Backlash – Steppers, Toothed Belts & Ballscrews

[TrevorGParticipant @trevorg]

[TrevorGParticipant @trevorg]

I am putting together a shopping list for my first pass at CNC. A Mill conversion.

 

Backlash is an issue I need to understand before spending money on parts. Assuming I install a ballscrew I next have to attach a stepper. The usual method seems to be with a flexible coupling of some kind this leaves the stepper sticking out from the machine. Which I am not happy about. An alternative would appear to be toothed belts.

 

Before going further I need to understand how much backlash is tolerable and what backlash different components will contribute. It is quite common to find suppliers refer to “low backlash” and “zero backlash” ( I suspect zero to be a non-zero number in this instance!). But it is very hard to find actual figures.

 

1) Ball screws – I have only found one figure 50um. Which sounds a lot.

2) Flexible couplings and Oldham couplings quoted as zero backlash. Are they ?

3) Toothed belts – steel reinforced T5 sounds promising but how much backlash?

 

If I can get some numbers and set a target then I can get on with the design.

 

Trevor

[NobbyParticipant @nobby]

Hi

For ref .Most C.N.C milling machines are fitted with ballsrews . This enables you to climb mill for better clean cutting . I dont know if this helps ?

 

Regards Nobby

[elanmanParticipant @elanman]

Trevor,

I have converted a Warco AIS mill using steppers, toothed belts and the original lead screws. The belts seem to have no backlash but I can see they will wear quite quickly and so will need changing from time to time. The screws work well but you have to set parameter to allow for the backlash in the CNC software. I am using Turbocnc.

Nobby is correct in that ball screws are the best way to go, just depends on the cash!

Cheers

John

[PekkaNFParticipant @pekkanf]

First question is that does ballscrew has low enough lead and stepper (and controller) enough grunt to move that machine? Most of the time you will need gearing to gain enough power out of stepper. This will sacrifice speed, but it is evitable. Z-axis will typically exhaust any reasonable size stepper without gearing, You might think of using counterweight, but that would add more inertia and limit acceleration.

 

Good couplings and toothed belt drives are relatively rigid. There are straight below or spiral cut couolings that are essentially zero backlash and rigid. Ball screw bearings and their mounting to the frame is very important.

 

Limitting factors for a accuracy stability It’s not only apparent play, but least equally flex, due to mounting and superstructure.

 

Also one important point is to understand that any manual mill is not made for CNC. Besides acme screw, least equally limiting factor is lack of linear bearings. Go easy. Go slow. And lube slide ways well.

 

PekkaNF

[Billy MillsParticipant @billymills]

Well how long is a bit of string? or how much money have you got?. Ballscrews generally have less backlash and can wear at a lower rate but it depends on their construction i.e. cost. Cheaper rolled screws are formed by rollers under extreme presssure, More expensive screws are ground and more accurate. If you don’t keep the junk out of the nut it can wear although they don’t seem to jam. The manufacturer can then fit different size balls to tighten it back up if local screw wear has not happened.

 

The Acme thread is a lot less efficient -harder to turn – but the nut should be adjustable to remove the backlash. Real issue is that after some time the screw wears in the most used parts then you can’t have uniform nut tightness along the whole thread. A CNC machine tends to have a lot more action on the leadscrews than a manual machine so some sort of pressurised central lubrication system is very desirable -even essential with acme screws.

 

Many commercial machines use belts to connect the screw to motor, the belts are not a problem, with home use you may not need to ever change the belt. Some designers use the belt drive to change the torque/ speed/ step size or to tuck the motor underneath.

 

Oldhams can wear, the bellows coupler can be better long term. The coupler tends to be longer than the belt solution so the motor sticks out more. Some use rigid or no couplers, nipping the stepper flange bolts up last and praying that the screw and motor end up on the same axis.

 

Software can compensate the average backlash, a good test is a nest of circles stepped by 0.1″ in alternate cutting directions or a square at 45 degrees to the X axis.

 

So how much backlash ? Well the stiffness of the machine is another big limiting factor. Given perfect ways and gibbs with great screws, if the machine is bendy then climb milling may end up badly.

 

You could fit the drives to the existing screws and add one shot oiling and linish the gibs then see how it runs. If or when screw wear is an issue then you can always fit ballscrews.

 

Regards,

Alan.

[Anonymous]

Bear in mind that most commercial CNC machines use servo motors in a closed loop system, ie, the position of the ways are measured independently. Hence any play in the drive belt system is of less importance. Most hobby style CNC machines use stepper motors in an open loop system, ie, the ways are assumed to be in a certain position from knowing the number of steps made and the mechanical gearing. In this case play anywhere in the system increases the positional uncertainty.

 

A good start would be decide what you wish to make, and hence what positional accuracy you can live with.

 

Regards,

 

Andrew

[TrevorGParticipant @trevorg]

Thanks for your replies. My decision before asking my question had been to replace the ACME X & Y with ballscrews.

 

The Z axis I had decided would initially be left as an ACME. Replacing with a ballscrew would be very difficult. Due to the weight of the knee backlash is unlikely to be a problem.

 

Reading your replies that still seems to be a good plan. I will double check the lubrication systems and ansure everything is free running. Being an old machine it is heavily built so I am not concerned about stiffness.

 

The point about ground versus rolled ballscrews was one I had nor previously picked up on.

 

You have helped me make the decison to use toothed belt drives. I felt it was the right way to go but the mill conversions I have seen in articles have gone with a direct drive. I wondered if there was something I was missing. Apparently not. The ability to trade torque for speed is a welcome bonus.

 

This leaves the issue of the acceptable value of backlash. I would still like some target value so I can check if my conversion is heading for trouble.

[MICHAEL WILLIAMSParticipant @michaelwilliams41215]

The amount of backlash in a CNC milling machine directly controls the limit of accuracy and finish which you can achieve . So to arrive at a numerical value for acceptable backlash decide first what level of finish and accuracy you want on your milled parts and work backwards . For instance if you decided that an accuracy of +/- 10 microns was acceptable then your backlash at the table needs to be something less – ideally an order of magnitude less but divide by two seems to work in most cases – so for +/- 10 microns accuracy you need +/- 5 microns maximum backlash .

 

 

Backlash at the table is a combination the direct backlash in the drive screw , nut(s) and any bearings and of the backlash in the rest of the drive eg pulleys , belts and couplings .

 

 

Remember the primary consideration of what accuracy you want to achieve also controls the more basic decisions of what table movement you want per input step from the motor and what level of pitch accuracy you need in the feedscrews .

Edited By MICHAEL WILLIAMS on 28/01/2011 10:59:03

Edited By MICHAEL WILLIAMS on 28/01/2011 11:04:35

[David ColwillParticipant @davidcolwill19261]

I have converted my bridgeport clone using ballscrews, steppers and toothed belts. I paid extra for double ballnuts which allowed me to reduce backlash to about .03mm (total including endfloat in the screws etc). The software is set to compensate for this.

My mill is fitted with a newall dro which allowed me to make comparisons between where the software (mach 3) thought the machine was and where the machine actually was. After running the machine for an hour the x axis was .02 out and the y .04 out. This is more than adequate for my purposes .

[duncan webster 1Participant @duncanwebster1]

I’m looking into putting a power feed on the X axis of my Centec 2B. Not CNC yet, but might as well go down the right road. I want to drive the leadscrew via a toothed belt from a stepper motor. The drive will incorporate a clutch so I can disengage for hand feeding. I built a motor drive set for my last miller, a Naerok, using a 12v DC motor which was successful eventually, but I had to fit feedback to keep the motor speed constant. Stepper will be easier and upgradeable. My old machine needed 55N.cm on the leadscrew to give a decent cutting performance, but had a 0.2″ pitch leadscrew. I would think the Centec needs less as less pitch screw. (0.1″).

 

Question is, what size motor do I need and what reduction ratio?. I need a rapid traverse speed of about 200 rpm. I can sort out the electronics I think.

[David Robinson 2Participant @davidrobinson2]

I converted my old Henry Milnes vertical mill last year. For those unfamiliar with it it is a very hefty lump. I used ballscrews and a two to one drive x and y from steppers and a bigger stepper with a three to one

to drive the existing acme thread of the knee and carriage on the basis that if I could get backlash lifting going on for a quarter of a ton of knee and carriage then I must be taking too big a cut!

The speed reduction obviously affects the feed and movement speeds but I find it more than adequate for me to keep up with. The ballscrews seem to have totally got rid of bachlash.

Dave Robinson.

[john kennedy 1Participant @johnkennedy1]

David , I used to have a Henry Milnes so I know how heavy they are. Now have a Harrison,which is a bit lighter. The leadscrews are very worn now and I'm going to have to replace them and new nuts.

The X axis traverse is driven by a separate motor which,through gearing,turns the nut while the screw is locked. … I 'm not sure if that can be replaced so thought of ballscrew and nut with a stepper motor to drive it ??? Sort of CNC in stages.

Everyone seems quite shy when it comes to motor sizes. Can you tell me what size you used ?

some pics would be nice, Thanks, John

[TrevorGParticipant @trevorg]

Despite buying the parts in 2012 I only found the time to do the conversion this year. The toothed belt drives have worked well (1:1.67) and do not seem to contribute anything I can measure to the backlash.

The Y has an overall backlash of <0.01mm. X is giving a little trouble and backlash is 0.12mm. I think this is down to the preloaded bearing mount. The design of which was determined partly by the machines 1930's origin and partly due to a previous owner who appears to have used a large hammer to smash part of a casting to increase travel! I feel a redesign coming on.

So overall I need not have worried ballscrew+toothed belt+3Nm stepper was a good choice.

The Z is another matter I used a direct drive 11.5Nm stepper. Power is marginal on the ACME . While it runs at speed when the speed is reduced it bangs and bumps and goes nowhere. But the solution will be OT for this thread .

The mill is a CVA 79 and while I don't know the weight of the knee I suspect it is not far short of the Henry Milnes 1/4 ton mentioned above.

Trevor

[BowberParticipant @bowber]

I have an old Denford Ezymill and the steppers are connected to the lead screws by belt, seems to be fine and it was made in the 80's as far as I know. I'm not running it very fast though and it can climb mill no problem.

Steve

[Hugh AllenParticipant @hughallen31308]

I am about to buy a Centec 2B. The backlash is rfeported as .004". I realise that some of this may be reduced by playing with the lead-screw nut, but can anyone tell me what you'd consider norm al or acceptable for this amount of backlash? A DRO takes care of a lot of the problem, but I think that unusually large backlash would be quite irritating and would limit climb milling options

Hugh

[Hugh AllenParticipant @hughallen31308]

Correction chaps. The backlash isn't 0.004 but 0.04" Seems like a lot to me but I'ved always worked on more or less new Bridgeports, and my Emco FB2 (virtually unused since new) has about 0.002, so I've been spoiled.

[duncan webster 1Participant @duncanwebster1]

I'd be ecstatic if mine was that low. I don't have any problems, but get involved with climb milling apart from just running back at the same setting to improve the finish, I think that is best left to big industrial machines

[duncan webster 1Participant @duncanwebster1]

Your last post beat my reply, 0.04" is a lot. If you can't adjust it out then it's new leadscrew and/or nut time. I've done away with the fast feed pinion on mine, if I were replacing the nut I'd just have it solid, permanently engaged.

[Hugh AllenParticipant @hughallen31308]

Probably sound advice Duncan, since in any case the fast feed capstan gets in the way of things and was deysigned for production, which isn't what I'll be doing. Many thanks. This forum is very helpful.

[Author]

Posts