Backlash

[N4NV]

I have been trying to chase down a backlash problem I have with my Hiwin ballscrews for almost a year. I even sent the ballscrews back to Hiwin and they said there is no detectable backlash in them. I made a video of the test recommended by Hiwin. In the test I attached a ball bearing to the end of the ball screw shaft so I am not measuring any out of squareness in the end of the shaft. I am using a tenths reading (.0001”) indicator on the end of the ballscrew and a thousands reading (.001”) on the table. I am measuring between .0003” and .0004” backlash in the end play of the radial contact bearing, and .004” on the ballscrew/table movement. The gibs are adjusted to that if I push as hard as I can anywhere on the table I can see only .0005” deflection. The mill was recently reground and hand scraped by a professional machine tool rebuilder.

There are only two places I can see the backlash coming from. One is the yoke, which was supplied by Hiwin (very large cast iron, I can’t see that flexing), and the other is the ballnut/screw combination. The ballscrew has a single nut. Hiwin says they take care of back lash in one of two ways, either with a different pitch between the nut and screw, or by adjusting the ball size in the nut. They won’t say which method is used in my installation.

Here is the video of the test:
http://www.flyingcritters.com/video/Backlash.wmv

Any suggestion would be greatly appreciated.

Vince

[BobWarfield]

I agree. Your video shows it isn't the mounting of the ballscrew in its bearings. It has to be downstream--the ballscrew to nut interface (i.e. HiWin is wrong) or the yoke, or possibly the attachment of the yoke to the table.

I am assuming you cannot access your ballnut and yoke without disassembling the machine? It would be nice to get that second indicator reading off the ballnut to rule that out. I suppose if you had to you could put together a test rig to allow you to measure that with the ballscrew removed from the machine. If you can't measure on the machine, I fear it has come to that.

One thought. I have read somewhere of someone that was having a terrible time diagnosing backlash. The took the machine apart several times. The problem was finally traced to a cracked mounting component. It was a hairline crack, and it could not be seen by inspection. It only showed up when you stressed it enough to move the table--hence only when everything was assembled on the machine.

I think if you do disassemble, you might check for cracks of this sort. Perhaps you can get an engine shop to magnaflux the yoke, for example. Just an idea.

I would take the machine apart and Try to build a test fixture where you can measure the component backlashes. I think I'd try to flip the table over so the yoke is on top and the table is face down and work from there with everything bolted up as it would be. Failing that, it has to be relatively easy to test just the ballnut to ballscrew backlash. You can use the machine mounting components to hold things.

Not easy, I know. Sorry!

Best,

BW

[NC Cams]

You fail to mention what type, if any ball screw support bearings are being used.

Same comment for the machine and the type of ways or gibbs involved.

We chased this problem down for weeks in our Hiwin/gibbs fitted EzTrak. A number of comments were made about the problem solving process on the 'Zone - It would behoove you to look them up via the search engine.

We first went so far as to preload the ball nut in our Hiwin screws via Roberts Ball Screw Service in Michigan (it helped, but didn't solve the problem).

Even with a preloaded ball nut, we had backlash - I forget the amount but it left dwell marks up to 0.003" deep in circular or elliptical parts when you did a direction change on either X or Y axis.

So we startind pushing and shoving on the table - that's when we found that the gibbs caused all kinds of backlash but not in the expected way. The side slop/clearance in the X axis gibb which allowed it to slide freely would cause backlash in Y and vice versa.

It took several days of iterative gibb tightening and loosening, with repeated checking with dial indicators, to make them tight enough to NOT have slop yet FREE enough to not have "stick-tion" when you change direction with either hand wheel or servo input. DON'T EVEN THINK that backlash comp will "fix" this - it flat won't. Been there, done that. When we FINALLY got the gibb clearance/stiction issue solved, we STILL had issues.

That's when we focused on the ball screw support (aka thrust) bearings. In our case, the OEM used some modified 6204's mounted in DB configuration with some offset/preload gound into the sides of the rings. From OEM drawings we were lucky enough to obtain, we learned that the bearings had about 75 lbs preload. We even increased that to nearly 150 lbs via shims and EVEN THEN, there was STILL backlash.

Why?

Because deep groove ball bearings are NOT designed or intended to accept appreciable amounts of axial loading AND STILL MAINTAIN ACCURATE POSITIONING - 75 lbs is appreciable regardless of what the OEM did/tried to do. Even at 150lbs (another "optional" OEM bearing configuration we stumbled acrosss) there was backlash, worse yet, the darn handles were too hard to turn!!!.

Fact of the matter, the sides of deep groove ball bearing raceways simply aren't contoured for that type or amount of axial loading. The effect on backlash is not unlike that of a bowling ball as it rolls down the gutter in a bowling alley - the thing never really can stop moving axially, it just gets harder to move further but move it will.

How do you eliminate the problem?

Simple. Replace the deep groove ball bearings used as ball screw support bearinsg with TRUE PRELOADED ANGULAR CONTACT BEARINGS. The higher the preload and the higher the contact angle the better.

Again, the Extrak thread in the Bridgeport section outlines the methods and bearing part numbers we used.

How much does it cost?

If you want backlash that will be a less than a tenth or less, plan on spending BIG bucks and lots of time. WE got our machine to make nearly PERFECT circles with dwell marks under 100 millionths as measured on our Heidenhein equipped cam measuring machine.

The ball screw bearings we use are true ABEC 7 ball screw bearings that retail for about $825 per PAIR and two pairs are needed for a typical Bridgeport mill.

But with 525lbs of preload, ABEC 7 accuracy and 60 deg contact angle, the bearings haVE NO SLOP yet the screws turn with noticeable drag but with absolutely no baclash - accuracy is now "shake your head in amazement" good and that was judged by a guy who tuned our machine who used to dial in the machines at the the Bridgeport factory.

If you want under 0.0003" to 0.0005" net backlash repeatably, you can get by for a bit less costly - probably several hundred per bearing set. One of the Extrak threads lists the alternate bearing numbers for a Bridgeport. SOme are available as cheap "off the shelf" bearing cores BUT they need custom preloading that we have done on a case by case basis.

The overall part cost depends on what type of bearings you need, how much space there is and what's in there now.

Part of our side business is bearing consulting. We'd be happy to provide or recommend bearings BUT, ultimately, there is some skill/expertise associated with assembling the bearings into the machine which isn't always doable as a DIY type of repair.

Anyway, the basics are provided above. It is doable but time consuming and depending on the accuracy required, it can get to be somewhat costly. We have a machine running in our shop that can demonstrate to the doubters that it is doable.

[Zumba]

Originally Posted by N4NV I am measuring between .0003” and .0004” backlash in the end play of the radial contact bearing, and .004” on the ballscrew/table movement.

Radial bearings work fine for manual machines but are worthless for CNC. You should've replaced them with A/C bearings at the time of your CNC conversion, as per NC Cams' advice.

This seems to be an important detail that a lot of CNC conversion people leave out. I guess it's because people go in with the mentality that for a few hundred bucks, you get a brand new, shiny, heavy ball screw. For a similar amount of money, all you get is a block with two bearings in it... not nearly as glamorous. Unfortunately, as you have clearly seen and demonstrated, they're equally as important.

Don't blame the ballscrew. Get yourself quality ballscrew bearing mounts. The easiest way is to "let someone else do it". That is, buy yourself one of these...



... and figure out a way to adapt it to your current setup. These particular bearing blocks are made by Nook. They'll cost around $300 a pop. They won't be as accurate as NC Cams' setup but they'll be a hell of a lot better than .004".

[NC Cams]

I'm not at all familiar with what is INSIDE the Nook pieces, but the pictured parts have beautiful housings for sure. Warning: The housings will essentially do nothing for backlash - it is what type of bearing INSIDE the sexy housing inside in concert with the amount of bearing preload that will/does determine backlash potential.

EDIT:

The following link shows the EXTERNAL housing specs but it's what's inside that counts!!!

http://www.nookindustries.com/endmou...fm?EM=EZM-1009

I'd contact them for details BEFORE YOU BUY. If they are not offering a double row precision A/C package with a heavy preload, you're not going to see much if any improvement in backlash.

END EDIT

If the housings come with inexpensive deep groove ball bearings or unpreloaded A/C's, you won't be any better off than you are. At $300, I'd expect some pretty decent bearings to be included - if not, it is a bit of a rip-off. at least from a performance standpoint.

At this point, YOU have to decide how much backlash you want to tolerate. Once you do, then you go shopping for the parts needed and then you perform or have the work performed to accomplish the task.

Sorry, none of the bearing applications work I've done lately or in the past had $$$ signs as part of the technical specs - we pretty much designed and configured bearings to meet the performance requirements and then let the salesmen negotiate pricing.

Rest assured, when it comes to A/C bearings, especially TRUE high contact angle ball screw support bearings, prices are NOT appreciably negotiable.

True preloaded ball screw support bearings are MUCH too specialized, they are not not high volume/low cost parts (like a 6203 bearing) and they are all pretty much custom made, hence they will NOT be inexpensive.

But if/when you ever get a chance to use a machine fitted with such precise bearings, you will be spoiled by the precision and repeatability FOREVER.

[Zumba]

Originally Posted by NC Cams I'm not at all familiar with what is INSIDE the Nook pieces, but the pictured parts have beautiful housings for sure. Warning: The housings will essentially do nothing for backlash - it is what type of bearing INSIDE the sexy housing inside in concert with the amount of bearing preload that will/does determine backlash potential.

The product is called "Ezze-Mount" (dumb name).

The "Universal-Mount Double" contains two AC bearings inside.

The "Universal-Mount Single" contains a single radial bearing.

Obviously, the OP needs to make sure he buys the Double for the Y-axis. While he's at it, he should probably retrofit his X-axis with a Double on the motor side, and leave the other side as is.

It's questionable how much preloading those bearings have. He'd probably have to call up Nook to find out. My guess is that it's probably on the light side. If he needs heavier preloading, he'll probably have to make his own housing and install the bearings himself. An imperfect machining/installation job could give an end result that's worse than the lightly preloaded Nook product.

[BobWarfield]

Here is my question, and perhaps Vince's too:

NCCams and Zumba, you guys are focused on the ballscrew bearings and or the housing. Vince's movie shows him measuring the movement of the end of the ballscrew itself relative to the stationary base of the machine and getting 0.0003" to 0.0004".

Doesn't his measurement eliminate the ballscrew support bearings and housings from contributing more than 0.0003-4" of backlash to the overall 0.004" number?

If so, doesn't that mean he should be focusing on the ballnut and it's downstream mounting to the machine rather than to the bearings?

Put another way, how can he measure so little motion of the ballscrew and so much backlash and still be facing a mounting bearing problem? I'm having a hard time visualizing that and would like to understand it better.

Best,

BW

[N4NV]

Originally Posted by BobWarfield mounting to the machine rather than to the bearings? Put another way, how can he measure so little motion of the ballscrew and so much backlash and still be facing a mounting bearing problem? I'm having a hard time visualizing that and would like to understand it better. Best, BW

That is exactly what I have been trying to tell Hiwin. I have spent $900 on bearings, specifically made for ballscrews. Three different match pairs of precision angular contact bearings. I have another mill that had .008" of backlash. I did the same test and found almost all of it in the end play of the ballscrew. I replaced the bearing with a precision match set of angular contact bearings and the backlash dropped to .0006". If the end of the shaft is only moving .0004", how can that multiply in the bearings by a factor of 10?

The yoke is held in place with 4-1/2" socket head bolts, the yoke itself must weigh 10 pounds and was supplied by Hiwin. I just can't see if flexing that much.

I am checking just the Y axis for now. The gibs are locked on the X so it does not come into play.

Vince

[N4NV]

Originally Posted by BobWarfield I think if you do disassemble, you might check for cracks of this sort. Perhaps you can get an engine shop to magnaflux the yoke, for example. Just an idea. I would take the machine apart and Try to build a test fixture where you can measure the component backlashes. I think I'd try to flip the table over so the yoke is on top and the table is face down and work from there with everything bolted up as it would be. Failing that, it has to be relatively easy to test just the ballnut to ballscrew backlash. You can use the machine mounting components to hold things. Not easy, I know. Sorry! Best, BW

Thanks for the suggestions. I never thought of turning the table upsidedown. Over the past several months I have had lots of suggestion to take more measurments, but none of them came up with the way to take the mearsurments.

Vince

[davo727]

Here is part of a NC Cams post in an eztrack retrofit thread ref bearings:


The bearings are size dependant and the bearing P/N's vary from machine to machine. I can provide info for a typical Bridgeport mill as that's what we tuned up.

Basically the OEM bearings are modified 6204's - barely adequate for a "drill press" but made suitable for a precision mill via "trick machining" (preloading).

First low cost option upgrade would be 7204BYG's. Have them preloaded DF to about 150 lbs by KAF Mfg in Stamford CT.

Then you go to 7204A5TYDUHP7. ABEC7's and they bolt in with a bit of shim/spacer work between outer ring and retaining clamp plate. Nothing fancy shim wise, a motor preload wave washer works fine.

Then you to to a true hard core ball screw bearing but these are in the $800/screw range. Not for the faint of heart BUT deadly accurate. With nearly 500lbs preload they turn with LESS torque than OEM's (long story explained on another thread).

Once you do all that, you need to do some iterative gibb adjustment (oops too tight, drat too loose, ahh just right) and you can turn a sloppy table into a darn accurate machine - ours can now hold within a tenth or so at mid table when machining masters...

[NC Cams]

I see plenty of references to A/C's in this and various other backlash posts but NO mention of preload amount, if any. Simply stated, you MUST have preloaded A/C's in the thrust postions of ball screws or they will have slop until you load the bearings enough to take up/out the axial clearance.

Edit: a quick review of the NOOK site did NOT find any reference to bearing preload if any in their double mount bearings.

End Edit:

RE-Edit

I didn't look at the WMV - this computer didn't have the WMV properly loaded hence I couldn't play the movie. THis is why i posted the explanation, perhaps redundantly, as I did.

END RE-EDIT

I've seen any number of guys buy some A/C's, mount them in DB or DF and then wonder why they have slop when they have absolutely no or definitely not enough preload - moreover, there are not that many folks outside of the machine tool service or bearing service industry who REALLY know what preload is, how to get it, how to measure it and/or how to properly requalify it.

In our case, we did NOT have a loose/sloppy yoke - it should be intuitive obvious that yoke slop would have to be addressed and eliminated sooner or later. Otherwise, this whole thread evolves into an exercise in futility.

If the fact been told that that true ball screw bearings had been mounted in the machine in the original post, as opposed to much later, I would never have mentioned the issue of bearings. The impression I got was thet root causes were being looked at/for so I provided some that often get repeatedly overlooked. I appologize for not being clairvoyant and wasting the member's time...

Since the member has already recognized that the yoke is/may be a potential root cause of his problem, I saw no need to re-explain what he has already diagnosed as a potential root cause of his problem.

The bearing stuff simply is NOT obvious. Besides, it has been overlooked/misdiagnosed/improperly assumed when the issue of ball screw backlash is raised (again and again). I appologize for even bringing it up now that we know more about what was done initially to service the machine.

When we locked our Eztrak spindle, mounted a dial indicator in it and shoved on the table we did NOT look at the ball screw shaft. We looked at how much motion we had in the table with respect to the spindle via the dial indicator when we had a 275lb guy shoved the daylights out of the table.

We then moved the handles by hand and looked at how much the dial on the handle moved before we saw the dial indicator move.

We then went about eliminating belt looseness, servo tuning, and uneven gibb wear which is when/how we found how lame the OEM ball screw bearings are/were. I'd suggest that these and any other machine specific sources of slop be looked at as part of a holistic approach to the solving of this and other backlash problems.

I know of a instance where the drive pulleys on the motor and the handles had sloppy keys which let them be loose as well - this was sort of a dumb thing but, things like that happen. HOw do I know? We mounted our $800 ball screw bearings BACKWARDS and got more slop intead of less. You should have seen how many times we tore into the machine until we found THAT stupid mistake. It took a keen eye for the obvious to fix that dumb out.

I saw no need to repost the Extrak thread verbatim or in part as most of the big issue items were explained in the thread if i recall. Perhaps DAVO727 can post the whole link - I don't have it as a handy reference on this particular computer.

Geting to the yoke is NOT an easy task - taking out sloppy machining is going to be even harder. Proving it to HIWIN, harder yet/still.

We did NOT go into the yoke in our Extrak because it was simply too deep into the machine - by doing the external, relatively easy to get at albeit expensive details, we got our machine to perform with no drag and easily less than 0.0005" table deflection when shoving on it and at or under 0.0001" error when we did dynamic trials on the profiles we were milling.

Our experience taught us the finding and eliminating the "big" slop was the easy part. Getting the slop to the point where you had to use tenths indicator to find/measure it was the hard part.

[BobWarfield]

Originally Posted by N4NV Thanks for the suggestions. I never thought of turning the table upsidedown. Over the past several months I have had lots of suggestion to take more measurments, but none of them came up with the way to take the mearsurments. Vince

Seems like the easiest way to try to take some more measurements. It ought to enable you to check the ballnuts at least. It will require some inventiveness to use this approach to look for flexing, but it may be a place to start.

Looking at your video again, it seems pretty clear there is no lost motion in the pulleys. You could see better directly, but it doesn't look like that's the issue. I still think it has to be the ballnuts or possibly something to do with the yoke.

When you get that table upside down, I'd almost be tempted to gingerly try to apply some side torque to the ballscrew and see if the nut is solidly mounted in the yoke. Not too much--don't want to damage anything, but just a little to look for issues there.

I suppose if you get that table upside down and can't find any ballnut issues, there is still a possibility of some kind of slop in the ways or gibs. I realize you were unable to move it by hand, but maybe you can think of a way to use your indicator to sweep the ways and do some other measurements to see if anything seems amiss in that respect.

I will be curious to learn what you figure out.

Best,

BW