[multiplex]

just wanted to get some feedback on an issue we are having with an older partner 5 mill.

As a simple test, i cut a circle in a piece of aluminum. Measured x, and y. X results are exactly what they should be. Y results are out by about .005 - .006" smaller

i thought it could possibly be backlash, so i changed this setting with no change in results.

What is odd is that i can pick up two edges off a known width and the machine reports the number exactly.

I guess its an issue when the machine is under load? Thrust bearings?

Any ideas on where to look would be great :rainfro:

[70cyclone]

could it be following error?

[One of Many]

If it were following error, that would become an error message on the screen.

If the parameters were messed up at some point, verify the encoder pulses per unit , the multiplier value, or ballscrew compensation table are set as shipped.

This could be mechanical and should be verifiable with an indicator to see where the lag between reversal of the table and lead screw.
- the motor/ballscrew coupler,
-the ballscrew nut and saddle,
-the ball screw thrust bearings and Y axis base,
-the encoder coupler and motor.

And in the rare case the encoder disk remains bonded to the hub.

Short of replacing the encoder if it were failing. The encoder disk and sensor area could be flushed with a mild brake cleaner, just in case it were losing a pulse periodically. Doubtful, but you are looking for a needle in a hay stack.

DC

[multiplex]

Thanks for the responses guys - i will check the parameters.

I did pull the screw out in the meantime. The thrust bearings are in pretty good shape, so I'm going to keep those as is. The rear support bearing was totally shot, so i have a new sealed one on its way. With that shot, it may have put some extra off axis load on the ball nut.

You can feel a little slop in the nut, but i would assume it could be taken care of in software? The slop is not so much in the direction of the screw, but more or less 'wobbly' for lack of a better term.

no pitting on the screw or anything like that.

Is it worth looking at the balls themselves?

Next step is to figure out how to take off the way covers. There is years of grime in the tunnel which needs to be cleaned out. Its definetly the reason the rear support bearing wore out.

Any hints on removing the ways?

[One of Many]

I wouldn't remove the ways, if that is what you intend to do.

The Y way covers should be bolted on to the face of the saddle and from under the side ledge front stationary section. Every model is a bit different.

How did you get the screw out without removing the way covers? Hatch in front?

Either be very gentle in cleaning the tunnel area or since it is this far apart, replace all the cables for the limit switches. Brittle insulation opens and once coolant gets in there, you will experience constant overtravel E-stop alarms. The home switch might get wierd too. Check ALL lube lines too.

Cheap insurance just to replace all the bearings too. If one seal went bad and coolant flushed the grease out of one bearing, the rest are not far behind. Unless you need the experience taking it apart again....

As long as the screw isn't scored, you shouldn't need to check the ball condition. The nut wobbling is a bit puzzling. Make sure some dunce has not lost some of them on a prior repair. Sounds like it has worn beyond its preload though. Possibly worse in some areas than others. You might send it out and have it gone through. If not too late, they can load it with oversized balls, resize etc. Or like we did, put it back together and ordered another, then installed it at a better time.

DC

[70cyclone]

If it were following error, that would become an error message on the screen.

only if it became 'excess error'(following error keeps increasing not staying steady).
if the following error on x is .05 and .075 on the y it would still run without errors and it would point to point(i.e. lines in one axis) move to the correct position but interpolated moves(i.e. circles) would not be correct.
this is how I remember it...but hey i have been wrong before.

[One of Many]

only if it became 'excess error'(following error keeps increasing not staying steady).
if the following error on x is .05 and .075 on the y it would still run without errors and it would point to point(i.e. lines in one axis) move to the correct position but interpolated moves(i.e. circles) would not be correct.
this is how I remember it...but hey i have been wrong before.

Good point. Looking at the parameter files I have here, Milltronics calls it excess error and in the error display screen they do call it following error like Bridgeport. This can be viewed as the part is cutting. The parameter is set to .300, so no message until commanded verses actual gets that high!

DC

[multiplex]

thanks for the tips.

i'm resurrecting this machine for personal use here at work, so the budget is somewhat limited. Sad thing is the machine has been sitting for a while since everyone thought it was junk. I know what will happen though - If i fix it, all of the sudden interest will peak

if you don't mind me asking, how much was a new ball screw assembly?

[multiplex]

think i made some headway here, and the problem ended up being mostly mechanical

i took apart the y axis screw. cleaned it, replaced the rear support bearing and thoroughly cleaned out the tunnel which houses the ball screw - it was filled 1/2 way with old coolant, chips, grease and whatnot.

big thing was that i don' think the thrust bearing was properly preloaded. once reassembled, i cut another circle and measured the results on our OGP smartscope.

best results i got from before the fix was .006" total difference between x and y.

i'm not slightly less than .002" total difference between x and y.

some of this could still be in the backlash of the y axis, but i think at this point its reasonable for what i need to do.

what kind of results should i really expect from a mill like this?

i noticed the x axis is making some noise now - very consistent with rotation. so that might be the next project.

[One of Many]

Now that a looked at the PO, this was a $2200 X axis ballscrew replacement. So, the Y axis might be in the $1700 range. That is through Milltronics, so there may be other sources that are cheaper if they can duplucate the configuration.

The most recent VK4 we have included a ballbar test printout with most spec's well within .0002 and circularity just under .0007.

On a worn machine +/- .001 isn't all that bad and may be slightly improved with parameter adjustments and servo amp tuning. If better is required on intereference fits, boring would be the wiser choice than reliance on interpolating.

DC

[multiplex]

thanks for the feedback.

things were going well until the other day. The Y axis was not able to travel its full distance. I decided to pull the screw back out to inspect it. It appeared as though the nut was jamming up at certain points.

Being desperate, i decided to pull off the return tubes, to examine the balls. I felt there might be something inside the nut, chips, etc..

One of the return tubes was damaged on the end - looks like a ball collided from the outside possibly (not sure how that is possible). It crimped off the end of the tube and would not allow the balls to circulate.

I contacted milltronics, and they no longer stock replacement screws. They offered to reload the nut however. But i'm not sure if the return tube is a deal breaker. It also appeared to me that some balls were missing. Luckily they quoted 300 for a reload.

i called a local company - they stated it would be 1,000 for a rebuild until i told them about the return tube. At that point it jumped up to 2,500.

any suggestions on moving forward? this whole project is driving me crazy. I've been putting so much time into it, i haven't gotten any of my real work done!

[One of Many]

You could probably make your own return tube, if the tubing is available. I wouldn't imagine it special. Certainly not precision for the most part.

$300 is cheap for that in terms of you not fussing with it. Although you can order your own ball bearings after mic'ing the ones in there now.

I thought I had seen total rebuilds of ballscrews for around $750, but that was some time ago. I'm sure if you do some digging, there may be repair houses that are much cheaper than custom replacement options. Ouch, they is pricey buggers!

DC

[multiplex]

yeah after taking a deep breath, i think i have a good plan forward..

the tube is dented in from the outside which makes 'hammering' it out difficult. One of the guys in the shop had the idea to weld the outside of the tube 'carefully', and then plunge mill the inside of the tube to open it back up. there is some funky geometry on the ends of the tubes, or i would just try to remake it.

moving on from there - i finally found a good source for balls.

http://www.precisionballs.com/catalog/index.php

we measured 37 balls from the screw to get an idea of what size they are. i was surprised at the range. maybe its wear related?

here is a histogram of the results

Bin Frequency
0.12340 1
0.12350 1
0.12360 6
0.12370 1
0.12380 1
0.12390 5
0.12400 2
0.12410 4
0.12420 10
0.12430 6

Most of them are .12420. I'm wondering if the .12355's are spacer balls that are meant to be smaller.

i'm worried if i reload with all .12420 it will be too tight. I read that .0002" increase in ball size can account for .001" of backlash.. I currently have about .0027 per the machine settings.

I might order some .1242 and some .1240 and see what happens.

I could always sort in some smaller balls if needed

[One of Many]

Well, in a mad dash out to the dumpster to look at the X-axis ballscrew we just scrapped. Dug it out from the bottom of the dumpster to see if it had the tubes you might need. No such luck. This one had the plastic cap transfer ports internal in the nut.

Sorry, I can't help on the ball reloading details. There is some good info around the web on reloading, but not much detail that can keep you out of trouble.

If you have not searched here, The late NC Cams had some posts on the subject here.

DC

[multiplex]

wow - thanks for trying to dig up the parts! appreciated even if they didn't match

today my new balls came in, so i spent some time tonight reloading the nut. I decided on a set of .12405 balls. which is slightly above the average size of my previous balls. I ordered 250, and ended up needed 258 so i sorted 8 from my old set.

i was kinda shocked to see that i was previously short by nearly 50 balls.

with the new balls the nut is much tighter, but still has some back lash. The screw is worn differently across the length so I don't think there is too much i else i can do on my own.

got the whole machine back together, and ran some rapids back and forth. Machine seems to function great. hopefully this is the end of my troubles!

i would like to however find out more about tuning the servos. When i move in .001" increments it really only moves around .0007".. does better in longer distance moves.

[zymo]

Multiplex, can you possibly share the thrust bearing reassemble diagram/pics/instructions? I just recently did the same and I'm not too confident I completed that correctly. Came across this thread researching it.

Thanks

Z

[ZZZZ]

I have always put the narrow faces of the inside races together and the wide faces of the outside races together on thrust bearings.

[zymo]

Oops, I should clarify. The specific installation to the partner v. I had to remove the head to fit it in the garage, and time has passed and I forgot the exact placement of the snap ring, and can't really find an example. The manual, is to antiquated, I would prefer to verify it rather then guess. I just need to see a diagram of it in more modern times.