[nova24]

I just finished my first 3-axis CNC about a week ago and I've noticed that all of my edges are bumpy. I've been cutting 3/4" plywood and MDF and even when cutting the depth in multiple passes the edges are still bumpy. The interesting thing is the bumps all line up in each step creating a vertical line the entire thickness of the part.

My X and Y-axis are both driven by NEMA 34 motors a with a 4:1 timing belt reduction. The timing belts are attached to shafts which drive 20 tooth #25 roller chain sprockets. I purposely used 20 tooth sprockets instead of the 9 tooth I typically see used to reduce the chordal action or deviation to less the 1%. I was hopping by doing this I wouldn’t have to worry about speed variations that, I would think, would cause exactly this issue.

I’ve changed the micro stepping from ½ to 1/8 with no change in finish. I've tried cutting from 35 IPM at .2 deep to 50 IPM at .4 deep, still same result

I’m currently using a 2 flute ¼” end mill running at 15,000 RPM.

The axes both move free and smooth.

Any help is greatly appreciated.

Thanks!

[Phife]

without knowing the details of your machine, i would have to say that the bumps you see are caused by the machine flexing during cutting. Or some sort of motion caused by your drive mechanism.

[nova24]

Phife, I just ripped apart the drive and through together a quick test with some left over ACME lead screw its completely smooth running. :banana: I looks as if the chain drive seams to be the issue. I'm going to change the Y-axix to a lead screw and see how it going. If I'm happy with it I'll change the X-axis.

Thank you for your input.

[PaulRowntree]

Nova24, could you please describe these bumps? How big are they, and how far apart? Do they depend on the feed speed?

Cheers!
pr

[nova24]

Paul,

I've attached a picture of what I'm seeing. Feed rate doesn't seem to matter. What did make a difference in what I felt what holding the cutter while running the machine was to replace the roller chain with an ACME lead screw. I had a little chunk left over from the Z-axis that I tried on the Y-axis and it seem to improve the vibration. I ordered some lead screw today to convert the Y-axis to. I'll let you guys know how it goes. If all goes good, I'll be converting the X-axis as soon as I can.

Thanks!

[PaulRowntree]

Congrats on eliminating the problem, but my interest remains because I have chains on X & Y, and want to understand this type of problem. I have some slight ridges when using a large diameter surfacing tool because my Z is slightly inclined wrt the Y axis. So if you don't mind,...

Are/were the structures seen on pure Y moves? pure X ? Coupled X/Y ?
If either of the first two cases is true, then there must be looseness or flex in your system, because the X & Y axes should be rigidly uncoupled. If they are only visible when moving X and Y together, then perhaps one axis is moving slightly irregularly (but still periodically); this should be measurable with an indicator dial and test displacements.
Because your gantry advances ~5" per rotation of drive shaft, you are extremely sensitive to the drive shaft being slightly out-of-round or the sprockets being not perfectly centred on the shaft(s), although the spacing of your ridges (~13/inch) seems too small for this. This is why I stuck to the 9-tooth sprockets.
Cheers!

[nova24]

Paul,

I hope I've eliminated the problem. I'll know this weekend when I put my Y-axis together.

The problem was on both axes, but just slightly more prominent on the Y. Coupled moves were the worst.

I do believe the looseness was a result of the chain bouncing. My chains are very very tight, which did help reduce the issue from some of my first runs when the chains were much looser. But with the long travels on my machine, X=106" and Y=52", I don't think I could pull the chains tight enough to ever eliminate the vibration in the chain.

I was going to change to a the 9-tooth sprocket to see if that made any difference, but my sprockets are too small for my 1/2" drive shaft.

-Dennis

[nova24]

Well, over a year into running this machine and a lot of upgrades later and the machine cuts almost perfectly smooth. It used to take 15 minutes of sanding on a part to remove the ridges around the periphery, now it takes maybe a minute.

The chains do work pretty well for a drive system, but they need to be tight...really, really, really tight. I used #25 chain which have a working load of I believe 180lbs. I was probably pushing that the chains were so tight. I noticed a definite correlation between the chain tightness and the quality of the cut.

I eventually replaced the Y-axis chain drive with ACME threaded rod and an anti-backlash nut. This improved cut quality, but not much more then the chain being extremely tight.

So, tight chains helped, but rigidity of your structure is just as important. I started replacing some of the plywood components with steel and aluminum. The first of which was the Y-axis trolley and the Z-axis since this was the area I measured the most flex. This made a huge difference. So much so, that a Y-axis only cut was almost perfectly smooth. I still got a few bumps from flex elsewhere in the machine, but very minimal.

This last major upgrade that I just completed included adding large steel plates to my X-axis, linear rails on both the X and Y-axes and ball screws driving each side of my X-axis gantry. This seam to take 99% of the rest of the flex out of my machine.

I'll post some pix as soon as I as can. Hopefully this information will help guide someone looking to improve the cut quality of their machine. Other factors beyond the rigidity of your machine and drive components also determine cut quality. But my experience tells me that having a ridged machine to start with is the ONLY way to have a chance of getting a smooth cut.

Regards,

Dennis

[PaulRowntree]

Congratulations Dennis, it sounds like you are getting to enjoy your machine.
I am still running chains, but looking to make the entire machine stiffer. For me, that means a complete rebuild, because my table is really not up to the task, and using steel where possible. Improving my chains (tighter, stiffer mounts) made quite a difference, but it is never clear to me how tight they should be, and how to make sure that each side is exactly the same working length and the same 'phase' as the other. So I am looking to go to acme, and giving enough space for a rotating nut in the future. Everman-style belt drives are another dream.
Cheers!

[nova24]

Paul,

I almost went with belts for my X-axis because of the length, but I was concerned with stretch over time because the belts would have to be so tight. Now I could have used some wide belts, 1" or more that have some crazy strength so the stretching would be a minimum or eliminated all together, but I don;t have that kind of money.

The chain tightness was pretty easy. I just cranked the chains down so when I plucked the chain it wouldn't sit there and vibrate, at least not noticeably. I would then tighten one side or the other to align my gantry to the table. Worked really well.

Right now I'm actually using 3/4" ball screw I purchased from an industrial supply company for $20 for a 12' length, got three of those. I then purchased 24" of 2" diameter Delrin from Tap Plastics for about $40. I cut a foot of the ball screw off one of the pieces and ground flutes in it and tapper one end to make a tap. I then chucked a 6" piece of the Delrin into my lathe and drilled a hole. Put the Delrin in a pipe vice and started the workout tapping it. Turned out great. For now there's no measurable backlash. Eventually I'll slit the Delrin nut so I can keep it tight on the screw.

I came up with a rotating nut design that's cheap and works awesome. See the attached pictures. The bearing is a wheel bearing, their like $8 each and can take loads that I'll never apply. Their not sealed, but I've got them buried I grease so I'm not to worried.

You can see the upgrade to the Y-axis trolly and Z-axis. X and Y axes run on linear bearings, but the Z-axis is still using 1" x 1" x 1/8" steel angle and v-groove bearings (love those things). The angle is bolted approximately every 1.25" to a 1" x 2.5" steel bar. The bar is then bolted to the side of the aluminum box. This is all scrap material I had lying around from other projects.

Anyway, take a look at the pictures and post any questions you may have.

Oh, sorry for the amount of junk in the background. Looks worse in pictures then real life.

-Dennis

[PaulRowntree]

That looks very good. Do you think that thrust bearings on each end would have worked? With that length of a nut, I would expect that the alignment to the shaft would be perfect. Your design is better (and far easier) than what I had in mind. I was going to have a 3:1 or 4:1 reduction on the drive. If I understand the pics, the inner sleeve of the bearings rotates with the nut, and the outside is fixed on the end plates?
What kind of rapids and acceleration do you get with the rotating nut?

[nova24]

Thrust bearings wouldn't work with this design very well, because part of this design requires the bearings to support a side load. The wheel bearings are a tapered roller bearing that have some of the same characteristics of a thrust bearing. But can also take a side load which helps in this design. The way I designed this setup to work is like this.

The ends of the Delrin nut are turned down so they fit snugly inside the inner race of the wheel bearing. (So yes, you are correct about the inner sleeve rotating with the nut) The end of the nut that has the pulley has a longer turned down area to accommodate the pulley. The ID of the pulley is turn so it fits snug on the turned down area of the nut and the pulley is held in place with four drywall screws that are screwed into the Delrin nut.

The steel end that the motor mounts to has a .1" deep pocket cut into it that's only a few thousands of an inch bigger then the O.D. of the wheel bearing. This is so when I tighten the belt, the wheel bearing will rest against the wall of the pocket and not against the screw. (This is where the thrust bearing is going to have an issue)

The other steel end has a similar pocket. The only real difference is the pocket is longer (up and down) to allow that end to "float" to help with alignment. Once everything is tightened down it really can't "float" so I don't think that features all that critical.

The two long bolts you see going from one end to the other are to pinch the nut assembly between the steel plates. I put these at the top instead of the bottom to help keep the end plates perpendicular to the mounting plate above while tightening the long bolts. My procedure to tighten the long bolts is to tighten the bolts that hold the end plates to the mounting plate just so you feel resistance as you tighten the long bolts. Once you get the long bolts tighten to where the nut spins with slight resistance, finish tightening the end plate mounting bolts.

My screw has roughly a .2 pitch, so with the motors I'm using my rapids are 100 in/min. Which I'm pretty happy with. If your motor had the torque to go faster, you could. The rapids with the rotating nut design are really only limited to what the material of the nut can handle. Stepper motors usually give up before the nut.

Sorry for the long winded explanation.

-Dennis