CNC questions

It's because they haven't made a good job of it !!!!

I doubt it

No, a stepper+ ballscrew is/will be far more accurate.

Most accurate will be a ballscrew + servomotor + feedback. Stepper motors can skip /miss pulses when cuts/feed rates are too high.. then you can watch your tool attack your chuck :)

Sorry, but you are entirely wrong about that - in terms of tool position, at least.

Over say 100mm an expensive good quality stepper/ballscrew combination will give a tool positional accuracy of perhaps 0.01mm or 10 microns at best - whereas a servo/scale combination will give an accuracy of as good as the scale is, which could easily be in the single micron range.

That's an order of magnitude better.

Maybe I wasn't clear - we, or rather the control electronics, measure the actual tool position at all times, and feed current to a motor so that the measured tool position is the same as the desired tool position. There is feedback, and a (servo)motor, just not an optical encoder - and thus any mechanical inaccuracies between the rotary optical encoder's idea of the tool position and tha actual tool position are eliminated.

I can't think of a more accurate way to work, other than measuring theworkpiece dimensions directly.

You have just answered your own question what good is it having a stepper ballscrew system that can't work to the accuracy of the scale. The scale tells it it's 2 microns out but it can't move?

Add to this you are further limited by mechanical accuracy that isn't as good as the smallest ballscrew stepper travel.

That's all real world results not teory.

John S.

?? I don't know what you mean.

If you are referring to my suggestion, then of course it can move, that how it works.

But this doesn't seem to be referring to anything I said, or anyone else said either.

However, my news server has been a bit hinky recently, and maybe you are replying to a post it and I have missed.

-- Peter Fairbrother

Sorry re-read your post and you do mention stepper / ballscrew and then servo and scale

I was replying to stepper / ballscrew / scale

OK now we are on the same page then yes it is a better idea. It is done in industry, not as much now as the servo's own encoder has got better. Heidenhain used to use this system together with some of the controller manufacturers.

Problems at hobby / low cost level is no one makes an affordable controller that takes back the encoder / scales signal.

Affordable servo drives read the encoder with closes the link between the motor and driver but not a closed link between the motor and controller.

John S.

ISTR that's exactly what the Prototrak CNC system does? In that it uses the DRO to provide feedback to the servos? Not that it is in any way hobby shop territory with the cost involved.

Peter

">>

You weren't clear.... maybe I wasn't either. I draw on my experience of having worked on some of the very first UK CNC machines fitted to Chuchill Batchmatic V Machines and Boehringer V series machines in Germany. We were the first company to pioneer optical tool tip radius measurement coupled with an automatic tool changer- designed for use with turning down titanium shafts for Rolls Royce jet engines and other applications in Boehringer. Then I worked for a shitty company that produced micky mous 'copy' lathes in Guildford.Forget their name now but they used steppers with no feedback and suffered endless tool to chuck crashes as inexperienced part programmers learnt their trade.... So I introduced a course step comparator/counter. If the stepper got out of sync by more than 1 rev of the ball screw, then the machine stopped......saved many a mash up :) They finally woke up to the idea that it was basically a crap system, but had sold shed loads as it was el cheapo .... Those were some of the best/most rewarding days of my life. hey ho.

The firm I used to design machines for used both stepper/ballscrew and servo/encoder or servo/resolver systems in various places. The steppers are limited to one step (or half step if multistepping) whereas the servos are limited by the scale. Servos also don't 'stall', steppers can and do and if it happens you may not know that it's lost its position till later when the workpiece is examined.

Siting the scale to ensure that mechanical distortions least affected the apparent reading was a carefully considered undertaking. There were no cutting forces involved as the 'cutter' was a laser, but holding 5um repeatability over 6m was challenging! Note _repeatability_, not absolute accuracy.

The resolvers were around £1500 each and the linear scales up to several £k. You can loose that in a half million pound machine, but it makes a hell of a dent in a £700 home mill hence the widespread preference for steppers.

Richard

The whole thing revolves around a cost performance issue and whether the obtainable lower cost solution is in front of mechanical limitations.

Non jargon speak means if the stepper resolution issue is greater than the machine ballscrew tolerances then go for that. Remember with either stepper of servo you still have the same ball screw errors and machine limitations. Often on the ower cost machines this is the stumbling block, not the method of driving.

What is attainable on paper often, no scrub that, usually, doesn't equate to real world solutions and all you finish up doing is an exercise on paper.

John S.

Actually Richard, stepper motors can be microstepped. Microstepping allows a single motor step to be divided into even hundreds of smaller steps. See this link:

Some stepper systems use microstepping to avoid resonance at certain step rates. Microstepping combined with software can result in extreme accuracy. See this link: Eric

Actually Eric, it is a big mistake to believe that just because the driver can microstep, that this will result in a corresponding increase in accuracy. It just doesn't work like that, UNLESS the torque generated by the motor is significantly larger than the torque it has to overcome in order to move the shaft it is attached to. For example, if there is significant "sticktion" to overcome before a leadscrew starts turning, then it may require several microsteps before the motor is generating enough force to start the screw turning.

For similar reasons, when you have actually managed to get the system to move, it can (and therefore will) overshoot when you decelerate it to a stop.

The relationship between the theoretical resolution of the stepper motor/microstepping driver on the one hand and the actual resolution of the complete system of motor/coupler/bearings/leadscrew/slide (or whatever) is not at all straightforward.

Regards, Tony

Tht's wot I said but I used smaller words...........

JS

Greetings Tony, You are correct and I realize the limitations of stepper systems. I have built some for my own use and built the mechanics of prototype and production units. One of the ways to overcome some of the mechanical limitations is with software and measurement. Measuring the actual motion and comparing to the commanded motion provides data that software can use to correct motion errors. The link I posted to the Nanomovers shows a system that uses laser interferometry to map the motion of a stepper driven leadscrew. This mapping allows the software to correct for mechanical position errors, such as leadscrew errors. Cheers, Eric