Official Luthiers Forum!

Guys I have to ask you a question. Do they use ball screws on those really large cnc machines....like the 12-200 ft long machines? No. Why, cost, and stiffness.

As far as accuracy, or back lash goes, I can attest to the fact that r&p machines can be very accurate. It matters a great deal how they are made and how they are used. I have tested my machine for backlash and accuracy different times to make sure, and it is as accurate as you can machine wood. I have a machine that the footprint is 8ft wide by 12ft long with a 6 in z axis, all of which use r&p. I can drill a hole anywhere on the table, run several programs, go right back and put a drill bit down that hole without even fuzzing the wood. How much more accurate do you need?

Now, if I lose power, or for some reason like a bit breaking etc, and I have to hit the home switches again, then I do have a problem going back to that hole. But is isn't the machine, it is the switches. The big problem I have with my machine is flex! The z axis is extended way out from x axis that carries it, and I cannot get a clean cut on the bottom of a pocket cut for anything. Part of that, might be r&p, but then again I have seen pocket cuts from high dollar ball screw machines that have little irregularites in the bottom, just not as much as from my machine.

Mike

Hi Rick - I'd love to take that off your hands, but I won't be starting this project for a while. If someone else can use the stuff more quickly than me, then please do. I will probably order a kit from Joe for my build, and will be using the leadscrews for the drive system. There's so much I don't know about it yest that I need to do some homework and study up on the nomenclature so I can make some good decisions on my build.

Thanks!

_________________
Mitch

Turmite you are also correct about the stiffness.....

Yes, if you want to run a machine with the dimensions you state, ball screws are not very effective, because they flex at some point do to their length, and the fact they are not supported over their entire length of travel. R&P machines, as you can see from various designs do not suffer from this, since the rack is supported along it's entire length, and hence there is no flex....

Everything reaches a limitation, and that is certainly true for ball screws. In addition, the screw tends to "whip" if it is long enough, and not stiff enough, given a certain distance, and a certain rotational speed (angular velocity/accelerations etc). This most likely will cause a stall, or worse, if using Servos as compared to steppers, and will make rapids and cutting speeds suffer. I guess if certain design considerations are taken, everything can be made to work, depending on the requirements for the task at hand.

Have you also considered Belt Drives? Some folks who want very large machines, use timing belts for their X/Y axis movements, and those designs are decent as well. If the belts are tensioned properly, there is little, if any, backlash.

However, for a machine about the size of the ones used for "smaller" jobs, and certainly those used for Guitars, I would never consider using anything but ball screws. Their benefits, accuracies, simplicity etc. outweigh their minuses....for larger machines off course, that is another story.

As far as accuracy, I agree with you as well...how much do you need for wood? I look at folks touting how they can get .001 accuracies out of a machine, and that is great, but wood will expand and contract with the weather well over that! So what is the point?

Do not get me wrong....I get great satisfaction when I cut a fretboard, given a particular scale, and then I take out a digital caliper, and measure the position of the cuts, and they are spot on. Same for placing a bridge.....but do I really care if the pocket I just cut if off by .001?? That is all I am saying....

I am sure whichever machine you decide to build, will be more than adequate...we might be splitting hairs....

_________________
-CyborgCNC
http://www.cncguitar.com
https://www.facebook.com/CncGuitar

Just a couple randoms:

1- Flex and stiffness problems aren't related to the drive, whether it's R&P or ballscrew, they're related to the rails and frame. The only exception is if the motors are undersized and the drive system is backdriveable, but this isn't the case most of the time.

2- There are issues with every kind of drive. Ballscrews are efficient by design, not accurate. Most routers are running crap ballscrews. An acme screw can be as accurate or moreso than a ballscrew, and I'm certain that there are lots of R&P systems out there which are much more accurate than budget ballscrews. I don't like the compromises inherent in either, so I'm going direct-drive on all axes on my next machine

3- Accuracy in wood matters to very few of the people machining it, but it does matter to some. It matters to Taylor when it comes to their neck joints. It matters to me in some of the projects I do, but that's mainly to do with inlay work or 'to be assembled immediately' joints.

4- If it's a personal machine (ie: it's there to be your shop helper) then go for accuracy over speed. Even a very slow CNC can make all the parts of a guitar that need be machined in the course of one day, and knowing those parts are all going to fit well and have a good finish saves a lot of time.

_________________
Bob Garrish
Former Canonized Purveyor of Fine CNC Luthier Services

Hi Cyborg,

I used the really large machine comparison to prove a point. R&P is used on lots of high dollar machines, and could as easily be used on smaller machines. Just like screws, r&p come in many different flavors. I would not hesitate to use r&p on a new machine, if I did not want to try the belts. Actually two belts per axis that engage each other with no backlash and tons of speed, but that brings us to Bob's comment about accuracy over speed, which I agree with.



Bob you going with linear motors?


Mike

First I must say thank you CyborgCnc for your backing of what I said ( Sorry about the spelling of your avatar in the previous post). I have learned a lot in the past month or so. That from many here. Now also I believe Mike has a point when it comes to LARGE machines, yes screws have a limit of size and length, and many Do Use R&P as well as Cogged/timed Belts. So I stand corrected in that. I do think for the home builder, ball screws are still the best for the dollar and ease of setup.

I also think more need to be aware of the inherent problems with using stepper motors and frequency resonance. While I chased a problem not knowing at first I now understand the problem and resolved the same to a reasonable level using proven designs.

Now I am working on a solution that resolves it completely. I'm sure that is not possible, but I can try. I have now researched and looked at/read 25 different patents on this matter. Most are the same with some improvement or difference to allow a new Patent because of design changes. Most are minimal changes just enough wording and such to allow someone to do it.

That's great and I commend them. Now you or I cannot make these available without paying a royalty to someone. Therefore unless you make it your self you cannot as no one makes them.

Well I hope to change that.

So If by chance you are having a resonace problem please PM me as I would like for you to test my design.

Mike

_________________
I'm outside looking in, just farther from the window than most.

Hey, somebody had to do it. The rotary axes might be geared, depending on the costs of torque motors, but they'll be using some form of zero or near zero backlash transmission.

_________________
Bob Garrish
Former Canonized Purveyor of Fine CNC Luthier Services

Bob,

WOW!

I had thought of a system like that, but I needed to do a lot more learning before tackling the task....hence I decided to go with something less "eloquent" as one might say....but non the less quite effective...

Have you looked at this?

http://www.baldormotion.com/products/li ... ts/lsg.asp

They seem to be providing the system already build for up to 78x60 inches of travel....I had started the research, but I never bothered finding out the cost, as it would have probably been out of my league....

Wondering, are you building your own from scratch? WOW again if you are! Talk about speeds with one of these puppies!

_________________
-CyborgCNC
http://www.cncguitar.com
https://www.facebook.com/CncGuitar

The Baldor with it's linear motors is exciting stuff!!

One thing I would like to mention as I don't believe it has been mentioned above:
The feedback method is probably as important or even more important than the drive method whether it's ballscrew, R & P, belts etc.
I guess that most CNC routers use rotary encoders on the screw or pinion to provide the feedback which is adequate is most cases.
Would like to suggest linear encoders as an alternative not knowing whether they are price competitive.
The beauty of a linear encoder is it takes out any ballscrew or R & P error thus allowing use of less expensive (less precise) drive components.
http://www.motioncontrol.com/index.cfm/Linear-Encoder/
http://www.newall.com/DROs/DSG_DMG.htm
The Newall linear encoder design has been around for a number of years and, I believe, uses precision balls placed side by side in a tube.
I think it's used also to drive DRO's on manual machines.
Interesting discussion guys! It usually gets down to cost if you're as poor as I.
Nelson

You can run dual-loop (or 'true closed loop') by using a shaft encoder for commutation and a linear encoder for position. There can still be error due to thermal change or mounting error on the linear encoder, but it certainly makes up for a lot of leadscrew error. That's what they do in really nice machines, and you can get it as an option on a lot of machines (Haas offers it as an option on most of their high-precision machines, for example)

Pre-built stages tend to have a very high premium compared to buying motors, rails, and encoders. On a machine, it's really best to mount the components directly to the frame rather than to eachother (saves space, gives you more choices). Buying components also gives you options on whose rails and encoders you use.

I'll be doing all of the non-component work. Motors, motion control cards, amps, encoders , etc will be bought but the control software, algorithms, and machine design are all on me. The budget is not very large, but I have access to some useful machinery and a lot of non-material resources that most don't. It's going to take the better part of two years of essentially full-time work (I'm submitting parts of the design for my masters in mechanical engineering).

_________________
Bob Garrish
Former Canonized Purveyor of Fine CNC Luthier Services

Hi Bob,

That sounds like a great plan. One thing I have been trying to research is the attachment of a rotary (can probably be linear as well) encoder to an existing machine for feedback. Through the feedback loop, one should be able to monitor the position of say a leadscrew.

Here is my challenge: The software. I am a user of Mach3, and I can find a plugin for it that will "stop" motion in case it finds any discrepancies between the leadscrew position data and the encoder data, but NONE that will correct it....

This might be different for servo machines, but it is certainly true for steppers.....

What software are you using Bob, that might be capable of this?

_________________
-CyborgCNC
http://www.cncguitar.com
https://www.facebook.com/CncGuitar

http://www.artsoftcontrols.com/docs/Mach3Mill_Install_Config.pdf

Mach3 has info on pg. 5-22 that might be of general interest regarding following error etc.

Nelson

Mach isn't a controller, really, more a front-end for controllers. The drivers and motion control hardware behind it are what you need to be dealing with directly if you really want to mess with stuff. In the case of steppers, you'll probably have to make custom software as they can't be controlled in a closed-loop manner the same way servos can with a PID loop. One of the hobby control companies was making some sort of semi-closed-loop stepper system, but it's more workaround than solution. I can't recall which it was at this point (for the price, it's easier to just get a servo instead)

I'm going to be programming the high-level software for the motion control board myself, which gives me direct access to the motion board parameters. The actual motion control boards have pretty full featured languages and command sets, and you can usually program them in C. It's not G-code easy, but it's not rocket science for sure.

_________________
Bob Garrish
Former Canonized Purveyor of Fine CNC Luthier Services

Bob in the sense that Mach provides the step and directions signals by using the timing in Windows, it is a controller. But in the sense of taking encoder signals from motor and linear encoders and doing something with them, it is not.

I don't know how far along with your build you are, and I don't know will be interested, and I don't know if it will be ready "again" anytime soon, but here is the name and the link to what is coming.

http://www.safeguardrobotics.com/defaul ... b=cncbrain

I have one and am waiting for the second generation software before I start my machine. All I want is stiffness, ATC, dual loop feedback and speed, where I can use it.

Mike

I used to be -really- excited about the CNCBrain...and then they dropped the ball so many times it's starting to look like they're dribbling. I think they were on to something great when they started, but others have made solid products to fill in that niche while they spun their wheels.

Check out the stuff by Dynomotion and Granite Devices. You can also get used Galil products that'll spank most other controllers, industrial and commercial, on eBay for a few hundred bucks. The Dynomotion is something to get seriously excited about

_________________
Bob Garrish
Former Canonized Purveyor of Fine CNC Luthier Services

Thanks for the post Bob. Where does Gecko fit in?

Mike--I'm a little confused about how Mach controls servo drives without receiving feedback from the encoders as there is no direct connection.
Can you explain how this all works?
Thanks
Nelson

Sheldon, Nelsen:

It's the difference between drives and amps (more or less). A 'real controller' takes the G-code, commands the motion, closes the loop, corrects motion, and sends signals through the amps to the motors. Essentially, the controller does 'everything' and the amps just amplify the voltage and amperage output to the level it can drive the motors.

A drive is an amp with some control logic in it. A drive closes the loop on its side, and therefore it can take things like step/dir commands from the 'controller'. Geckodrives fit into this category. A pure amp would just take a +-10V input and amplify it, rather than being able to accept things like step/dir commands or step/dir/velocity etc from a higher-level system.

So, essentially Mach controls drives but Mach can't control servos. A true controller can control servos directly, the amps (or drives) are just there to amplify the signal to run bigger servos or possibly take some of the work off the control (commutation, for example).

_________________
Bob Garrish
Former Canonized Purveyor of Fine CNC Luthier Services

So, one could say that those Geckos are pretty smart lizards, eh mate.
Sorry, couldn't stop myself from saying it.
Thanks for explaining it, Bob.

http://www.nexengroup.com/nexen/index.jsp
Ran across this in New Equipment Digest. Thought it might be of interest as a positioning alternative.
nelson

Here's a video for those with a fast web connection:
http://www.nexengroup.com/nexen/prod/prod_main.jsp/product?pn=966687&acc=Pinion_Preloader

Edit: Sorry, looks like one has to register to view the video

I'm curious how they're getting zero backlash.
You guys have any ideas?
Nelson

Nelson I have no idea, but here is another such app with same claim. I cannot for the life of me remember who makes this, but it could easily be diyed.

Mike

That is very interesting, have seen nothing like it.
Looks as if the "idle" upper and lower belts are in mesh which should stiffen the drive considerably.
I wonder if the cam follower bearings have eccentric shafts for adjustment?
Where do you find this stuff, Mike?

I've seen the belt-drive thing before. It doesn't suffer from 'mechanical' backlash if it's pre-loaded, but there's still the stretch and tension 'ripple' in the belts to deal with as well as the actual stiffness and fit of the 'idle' belt portions on either side.

The 'roller rack' system is probably pre-loaded from the top and shaped to have perfect contact through the motion of the roller. The actual backlash depends on the accuracy with which the rolling portion is machined, and if the preload is sufficient. It looks like any amount of wear instantly adds backlash to that system, though. It's essentially a 'perfectly fitted' R&P system with some wear compensation built into the pinion.

_________________
Bob Garrish
Former Canonized Purveyor of Fine CNC Luthier Services

Guys,

Spending a lot of time researching CNC lately and lurking around here. Appreciate being able to learn from all of your collective experience. Found a link to the Bell-Everman Servo Belt system mentioned above. There is a video of a prototype in action. Pretty impressive.

Kevin

http://www.bell-everman.com/servobelt.html