Serious Accuracy problem Attention HiveMind!

So I have really been working out with the 4th axis making some statues. I’ve found the best results balancing time and quality by using 3 finish toolpaths with various bit using the touch plate for dead nuts accuracy and completely trusting the avid script to do a “Z axis touch”. I noticed that the 2nd finish tool path was not cutting, so back to the touchplate and as a test I switched to machine coords to test repeatability. Well I got a different number each time. (See pic) I did this 5 times and the number was getting larger each time for a difference of .066. Anyone else encountered this. Can someone else see if they get the same results.

Hmmm, did you remove the touch off plate from the 4th axis mount and try to touch off again somewhere else to see if the numbers keep changing?

Eliminate the mount from the picture.

Re-home the Z between each and make sure it’s not losing steps too.

Eliminate the easy things first.

1 Like

good advice, Yeah as soon as I discovered this I stopped and made the post. It is aggravating, I think maybe I’ll get a reference height off a spot on the chuck or better yet the gearbox.

Darn the tool path window locked up again. I am running ver 4768 of mach and I have already knocked Walter White’s nose off his face, so I think I may jump to ver 5000. You have any advice to the contrary?

You might also want to check the Z access coupler. Make sure it’s still attached and not loose.

There is a new version of Mach4, I have not tested it yet. I can’t be introducing more variables into my testing regime but I also doubt it would be fixing anything if note.

when using the touch plate after it makes contact it makes a loud KLUNK and then raises up, it has always done this when using the 4th axis script. The results are pretty good running 4 tool paths 1 rough and 3 finish paths. I would definitely see .06 error, heck I can tell if it is .02 off. And look I fixed Walters nose.

1 Like

Awesome!! Well done! :wink:

1 Like

Apart from the initial setup, I’ve avoided the 4th axis scripts. I find them to be less accurate than cutting and measuring a test piece to find the offsets, and I’ve done projects that needed the accuracy. When I do a Z rotary touch-off, I do an extra step where I manually move over an empty space and command downward by my own measured offset (via MDI command), then zero the Z-axis.

My touch plate rocks a bit in the mount on the rotary, so I push down toward the non-corner-hook side before I use the touch plate. I can take a video if this isn’t clear.

I’ve noticed recently that V-bits haven’t performed as well for me with the touch plate as straight bits. Touching off with a straight bit, I can incrementally step to zero to check the offset and find that it’s dead on. Doing the same with a V-bit, I sometimes get a gap between the surface and the bit, and I switch to using the paper method to zero Z. I wonder if ball-nose bits are similar.

When I get some time I’m gonna look into it I installed mach4 ver 5036 and so far so good. this is how my touch plat attaches to the 4th axis, is yours the same? The ears on the aluminum block fit in the plastic mount.


I quit using that touch-off plate thingy a long, long time ago for just what I see in those pictures of yours…

Looking at the plastic mount, all that dust and swarf in there? That would be producing a whole lot of scrap if you were using a ATC.

Are you sure your error isn’t coming from all that swarf lodged in that mount? If I remember correctly that mount was very close tolerance. Even minor swarf would be a huge variance.

The fact you touch off on it and it changes, and growing larger would make sense if the swarf was the problem. The touch-off plate isn’t fully seated.

I would be blowing it out, digging out ever speck of dust and dirt like some kind of OCD freak… well, because I am :stuck_out_tongue:

John, I would ask you to zoom in on that pic, I know you work in a “hospital clean” enviro in your shop, but in the real world there is not even .002 of dust in that bracket.

Well only if the swarf were spontaneously multiplying in the 5 minutes it took me to take those 5 Z axis touch offs.

Like I said, OCD… :grin:

The tighter tolerances and the slippery nature of the Delrin and HDPE makes for a strange beast.

You give it an opportunity to slip and slide it wil take you up on the offer.

1 Like

my older non Avid CNC also does this in Mach 4. I have never been able to discover why, but repeated Z touchoffs under otherwise identical circumstances are inaccurate for me too. I’m completely guessing here but is it possibly related to the “look ahead” parameters?

Here’s what I mean by that:

You can see from measurements why this happens. The pocket for the hook is deeper than the hook. There’s no support on the hook side.

Before I use the touch plate on the rotary mount, I press down on the supported half and make sure I touch off on that half.

I just took the same measurements on my machine and I am within .001 of your readings and no rocking of the plate. The aluminum housing is very subject to get dinged up so you may have to stone the surface. Hopefully you can find out the problem.

I don’t know too much about that parameter, I’ll dig into it.

1 Like

Ive worked with UHMW and man stainless steel will wear out before that stuff does. Conveyor belt slides in a chicken plant.