Diagnosing backlash: ball nut or thrust pack?

[dahui]

I'm having some backlash issues and before forking out 1200 bucks per screw I'm wondering if I can (on my own) diagnose the source of play. To get the numbers below I put a mag base on the table and indicated on-center on the spindle housing. Feed indicator in and zero out. Set axis origin (operator) to zero. Run the table back .5 inches and then back in to operator zero. Observe reading on indicator to determine backlash.

I checked the table play with the same indicator setup, and then pushed on the table and observed the free play (not momentary deflection).

Here is what I know so far.

X axis:
1) Backlash comp set to zero = measured backlash @ .0008
2) Table play = .00035

y axis:
1) Backlash comp set to zero = measured backlash .0019
2) Table play = .0009

I called my HFO and they suggested pulling the way covers back and indicating off the ball nut to determine whether it's the nut or thrust pack. There are two issues. One, I don't get how this will isolate the problem. Two, the way covers, where they attach to the saddle, have been installed with silicone and I don't want to break that seal if I don't have to.

So, I'd like to know if the screws are worn, or just the thrust packs before getting the tech out here. Any advice? Oh, the machine only has 6000 spindle hours and 8000 motion hours. The 2k motion hours were digitizing with a probe, and the 6k spindle hours were mostly light work and a lot of fine milling with a tiny tool and spindle speeder. (I got this machine used) I suppose that might mean there is a lot of wear right in the middle of the screws, now that I think about it. Any advice?

[Machineit]

Well, without taking a cover off your options are limited.

1. If the ballscrews themselves are worn from heavy use, they will most likely be more sloppy/worn in the middle where most machining is done. So check your backlash in the center then check it at one extreme end or the other.

2. If it is in the ballnut is bad it will be much more consistent everywhere.

3. If it is the thrust-packs it will be the same everywhere.

So without removing a cover and checking it out you can only get a definitive answer if it is the ballscrews.

Removing the cover is very easy and it could probably use a good cleaning under there anyway. Silicone is inexpensive.

Once the cover is removed you can mount an indicator on the screw itself, then apply force to the table and see/measure if the screw moves back and forth. Do this without turning the screw, just try to move the table back and forth to see if the screw moves.

One last suggestion. Did you run these tests because you noticed big errors in your work, or just decided to check the machine? What I'm getting at is to make sure that the gauge you are using is very sensitive and overly capable of measuring these small moves. Many gauges will repeat OK, but not make two separate accurate reading. They all have some friction/resistance and also can deviate/move depending how it is being held or mounted.

Good luck,

Mike

[dahui]

Awesome! This is exactly what I was looking for. If I had a brain I might have even figured that out myself I'll check for backlash at the travel limits today! Thanks.

[Fabio1]

What type of machine is this.Like he said removing the covers and checking off the screw is the proper way to go.However these readings aren't bad unless you are circle interpulating a high tolerance bore...

[dahui]

This is a VF1. I'm doing interpolation around an OD and I'm getting little ridges when the machine changes direction. I can comp out most of it but the finish is cosmetic.

[Machineit]

Originally Posted by dahui This is a VF1. I'm doing interpolation around an OD and I'm getting little ridges when the machine changes direction. I can comp out most of it but the finish is cosmetic.

If you are not actually using precision gauges to check the machine you will not get usable readings.

In interpolating holes or OD's no machine will make perfect circles or OD's. There are too many variables, such as how it is held, the condition of the endmills, coolant spray and on and on.

Even a machine that is in factory specs can be .0004" off, as it has a .0002" location tolerance. Now generally it is much less, usually stays in the .0002" range, which is in the repeatability tolerance.

The machine really does not change directions as much as it simply speeds up and slows down in each as you go around, and the computer must constantly adjust the speed of both motors. Sounds easy, but it is not. Trying doing a hole on a conventional mill and make a circle. So what speed you are traveling at means a great deal too.

You said the "Y" deviation is .0019 and that is quite a bit, but it is not if you are interpolating an OD where your tool start and stop point is on zero in the Y axis.

Put a tenth's or better dial indicator against the back vise jaw and make sure it will repeat well. Then simply try to push and pull the table itself and see if the indicator moves. My guess is that unless you are a big weight lifter it will not move.

One last thing, make sure you write a simple OD or ID program by yourself. No cad system, they like to break circles into quadrants for example and I have seen them make small errors. I would suggest a ID circle with a G12 or G13 line on an easy to machine material like aluminum. They have a circular lead in and out. Run a decent size circle maybe twice, then turn off the coolant and run it once more taking another .005" and no coolant.

A machine that has real wear will still show it then. In accuracy, speed kills.

If you need to do an OD, because you don't have the tools to measure the ID, run it multiple times as previously mentioned, and do not measure at X zero or Y zero, where the endmill starts and stops.


Good Luck,
Mike

[CNCtech 606]

Make sure the ball nut bolts are all securely tightened to the table in X and the saddle in Y.