would cut) climb mill your way back to the other side. This should take up any backlash in the X axis / Y gibs.

Could the vise be lifting or vibrating or the jaws out of parallel and only clamping on the other side?

I don't have my NM-145 yet so I can't help you with the exact proceedure for adjusting the gibs. I think I saw it in the NM-145 manual though and it should be very similar for the NM-135. I think grub screws press against a moving gib. If it wouldn't cause a problem I would try to remove the moving gib and make sure it is clean and smooth and doesn't have a bit of debris rolling back and forth as the axis moves.

Charlie.

[kvom]

FWIW, G-Wizard suggests a maximum cut width of .09" for a finish pass in order to limit the tool deflection to .0002. (Carbide 4-flute). For HSS, it recommends a .015 width.

[StillLearning]

Thanks everyone for the help.

As I was adjusting the Y axis I noticed a 'click-click-click' when translating quickly in the direction that has been wash-boarding, not in the other direction. Major Bummer I think I know what that means, but has anyone had this happen?

Normally I use the step function and move at a fairly slow feed, so I have a feeling that click has been there all along, because of the slow motion it was not audible.

I have a call into Novakon. We'll see how it goes.