New Machine Build- My first CNC Router - need some help.

[DonW]

Hi folks,

Just finished up my first router, which is based on SteveSpo's design.
(Thanks to SteveSpo and Clockwork for their invaluable help!)

It features 8020 aluminum & aluminum plate for the frame members, and the following parts:

Gecko G203v's
Gecko G100
HSCNC Rex ballcsrews w/ Zero-backlash ballnuts.
HSCNC Rex RS600 motors
THK KR33 for Z-axis

I'm using a PC 890 router along with a small Wolfgang Engineering mini-spindle for inlay work.

Pics to follow...

Anyway,

I just started some test cuts, and am having some accuracy issues.

I cut a simple square and a circle using a 1/4" router bit in my PC 890 router. The grooves measured a perfect 2.500, so the runout is pretty darn good on the new router...I had my doubts, but it tests out fine.
My issue is that the 3" square that I cut measures 2.950" instead of 3".
Looking at the G-code, it seems to be compensating properly for the bit thickness properly, and the dimensions make sense. Still, I'm off by -.05" on the circle size.
On a 1.5" circle, it measures 1.467" - slightly less error, but still pretty bad.

Any thoughts? Could this be ballscrews? Or perhaps I need to tweak the formula I created for velocity/steps etc? Am I missing some sort of setting in Mach?

My velocity settings in Mach 3 are as follows:

X - 10,000 steps, 100.02 ipm, acc = 50, step & dir both show 0.
Y - same as X
Z - 5,080 steps, 100.02 ipm, acc = 40, step & dir both show 0.

Any thoughts why I'm off by such a large amount? This is a very unacceptable error for my purposes.

Thanks for any help you kind folks can offer.

Best,

Don Williams

D.E.Williams Guitars
http://www.dewguitars.com

[ger21]

Try changing your X and Y to 10225 steps/unit.

[ger21]

Oops, have it backwards. Maybe 9775 stpers/inch.

[DonW]

Hey Gerry,

I currently have it set to 10223 on both axes, and it's closer, but weird things are happening...

The 3" circle measures 3.006" in the X direction, but 2.994" in the Y direction.
The 1.5" square measures about 1.498 either direction.

You sure I should try it at 9775????

Thanks,

Don

[DonW]

Here's a couple pics of the cnc...

[ger21]

Make sure you close and restart, as well as retune the motors after changing the steps per inch. Also, make sure you take very light cuts, to try to rule out deflection. Also, cut some parts clockwise, as well as CCW and check if there's a size differrence. With your Z hanging down that far, you could be seeing some deflection.

As for the steps/ unit, I guess I had it right the first time. I was thinking John on the Yahoo group was right when he said 9800.

[DonW]

Thanks Gerry.

Well, I discovered pretty quickly here that my gantry is out of square with the table, i.e. the right side is a little bit forward of the left. I guess that's a leftover of racking the thing a half dozen times from overloading the G203v's with voltage. Had to rewire in series. Long story.

So looks like I need to do a couple things here to adjust the gantry, then move a limit switch on that side (technically, the A axis, slaved to the Y.)

It's out over 1/4", so once I get that fixed I'll report back with new findings.

[DonW]

As you know, I discovered that the gantry was woefully out of
perpendicular in relationship to the X axis. So I routed holes at
0-0, 0-24, 24-24, and 24-0, and discovered just how out of square it
all was. Better than 1/4" off. After adjusting this, I then started
tackling the number of steps to try to get a 24" spacing to actually
measure 24". I finally settled on 10160 steps, and ended up with a
24" grid that measured 24" (using machinists ruler, visual
measurement). The diagonals also were identical, or as near so as I
could tell visually with the 36" ruler. So at least it's ballpark.
Even a few more steps made it too big, and a few less too small. I
figured if I can nail the correct number of steps for a large
dimension, it should increase the accuracy of smaller entities.

Wrong.

I then ran my 1.5" square and 3" circle program again.

The circle was circular, but it only measured 2.988", in all
directions. I was happy with how circular it was, but being .012" off
is not going to be acceptable for what I'm doing. It's possible that I
ran this too fast at 50 ipm, but that seems rather slow to me.
:-)

The square was another story. The measurement along the X-axis was
1.487" (off .013"), and the measurement along the Y-axis was 1.491",
off .009", certainly the better of the three errors.

Both diagonals measured 2.101", so at least it would seem that the
corners are square, which is better than it was before.

Looks like I have some more work to do, as I am starting to assume
that I have some step loss errors somewhere in the system.

Again, I would assume there might be some runout in the router
spindle. This could possibly account for some of the error. I used a
brand new solid carbide Whiteside 1/4" spiral upcut bit for the test.
I may try a smaller bit, or an end mill to see how that performs.
Also I may replace the chuck with a precision made one to see how that
refines things.

I need to cut for accuracy as I'll be doing a lot of small intricate pocketing and profiling for inlay work. I really need to get the tolerances as
tight as possible. Being off .012" is not a good thing for this.
I'll need to cut an inlay, and drop it into a pocket just a hair
oversized so that it's almost a perfect fit. There's no room for much
error there, so I'm going to have to keep tweaking things until I can
get things as good as the machine can possibly be, and hope it's good
enough.

I'll also take a look at how I might possibly lower the gantry a bit. The issue I have is wanting to get an acoustic guitar body under there, but it may not be a reality. I knew I was pushing the limits of the design with the height, but was hoping that any deflection that might happen would be minimal.

[mlammert]

Don,

From your pictures it looks like you are running and Apple laptop...

Is this correct???

If so, are you using that for your CAM and controller software???

And, again if so, what software products are you using...

I am a Apple person primarily but have had to dive into the PC world due to lack of CAM software for the Mac...

Good luck with your issues!!!

Mark

[ger21]

What you could be seeing is screw error. The important thing you need to look for when doing your inlay work, is repeatability. If the machine is repeatable, then you cut your inlay in the same place on the machine that you cut the pocket. Even if the machine's off a little, the inlay and pocket will be off by the same amount and should still fit together fine.

What you might want to do is cut your 3in parts in different areas of the machine, and see if they are all the same size, or closer. If your screw error is consistent, then small parts will be the same anywhere on the machine, and only large parts would show the larger error. So you may want to set it up for smaller parts, which may be more critical for you

[DonW]

Originally Posted by mlammert Don, From your pictures it looks like you are running and Apple laptop... Is this correct??? Mark

Hi Mark,

Yes, it's a Macbook Pro. It is however running XP Pro via BootCamp.
I'm using Mach 3 (V. 2.63) and BobCAD for my cad/cam software. (no bashing, please!)



Gerry,

Yes, this could be screw error, it could also be that I'm pushing it a little fast. Most of what I will be doing will be in the 6 ipm range, and that could possibly tighten things up a bit too. I really don't have a need for speed with this thing, as most of my work will be of the nature where if it takes a little more time and gets cleaner, more precise cuts, it's all good.

I'll try to find some old pics to load from when I was constructing this thing.

[mlammert]

Thanks for the info, Don...

Yeah, I am using a white MacBook and XP on Parallels... Just bought the computer with 1Gb of RAM... I think as soon as I max out the RAM at 4Gb my CAD/CAM software will be much happier...

Again, good luck with your stuff,
Mark