Backlash Compensation - your opinions please!

Hi all,

fairly new to CNC, and so I'm hoping for your help/opinions about using backlash compensation. I'm still using EMC2...

I'm both unsure as to it's full intended use and limitations.

Would it be your opinion that it is better to not use it at all, and simply plan paths (if possible) to negate it...

It seems that where one is interpolating such that an axis needs to reverse... there is no alternative than to use backlash compensation... is this correct?

Finally, are there any "gotcha's" with BC? By this I mean... one must probably zero all axis in a standardized way... but I am unsure how to communicate to the system which direction I used for machine zero. Does EMC2 force me to zero in a specific direction?

Sorry about my obvious confusion... CNC is certainly promising, but to a newbie, it's sure got a learning curve.

And if there is something more to this that I'm simply missing or not anticipating, a heads-up would be a great help.

Thanks ever so much for your kind help.


R

Hi JT,

x = .005
y = .003

I need to hold .001" when cutting a rectangular grid pattern... the operation will free parts from a pc board (a process called singulation).

So you can see that if I pattern it carefully, I can always manage the direction that I enter the cut... avoiding BC, but since I will be having to define several different patterns, and to keep things simple I would rather not lift the cutter out, and it would be better to avoid overshooting in the reverse direction to reset the backlash...

I guess I am saying that it would be *nice* if the BC worked the way I hope it does.

I sure don't want to recode by hand if the backlash changes (or I tighten the machine)... not fun. I figure BC could help me there.

I still wonder about how to zero the machine with BC turned on.

What will end up being less of an issue? BC, or managing my g-code?

Thanks ever so much.

R

I still wonder about how to zero the machine with BC turned on.

What I normally do is make a move away from the direction that I want to touch off to.
For example on the X axis...
G1 X-1(move left)
G1 X0(move right to take up backlash)
Jog further to right then touch off on the target material. Without backing up or moving left.
These moves have to be made slowly so the inertia of the machine does not allow it to coast to the other side of the backlash. This sets you up.
As John said the machine can still float around within it's backlash, but it should move to the correct location.
In theory if the mass of the machine and it's resistance to movement is greater than all the forces including momentum and cutting forces it might be OK.

Rick G

My machine also has considerable backlash (0,16mm) on the X-axis due to the method the manufacturer used to connect the screw nut to the saddle.
I adjusted the gibs on the tight side and use software backlash compensation. This works OK as long as you are not using climb milling.

Rick wrote:

I still wonder about how to zero the machine with BC turned on.

What I normally do is make a move away from the direction that I want to touch off to.
For example on the X axis...
G1 X-1(move left)
G1 X0(move right to take up backlash)
Jog further to right then touch off on the target material. Without backing up or moving left.
These moves have to be made slowly so the inertia of the machine does not allow it to coast to the other side of the backlash. This sets you up.
As John said the machine can still float around within it's backlash, but it should move to the correct location.
In theory if the mass of the machine and it's resistance to movement is greater than all the forces including momentum and cutting forces it might be OK.

Rick G

Thanks, Rick.

What will probably save me is that I am cutting softer material (LCP and a material not unlike teflon) Also, I am using a small tool... .0625 or less. This should prevent too much floating.

It sounds like perhaps you are saying that it does not matter which direction you touch off on... but this still confuses me. If I touch off in the + X direction (meaning that the screw is loaded in one direction, and further moves in the +x direction will not need BC... But how does BC know this? It could just as easily be that I touched off in the -X direction to zero. There seems to me to be a 50/50 chance that my first X-Move will include BC, and not need it because the screw is already loaded.

Does that make sense?

Thanks for the assist!

R

BigJohnT wrote:

Look at it this way even with backlash correction your machine will not be capable of holding 0.001" tolerance as the cutting forces will also move the spindle around. The problem with using backlash correction is your axes can still float the whole range of the backlash as cutting forces change. If you have 0.005" backlash then I suspect your not using ball screws and in that case your acme screw will be out of the 0.001" tolerance by quite a bit.

Have you mapped out the X and Y to see how far out the screws are?

There is screw compensation that can be used to "map" a screw and that might be a better solution to get you closer.

www.linuxcnc.org/docs/html/config/ini_co...d_sub_axis_section_a

Can you fit a spring loaded nut that reduces backlash to at or below your tolerance?

JT

As just mentioned in my reply to RIck... I'm cutting soft materials with a small tool... and a small overall material thickness to boot. (about 1/16")

I have not mapped out the screws... I probably don't have the tools for that. The dimension of the part are under 3" x 3", so hopefully I won't stray too far in that short a distance.... but I had no idea that the system had screw compensation

Amazing!

I have a lot to learn still.

As far as discovering the behavior of BC (in which direction it applies it first)... I guess I can do a few tests... but if there is a standard way to force it... or an expected way to zero, I was hoping to discover the "right way" to go about it... instead of a bit of fumbling in the dark.

Thanks a bunch!

R

It sounds like perhaps you are saying that it does not matter which direction you touch off on...

It does make a difference, the idea is to make sure when you touch off all the slack is away from the touch off.
If you have already made a move in the positive direction the next move if it is also in a positive direction will not have backlash compensation (BLC) added to it and so on. However if you make a move in the opposite direction the BLC will be applied.
In short the BLC is applied when you change directions.
Might be simpler to set the backlash to a high number just to experiment with the results. Once understood set to correct number.

Rick G