[cnczoner]

I've measured 0.014" of backlash near the middle of the X-axis of my mill (converted to CNC), and I understand that eliminating that will give me a noticeably better finish in aluminum. I can also see that my circles are not exactly circular, btw.

I need no better than 0.004" error per every 4", (= 0.012" per foot) and from the specs I see on the different options, that is not a problem with any of the options. So eliminating the backlash (as much as reasonably possible) is the only real issue here. Also, I'm going to focus on only the X & Y axes for now.

Existing setup: 23mm (0.91") dia. acme screws. X-Axis total screw length is about 3ft (supported on both ends) and Y-axis is about 2ft. Travels are a bit shy of 20" x 7". Realistically, about 5-10 hrs of use per week, and slow (about 15-20ipm for cutting and 5ipm for finishing).

Possible options...
(1) Anti-backlash nut. I think I understand how these work but I haven't found any for my thread size, and it seems like it would wear out fast. It would not be easy to get under the base of a 700lb machine regularly, so I don't want to have to replace or adjust it often (it was hell getting the mill up on the workbench with 3 guys and some jacks, etc).

(2) Get 2 nuts on the acme screw and tension them somehow -- I haven't figured out the details on this yet, but tensioning a nut on an acme screw seems like it would wear it out faster, wouldn't it? And as I understand it, wear occurs mostly in the center of the table's travel where most parts are located on the table, then if the nuts are tensioned, they bind on the ends when a larger part is machined someday. Ugh! Or am I being overly concerned with my low usage level?

This article touts leadscrews being a viable option, but I understand that costs can exceed that of basic ballscrews.

(3) Ballscrews. Seems like rolled would be the better option here, as even C7 or C10 meets my accuracy requirements. But what diameter? I'm fairly sure that a chinese machine wouldn't go above and beyond in their acme-screw selection, so should I go with something at least 23mm (say 1")? Or would 5/8" or 3/4" be fine due to less friction, etc? At my low speeds, say even 30ipm with 5 rpi, that's only 150rpm, so whipping should not be a problem on the y-axis which would have only about 1 ft hanging. (The x-axis is supported on both ends). I'm worried with larger diameters I'd have to start grinding/cutting parts of the mill base for clearance. Also, seems like Ind. Hobbies is using 5/8" or 3/4" dia and their mill is larger than mine (though it seems to be the same motor/head). Tormach uses 3/4" dia. and is pretty much the same size/hp as my mill.

Assuming I'm still on the right track here, what else do I need to know? Ballscrews/nuts still have some backlash, but I see some people change the balls to get rid of the backlash. Is that something I need to do? Or can I get them with zero backlash already. I've not seen backlash specs from any source yet.

Also, I see single-nuts, double nuts and pre-loaded nuts, but not sure what this is. Single-nut seems straightforward, double-nut seems like 2 nuts that should be tensioned against each other, but I don't see any mechanism for tensioning. Or is that internal somewhere? Also what is pre-loaded? Does that mean that the balls are pre-loaded into the nut? Or that there is some pre-load tension on the ballnut mechanism?

Right now, I am looking for what direction to go, and I'll get into the specifics later, such as what type/size/load of bearings, end-blocks, etc I'll need.

Thanks,
-Neil.

[harryn]

I am still very much a newbie at this, but there is some excellent info in the linear motion FAQ section threads.

What you are seeing "sounds" about right for the setup you have.

[cnczoner]

You mean the sticky FAQ? I had read Swede's intro, which was very informative, but didn't answer the questions I'm asking here. Another link in the FAQ was just a glossary. And then there was some regarding rails.

I'm trying to figure out what's the best recommendation (or a way to calculate the best options) for MY setup. Most everything I've found talks about mini-mills or bridgeports. And setups are so different that it's tough to get a trend for any given machine size.

[Al_The_Man]

I see single-nuts, double nuts and pre-loaded nuts, but not sure what this is. Single-nut seems straightforward, double-nut seems like 2 nuts that should be tensioned against each other, but I don't see any mechanism for tensioning. Or is that internal somewhere? Also what is pre-loaded? Does that mean that the balls are pre-loaded into the nut? Or that there is some pre-load tension on the ballnut mechanism?

In my opinion, for CNC applications some kind of preload is necessary, this can vary from sizing the balls for minimum backlash , or my preference is preloaded nut, this is usually done with two nuts and something that resembles two belville washers to load one against the other and for a pretermined preload force, usually published by the manuf.
Rolled is obviously cheaper than the precision ground and I have found vary in quality, the bulk ones sold on the auction site here and ebay, are very good quality precision rolled from N. Korean Hiwin, dia is a bit small though 20mm.
Al.

[Zumba]

Get ballscrews for starters. Then get an antibacklash nut. Finally, your ballscrew support bearings should have angular contact bearings on one side, deep groove radials on the other.

Since your accuracy requirement isn't very high, and since your speeds are slow, you can probably get away with 5/8" diameter screws and nuts (about $12/foot for the screws, $120 per nut), and VXB brand bearings from China. 5200 angular contact bearing ($10) on one side, and a 6200 radial bearing on the other. Bigger is better though.

That's what I'd do to achieve this goal with minimal expense and labor. If you've never machined a ballscrew before and you don't have access to a pretty heavy duty lathe, consider getting them machined by someone else. I believe homeshopcnc.com sells screws and offers end machining service.

P.S. I'd spend the extra bucks for 1" screws and nuts and name brand 7203/6203 bearings. Much better accuracy and higher possible speeds... maybe your needs will change in the future.

[ckm]

I've measured 0.014" of backlash near the middle of the X-axis of my mill (converted to CNC), and I understand that eliminating that will give me a noticeably better finish in aluminum. I can also see that my circles are not exactly circular, btw.

I need no better than 0.004" error per every 4", (= 0.012" per foot) and from the specs I see on the different options, that is not a problem with any of the options. So eliminating the backlash (as much as reasonably possible) is the only real issue here. Also, I'm going to focus on only the X & Y axes for now.

Existing setup: 23mm (0.91") dia. acme screws. X-Axis total screw length is about 3ft (supported on both ends) and Y-axis is about 2ft. Travels are a bit shy of 20" x 7". Realistically, about 5-10 hrs of use per week, and slow (about 15-20ipm for cutting and 5ipm for finishing).

Possible options...
(1) Anti-backlash nut. I think I understand how these work but I haven't found any for my thread size, and it seems like it would wear out fast. It would not be easy to get under the base of a 700lb machine regularly, so I don't want to have to replace or adjust it often (it was hell getting the mill up on the workbench with 3 guys and some jacks, etc).

Have you seen the anti-backlash nuts from DumpsterCNC? Those might give you some ideas. I think that before you do the ballscrew conversion, you should try to see if you can eliminate the backlash in your acme setup.

I've got an HF MicroMill converted to CNC and the X/Y axis' have little if any backlash on the stock acme screws. The backlash is taken care of using the stock split nut, which is basically a metal block with a slot cut in it and two set screws, and a thrust bearing arrangement at the motor end of the screw.

The thrust bearing was definitely the most critical part of getting rid of the backlash, and tightening it up worked very well. You might go over your setup again and double check if you actually have thrust bearings and that they are properly tightened. I used a small pry bar to snug them up real good...

As far as finding acme screws in that size, I would check McMaster Carr, they stock a lot of acme screws....

HTH,

Chris.

[cnczoner]

FWIW, I did some more backlash measuring: X =-> 0.014", Y ==> 0.011", Z ==> 0.001".

Still haven't found anti-backlash nuts for my size leadscrew, but in that range they're really expensive ($190 at Mcmaster!). I don't know the formal procedure for checking axial play, but pulling back and forth on the X and Y leadscrews did not show me any play in the system. So I think ballscrews are the correct direction now.

Still curious why the ballscrews don't already come with large enough balls to eliminate backlash...

Whereas I don't want to waste money, I'd step up to the next larger diameter if it's worth it for my setup. I'm more concerned about the time/skills/ordeal in retrofitting this, but while I'm in there, I might as well do it with the proper-sized parts.

Larger diameters would probably run me into clearance issues, but would have enough meat so that I could maintain 17mm on the ground-down ends to fit my existing pullies/handles, etc. But my puny 7x10 lathe can turn 5/8" max thru the spindle/bore IIRC. But I had another revelation -- I could instead machine a custom "coupler" from round steel rod that would have a hole through one end to snugly fit the (unground) ballscrew, and the other side of the custom coupler would be the shaft for the bearings, handle, etc...??? I'd have to cross drill through the coupler and ballscrew to secure it with set screws or spring pins, but it should work. Thoughts?

As for bearings, the current acme screw system holds the load well, right? So why not re-use those? What makes that undesireable, so that we go through the hassle of creating custom endblocks, etc?

BTW, homeshopcnc will turn down the screws they sell (Nook, .631"), but I found a couple negative comments about Nook in my research. What about Roton? Since they had prices published, I'm been estimating from that, but they have no error specs published, so that worries me. I can call, but why would they not want to publish it? McMaster is generally more expensive, but without proper competing data (prices and/or specs), I can't say for sure yet in this case.

Also, would the larger double/preloaded ballnuts cause me any loss of travel? I want to keep, or extend to, as much travel as possible..

Thanks,
-Neil.

[ckm]

If you don't have thrust bearings, or they are not tight, you will have large amounts of backlash. Most of the Chinese mills come with no bearings, just a lubricated surface. You can spend a lot of money on ball screws and other expensive hardware, but if you have play in pillow block that supports the motor end of the screw, it'll make no difference at all.

As far as not being able to feel any movement, well, the amount of pressure your arm can apply is no where near what the machine is capable of....

Feel free to ignore this, but I can definitely say that in my mill, thrust bearings made it possible to adjust for zero backlash, which was pretty much impossible to do before.

Chris.

[cnczoner]

I was just looking thru an exploded diagram of my mill and you're right -- there is a "thrust bearing" but seems like just a sleeve.

Just spoke with Roton... They claim .006" backlash and .003"/ft accuracy, for the 5/8", 3/4" or 1" sizes. They can also re-load with larger balls for and additional $10 per nut, but that would probably get it down to about .003" backlash. With larger balls, the assembly would "behave differently", but no clear info on how exactly. But if I still need a pre-loaded arrangement to get to zero backlash, I don't see why I'd bother to change the balls.

Zumba -- you said "Get ballscrews for starters. Then get an antibacklash nut.". What exactly are you referring to as the anti-backlash nut here? Is there a special anti-backlash nut for ballscrews? Or were you referring to a pre-loaded double-nut setup?

BTW, Roton and McMaster do not carry pre-loaded nuts, so I'd be interested to hear of some sources for these.

THanks,
-Neil.

[Zumba]

Nook and Rockford both sell anti-backlash double nuts. The second nut has adjustable preload for your application.

Larger balls results in added friction, and even with these, you still get .003 or so backlash. A quality adjustable preload double nut should take it down to zero.

You can reload your own balls if you want to. Probably cost you $2.00.

[ckm]

I was just looking thru an exploded diagram of my mill and you're right -- there is a "thrust bearing" but seems like just a sleeve.

Adding a thrust bearing will get rid of most of your backlash for about $20 plus a little creative engineering. McMaster has them for about $8 ea in stainless, don't forget a shaft clamp to tighten it up....

Or you can spend $500 on ball screws.

Chris.

[cnczoner]

Adding a thrust bearing will get rid of most of your backlash for about $20 plus a little creative engineering. McMaster has them for about $8 ea in stainless, don't forget a shaft clamp to tighten it up....

Or you can spend $500 on ball screws.

Chris.

Isn't this something I can test for? I expect that if I do my regular backlash test (bring the spindle to zero position from both directions and measure the difference with a dial indicator), then I could measure the play/slop in the thrust bearing with a dial indicator on the end of the leadscrew....right? Because if this really is the main problem, yet it would be a no-brainer to try this first.

THanks,
-Neil.

[Zumba]

A dial indicator placed at the end of the leadscrew will tell you whether it's flopping about (when the direction changes). There should be no movement.

[ckm]

Isn't this something I can test for? I expect that if I do my regular backlash test (bring the spindle to zero position from both directions and measure the difference with a dial indicator), then I could measure the play/slop in the thrust bearing with a dial indicator on the end of the leadscrew....right? Because if this really is the main problem, yet it would be a no-brainer to try this first.

As far as I know (and that's not much...), you can test for backlash overall, but there is no way to seperate backlash from the nut vs the thrust bearing. You can try tightening down the bearing nut until it's almost unmovable and see if that eliminates most of the backlash, I guess.

All I know is that in my mill, the thrust bearing (or lack there of) was a major source of backlash.

Chris.

[NC Cams]

It may interest you to know that even preloaded ball screw support bearings fitted to GROUND Hiwin's will result in pretty sloppy backlash depending on the condition and tune of associated hardware.

Do a search for "EzTrak" on the Zone. In my earlier postings, I explain in chapter and verse how I went about methodically eliminating the slop from my mill. Yes, this is a bit different than yours but the methodology remains the same - get rid of sloppy, motion losing and/or stifling members.

First, you need to establish a budget - then you need to select the best parts that are within your budget and proceed methodically.

For ball screw thrust bearings, you want THE HIGHEST CONTACT ANGLE, HIGHEST PRELOAD PAIR OF BEARINGS YOU CAN AFFORD. These will pretty much eliminate the bearings as slop generators - period paragraph.

After that, you'll find that ball nut preloading adds a bit of backlash and/strangely enough, gibb tighness and/or looseness adds a bunch more - even with horribly expensive ball screw bearings and/or ground screws.

Getting to a thou or less is not that difficult. Getting to 0.0005" takes a lot of detailed work and good parts. Oversized balls may improve backlash potential on a rolled screw but it won't do anything to fix lead error - only a ground screw will do that.

Check with Hiwin as they also sell ground screw retrofits for many mills. Once you run/operate a mill with deadly smooth and backlash free lead/ball screws, you'll never want to run something sloppy again.

And don't forget to adjust the gibbs - you'll be amazed at what gibb tuning does for backlash elimination.

[cnczoner]

Still on this, though I've been away for a bit recently. I still have not checked the slop etc, and I need to re-build the power supply as the one in the controller got fried by a power outage/surge last week during a thunderstorm. But it was a cheezy 24V switcher and I had already ordered the parts to build a linear 70V, 10A supply.

I did setup an automated ebay search for ball screws and so far nothing in my range that I like. But it does help me identify brands that I can then research. So far, double nuts cost far far more than single nuts.

Zumba, you mentioned "about $12/foot for the screws, $120 per nut". What brand or source is this, if the nut really is preloaded?

Cheers,
-Neil.

[cnczoner]

BTW, is there a simple way to make a preloaded double-nut using 2 single nuts? Easiest design I can think of would be to have one with the RH flange threads and the other with LH flange threads, then have a F-F coupler with LH and RH threads which would be used to adjust the preload. But flange threads only seem to come in one direction. Other thought would be to have flanges on both nuts, then run two screws through them on either side (with the screw threaded on one, and free to slide on the other), and a spring between them on the screw. Any better/simpler/proven ideas?

[Zumba]

Neil, the last time I checked, Rockford sold their double preloaded nuts for around those prices, give or take. Nook might be marginally more expensive, but not much.

Yes you can make your own preloaded assembly using two single nuts, but you need two flanges. I make my own flanges so they cost me about 3 bucks in materials using A36 steel. If you buy them, they're like $30 and up. Ouch.

You don't need to mess with RH/LH. They should both be RH. The flanges are locked to the nut with a setscrew. In order to preload, one nut remains fixed (flange securely bolted) while the other is preloaded using disc springs of some sort. How much preload you need is up to trial and error. More preload = less chance of backlash from heavy loads, at the expensive of added friction. Calculate your loads to figure out just how much preload you need. If the combined load of acceleration + friction + cutting force is say 50 lbs at max speed, no need to add 200 lbs of preload.

Your thrust bearings should be preloaded the same amount.

[tpworks]

plan to make your own preloaded ball nuts
http://www.homecnc.info/ballnuts.htm

Tom

[harryn]

Interesting thread on ball screw preload and backlash