So I've read a couple different methods of testing for backlash and after trying a couple, I've come up with some numbers. I'd like some feedback on whether my test method and results appear accurate.

I set up a digital dial indicator for both the X and Y axis. For each axis I moved into (negative X) the indicator about .4in, zero'd the indicator and then used mach3 to move it .25" (positive X) and recorded the indicator reading. I repeated this five times.

So, for the X axis i have .0182, .0170, .0175, .0178, and .0170. I added those and then divided by 5 for an amount of .0175. Did the same thing for Y and got .0105.

Should these number reflect the amount of backlash on my machine or did I not perform the tests correctly?

Thanks!  

I'm confused on your readings. If you had Mach move 0.25" and your readings are only (roughly) 0.02 then you have more than a backlash problem. Or are you meaning that the error was roughly 0.02"
If so those, are pretty high backlash figures and I'd be looking for problems. It's unlikely you will get satisfactory operation with that much backlash even with the compensation. A few thou would be a more reasonable correctable figure (IMHO anyway)
 Basically you have the process correct but you could / should probably measure the backlash in both directions.
i.e.

Move in one direction.
Zero the indicator
Move in the other direction a known commanded amount and measure the actual movement error.
Zero the indicator
Move back in the other direction a known commanded amount and measure the actual movement error.
Enter the average error into the backlash settings.
Repeat the test to see if it got any better and adjust the settings to eliminate all error.

Sage

Sorry that was not clear. Those numbers are the errors. So I had Mach3 move the machine .25, but the dial indicated that it only moved .02318, a difference of .0182 in that case. I added up all the errors then divided by five for the average.

My indicator has a range of only .5". Is there a minimum distance I should move to accurately test this?

>>>had Mach3 move the machine .25, but the dial indicated that it only moved .02318

I think you still have a decimal point problem in your response. Do you mean you issued a move for .25 and it actually only moved .2318 (an error of 18 thou)?

Regardless, there is not much point in moving any more than a you are. Once the backlash is taken up, the rest of the move is assumed to be accurate. The backlash is only the error present when changing directions i.e steps that are issued but do nothing to actually advance the axis.


Oh, BTW be sure the dial indicator is EXACTLY parallel to the direction of movement. If it isn't then it will not indicate the correct measurement.


Sage

That is a huge amount of backlash (if your measuring things right), I suspect you have other problems though as backlash is repeatable. If you are getting readings all over the place then you have problems with something flexing, lots of slop, etc. Backlash comp won't do a thing to help a loosey goosy machine. I'm not trying to sound negative, or like an ass, but I hate seeing people get off on the wrong track. Take care of the mechanical things first, make things consistent and then worry about backlash.

Oh, BTW be sure the dial indicator is EXACTLY parallel to the direction of movement. If it isn't then it will not indicate the correct measurement.

Good  advise, Sage. That's one thing that is commonly overlooked. I've seen so called "professionals" make that rookie mistake. There's a video on Fine Woodworking about indicating a table saw blade parallel to the rip fence, and the instructors indicator is probably off 20-30 degrees from perpendicular. All I could do was sigh with disappointment.

dphens: I agree with the others. You are facing other issues that should take precedence over compensation. Getting a tighter machine movement will yield a better overall result than comp could ever achieve by 10 fold.

Sorry everyone, was tired last night (long day) and the old noggin was not at 100%. 

You're correct Sage - the reading after moving .25 was .2318.

I performed the test a little differently and had similar results. This time I moved into the indicator until it registered movement, zero'd it out and then kept moving in that direction (X-) at .001 increments seven times. Mach3 showed .007 in the DRO while the indicator only showed .0066. Then I moved in the X+ direction at .001 increments until the indicator registered movement. I reached .012 in the mach DRO before the indicator changed. Ran this twice with results of .018 and .019, so something is really off.

I think I've done as much as I can to mechanically eliminate any slop, but I'll go over it again. There was some minor damage from shipping when I received the machine, so it's possible there's something wrong that I have not found.

I've contacted the vendor too for his input and he agreed that these numbers are way out of range. I'll see what he recommends.

Thanks for all the help! I'll post back with an update when I have one.

Dave

Dave, there is an easy way to measure backlash directly that you might like to try. Using the same setup that you described, move into the indicator 0.25 in or so, (approx. mid-stroke). Now set BOTH the DRO and the indicator to ZERO. Continue in the SAME direction for another 0.1 or 0.15 in, (just a bit more than the expected backlash). Now reverse direction and step back to the DRO zero.  Now read the backlash directly from the indicator. As a check continue movement for a little more than the backlash and then step back to the DRO zero. The indicator will (should) read zero. This should repeat precisely all day long, (for that location of the lead screw). If not then as others have suggested, there is definitely some other problem in addition to backlash.
Note that the backlash probably will not be consistent throughout the length of the lead screw.  It is usually greater in the center (due to the most wear) than at the ends. You will want to take that into account when you do set your backlash comp.

Another thing that is troublesome is that you commanded a .007 move but only got .0066 movement. That is unrelated to backlash since there was no reverse movement involved. It would imply that it is a calib (steps/in) error, but that is too short an interval to draw that conclusion. But the bottom line is that the DRO and indicator should agree. Possibly another clue to a mechanical problem? Have you checked your calib over a longer distance?

Al

DAVE,
What was the minor damage?
As others have replied, you just move in one direction, set the DRO's to zero, then move in the opposite direction and see the difference between commanded move to measured move. It's assumed you have the steps per unit correct. There are a few things to be aware of:
1.Make sure that the screw is parallel to the axis, gibs are properly adjusted,the nut and it's associated assembly have no movement, and that the screw bearings or collars are preloaded if that is possible.
2.There is no such thing as zero backlash but can be minimized to a small amount. ie; bearing preload......the bearings or collars can affect backlash
   and depending on how preloaded can account for differences in backlash.
3.Don't know what kind of screw you are using, so the backlash can be different depending on where you are measuring along the screw length.
   The backlash can end up being different from one direction to another, depends on the screw, nut, and collar / bearing preload.

So you check and adjust the above as best you can for what you have.

The indicator should be perpendicular to the surface and suggest using a tip which has a flat face on the end. A fine point can easily land on a surface mark and give you a false reading. Don't use the end or beginning of the indicator travel but rather after is has indicated some. Suggest you move
the axis , say 0.1" and not jog / move in 0.001" steps. Then move in .1" steps a few after the backlash was taken out and the movements should be the same.

Need to be meticulous when checking. There is more that affects the backlash , a few settings on how it's implemented etc, but for now this should
get you in the ballpark.

RICH

    

Thanks Al! I'll give that a try.

I am in communication with the vendor right now to verify my settings and that I've got no other issues. Hopefully he'll be able to determine if something is either configured incorrectly or if maybe there was more damage during shipping than first appeared.