Backlash config question Mach3 Mill

scdiveteam

scdiveteam

Rick Geschrey
Silver Member
Hi,

Last year I installed new anti backlash screws on my Cue Monster. Before doing so I compensated the backlash error on Mach Mill 3.

When I got the new units installed I put the Mach Mill setting back to zero.

I have noticed that I still have very small amount of error on my X and I have been getting around it by moving into the zero position from one direction every time before rotating my A axis. This has been working very good but I must eyeball the mill end to a line on the cue.

My questions to the experience CNC brain trusts here is can anyone give a detailed step by step explanation of the use of the indicator and the plugging in of the compensation error into Mach Mill. I did it before and understand the where's and how's to do the deed but I can't remember if I should plug in the backlash error number I see after 3 movements or do I use 1/2 half that number as an average for calibration. Also what to do with the Speed % of Max number?

The last time I did it my error was way off and now it is pretty close. I went To a Mach Mill forum last time but I thought this might be a very good topic to kick around here On AZ and I am sure most of the other CM with intermediate knowledge in this area like me would love to listen to any posts on this detail with all ears on.

Rick
 
Not an expert but this is how I set mine in my old taperer.
Went to Y.100 then to Y-.100 back to Y.100.
Measured the backlash . Router had 1/4 drill blank as "cutter".
In mill mode , I didn't cut it to half.
In tapering , I divided all diam. in half.
If you inlay, you have to stay 1 to 1 .

Mach3 does have lathe mode settings too.
I never used it.
It was easier for me to write XY for tapering and using safe Z.
 
Rick,

I hate to keep going back to this, but it sounds like you have a screw loose.
How much of a measurable error are you seeing? The reason i ask is because you would likely never notice a couple of thou if your cutting with any relief tolerances for fit, and especially if your cutting large parts.

I cant imagine that new screw having that much play already. The other major red flag is that there are more than a handful of people who have the same machine and are apparently not experiencing the same problem.

I keep trying to find a picture online of an old cuemonster and its end supports but cant find any.

I had the same problem, thought i had backlash but the endsupport was not tight within the bearing, allowing the screw to move with the axis until it took up the slop within the bearing support.

hope you figure it out without using compensation.
 
Have you checked your thrust bearings on the threaded rod for proper adjustment or wear?

Dick
 
LGSM3 said:
Rick,

I hate to keep going back to this, but it sounds like you have a screw loose.
How much of a measurable error are you seeing? The reason i ask is because you would likely never notice a couple of thou if your cutting with any relief tolerances for fit, and especially if your cutting large parts.

I cant imagine that new screw having that much play already. The other major red flag is that there are more than a handful of people who have the same machine and are apparently not experiencing the same problem.

I keep trying to find a picture online of an old cuemonster and its end supports but cant find any.

I had the same problem, thought i had backlash but the endsupport was not tight within the bearing, allowing the screw to move with the axis until it took up the slop within the bearing support.

hope you figure it out without using compensation.
Click to expand...

Hi Jake,

I never set backlash adjustment with the new units in the config area of Mach3 Mill and I think this is the problem. Just put them on and set on 0 in config area for backlash. I am now thinking that even new they should have been measured numbers entered into the config area.

My parts and pocket compensation geometry is absolutely perfect and repeatable. Every time even when nesting many parts. So I doubt of my screws are loose, also checked them.

When ever I rotate my A axis it steps off a little when I "go to zero" location after an operation. If i had a loose screw I would think that my parts would not fit the pockets especially on long floating points that are also fitting absolutely perfect.

The problem was real bad before I changed the backlash load units. They were physically bad, no doubt.

If I program a routine where it repeats on a rotary output axis, they step out a little each time but the end result is a summation of the small deviation that has amplified each time the A-moves. You can't notice it very much until it goes around and the 1st and last pocket is next to each other. The only reason I see it is because i make a line in the cue when it is in the lathe with a pencil as the zero daturm for my X-0 start point.

By reloading the backlash in the same direction and adjusting the X-0 to a datum line by eye with te mill and the line I make on the cue, I can make it work just fine but it is a pain as I have to babysit a machine that should work without my intervention.

Frustrated,

Rick
 
yah, i understand your frustrations...sometimes it helps to stop, get real intimate with your machine and polish the hell out of it. Sometimes you run into the obvious and sometime you just get a nice clean machine.

Maybe if your zeroing on your first op and loading up in one direction all is well. Then when you make a hard rapid at the end, its forcing the mechanical discrepancy to show up

Why are u re-positioning with A? Did i read that correct? maybe rough all 4 then clean up all 4?
 
LGSM3 said:
yah, i understand your frustrations...sometimes it helps to stop, get real intimate with your machine and polish the hell out of it. Sometimes you run into the obvious and sometime you just get a nice clean machine.

Maybe if your zeroing on your first op and loading up in one direction all is well. Then when you make a hard rapid at the end, its forcing the mechanical discrepancy to show up

Why are u re-positioning with A? Did i read that correct? maybe rough all 4 then clean up all 4?
Click to expand...

Jake,

I have to reposition a hair because it does not come back dead on the line when I go to zero after the operation. Maybe if I load up from the other side of the last move on the X, it will be ok and keep repeating correctly. It may be the ticket. Since the nesting routines are perfect on the table this maybe a a simple operator error.

I will run a test that way and see what goes on.

Thanks,

Rick
 
i think i've been a little confused to this point. The pencil mark your reffering too....for visualization purposes it could be the same as an a joint ring? your x0 reference?
 
Thats what i thought...if you can see visually a discrepancy that large, im of the opinion that its highly unlikely that its backlash concerning the screw and nut, especially with the kirk style compression nut.

Im betting its in the end supports, the nut mount to carriage, or the nut bolts to nut support. Seeing or feeling a few thou or maybe more can be kind of hard. Hope this makes sense...basically i think your screw is moving with the axis the distance of the amount of play thats somewhere. Of course as usual i could be out in left field too, just trying to offer advice based on my experience.

You can think of these problems as a blessing, the experience in fixing them is well worth it.

FYI, i'm not sure if i even know where backlash comp is in mach....Royce said build it right and you wont have to worry about that.
 
LGSM3 said:
Thats what i thought...if you can see visually a discrepancy that large, im of the opinion that its highly unlikely that its backlash concerning the screw and nut, especially with the kirk style compression nut.

Im betting its in the end supports, the nut mount to carriage, or the nut bolts to nut support. Seeing or feeling a few thou or maybe more can be kind of hard. Hope this makes sense...basically i think your screw is moving with the axis the distance of the amount of play thats somewhere. Of course as usual i could be out in left field too, just trying to offer advice based on my experience.

You can think of these problems as a blessing, the experience in fixing them is well worth it.

FYI, i'm not sure if i even know where backlash comp is in mach....Royce said build it right and you wont have to worry about that.
Click to expand...

Your right Jake anything that is too easy is no fun.
 
I'm a little confused as to what is happening.

At first, I thought you had backlash issues on a leadscrew driven axis.

Then, I got the feeling you are having an A axis, or rotary index, backlash issue.

Now, I seem to be understanding that your stock moves along the X axis when your rotary or A axis rotates through your cuts and then returns to zero. If this is what's happening, you would be able to see it merely by rotating the A axis. Never even moving the X axis. If you rotate the rotary through a couple of rotations and then back to zero and your stock moves along the X axis then you most likely have an alignment issue with your indexer and it's tail stock. How do you hold your stock? In a chuck? A collet? Is the part of your stock that is clamped a true cylinder, or is it tapered? Consider this. If the stock is pulled to one side because the tail stock is off center, as it rotates the stock will "walk" in the jaws of the chuck or collet. Does the stock move towards the indexer, or away from it. I'd bet that it moves away because it's "walking" out of the chuck.

As for backlash compensation, I don't use it. Not that it's a problem as I'm sure it works pretty well, but I just don't have any issues that would require me to mess with it. I use Smoothsteppers on most of my machines and not all levels of the SS plugins work well with backlash comp. I just leave the "Backlash Enabled" check box empty in the config/backlash settings.

Backlash itself can be found in lots of places. Of course, the leadscrew to nut interface, the leadscrew anchor bearings, couplers, even the stepper motors themselves. Depending on the linearity of your motor, the full steps my be the only place you can count on accurate positioning. Since most of us use the microstepping to get resolution, you can get an effective backlash just from the motor not responding well to microstep commands. Usually, if you have all the bolts tight, good bearing mounts designed right, and good anti-backlash leadscrews and nuts you just don't have to worry about anything else. The linearity issue only really comes in to play with steep screws. Something like the 1/2"-10 thread 5 start screws I use on the X of my turning machines. At 1/2" pitch, the full step resolution is pretty large, but X positioning on a turning only machine isn't that critical.

I hope that all makes sense. It seems like I rambled on, but I was trying to cover the thoughts I had. If I'm way off on my assumptions of your issue, let me know.
 
I would have to say that I agree 1000% with Jake on all issues. Rick is X axis your long travel or short across the cue?


Jim
 
Mc2 said:
I would have to say that I agree 1000% with Jake on all issues. Rick is X axis your long travel or short across the cue?


Jim
Click to expand...

Hi Jim,

Long it is a Cue Monster.

What is puzzling to me is all of my parts a pockets are perfect and repeatable and fit with no glue line after I compensate my pockets.

When I do long point geometries or nest many parts on the table everything is very cozy.

When I do a feature and the routine finishes when I hit go to zero it is off visably to a datum line on the cue I penciled in. If I rotate the A and repeat the pocket is off a little on the X. If I keep repeating that process when it comes around the offset of the pockets is way off.

Rick
 
scdiveteam said:
Hi Jim,

Long it is a Cue Monster.

What is puzzling to me is all of my parts a pockets are perfect and repeatable and fit with no glue line after I compensate my pockets.

When I do long point geometries or nest many parts on the table everything is very cozy.

When I do a feature and the routine finishes when I hit go to zero it is off visably to a datum line on the cue I penciled in. If I rotate the A and repeat the pocket is off a little on the X. If I keep repeating that process when it comes around the offset of the pockets is way off.

Rick
Click to expand...

Rick, are you saying that just by rotating the A axis that X is showing off in reference to your line. Does this happen with A axis being the only axis that is moved? Or is it happening after you have finished a full program and X axis is rapid ed back to zero position?

Jim.
 
Mc2 said:
Rick, are you saying that just by rotating the A axis that X is showing off in reference to your line. Does this happen with A axis being the only axis that is moved? Or is it happening after you have finished a full program and X axis is rapid ed back to zero position?

Jim.
Click to expand...

Jim,

It happen when I Go to Zero command before rotating the A. Again all of my geometries for pockets and parts are perfect without any sizing error even when multipul nesting were all the clean up passes are done at the end of the routine. The problem only happens when the job ends and I hit go to zero to get ready to move the A and repeat the pocket.

I am working around this problem by setting the zero in the same direction by eye and it is repeatable around the A. Maybe I need to set the X from the opposite direction from the last X move on the X in the routine to load it any possible backlash that may be present.

Thanks for you involvement. It is appreciated. I will test today.

Rick
 
scdiveteam said:
Jim,

It happen when I Go to Zero command before rotating the A. Again all of my geometries for pockets and parts are perfect without any sizing error even when multipul nesting were all the clean up passes are done at the end of the routine. The problem only happens when the job ends and I hit go to zero to get ready to move the A and repeat the pocket.

I am working around this problem by setting the zero in the same direction by eye and it is repeatable around the A. Maybe I need to set the X from the opposite direction from the last X move on the X in the routine to load it any possible backlash that may be present.

Thanks for you involvement. It is appreciated. I will test today.

Rick
Click to expand...

Rick, put an indicator on one side of your X axis and load it up half way. Then physically push against it in both directions. The indicator will move, however your are looking for the difference after the x Axis springs back. If you have something loose it may show up there. You will want to remove all mechanical play that you can.

Jim.
 
scdiveteam said:
Jim,

It happen when I Go to Zero command before rotating the A. Again all of my geometries for pockets and parts are perfect without any sizing error even when multipul nesting were all the clean up passes are done at the end of the routine. The problem only happens when the job ends and I hit go to zero to get ready to move the A and repeat the pocket.

I am working around this problem by setting the zero in the same direction by eye and it is repeatable around the A. Maybe I need to set the X from the opposite direction from the last X move on the X in the routine to load it any possible backlash that may be present.

Thanks for you involvement. It is appreciated. I will test today.

Rick
Click to expand...

Rick

I think I finally saw something that might indicate what the problem is.

Do you have home switches on this machine, and if so is it referenced to home? You should reference it to home, even if you don't use home switches.

What work offset are you in when you hit the "Go To Zero" button, and did you save that offset?

If your current work offset is not saved, it might not go to the position you think it might should. If my memory is correct, the "Go to Zero" button will go to the work zero position of the current work offset. If that position is not where you think it should be, then it won't go where you think it should.

I know that, on machines I don't use home switches on, the whole work offset versus machine coordinates thing will do some strange stuff. It does what it's supposed to do, but because I'm not paying attention to both the machine zero position and the current work offset position I can get confused.

Royce
 
Last edited:
Just my advice...

Call Unique Products... They will work through the issue you have I PROMISE you :wink:

To add to this for everyone following this post... If you want to check backlash on your cnc machine this is the easiest way to check it. Place the indicator on the right side of the Router mount and bump into the indicator (1 revolution or more) to the right. Adjust the machine to only .001 step and move away from the indicator, which is to the left, until you see the indicator land on a solid hash mark. Once you have this on the hash mark and zero the machine axis. Write down your number so you remember when it returns. Now run the machine away from the indicator, left in this example is away from the indicator. I usually use 1 inch to make sure it is away from the indicator. Now go to zero (home) with the machine. Where the indicator lands is how much lash you have in that axis. If you started on 5 and ended on 6 you have .001 lash in the system.

Hope this helps.
 
Last edited:
also a good idea to have backlash comp turned off when doing this so that you can get a true reading, or else move a very small distance so that it never has a chance to get applied.

I would agree that Unique probably knows the machine better than anyone
 
You must log in or register to reply here.
Back
Top