Need Help!- Pocket and Island Perimeter Finishing

[SC250]

Hello,

I am a mechanical engineering student, somewhat new to the world of CNC. I've been learning how to use MasterCam X5 over the past year or two and feel pretty comfortable with basic operations, but have a problem that I can't figure out. A great teacher at my school and I are working on figuring this out, but I’m hoping that you all might be able to help sort this out.

I'm cutting a pocket with letter-shaped islands out of 6061 aluminum using a 1/16" carbide endmill. The pocket is 0.020" deep; I use a 0.017" roughing pass and then finish the floor, island walls (contour) and boundary with 0.003" finishing cuts on each. The machine is not top of the line, but has a positioning accuracy of better than 0.0004" and is quite rigid.

The problem is that I always end up with a "shadow" around the perimeter of my pocket; it seems to cut slightly too deep when it finishes the island and boundary walls. I have no idea why this is happening or what I can do to fix it. I don't think that it has to do with the accuracy of the machine because the parts are coming out consistently; this is NOT a random phenomenon.

Here are the basics of the program that I'm using:
- zig-zag roughing pattern with helix entry, 75% stepover.

- 0.017" DOC for roughing. 4,500 RPM and 3 IPM feed (kept it slow to minimize tool marks and avoid breaking the endmill).

- Climb milling for pocket

- 1 finish pass on the floor of pocket, 0.003" DOC

- 2 finish passes on perimeter of pocket + island contours. First at 0.017" DOC and then one at 0.020" DOC. Each is removing 0.003” from the wall of the part.

- I am currently not using my flood setup; I figure that chips are not the problem here.

Do you have any ideas as to what might be going on? What should I do to fix this?

Attached is an image of the part I've described after wire-brushing it. I don't think that the small step around the letters is normal. Granted, it is not large at all -- probably on the order of 0.0005"

Any help would be very much appreciated!

[cadcam]

This is the finish pass coming back to finish the profile if you measure this do you find a diffrence if you do it is tenths I bet as the tool comes back for the profile the tool is not repeating axactly. you can rough the pocket and come back with a contour to finish the letters and define the depth 2 tenths shy.
I alwas suggest you run coolant in the Alum by the way this can help with finish along as tool life.

[SC250]

Thanks for the advice, Cadcam.

I agree that the depths are not equal, I'm just wondering how they got that way. I've made many of these parts and they all come out exactly the same -- why would the machine error in repositioning the z-axis for the finish cuts always be in the same direction (i.e. too deep by ~0.0002")? The repeatability of this error makes me think that there must be a parameter that I'm missing or mistake that I'm making in the setup.... or is this a common problem?

In the meantime I'll try the part again with coolant and a fresh endmill to see if that helps.

[TXFred]

What type of CNC machine are you using? If it's a benchtop, then it's very possible that the column is pushing away 0.0002" during your bottom finishing pass, or the head is lifting by the same amount. Then when you run your perimeter finish pass, the column/head springs back due to the cut being lighter.

On your next part, if you have the same problem, rerun the finishing pass along the bottom of the pocket and see if it eliminates the step.

Without knowing more about your machine, that's my best guess.

[SC250]

Thanks for the response. The machine is a Syil X7 that I'm controlling with Mach 3. It is a big machine -- not at all a benchtop -- and the quality of cut is very good with the exception of this weird problem. I figure that for a machine of its size a 1/16" endmill job should be a cake walk. Also, since the finishing pass on the floor is 0.003" I don't see how there would be much deflection due to that cut.

The bit appears sharp to me, but could it be that the bottom has dulled a bit and is responsible for this? One other possibility is that the connection between the driving computer and the machine is making for a lag or step-skipping, but that seems unlikely to me since everything else works perfectly.

I'm really stumped!

[TXFred]

I consider the Syil a benchtop. Bear in mind that I own a Tormach 1100, and I also consider that a benchtop. My definition of benchtop is "a machine that requires a table, workbench, or stand to place it at a usable height." That definition also defines the overall size of the machine.

You're right that something the size of the Syil should be able to do the job without flexing. But it's possible that the head is lifting slightly against the ballscrew backlash. The effect would be the same as flex. With the Syil's linear rails, there is no friction at the ways to keep this from happening.

I'm pretty convinced that this is a machine issue.

Frederic

[SC250]

Hi Frederic, thanks for the prompt follow up. I understand and agree that my machine is not great compared to most serious CNC setups. That said, my machine is not a benchtop, even by your definition. It weighs twice as much as the Tormach 1100 and has an integrated base casting so it stands on the floor, not a benchtop. The machine is very rigid and I have not noticed slop, blacklash, or excessive chatter of any kind since I've owned it.

That said, it still is a definite possiblilty that machine deflection and accuracy are the issue here. I just don't see it, though. The finish cut on the floor of the pocket is just 0.003." Then I did a finish on the island and boundary sides with a spacing of 0.003," which was broken into two operations: one at 0.017" DOC and the final one at 0.020" DOC. How much could the tool possibly deflect (or relax) during cuts of that size? Do you see this happen on your machine?

Could this be a chip build-up issue? I was not running flood....

[TXFred]

Another possibility, since you're not running flood, could be heat expansion. Could the end mill be getting hot enough to grow slightly during the finishing passes? If so, then adding coolant should fix it.

I've only had my Tormach for a month, and have not gotten much use out of it yet, so I have no comparison there.

But every machine has some flex, no matter how big. For example, I can stick a test indicator in the spindle of the Haas VF-2 at work, touch it against a T-slot's vertical wall, zero it, and then lean on the spindle and have it indicate 0.0001" of horizontal movement.

If I can get 0.0001" out of a Haas, then 0.0005" from a Syil is not outside the realm of possibility. Syil's website says that your machine has a repeatability of 0.0004", so that may be the issue.

You can try to compensate with a programming change. I assume that you rough pocket, retract, rapid move, finish the pocket floor, retract, rapid move, and finish the pocket walls. There is an option in the Pocketing configuration called "Keep Tool Down." If you select this, then the machine won't retract between passes, and so shouldn't introduce any Z error.

Anyway, those are all the guesses that I have. I hope you get it solved. When you do, please post the solution so others, including me, can learn from it!

Frederic

[SC250]

Thanks, Frederic. I just took a class on precision machine design and know that you're 100% right about all machines flexing and about tool expansion. Small stuff like that, that one never thinks of as problems, can have a way of adding up. In my case I don't think that the tool expansion is the culprit because the step in height is very specific to the finshing pass along the perimeter of the part -- it doesn't happen gradually.

In any case, I'll give all of these suggestions a try this weekend and post the results afterward, whether they are good or bad!

If you or any others have other ideas in the meantime, please post them too. Have a great week.

[cnc-king]

what you are seeing there is just toolpath blends from the circular to the straights cuts and as cadcam pointed out you probably will not be able to measure the difference in depth. also the other cause of this is the spindle is not perpendicular to the table. some one probably whacked the head at some point and time.

[nda22]

Looks like it could be tool flex too me. Try the floor pass with XY stock to leave set to .001

[SC250]

Hi Everyone,

Just wanted to update you all re. this pocketing finish issue. Over the weekend I was able to try multiple solutions and evaluate how much they helped the situation. I've come to the conclusion that the issue is linked mainly to tool flex and to toolpath selection. Here's why:

- Turning flood coolant on did not help out much

- Keeping the tool down for finish passes also did not eliminate the problem, which means that z-axis error is not likely to be the cause of this. I'm thankful for that!

- Reducing the floor finish passes to smaller and smaller values (finally to 0.001" DOC), and creating more and more passes DID help.

I think two things are happening:

1) Tool deflection, both during the floor finish passes and during the perimeter finish passes. Frederic, you're right that tiny deflections (or thermal expansions) stemming from multiple sources could easily cause small marks like this. Lateral deflection during the perimeter finishing pass would tend to push the tool away from the wall and push one corner of the tool into the floor of the part, which would explain the shadow that we see in the original picture.

2) Toolpath blends. Cadcam & CNC-King.... I think you may have been right on. Bravo! Going from a zig-zag rough to a arc-based perimeter finishing pass tends to leave marks unless you take things ultra-slowly. Changing the roughing toolpath to a morph spiral helped to reduce the appearance of the shadow, most likely due to the fact the roughing and finish toolpaths blended together better.

In any case, the issue is still open for debate and I'd certainly still welcome any new theories or ideas for improvement. Thank you to all who helped me solve this -- I very much appreciate your time and hope to keep in touch.

Best,
Evan