View Full Version : Some help with measuring backlash needed

02-26-2009, 06:08 AM

I now have adjusted the headstok of my taig and when I sweep the table the needle just bounces a bit over a 0.01 mm range (0.0004 inch) so that's pretty straight.

When I measure the distance traveled on the x-axis I get the following results:

I move the x axis in 10 steps of 1 mm each and on the clock it measures just that so it ends on 10 mm.

Now I reverse in 1mm steps, and with the first step it ends up on 9.03 mm, so it's 0.03 mm short, all the next steps are 1mm again.

Am I right that on the x axis my backlash is 0.03 mm (0.001 inch)?

02-26-2009, 06:15 AM
Yes! :D

02-26-2009, 07:31 AM

I have now measured the backlash on my x-axis on 0.03mm (0.001 inch) and on my y-axis it's 0.06 mm (0.002 inch) pretty standard I believe on a taig?

Does setting backlash compensation in Mach cancels the backlash effectively?

If so, what should I enter for amounts? Motortuning is setup in inches (40.000 steps on the x and 40.100 on the y).

cheers, Daan

02-26-2009, 08:55 AM
Changing your steps on the Y won't fix backlash. You do that if your screw isn't exactly 20 TPI. Say your screw was in reality 19.5 TPI (over the length of the screw!!!) then you would change the steps per inch to compensate. If that were the case you would have an increasing error over your movement, the further you go, the more error you get.

While the amount of backlash can vary slightly over the length of the screw due to factors like screw wear etc... it is essentially constant and is only a factor on the change of direction.


02-26-2009, 10:10 AM
Thanks Bob.

The reason I mention the number of steps in motortuning in Mach is that I suspect I need these to somehow calculate the number I have to fill in for backlash compensation in Mach.

Mach asks for a number 'in units' and motor tuning etc. is in inches, but the rest of the program (and I) are set to mm.

So now I am wondering if I should enter for backlash compensation on y either A: 0.06 mm - B: 0.002 inch - C: the number of steps for A and B.

02-26-2009, 08:19 PM
I'm a newbie, but I just set my backlash compensation earlier today, so I think I've got a grasp of the process...

Mach3 only works in one measurement system at a time (as far as I know), so since you're using a TAIG everything you do in Mach3 will be in inches. That being the case you should put option "B" in for backlash compensation.

02-27-2009, 02:52 AM
Thanks! That makes sense since I set up motortuning etc. in inches.

I do run the rest of the program in mm though, I switch to mm in the settingsscreen and all dro etc. are in mm then.

TOC Admin
02-28-2009, 09:02 AM
Let us know how the Backlash Compensation in Mach worked out for you..

I tried it Awhile back on my SYil X3 and Didn't get very good Results....

Seemed fine Jogging around but while Making a Part it did all kinds of Strange things.......I don't remember the Version of Mach I was using but it was about a yr ago....

03-03-2009, 07:22 PM
Everything I recall reading about Mach3 backlash compensation was that the principle didn't really work reliably, and could make things worse. Not that there's anything wrong with Mach3, the whole idea of software backlash compensation may be dicey.

I suspect what it comes down to is the torsional vibration of the leadscrew as it moves in steps create uncertainty as to which direction the backlash counts off of. +/-0.001" backlash would mean that a commanded position of 5.000 could be anywhere from 4.999 to 5.001. In a perfect, smooth world the backlash would have it lagging behind whichever direction it just moved in. The commanded is to move from 4.9 to 5.000", and you'd expect backlash to leave the table at 4.999" and AFAIK all the software does is modify it to move to 5.001. But if the actual position is subject to leadscrew vibration, it might bounce forward and end up at 5.001 from a commanded move to 5.000 and the +0.001 added in software would only amplify the problem, not reduce it.

If my reasoning is valid, I'm sure it depends on mill type, motor drive configuration, speed, even mass on the table.

Is there some reason you need >0.001 accuracy for sure? This might be more trouble than it's worth.