Ball screw vs. rack & pinion


Thread Starter


Hi all. I want to construct big 2D plotter (working area 1,5 x 2,5meters). It should be capable to do: linear, circular and spline contouring. I wanted to use ballscrew linear systems, but price was too high for demanded accuracy (aluminum profiles + linear modules + ball screws all pre-assembled). The next choice was rack & pinion, but there is question of clearance. So the questions are: Is it possible to compensate progressive clearance in rack & pinion by means of incremental linear encoder or LVDT? (position loop). What motion controller would You recommend: Bus based (PCI or ISA card) or standalone device for plasma cutting systems? Accuracy should be 0.1mm per axis. Regards, Ivan.

Cameron Anderson

I would say ballscrews are probably out too for what your trying to do. Maybe a ball screw Cartesian system, but might not be able to get any speed due to the length of the screws. How about Linear Servo Motors? Expensive too but they you have a linear encoder the whole length and very smooth velocities. Rack and Pinion, I guess, is the next best choice. I guess you could look at doing a belt drives depending on the loads and acceleration requirements. The bad thing about Rack and Pinions are that they are noisey and if you want the precision, you will have to pay the price for the components. I would suggest a magnetic linear encoder vs. a LVDT. On controls, I guess it all depends on what you need to interface with (DXFs, DWGs, TIFFS, G-Code, others). I have done similar applications and the best solution was a PC-based control card. Another was using a CAD-to-Motion software on a stand-alone controller. Here are a few site's I suggest looking at: PC-Based controls: <a href=>DSPCG</a> and others like Delta-Tau, Baldor (Optimized), MEI, Acroloop, and .... Stand Alone Controls <a href=>Emerson</a> <a href=>Superior</a> Take a look here for actuators and cartesian systems: <a href=>IDC</a> and Warner Electric has some too. Good luck and hope this helps Cameron Anderson Senior Motion Control Specialist
Have a look on low-stretch variety of timing belt. They cost significantly more than common/garden variety of timing belt but they are very good. They come in any length, accuracy should be sufficient for your purpose and overall design is likely to be substantially cheaper. Petr Petr Baum <[email protected]> --------------------------------------------------------------- Niksar Pty Ltd Unit 135/45 Gilby Rd, Mount Waverley, 3149 Phone: +61-3-9558 9924 Fax: +61-3-9558 9927

Charlie Griswold

In my experience the accuracy of a rack and pinion system is (+/- .005in) +/- .12mm over a travel of 3 ft or more. To my knowledge there is no way to compensate for this mechanical inaccuracy with controls. Linear encoders are expensive anyway, and can be easily damaged. I would look into a large variety of manufactures on ball screw products. NSK, THK, ACME....etc. A ball screw system can be bought as an off the shelf product with extrusion, yes, but you could also design your own. Buy the drive screw and coupling, motor, alignment coupling, bearings and shaft, but fabricate the rest. Most people do it this way. Design, component and assembly costs may work out cheaper than the actuator as a package. Charlie Griswold Automation Engineer SEQUENOM, INC.

Gene Wilkins

Balluff makes a very cost effective linear transducer that is sufficiently accurate for this application. (repeatable < 6um, resolution < 2um, hysteresis < 4um) probable approx. $800. US. Check it out at Gene Wilkins
Ivan, Here are some cool links for what you want to do, including freeware and shareware G-code interpreters: Also, check out . They have a lot of surplus stuff that can be useful, especially if you're building a one-off design. I got a couple of 60" Thompson 1" diameter lead screws with heads for less than $100 each back in 1989, and I think they still have stuff like this. Their website has a lot of pages that are "Under Construction", but they have a catalog in print also. You can also call or e-mail them to see if they one or two of what you need. Usually the cost for these items is around ten cents on the dinar, so you can save a lot of money even if it has to ship a long distance. Regards, Willy Smith Numatics, Inc. Costa Rica
Balluff update rates (too slow) for any stroke length beyond about 30 inches preclude use for dynamic motion feedback. Ken Brown

Gene Wilkins

Hi Ken, The SSI version of the Balluff linear transducer has an update time of .5ms regardless of length. Gene
Dear Cameron, Regarding magnetic linear encoder: Since my application could be related to the plasma cutting process, I thought that some interferences might occurred with magnetic sensors. I spot, but late, that LVDT would be wrong solution. I understood , from Your message, that rack and pinion could be used for my accuracy with external encoder. After examination of European's manufacturers of belt drives (ISEL, Rexroth and some others) I realized that their linear modules belt driven provide accuracy of +/- 0,2mm. That's why I didn't mentioned belt drives. Regards, Ivan.
Hi Gene, This is news to me. We helped a customer develop a predictive rate based feedback algorithm to deal with the latency of the Balluff transducers (quadrature output). The output was updated very quickly, but the source of the data was a magnetorestrictive device with a signal propagation coefficient of about 10usec/inch. So, while the output was fast, it was actually a *filtered* output with delays associated with the velocity of the moving magnet along the sensor tube. Ken Brown
Charlie, Time to update your technology. Helical rack and pinion with magneto-restrictive linear encoders make an accurate and nearly indestructable positioning system. Couple it with a UMAC or PMAC motion controller from Delta Tau and accuracies required for this application are easily achieved. Rack and pinions do suffer from backlash, but the DT controllers do a great job using backlash compensation. If you really want accuracy and can spend a little more money, anti-backlash drivetrains are not that tough to build. Ken Brown Applied Motion Systems, Inc.
In the 'thought you'd like to know dept.' Magneto-restrictive a good solution in many cases. I have used many of them. But I do not prefer them on long lengths. I am replacing them on an application because of poor update time. The machine designer did not take into account the update time of the transducer. Transducer length is 144" The particular press application has high vibration and is a nasty environment. I am going to Spherosyn Linear Transducers Lifetime scale warrenty & rated IP67 Priced in the range around linear scales David Kane Kane Engineering Group Inc.