[erikjgreen]

Hello, all;


In the ongoing saga of repairing/replacing parts and electronics on my Lagunmatic 250 CNC, I've discovered that the ball screws have some backlash - when I turn them they rotate without load for a little bit.

I verified this with a dial test indicator to make sure it wasn't a movement too small to see, and it's not. I can rotate the timing pulley that is directly attached to the ball screw in both cases by about 1.5 - 2 notches before the table starts to move. I can wiggle it back and forth that much without moving anything.

The pulleys on the ballscrews have 50 notches total. If I'm doing the math right, that means my ballscrews don't engage when reversing direction for 14 degrees of rotation. My ballscrews are metric 5 lead, so they move 5mm per turn, so that is about .2mm of linear movement backlash (8 thou), which seems like an awful lot.

I know some CNC controllers compensate for backlash when reversing in software, but I don't know if this is a normal amount or not.

Once engaged and moving the table, obviously the ballscrews move every time they are turned even a little.

Anyone know if this is a normal amount of backlash, or if not what could be going on? Worn ball nuts, worn screw, or was it just designed this way?

With this much backlash, how would it be possible to eg. interpolate a circle, with all those direction reverses?

Thanks,
Erik

[underthetire]

Too much!
Check end play of the ballscrew. Could be bad thrust bearings.

[erikjgreen]

Okay.

How do I do that? Try to move the ballscrew in its mount with my hand? Or try to move the table by hand and see what moves?

Edit: Following some instructions on line, I put a magnetic indicator on the ball screw itself, and the point on the thrust bearing housing. Attempting to move the table back and forth got me no different readings, although I could try using a longer lever or something.

It sounds like maybe the play is in the ball nut itself? I haven't taken the table off this thing yet, probably I have to build a crane to do that.

Also, I think it's weird that both the X and Y axes have the same amount of play in them. The Z axis is broken at the moment and disassembled, but there is no play in the Z ballnut. The thrust bearing housing on the Z is what's broken, can't test that.

Just to see if anything made noise, I used a car battery to run the (disconnected) X and Y axes back and forth. The only noise was a slight motor hum plus a small tick here and there from the timing pulleys... I don't *think* it was the bearings. No other noise, including grinding or squealing.

Erik

[erikjgreen]

I don't have any other ideas on things to check, so I'm going to put together some kind of crane and remove the table to check the saddle and ball nuts.

From what I can see of the ways, they're smooth and shiny. The X ballscrew (which I can see) is shiny in places and somewhat dull in others, but doesn't appear to have damage. I'll clean it off better when the saddle is off and make sure it's ok.

X is getting lubricated well now that I have the lube pump fully powered, I'm getting drips of way lube now and again on the floor.

I've been working on a problem with the servo drives, I blew a MOSFET on one so I'm replacing that, and I'm tracking down an encoder problem on the Y axis that seems intermittent.

If anyone has any other suggestions for things to look for/ways to test for backlash I'm all ears. Hopefully I don't open up the saddle and find the ballnuts hanging by one bent bolt each....

Erik

[RICHARD ZASTROW]

You might also check the ball screw support itself to see if it is securely in place. Ball Screw support bearings locked in place within the support bracket. Ball nut securely in place.

We had similar problems years back on a large Pratt & Whitney machining center. Nearly every bolt on every bracket had come loose to some degree. Held good tolerance after tightening everything.

Dick Z

[erikjgreen]

Dick -

I'm hoping it's something like that. If the ballnuts were worn, I don't think they'd be as quiet as they are or as accurate when they're not reversing direction.

I'm working on assembling (from spare parts) a shop crane so I can lift the table off the machine and check on the bolts on the ballnuts and brackets. From the looks of things I doubt it's ever been disassembled.

I'll post back here with what I find.

[erikjgreen]

Well, I haven't managed to disassemble the table yet (still working on the crane, plus figuring out how the thing comes off the mill) but I did get some more information.

I crawled under the knee and looked up with a flashlight, then watched the Y nut in the yoke as I moved the Y axis belt. The screw begins to turn first, then a little bit later the axis begins to move. It's easy to see under there because you're looking up at the underside of the table.

So I think the backlash IS in the ball nut, at least in the Y axis. I can't see the X yet, but I'm betting it's the same, because it behaves the same. Both ballscrews are in similar condition, they look ok to my inexpert eyes:



There's some grit and junk in the oil on them that I think the wipers on the screws might handle, but I'm going to clean them off by hand when I take this thing apart. Looking at the ball screws, I don't see any damage to the threads.

Here's a bit wider picture of the Y screw, plus some of the nut in the yoke (it's hard to take a focused pic under there). The ballnut is firmly in the yoke, no movement in the SHCS fasteners at all:



It's remotely possible that the reason I saw the screw turn without moving the table is that there's some play in the thrust bearings. I haven't taken the ballscrew pulleys off yet to check these, but I will soon, because any way you look at it I'm going to have to pay some $$ to get this backlash gone.

I did find out one other fact - the backlash is quite consistent over the whole length of the axis. It's between 8 and 9 thousandths linear as measured with a dial test indicator (I didn't care enough to get out the good DTI last night, so this is approximate).

If the problem is the thrust bearings, I'll replace those.

If the thrust bearings are fine... I dunno. It's almost like someone has loaded the ball nuts with balls that are .008 undersized?

If that's the case, maybe I could have someone replace the balls for not a ton of cash (under $400?).

Another possibility is replacing the entire ball nut with another nut from the same size/pitch NSK screw... cost is about $300 surplus for a C3 grade no backlash nut ATM. I'd probably have to replace/mod the yoke though, since I probably wouldn't get exactly the same nut shape. Hard to do without a milling machine available.

Or, I could replace the ball screw with another 32mm one with a 5mm or better pitch. As long as the diameter is the same the thrust bearings (if they're good) would work as would their mounts. The sticky bit would be getting a screw that would have the right length with sections machined to fit the bearings, pulley, and nut with no extra. I have a mini lathe available, but machining a 32mm ball screw four feet long is probably beyond its abilities.

Suggestions welcome,

Erik

[erikjgreen]

Silly me.

I think when I did my levering of the table earlier, I must have had a short lever AND a bad leverage point. Or something.

Tonight, I took the chip guards off to get behind the table better, solidly mounted a dial test indicator with a C clamp on a sturdy ladder next to the mill, weighted the ladder so it wouldn't wobble, tested the wobble, put the point of the dial test indicator on the tip of my Y axis ballscrew, and levered with a piece of aluminum scrap between the table and column.

The indicator moved about 0.01.

Given that the ballscrew is *supposed* to be axially motionless relative to the mill base (and floor, and ladder) when the table pushes against it (or vice versa) the fact that it moves a very similar amount to my measured backlash tells me the ballscrew support bearings are worn (by the way, I did do what was suggested earlier in this thread and checked the support bracket. It was nice and tight).

I decided to pull off the cover and pulley and check the Y support bearings directly. With some effort I got it apart (I chewed up the nut on the ball screw end a bit, but it was already chewed up. I need new ones probably).

When I first looked at the outer bearing, I thought it had a ring of grease extruding out of the face. Turned out it was one of the retaining rings.

The bearing came out in more or less one piece, but then rapidly came apart into inner and outer races, two glass fiber retaining rings, and a bunch of loose balls.

I cleaned the grease off and checked the balls with a micrometer. They showed variations in diameter depending on which way I turned them. Not huge, but something like 0.0005 differences or so. The inner and outer races look ok to the naked eye, but would I be able to see problems?

Given that the outer bearing came apart so easily and the fact that I measured the ball screw moving in the mount, I'm shopping for replacement bearings for both axes... they're Barden L078H, basically described here:

http://www.consbrgs.com/page.taf?detail=123_1_17

Looks like I can buy each one new for $175 , or get em somewhat cheaper on e-bay. Oh, well, I wasn't going to do anything fun with the cash anyway.

If anyone sees a flaw in what I did above or can tell me a way to check the bearings (I have one still in one piece, and one in pieces) to see if they're truly the problem now that they're out, I'd love to hear about it before I spend $700 on new bearings....

Thanks,
Erik

[RICHARD ZASTROW]

Review your bearing description. The DU at the end is "duplex pair". The text says they are face ground. That determines the preload for the PAIR of bearings.

That indicates to me this requires a new PAIR of bearings.

Dick Z

[erikjgreen]

Yes, they are used and mounted in pairs.

The existing bearings were mounted face to face.

Fortunately they don't seem to require matched pairs, but rather "universally flush ground", so I can buy four bearings and use them in two pairs and be ok, instead of buying sets, which would be even more expensive.

There are some about the right type on ebay here, have to investigate more to make sure they'd work:

New NSK Ball Screw Bearing Super Precision Bearing 20tac47bsuc10pn7b Free SHIP | eBay

Erik

[underthetire]

Call all world machine parts. Those are a standard tac thrust bearing. You may need to re grind/cut collars if you go to a universal bearing like that or a different brand. Years ago, bearings were sold as a matched set and big money, some were 1200 per set of three. Now with better cnc grinders, they are usually sold as universal. What I have found is typically the bearing measure about .0012 thinner on the new universals, so pre load spacers need to be cut by the .0012 per bearing, or shimmed depending on set up.

[erikjgreen]

Ah, ok.

I'll give them a call later today, although I can't order any bearings this expensive for a week or so.

There are some new old stock bearing sets on ebay too, so maybe by the time I can order two sets will pop up.

[underthetire]

A guy at work has a bunch of old stock thrust bearings from a previous job. I can ask him Monday if he has the small ones like those. He's been bugging me to help sell all of them off anyway.

[erikjgreen]

That'd be fantastic! If he has the right type, I'll buy a bunch

[erikjgreen]

I'm finding that the exact bearings I need are hard to find at any price because they're an older sizing/not made any more.

For instance, I can find NSK 20x47x15 ballscrew support bearings here on e-bay: LINK

However, the exact NSK bearing that would be equivalent to what I need would be an NSK 20 TAC 47 XB. Not listed on ebay, or anywhere else I can see.

The problem is the thickness or width. The Barden bearings I removed are 15.86mm thick. Most of the modern 20x47 ballscrew support bearings are only 15mm thick. So I'd have to make up about 1.72mm of space to get preload (actually a bit less, since the original bearing mount is only 31.2mm deep).

If I can't find the exact bearing in new old stock, can I use a modern bearing pair (30mm total) with a 1.7mm shim? What would I use for shim material? I have brass shim stock but it's only 0.005in. thick. I'm guessing the shim thickness has to be uniform on both sides, but could I use any uniform non compressible material?

I suppose another option would be to turn a new cap for the bearing recess, which has the contact surface 1.7mm thicker but otherwise has the same dimensions as the old one.

Erik

[RICHARD ZASTROW]

If you have to install a shim/spacer do not use brass. Brass will deform under impact or load. Best is ground, hardened steel spacers.

Those spacers MUST BE exactly the same thickness and have exactly parallel faces if they are placed between the bearings.

I do not know how the preload is applied in your situation. You stated a face to face bearing arrangement. That would mean the preload is applied to the outer rings. Is a spacer transferring the preload already there? If so, you could replace it with one 1.72mm thicker and thread the lock-nut that much further on the shaft end.

Dick Z

[erikjgreen]

Right now, the bearings are installed face to face (the lettered faces are together) in the cavity.

There's no spacer anywhere. The cavity is 31.2mm deep and the bearings are 15.86mm thick, so the outer bearing's outer race edge sits about .5mm above the cavity edge. The "cap" on the cavity which preloads the bearings is a 3/8 or so circular piece of steel held in place by 3 cap screws, and has a hole just big enough to pass the ballscrew shaft, IE about 25mm, in the center.

The new bearings (15mm) installed face to face would sit 1.2mm below the depth of the cavity edge, thus getting no preload from the cap unless I shim them.

Sounds like if I made a shim it I'd have to obtain a 1.5mm piece of ground steel stock with parallel faces and turn/bore it to size? If I replaced the "cap" I'm guessing I'd have to do the same thing with a bigger piece of metal. Maybe it's worth the time to find a local machine shop to make this, since my tools are limited atm.

Or is there a source for 47mm diameter shims for bearing outer races that are at least 0.5mm thick? McMaster has some for small bearings only. I could get some steel shim stock from them and cut it into five shims to stack I guess.

I suppose the other approach to all this would be to bore out the old bearing support cavity enough to install a standard WBK20 mount with bearings, but that's a lot harder to do, especially without a working mill...

Erik

[erikjgreen]

I think I've had some luck... waiting to see if I get a set of bearings I found.. they're Fafnir MM 9306WI 2H DU, which is the exact equivalent bearing I need for my mill, and is the correct 15.86mm width. Plus they're matched pairs, new old stock, and he even has two sets, which is what I need.

Crossing fingers....

[erikjgreen]

Yep, they're on the way.

Exact match for the original bearings, only paid $320 for two matched pairs, so I got a great deal.

Looks like I won't have to worry about shimming 'em either. I'll update here when they're installed and if they solved the problem.

Erik

[erikjgreen]

I'm getting ready to put these in, cleaning and repainting stuff while I wait for them to arrive.

When they do arrive, what grease should I use to lubricate them? Molykote? They might have some grease in 'em (they're in the original package) but it'll be 25 years old at this point.

I'm also ordering new socket head cap screws and washers for the mountings, the old ones are a bit worn out.

Erik