Ballscrew Accuracy vs. Repeatbility

[mesa5]

I'm working on an application involving ballscrews and I want to make sure my knowledge is up to snuff. I've been reading a lot about them, and I often see statements that say accuracy vs. repeatability requirements are often mis-understood, but don't provide a good context on why, and how you determine what your accuracy spec should be.

I understand the repeatability is how much the linear distance traveled can change when you try to return to a position you've already been to. In addition, I believe that I understand that the accuracy of the ballscrew refers to how much the pitch of the screw varies from the nominal value, due to the screw manufacturing procedure. For example, if the screw is a 5mm pitch, the accuracy spec allows you to receive a ballscrew that actually moves 4.999mm/rev, or 5.001mm/rev. Is that understanding correct?

In addition, I don't understand how you determine your ballscrew accuracy requirements. Is accuracy only critical in open-loop systems, since encoder feedback would monitor true travel distance and "self correct"? What are the circumstances in which a high accuracy ballscrew is necessary?

Finally, when the accuracy spec is called out as 5u/300mm, can the pitch vary within that 5u on each turn of the screw? For example, can the first turn be 4.995 mm/rev, the second turn be 5.005mm/rev, etc...?

 

[IRstuff]

Accuracy is application dependent. Since you have an application in mind, that application must have an accuracy requirement, regardless of how you solve your position problem. Inkjet printers typically must have accuracy/repeatability of better than ~0.001 inches, but integrated circuit photolithography systems need accuracy/repeatability of better than 5 nm.

5u/300mm is a runout spec, meaning that a full stroke travel of 300mm could be ±5um off at the end of the travel.

You should also be concerned about backlash, particularly if your repeatability requirement is stringent.

TTFN
faq731-376

Need help writing a question or understanding a reply? forum1529

 

[BrianE22]

For accuracy, some feedback systems use the motor rotation for feedback. For that you would want an accurate ballscrew. Other systems use feedback on the moving table itself. For those, accuracy is not as important in a closed loop system.

 

[mesa5]

@ IRstuff

I guess what I'm not understanding is why the accuracy matters if it remains constant. Can't the degree to which it is off be measured, and then have the software account for it? I.E. program a move of 5mm when in reality only 4mm is needed, since we know the ballscrew allways overtravels?

The source of our accuracy requirement is somewhat inventing numbers, so I'm just trying to educate my self on making a realistic determination.

@Brian

Yes true, didn't think of that.

 

[IRstuff]

Classic accuracy is not repeatable, in that sense. What you are describing is a scale factor error. And you can have convoluted scaling that changes as the measurands change. Optical encoders have that type of error, which is cyclical over the full circle. But, encoders also have random errors, due to noise, photolithographic errors, etc. These are not calibratible.

Backlash, for example, is bi-directional, and some people have to always arrive from one direction only, to mitigate backlash.

As to your requirement, you cannot derive the requirement from the candidate solution. A ball screw system can be coarse and inaccurate, because it only uses a micro-stepped stepping motor, or it can be highly accurate, because it uses an interferometer, or full-travel linear encoder, to measure absolute position. You can infer reasonable accuracy requirements from the application, but you cannot infer requirements for an application from what a ballscrew can, or can't, do.

TTFN
faq731-376

Need help writing a question or understanding a reply? forum1529

 

[Compositepro]

http://en.wikipedia.org/wiki/Accuracy_and_precision

 

[BrianE22]

Machine tools builders measure the accuracy of the table position and store the errors in a table which is then used to compensate the command (or feedback, I'm not sure which).

 

[IRstuff]

See

http://physics.nist.gov/cuu/Uncertainty/index.html

instead of Wikipedia.

TTFN
faq731-376

Need help writing a question or understanding a reply? forum1529

 

[BillPSU]

On machining two bores a specific distance apart programming changes can improve the accuracy however there are other thing besides ball screw accuracy which can diminish the accuracy of the machine.
1. Squareness of the axis to one another.
2. Tempearture of the slides/part/components.
3. Spindle runout.
4. Spindle rigidity.
5. Slide yaw and pitch as it moves.
6. Induced vibration.

On a lathe you are normally trying to maintain a radius being machined.
The same conditions apply but now throw in turning tool radius tolerance, dynamic axis flex/movement between the cutting tool/slide and the spindle.

Think about machining a radiused laser mirror to .0001" radius accuracy.

Stepper motors increments are usually .0001". You cannot hold the accuracy.

Think about using a machine tool in a temperature controlled room. Lubrication oil temperature controlled. Coolant temperature controlled.
Vibration isolation. Air spindle for rigidity and accuracy. Stepper motors with .00005 or less step capabilities. Controlled slip/stick motion control. No people allowed because of vibrations from talking and heat from the body. Tool radius control of 10-20 millionths.

These machhines currently exist but you better have a large budget. Could be military.

Bill