[cnczoner]

Hi all,

This was a finishing pass taking 0.010" off the perimeter of this part with a 0.5" dia M42 cobalt endmill at 1500 RPM and 5 IPM. What's very odd is that the straight sides have a very nice finish, but around the rounded sections it looks like scallops or scales and also looks like the curve was interpolated with a low number of straight edges. You can ignore the "dent" at the cardinal point as I need to fix my backlash compensation settings.

Thoughts?

BTW, I'm using the G02 circle code in Mach 3 and wondering now if there is some setting the makes it interpolate a circle with a certain number of divisions. I'll go investigate that in the meanwhile.

Cheers,
-Neil.

[cnczoner]

Surely someone here knows...

[Ken_Shea]

Neil,
Well I guess "knowing" is a relative term, however, if that were on my machine that little divot in the center would be very disturbing, perhaps excessive backlash in the ball screws, something loose ???, I believe Mach III has software compensation for backlash. The little cutting segments leading up to that could be from too low a tolerance setting.

Ken

[HuFlungDung]

It does look like an interpolated arc, but if your gcode used a G02 then it cannot be the fault of the code.

Is this a stepper driven system? How many steps/rev are you using? A coarse setting there will likely show up as a facetted circle.

What quality is your axis drive system? Is it a ballscrew?

[Mitsui Seiki]

Post a copy of the program.

[cnczoner]

Ken, I started tweaking the machine and seem to have reduced the backlash, but had not decreased the backlash compensation in Mach 3 yet, so I'm not worried about the divot because I know what it is.

Hu, it's an acme screw, and it has a backlash adjustment screw, but I haven't properly eliminated the backlash yet (until I make an extension for the allen key so I can get in there). See this photo... http://narwani.net/neil/mech/X-Axis_Screw_Nut.jpg

Stepper driven, with many thousands of steps per inch, because the belt drive has a 1:10 reduction IIRC (so 10 stepper steps makes one acme-screw step). I need to check on that machine tomorrow for the actual numbers. Yes, G02 and G03, but perhaps some setting I have in Mach 3

Mitsui, it's hand-written g-code. Here's the finishing pass snippet...

(----- Profile outside -- 0.1" steps -----)
G01 Z2 ( Clear over top )
G01 X0 Y-2 F30 ( Clear of material )
M6 ( Load 0.5-inch finishing endmill )
T03 ( For CNC simulator )

G01 Z0.1 F30 ( Over top of material )
G01 X7.25 Y0.625 F30 ( Start point on R.H.S. )

G01 Z-0.1 F1 ( Plunge 0.1 inch )
G02 X6.375 Y-0.25 R0.875 F5
G01 X0.625 F5
G02 X0.625 Y1.5 R0.875 F5
G01 X6.375 F5
G02 X7.25 Y0.625 R0.875 F5 ( End on center of R.H.S. )

G01 Z-0.2 F1 ( Plunge 0.2 inch )
G02 X6.375 Y-0.25 R0.875 F5
G01 X0.625 F5
G02 X0.625 Y1.5 R0.875 F5
G01 X6.375 F5
G02 X7.25 Y0.625 R0.875 F5 ( End on center of R.H.S. )

G01 Z-0.3 F1 ( Plunge 0.3 inch )
G02 X6.375 Y-0.25 R0.875 F5
G01 X0.625 F5
G02 X0.625 Y1.5 R0.875 F5
G01 X6.375 F5
G02 X7.25 Y0.625 R0.875 F5 ( End on center of R.H.S. )

G01 Z-0.4 F1 ( Plunge 0.4 inch )
G02 X6.375 Y-0.25 R0.875 F5
G01 X0.625 F5
G02 X0.625 Y1.5 R0.875 F5
G01 X6.375 F5
G02 X7.25 Y0.625 R0.875 F5 ( End on center of R.H.S. )

G01 Z-0.5 F1 ( Plunge 0.5 inch )
G02 X6.375 Y-0.25 R0.875 F5
G01 X0.625 F5
G02 X0.625 Y1.5 R0.875 F5
G01 X6.375 F5
G02 X7.25 Y0.625 R0.875 F5 ( End on center of R.H.S. )

G01 Z-0.6 F1 ( Plunge 0.6 inch )
G02 X6.375 Y-0.25 R0.875 F5
G01 X0.625 F5
G02 X0.625 Y1.5 R0.875 F5
G01 X6.375 F5
G02 X7.25 Y0.625 R0.875 F5 ( End on center of R.H.S. )


(----- Full depth at once finishing pass, 0.010 inch into side -----)
G01 Z-0.65 F1 ( Plunge 0.65 inch, through complete depth )
G01 X7.24 Y0.625 F5 ( Cut in a bit )
G02 X6.375 Y-0.24 R0.865 F5
G01 X0.625 F5
G02 X0.625 Y1.49 R0.865 F5
G01 X6.375 F5
G02 X7.24 Y0.625 R0.865 F5 ( End on center of R.H.S. )

(Return to home)
G01 X7.25 F10 ( Clear away )
G01 Z2 F30
G01 X0 Y2 F30

[Mitsui Seiki]

I suggest you change your start point to the lower straight part of the piece instead of in the middle of the curve.

[cnczoner]

Stefan, I did that so that it would break off closest to where it was clamped. I'm guessing you're recommending that for the divot though, right? Even if I did a complete circle, I get that segmented/interpolated effect.

[HuFlungDung]

We don't know what type or quality of machine this is. It is possible for drive screw inaccuracies and/or drive screw bearing eccentricity (in the bearings) or intermittent stick/slip of the screw or slides to combine with excessive backlash and produce a wavy surface. The prime indicator of this would be waviness that is some multiple of the lead screw pitch.

[Mitsui Seiki]

I'm guessing you're recommending that for the divot though, right? Even if I did a complete circle, I get that segmented/interpolated effect.

Yes,that's the reason. And you have tried to do a complete circle and the same result?
Have you tried different feeds and speeds?

[cnczoner]

We don't know what type or quality of machine this is. It is possible for drive screw inaccuracies and/or drive screw bearing eccentricity (in the bearings) or intermittent stick/slip of the screw or slides to combine with excessive backlash and produce a wavy surface. The prime indicator of this would be waviness that is some multiple of the lead screw pitch.

Okay, I can see that. But I'm thinking about a straight diagonal line. I should see this effect then too, unless both axes have the same problem and in sync with each other. I can run a test of that easily tomorrow. But with the backlash reduced/eliminated, you're suggesting I should not see this effect, right?

The mill is a converted (by someone else) ... HF geared-head mill. It's a bit larger than an X3. Conversion really being only the addition of stepper mounting plates and drive pullies/belts. I was planning to do a ballscrew conversion with some Roton screws, but with all my other electronic and automotive projects, I decided to just adjust this, reduce/eliminate backlash and run as is for now. Meanwhile, I keep my eyes peeled on ebay for an appropriate ballscrew. But I've barely begun to tweak it.

[dapoling]

I have seen this before and it was bad ball screws.
As to the program using a G2 as others have said is not the problem but I would change your tool path to enter and exit part before feeding in Z.
Your feeds and speeds are fine, depending on the machine can be increased.

good luck

[cnczoner]

Yes,that's the reason. And you have tried to do a complete circle and the same result?
Have you tried different feeds and speeds?

In different applications, yes, and I always see some effect like this. I haven't really run an apples-to-apples comparison with the metal, same endmill, same radius, same doc, but with different feeds, speeds, etc. I can try that easily though.

Thanks,
-Neil.

[Mitsui Seiki]

I guess it's the machine then.

[cnczoner]

I have seen this before and it was bad ball screws.

Hmmm... that somehow is good to know. Yes, seriously. Because that gives me an answer of how it can be solved (I have acme screws now). And because I've been contemplating that ballscrew conversion anyway. Perhaps it's time to do this now. I'll still adjusting out the backlash first, so I can see that effect.

As to the program using a G2 as others have said is not the problem but I would change your tool path to enter and exit part before feeding in Z.
Your feeds and speeds are fine, depending on the machine can be increased.

good luck

Agreed. Actually on the second unit of this same part I made, after the overall cuts, I finish-passed the long straight sides, then moved the clamps to the middle, and finish-passed the arcs in one pass each side. But the divots are still there due to the incorrect backlash compensation.

I'll do some tweaks/tests tomorrow and post my results here.

Cheers,
-Neil.

[neilw20]

You will always get this effect with backlash compensation turned on.

Using metric here cos I live in a place they tried to ban inch rulers: (flame2)
With backlash X and Y compensation set to 0.1mm when X changes directions the screw stops, reverses an amount giving 0.1mm.
During the reverseing action the axis is NOT MOVING.
For linear interpolation to work correctly the tool moves along the tool path at the programmed feed rate and each axis moves at a rate to maintain the required feed rate. For the machine to maintain the correct path all of the axis must move in the correct proportion to each other. I one axis stops then ALL axis must stop. While stopped the cutter rubs and cuts a little bit making the marks you are observing.

1). Turn OFF backlash compensation.
2). Fix mechanical backlash. You can't fudge it with the electronics, unless you just want to drill holes in the correct places.
3). On metal, climb mill. On plastic don't climb mill.
4). Plenty of compound or air so that the cutter can't eat the swarf.
5). When machining an outside radius the feed rate must increase form the straight.
6). When machining an inside radius the feed rate must decrease from the straight.
7). Beware that the cutter load can bend the cutter away from the job, even on light cuts, and this can give non vertical sides.

Been there. Done that. :cheers:

[cnczoner]

You will always get this effect with backlash compensation turned on.

Using metric here cos I live in a place they tried to ban inch rulers: (flame2)
With backlash X and Y compensation set to 0.1mm when X changes directions the screw stops, reverses an amount giving 0.1mm.
During the reverseing action the axis is NOT MOVING.
For linear interpolation to work correctly the tool moves along the tool path at the programmed feed rate and each axis moves at a rate to maintain the required feed rate. For the machine to maintain the correct path all of the axis must move in the correct proportion to each other. I one axis stops then ALL axis must stop. While stopped the cutter rubs and cuts a little bit making the marks you are observing.

Correct. Though I've seen some nicely machined examples with backlash compensation, though I believe those were ball screws.

1). Turn OFF backlash compensation.

Not an option yet, until I adjust the backlash out of the machine.

2). Fix mechanical backlash. You can't fudge it with the electronics, unless you just want to drill holes in the correct places.

Gotcha.

3). On metal, climb mill. On plastic don't climb mill.

Doing that currently.

4). Plenty of compound or air so that the cutter can't eat the swarf.

What is compound? I have been using WD-40 very generously and had originally been blowing chips away with an air compressor, but lately I think holding the shop-vac near the endmill works great to keep the chips away and saves me a bunch of cleaning afterwards.

5). When machining an outside radius the feed rate must increase form the straight.
6). When machining an inside radius the feed rate must decrease from the straight.

Hmmm... see my next post after this.

7). Beware that the cutter load can bend the cutter away from the job, even on light cuts, and this can give non vertical sides.

I've read about that and think it's responsible for my other finish issues... see next post after this.

Been there. Done that. :cheers:

Thanks,
-Neil.

[cnczoner]

So I ran a few tests today. I first ran a straight 45-deg (not sure of the metric for 45-deg ) diagonal line (as a 0.010" finish pass at 5 IPM) and got the scallops again -- a regular pattern of lines (similar to the polygon-like problem on the circles in my original post, but in a straight line now). This photo doesn't really show the problem as much as it is, but it's definitely there, and I can feel it easily with a finger. The spacing between each of this ridges is consistent. So the problem is confirmed not due to circle interpolation.




Next I worked on correcting the divot at the cardinal point due to the backlash compensation. I slowed down the backlash compensation speed from 50% of max feed to 10% and that worked beautifully. No divot visible now.

With this, I also ran a number of arc tests, finish pass of 0.010", climb milling at 1500 RPM, but with feeds of 30ipm, 20ipm, 10ipm, 5ipm, 3ipm, 1ipm, and 0.5ipm. The scallops are still there, and the spacing between the scallops doesn't really change with feed changes, but they are less pronounced at slower feeds.

I am also seeing a different finish problem, which is some vertical streaks, very closely spaced together. There is a direct relationship here that the slower the feed, the better the finish. This photo is a 0.010" finish pass at a fast 20ipm, purposely so, so that the streaks are more pronounced for the photo. The polygon-like scallops are still there, but it's hard to tell with the streaks over-powering them.




This is the best finish I could achieve, which is a finish pass at 0.5ipm, but this time I used ATF as the lubricant rather than WD-40. The scallops are still present, though less pronouced. Somehow I can't get a good photo of it.




I need to run more comparisons, but I'm much more pleased with the finish after all these tweaks. I still have a way to go though. Next steps will be adjusting out the backlash on the leadscrews, and adjusting gibs etc.

Cheers,
-Neil.

[cnczoner]

BTW, just for reference, in those photos above, the material is 6061 T6, about 2" wide x 0.375" thick.

[dapoling]

cnczoner I do not see any programming trouble or with your feeds and speeds, I still feel it is a machine problem.
You can keep chasing feeds and speeds but I think it will just be frustrating to you until the real problem is found.