[IMT]

Turbo,
So the step increment was used in the video was 0.0001in instead of 0.0025mm?

[philbur]

That's why I changed to inches for the video
Step

That's strange, if the 40 steps shown were microsteps with a 5mm pitch screw then it would be 100 microns, which is exactly what the video shows. Wheres 40 microsteps with a 5 tpi screw gives 101.6 microns. Or have I misunderstood what the test shows.

Phil

[TurboStep]

@IMT: correct.
@Phil: Fist you say that I shouldn't expect accurate control of the table at micron level but you then expect the DTI to be that accurate! Actually I wouldn't be surprised if were just an imperial DTI with a metric dial. On the other hand, measuring 4 thou to an accuracy of 1.6% isn't that bad Which ever way you want to look at it the video wasn't intended to demonstrate the accuracy of my DTI <-- notice the smilie
Step

[adamvs]

The ball screws are imperial: .2" pitch ( even tho all the hardware, bolts, screws and so forth are metric ). So doing very fine stuff in metric will show some aliasing.

[mountaindew]

Well that was a wash and a waste of time. I got the ball screw bearings adjusted per Tormach's manual. Managed to take a woping .0003 backlash out of the system Guess Ill have to live with .003 on all axis besides the z until I can replace the ball screws. what a pain in my a$$. Should have bought a real VMC, honestly need better tolerance than what Tormach says it can do from the factory anyhow so time to save pennies.

Wow!
While I didn't get much out of adjusting bearing preload on both x,y axis I did as I mentioned above worked out a far more consistent way to measure what I was doing.
As I Noted above I used 0.0005 for step and moved the target axis 0.002 or 4 steps in both directions evaluating the difference. I found a 0.0012 in movement of dti hand and thus ending up with about 0.0008 of backlash or "less then 0.001" at any rate on both x,y, and z with a little more 0.0009 to 0.001. The numbers your talking about are 3x this and out of spec according to my manual and inspection sheet.
Wonder what the deal is?
md


Would like to add I performed the measurements in a 3x3 grid of locations on the table to confirm nearly the full range of motion of both x,y, axis was returning the same consistent movements.

[jake hoback]

Wow!
While I didn't get much out of adjusting bearing preload on both x,y axis I did as I mentioned above worked out a far more consistent way to measure what I was doing.
As I Noted above I used 0.0005 for step and moved the target axis 0.002 or 4 steps in both directions evaluating the difference. I found a 0.0012 in movement of dti hand and thus ending up with about 0.0008 of backlash or "less then 0.001" at any rate on both x,y, and z with a little more 0.0009 to 0.001. The numbers your talking about are 3x this and out of spec according to my manual and inspection sheet.
Wonder what the deal is?
md


Would like to add I performed the measurements in a 3x3 grid of locations on the table to confirm nearly the full range of motion of both x,y, axis was returning the same consistent movements.

I was using a .0001 step, Ill try the .0005 and see if it changes much. Sometimes hard to check when the step doesnt move the full .0001 until three clicks later on the jog shuttle lol. Or no movement at all since the jog shuttle doesn't do anything for a half revolution anyhow sometimes.

I have a new DTI coming in tomorrow to double check my numbers with. I used both a DTI and DI both gave the same readings though.

[jake hoback]

after all the adjustments, I can still pull the table .005 by hand (maybe 10-15lbs of pressure) got it down from .007 at least! lol.

[mountaindew]

after all the adjustments, I can still pull the table .005 by hand (maybe 10-15lbs of pressure) got it down from .007 at least! lol.

As noted above I found the 0.0001 steps to be nothing but a total waste of time! I could move as much as 0.003 or 30 steps with no registered movement, Causing confusion and errors in my adjustments. Changed to 0.0005 and could measure movement on second step indicating the backlash being between 0.0005 and 0.001 on all 3 axis.


And Something is way wrong here!
There is no way to move any axis on my machine 0.005 in any direction on any axis by hand! Maybe 0.0005 with a 2x4 or pry bar still this is 10x less movement.


md

[Mr. G]

Does the Saddle move 0.005" in the Y direction? Not the table, the piece below it.

[philbur]

@Phil: Fist you say that I shouldn't expect accurate control of the table at micron level but you then expect the DTI to be that accurate! Actually I wouldn't be surprised if were just an imperial DTI with a metric dial. On the other hand, measuring 4 thou to an accuracy of 1.6% isn't that bad Which ever way you want to look at it the video wasn't intended to demonstrate the accuracy of my DTI <-- notice the smilie
Step

I'm not sure where I have said I expect the DTI to be accurate at the micron level. I was comparing 40 steps with the 100 micron reading on the DTI. Your DTI like your table may have significant error at the micron level but like your table it will not be cumulative, so your DTI when measuring 100 micron movement will be much more accurate than trying to measure 1 micron, this is of course if you set it up correctly. With regard to a metric dial on an imperial unit, I can't speak for the quality of the tools you purchase. I do notice that you have everything on the gauge but the makers name.

Phil

[CNC-Dude]

As noted above I found the 0.0001 steps to be nothing but a total waste of time! I could move as much as 0.003 or 30 steps with no registered movement, Causing confusion and errors in my adjustments. Changed to 0.0005 and could measure movement on second step indicating the backlash being between 0.0005 and 0.001 on all 3 axis.

Yeah man, what the heck??? I would like to understand that as well. When I did the 0.0001" the readings didn't make any sense. Then at 0.0005 it becomes much more meaningful. So what's going on here? Is MACH3 and a Series II machine capable of moving 0.0001 steps in the first place? Is moving 0.0005 the same as five 0.0001 microsteps? If so, at what rate? I wonder how they are doing this because at the end MACH3 can't command a motion in distance. All it can do is send STEPs to the stepper driver and specify direction. So in order for this to work, MACH3 and the drivers must be configured to move the smallest it can move, which apparently is 0.0001.

Then it all becomes microstepping. And it is my impression a Series II machine can't even dream about moving 0.0001". Not well, at least.

[jake hoback]

Does the Saddle move 0.005" in the Y direction? Not the table, the piece below it.

The .005 is at the saddle! freaking ridiculous! was .007-.008 before I got everything else adjusted

[mountaindew]

Hi Jake.
What does tormach say?
Can you post some pictures of your setup and how your making measurements?
A cosine error when using a dti in a position other then 0 deg could also impact your measurements.

Here is a quick chart copied from long island indicator.

contact point angle correction factor
10° reading times 0.98
15° reading times 0.97
20° reading times 0.94
30° reading times 0.87
40° reading times 0.77
50° reading times 0.64
60° reading times 0.50
Posted this also because of comments above about some dti accuracy and what effects it.

Don't think this is a factor for you because so many different setups on 2 different machines with 2 different amounts of use on all 3 axis would not provide the error consistency.

Hope you get it figured out! I finished up my yearly tune up on my tormach and it works better then ever! Like wine and dynamite it gets better with age.
Lube system needed the most attention and going over every inch of the machine helped me get to know it better and a good learning experience.
I also logged all the results for future reference.
md
.

[TurboStep]

Posted this also because of comments above about some dti accuracy and what effects it.

@MD: forget the DTI accuracy stuff - that's totally irrelevant.

@Jake: When the saddle moves .005" does the ballscrew also move with it, or does it remain more or less stationary?
If it doesn't move (as much) then the problem is somewhere around the ballnut/nut mounting. If is also moves with the saddle then the problem lies around the angular contact bearings / mounting - in which case you might want to check whether the spacer between the two bearing is missing. This issue was a reported a few years ago but shouldn't be occurring any more, however it's probably worth checking if the ball screw is moving and based on the values you're quoting.
Attachment 274498
Step

[mountaindew]

Well that was a wash and a waste of time. I got the ball screw bearings adjusted per Tormach's manual. Managed to take a woping .0003 backlash out of the system Guess Ill have to live with .003 on all axis besides the z until I can replace the ball screws. what a pain in my a$$. Should have bought a real VMC, honestly need better tolerance than what Tormach says it can do from the factory anyhow so time to save pennies.

@MD: forget the DTI accuracy stuff - that's totally irrelevant.

Step

Cant , wish I could forget a long list of things


Back to the bearing issue you note above ! Reading this thread, jake mentions this is a problem on x,y,z, and from what I can tell on 2 different machines.
Could the spacer be missing on all 3 axis causing backlash of 0.003 or more on all 3 axis?
I just don't understand how you get the numbers he is talking about. They are nearly 5x more then my numbers on all 3 axis! "imho that's huge"

As jake mentions above wore out ball screws or something. And I agree maybe should have purchased a real machine vmc
md

[philbur]

Have you measured the backlash with the table at both extreme ends of the ballscrew as well as in the middle. If all values are the same then it is unlikely to be ballscrew wear.

Phil

[Beezle]

My Y axis has always had backlash of about 0.002 or so, maybe a tiny bit less.

So for grins, I tried pushing on mine and was able to move it about 0.005. I was surprised.

But I had a thought and tried it again with the machine energized. If I really pushed hard, and I'm a large man, I moved it a bit less than the backlash.

So? I wonder why that would be. Maybe one can actually turn the screws a wee bit when it's turned off?

[Zetopan]

Maybe one can actually turn the screws a wee bit when it's turned off?

Of course you can. Ball screws are typically at least 90% efficient and as a result they can be easily back-driven. There are only two things besides the ball screw efficiency that fight against back driving. At low RPMs the residual magnetic field caused by the permanent magnet rotor through the air gaps and the stator exhibits a small detent torque. You can easily feel this by manually rotating an unpowered stepper shaft by hand. At higher rotation rates the back EMF will try to prevent any rotation if the winding is electrically connected to an external circuit (including just shorting the wires together).

[Zetopan]

I failed to mention that the unpowered detent torque ripple that you feel is one full step per detent. Hence, with the power off you can back drive a stepper driven ball screw by as many full steps as you desire.

[adamvs]

I finally threw down and got a new X-axis ball screw. It was a process of elimination, but the fact that I was able to 'push' the table in the X direction about a thou and a half, pointed to the ball screw. All other possibilities had been eliminated: bearing preload, saddle movement.
Results were better than expected:

Old backlash/lost motion: .002
With the new screw in and the six 4mm screws torqued to about 6Nm ( not much ), the gib set loose, backlash was .0001-.0002! Yes that's right, that's tenths! I relaxed the pre-load just a hair, and tightened up the gib; backlash is now about .0003 - .0004. I can live with that!

Adam