Anti-Backlash

"What is the best anti-backlash nut to use."

CR4 member Walks in Storms comes to mind.

^{Yeah, I know... bit it's funny you have to admit.}

Made me laugh

Hey look

WWL came back for a minute in June

A double nut with axial spring pre-load. One half should not rotate with respect to the other (key in between). If you give input about what you found I can give you hints.

You easily can realize in programming an old mechanic's trick to deal with worn out acme screws. At every bite, first back out a bit, take up the slack, and advance a bit. Repeat. Problem solved.

http://www.universal-thread.com/

Backlash is one issue, stability of backlash another and low and unpredictable axial stiffness a third one.

And: there seems to be no screw without an once per revolution axial error, that causes the nut to tumble and either be loose or transmit the tumbling as an error to the driven carriage.

Most of these are successfully tackled by the unique spindle-nut combinations of the above mentioned company. This is very likely out of range of permissible cost but have a look.

I have only one linear drive that is better: from a very old Moore hard disc turning machine, built near 1984 to produce 10" ? hard discs. These are above 200Kg weight per axis. But still have superb quality: straightness well below 1µm, 1n error 0.3µm, backlash (after replacing the axial roller bearing by a sliding bronze bearing) near 2µm. Backlash is here from axial stiffness and friction of heavy carriage on slide!

RHABE

Well I guess you know you will have anti-backlash compensation in your software, and that there are some great build your own cnc forae out there, but I am curious about which ab nut you have on the x and y, and how that decision was made, and why that don't help you with this question.

CNC

If Your looking for precision You need a ball screw and nut.

oilcan13

GA Oilcan

In addition to #5 and #6, you can put into software a calibration routine, to map backlash / slack, and out of round and similar problems. All it takes patience, a good gauge and entering the reading at every stop. It is running against the grain for old machinists, but all it does semiatomatically, what they do after a lot of learning.

A calibration run - on any of the x, y, z axis - can make such errors reduced to a few percentage of the actual mechanical error.

For the Z axis, with its high force, finer movement, like recirculating ball bearing, is inappropriate.

On the other hand, does the OP really needs precision on the scale of 0.001" or a few microns? Really??

I'm not sure to what precision ronseto needs, but he makes some pretty nice scale model trains. I would think it is important to his projects if he is asking our advice.

"For the Z axis, with its high force, finer movement, like recirculating ball bearing, is inappropriate"

Do you mean a ball screw is inappropriate for a CNC z axis?

That's going to shock a lot of mill makers

This is my first attempt into the world of CNC so I have a lot to learn. I started with a basic plan; see here http://buildyourcnc.com/default.aspx. I have redesigned it to suit my requirements. I've looked at anti-backlash nuts from Berg and Small Parts Inc. One is called a trantorque, one uses two nuts and a spring and one is a six beam flexible coupling. My plan is to get a prototype model up and working. From there I want to evaluate it as to accuracy and structural stability and make modifications as needed. The basic machine uses a wood router for cutting wood, but my ultimate goal is to work in Aluminum. 0.001 tolerance would be nice, but I will settle for whatever I can get at this point. I'm thinking I can slow down the reversal rate with the software so as not to overtorque the connections. The learning curve for me will be a long one, but I love challenges. Except for the basics I learned in college, my engineering knowledge has been mostly self-taught. CNC will noy be any different. I'm also browsing CNC websites for ideas.

The ball and screw device is too expensive at the moment.

Google has many CNC forums.

Also look here:-

http://www.cnczone.com/

Please post photos of the couplings,

these couplings may be very important to achieve accuracy. I designed some myself so I am interested in these.

Then think about rigidity of your machine: how much deformation by machining forces and vibrations?

RHABE

Here are two anti-backlash nuts I've found.

http://buildyourcnc.com/CNCMachineMechanicalParts.aspx

http://www.smallparts.com/kerk-motion-anti-backlash-diameter-efficient/dp/B000FSJZ1M/ref=sr_1_3?sr=1-3&qid=1320697930

I have the feeling that the 1st one is a double nut with a thread between the 2 parts so that the spring makes one part advance with respect to the other thus making the backlash nil. The second seems to be a cut nut which is tightened by a conical sleeve under the spring force on the threads flanks (as an edge pushed radilly in the thread). Both compensate for wear as well of the thread as of the nut. I have the feeling that nuts will wear more than than threaded shaft. The problem is a bit more complex since the wear will be proportional to the applied pressure and to the sliding speed. The other problem you are confronted is the axial force on the flanks due to the spring pre-load and the axial force ecountered when the shaft pushes or pulls the nut (if the shaft is the active part). The spring force must be greater to avoid any relative movement between the two parts. As well important is the axial nut stiffness since it will influence the behaviour during accelerating (positive or negative) periods and also the position accuracy under load. It is possible that the second one could have a higher axial stiffness since it has no motion between two parts.

Thanks all. I guess I will have to experiment, using what others are using. I wish I could afford the ball-screw ones.

If have access to and can use a lathe, you can make your own in Delrin.

There is even a "quick & dirty" method where you take a long piece of Delrin, drill a hole the size of the inner thickness of the thread (without grooves), cut it into two parts lengthways along the hole, heat the thread up after coating it in silicon oil, then press the Delrin around it to take the shape with a vice, leave till cold....cut up into generous lengths for usage.....

Works really good I am told.

Now that's what I like to hear. I do have all the machines, lathe, mill, etc; more in keeping with my hands-on approach to projects. It won't make anyone rich if I make all the parts, I can myself. If it's within my capability and equipment, I will do it myself. I draw the line at nuts and bolts.

At last i've got back on. Been stuck at 'loading editor contents' for a week.

If you're up for making them, there are plans and descriptions at various places. Have you looked at artsoft's (Mach3) forum? a very friendly and helpful bunch.

basically you can take a threaded object (don't forget you need to fasten your carriage to something) slice it mostly through at 90 to the axis, then drill and tap for a small grub screw through one of the pieces to poke out against the other, so forcing it slightly apart. This is the way to take out the slack

I know two guys who went for making their first in wood, and their second in aluminum, using the first to make the second, and using all the 'I wish I had.., I wish I hadn't's... as well.

cnc

Many of the older computer printers used 6mm Ø ball screws. You might find something chunky enough in 'plan printer' sized machines.

To pour some cool water on the discussion. Please do not forget Young's modulus. That is the measure of the flexibility of the matter. Plastic, any plastic quite low. Steel, brass quite high. It describe the "give" when mechanically loaded. No load, no problem.

Handling of dust and such, excellent.