Problems with backlash comp

[Green0]

I started up my lathe today and decided to update software to the new release 2.00065 or whatever.

my machine in rapid wants to slow down about an inch into a move and do something wierd. Sometimes it locks up and misses steps, sometimes it just slows down.

I noticed my mill doing the same thing- cutting arcs it would slow down every 90 degrees and deviate from the feed rate.

Why isn't backlash comp instantaneous? I don't understand why it appears to have a lengthy delay when I make a jog movement either. It's like over a second at 100%.

[jstevens]

was is o.k. before you updated the software?

[ScottC]

Hi, I have exactly the same problem using mach3 mill. Everything was running fine until i upgraded to mach 3 2.0 this morning. Now there is a huge dwell when changing direction. When you untick backlash comp it goes away. Nothing has changed in motor tuning and turning constant velocity on or off doesnt make a difference...anyone else have simialr problems?

[Green0]

Yeah. This is actually a huge problem. The mill is slow as dog**** because of this, and the lathe doesn't work with backlash comp on, and it produces garbage parts without it.


I wish mach support had a phone number.

This is messed up.

Anyone who could help would be greatly appreciated. I could really use a solution to this. My lathe has a couple grand of backlash.

[ScottC]

i solved my problem with the backlash compensation today. I read a post of someone with a similar problem and the suggestion was to change the shuttle acceleration found in config: general config to 0.1.

Scott

[Green0]

I tried that. Thanks. It seemed to help.

I was too busy to test it but I'll hopefully try it on Thursday, and maybe it'll be better.

It's deffinitely smoother.

I wonder how that delay effects threading? Or if it's even a delay?

[NC Cams]

Here's the fallacy (as I see it) of backlash comp:

Lets say you have backlash of 0.001" at a full go, stop reverse movement. So you dial in 0.001" comp and off you go.

Now, you have a wierd deal where you're intentionally trying to generate a "wavey" finish - we used to do just that when we diamond turned piston skirts.

I doing so, you'd rapidly go in and out (changer direction) very quickly in order to generate at "wavey" finish that was 0.0005" to 0.0008" deep.

Imagine the nightmare of comp stacking if you dialed in 0.001" comp each time a directio change was called for when all you were doing was shifting back and forth 0.0007 on average - your finish and size would go to heck in a hand basket very, very quickly.

The use of comp at direction change would also cause fits if you were intentionally trying to generate a very subtle a "barrel" or oval shape on a turning or milling - Again, another piston skirt turning scenario.

In my not so humble opinion, backlash comp is a computer science "solution" for a mechanical problem that can and should be eliminated in the first place.

When you get rid of the backlash (use high preload, high contact angle, A/C ball screw support bearings, slop free ways, etc) you'd be amazed at the gyrations you DON'T have to go thru with fit, finish or finnicky control codes.

Been there, done that, learned my lesson with "comp" and won't do it again.

[Green0]

My machine has .004 of backlash.

I agree- on my mill I'm going to take the X and Y values out (which are .001 and .002) and simply use the Z since it is .010"

There's nothing I can do with my lathe, however, short of buying a new machine.

Without the comp, the tool positioning is accurate to about .003 or so if everything is ideal and the movement to get to position is more than .015 or so. with comp it's within .001 predictably wherever I am.

The only issue I see is once in a while I lose a step or two in a rapid move- I think it's just when I hold shift and tap an arrow key twice while jogging.

For some reason it seems that's where I lose steps. If I let off shift to retap the arrow key, the machine seems to keep it's zero point more reliably.My lathe runs real smooth without comp, but occaisionally a move is .010 or .015 out of place because .004 of backlash comp is actually comping for .008 total, and the remaining error I don't know where it comes from.

>.010 or .015 on the diameter will ruin a part... More than a couple thousandths in Z and I end up having to hand fit semi-production parts.

[NC Cams]

Lost motion due to step loss is NOT the same and should not be confused with BACKLASH COMP.

Different root causes and, therefore, different fixes.

You might want to decide which problem you're having as the solutions and not necessarily interchangeable.

[Green0]

That's great- the difference is I don't lose zero without backlash comp. My mahine even sounds different at rest with backlash comp on. It's quiet without, hums with it.

It's a pro-con game. I can only produce quality parts with it, but occaisionally the zero goes and I need to re-set it.

So your damned either way, but less damned with backlash comp on.

[wjfiles]

I also have a strange result when I activated backlash comp.
I am using PhotoVcarve and to date have not previously used backlash comp.
I measured and set the X and Y axes and they work fine.

Assuming that the weight of the router and Z axis parts that it would not be a problem . I am only carving with the point on a 60 degree cutter so I left the Z setting at 0.

Now when I attempt to carve the Z will increment away from the work piece with every change of direction.
Starting with the cutter touching the material, after about 20 passes the cutter will be 2 mm above the work piece.
Turn off the BL Comp and no problem apart from a minor effect of backlash that I can live with in Photo's but would be a problem making parts.

[ger21]

Originally Posted by wjfiles Now when I attempt to carve the Z will increment away from the work piece with every change of direction. Starting with the cutter touching the material, after about 20 passes the cutter will be 2 mm above the work piece. Turn off the BL Comp and no problem apart from a minor effect of backlash that I can live

You're losing steps when the correction is being applied.

It's not documented (except for a few posts on the Yahoo group), but the Shuttle Accel setting controls the backlash accel (I think). A recent poster had good results setting it to .01 while cutting at 30ipm. I think the default is 1. You might try setting it to .1 and see if it helps. A bit of trial and error are probably needed to see what it does exactly.