Backlash compensation
 

This may be of interest to others.

My machine uses EMC and 20t pinions, and fuller steppers with 7.5:1 gearboxes.

I have had a chronic problem with inaccurate small parts. For instance, I've been making boat models from 3/16" ply. The bulkheads slot together, egg-crate style.

The major dimensions of the panels and bulkheads come out quite close to as-designed, but fine details do not. The worst manifestation of this problem is the slots (mortises?). I designed the slots .188" wide, but the parts would not fit together. I then redid them at .200" wide. Still no fit... again @ .210" and they fit *barely*. I measure the actual slots at .190.

After some circle testing, I verified that backlash was a significant problem. A 1.500" circle routed in a piece of ply (hard to measure accurately because of grain orientation) measured 1.450" (y axis) x 1.510" (x axis). The diagonal measurements were similarly distorted.

Through trial and error, I've settled on .015" backlash compensation x and y. On EMC you input this directly into your machine's .ini file with a text editor.

I have some more testing and fine tuning to do, but accuracy appears to be much improved. The diagonal measurement of a circle is now within .010" of the perpendicular measurements. One of the things I need to do is to redo this test using a piece of something measurable - like plastic.

I suspect that the x axis needs less compensation than I've input and the y needs more, but I'm a lot closer than I was.

Compensation has no obvious effect on the performance of the steppers.

Jeff that is good information. Mach3, which I think a lot of Mechmate owners use, also has backlash compensation available.

But backlash compensation, while it can help, is no substitute for reducing mechanical backlash to appropriate tolerance levels.

15 thou is an awful lot of backlash for any CNC machine, even a router like the Mechmate.

I would be looking very carefully in two areas.

First the backlash in the reduction gearboxes.

Second the motor mount pivot point.

It will be well worth taking the effort to eliminate as much mechanical backlash possible.

Gerald and some experienced Mechmate users would need to confirm this but looking at the machine
design I think it would be quite reasonable to expect less than 5 thou.

Greg

On the X-axis I think the gearboxes are the main contributor, with my tiny pinion gear also contributing.

Looking at my measurements, I think I have something approaching .020 backlash on Y and .010 (or maybe a bit less) on X. I think you're right about suggesting the router mount to account for the Y.

Because of the nature of the work I'm doing, I can live with it the way it is. I'm not going to do anything with the gearboxes in the forseeable future, but I do intend to fix the other causes (pinions and router mount).

Jeff,

I would like to see some pictures of your models.

Jeff,
you are wrong about the pinion gears, if you use a bigger pinion your backlash will also get bigger, you will want to run the smallest dia. pinions as possible, what happens is the bigger the pinion the larger the movement per step, which will magnify you back lash, smaller pinion gives better resolution than a larger one..
//chopper

Thanks Chopper,
It seems to me that a bigger pinion radius would yield improved contact between the pinion and the rack, but probably not enough to mitigate the aggravated gearbox backlash which you point out.

A quarter of a degree of gearbox backlash produces 50% more gantry movement on a 3" pinion than a 2".

Thanks for the correction.

Nils,
Hopefully, I'll have something for you tomorrow. :)

Jeff,
all I Am doing is sharing what I have learned if I can, I went through the same stuff you did and I actually bought bigger pinions with the same intention as you, I thought that the 30 tooth pinions would smooth out my machine since it would create a larger meshing surface for the gear and rack,
if you have a set of pinions try them and see what happens there is nothing like learning it first hand, and that is how I came to realize my backlash problem.
//chopper

Maybe this is why I am happy with the performance of old BBB running with Chinese steppers without reduction. When I started I read somewhere that 20 tooth pinions is needed....so I bought 20 tooth pinions. When I tried to fit them onto the stepper I found that there is hardly any meat left for the grubscrews. I then had some sleeves turned to beef up the pinion. Later I found out that South African and American 20 tooth pinions are not the same and that I should have gone for a 24 - 28 tooth pinion.

Well my cutting quality is OK and the sleeves on the THIN WALLED pinion is still holding after more than a year in service.

After some mechanical debugging (among other things, one of the pinion grub screws was hitting the y-car) I've eliminated my apparent x-backlash, :D and gotten the y-backlash down to about .008.

Procedure: measuring 1" squares routed into a piece of uhmw.

There is still some detectable slack in both x and y, but apparently because of the relatively forgiving nature of what I cut, it's not apparent in the finished parts.

I'm sure that when I start routing aluminum for a new dustfoot, I'll find the limitations.

Bigger pinions will magnify the apparent backlash of the drive system behind the pinion. But, smaller pinions do wear out quicker than big pinions, and then they develop their own backlash where they mate with the rack. . . this is probably the origin of the story that smaller pinions have more backlash than bigger pinions.

Also, if one goes for less than 20 teeth, there comes a point where the pinions are no longer suitable for spring loading because their teeth become "too sharp" and ride in the valley of the rack. . . . . similar to worn pinions with some rattle.