How much backlash is too much?

[TOTALLYRC]

I am getting ready to put together a lathe for use with Mach3.
I am wondering what is the maximum amount of backlash that Mach 3 can deal with efectively. The reason I ask is that I have a Tsugami turret lathe that has less .010 backlash but I am not at the machine right now to check. It seems to me that you might be able to get away with more on a lathe than a mill but I could be mistaken.
I realise that 0 backlash is best but if I can get away with .005" or .006" and not mess things up I would give it a try. I can always reduce the backlash after the machine is up and running.

[HuFlungDung]

Backlash on the cross slide is pretty tough to deal with, whereas on the Z axis, its not so bad. This is because diameters are more critical dimensions than lengths in most jobs, plus the fact that the X axis motor cannot effectively back the tool out of the cut (to counteract a slight taper) if there is a .01" gap to be crossed.

One of my retro fit cnc lathes does have .002" backlash in X, but, because it is a servo machine, I used a linear scale for feedback on that axis. You would think this would remove the effects of backlash, and it does to a certain extent. However, tuning that axis motor is almost impossible to get right, because the lost motion makes the PID filter impossible to get right: it has to be left a bit 'spongy' because the physical backlash does not logically fit the filter scheme.

Any controller without linear scale feedback would be even worse to deal with. About the only option in that case would be to preload the axis: for example, put a pulley on the cross slide, and hang a significant weight that would pull the tool against the nut. This would only 'work' (if you can call it that ) if you do only OD turning on one side of the work. For boring, you'd need to change the weight to the other side. This is because you always want to force the cross slide screw to bear against the nut, and not against the weight.

[chester88]

With careful programming you can get away with backlash in most situations.
just like a manual lathe U move the x axis to remove backlash before taking a finish cut. The trick is to always remove backlash in the same direction. Then Diameter sizes can be counted on. Of course if you are trying to turn something that has to switch directions while cutting...your out of luck sort of - you can try to cut from each side (I'm think of a ball shape ) Eg start at the middle of the largest OD -machine to the left then go back to the middle of the largest OD and machine to the right. I ran a large Okuma cnc lathe with .007 backlash_It was possible but a pain. backlash sucks.

Not sure about MACH but In EMC2 one guy set up a mill with a linear encoder and a motor encoder using two PIDs. The first PID with the linear encoder- only used P. the second PID for the motor encoder -only used I and D. then added them together before sending to the amps_ It was successful . That was with closed loop servos of course. Does MACH do closed loop servos?

[TOTALLYRC]

I went outside today and checked the backlash by hand as that was all the time I had. the Z which is the long axis has about .015", while the X has almost none. I will put an indicator on it and get some real numbers.
If I can get away with what I have I will give it a try. I will be using the dspmc/ip like I am using on my mill with closed loop servos.

It looks like the servo motor installation won't be that tough to do. Converting the tool turret will take a little more time but I will do the tool turret after the rest of the lathe is up and running.

I will post the measured backlash later when I get some real readings.

Thanks, Mike

[TOTALLYRC]

I went outside today and measured the backlash on both axis.
.005 on the X and .011 on the Z.

I found the paper work and will see if it can be reduced even further through adjustments.

Anyone see major problems with these numbers.

Mike