New Machine Build- Accuracy/precision, etc.

[mhmslm]

Hi,
I'm new to this forum. I am going to build a CNC in the near future. I have a question, though. Before I get way ahead of myself in design I would like to hear some opinions about the minimum accuracy, etc. that would be required to build a good sounding hollow body instrument, such as a guitar, or mandolin.
For instance, how many thousandths of an inch is it ok for the table to sag under the weight of things while cutting? My first inclination is to use 80/20 extrusions for framing. But when I run my ideas through their Deflection Calculator program I can see that they flex more than I would have guessed. After fiddling around with different ways to construct things it seems I may have to just make a smaller machine than I wanted since size is somewhat less important than accuracy (to a certain point). I have many other projects in mind, some of them requiring a larger machine than mandolin would.
Anyway, your opinions will be appreciated as well as some ideas on how best to attain the accuracy needed.

Mike

[Cartierusm]

I'm not sure what others here will chime in with but I've been making guitars for 15+ years and just started using a cnc to do the work. I would shoot for .001 if you can. It doesn't seem to hard to be able to build within this tolerance, but .001 seems to be the base minimum for home built CNCs.

[BobF]

I have never built a guitar, but I have 3 friends who have and still do. One is a famous luthier and has been building guitars and mandolins for 30+ years. All of them build their instruments by hand. Woodworking typically has tolerances of 1/32 to 1/64 inch. Wood moves more than this with the moisture changes that take place constantly in wood. You are going to have some inaccuracies. Shoot for .001, but don't let the specs of some components get in your way. Typical ACME screws list .009 per foot as tolerance of the threads. This should not force you to go to higher precision because someone said you need .001. The guitar won't know, and I don't think most humans can achieve this, but plenty of them make fine guitars.

[ger21]

Originally Posted by BobF Woodworking typically has tolerances of 1/32 to 1/64 inch.

Try cutting dovetails or any other exposed joinery with those tolerances. I'd be embarrassed to show them to anyone. Imo, if you can see a gap, it's too big. And you can see a lot less than 1/64.

What you need to be most concerned with is repeatability, which is relatively easy to achieve with a homebuilt machine. As long as you can cut parts over and over and they all come out the same, then you can make parts that fit together tightly. Ideally, you don't want to be dealing with a sagging table. I'd recommend something that is fully supported so you don't have to worry about it.

[BobF]

Gerry,
You are certainly correct that you can see less than 1/64". I am not talking about 1/64" gaps, but absolute measurements. If you cut a row of dovetails by hand and they all fit as you expect, it is still possible and maybe likely that the width of the tails vary by 1/64". As long as the joints are tight, you may not see that the tails are not the same width, especially if the widest and narrowest are at opposite ends of the row.
I guess I need to choose my words more carefully.

[ger21]

I was going to elaborate and say what you just did, but I was too lazy.

[mhmslm]

Thanks for your input. I had been trying to stay around .001" but was not sure if the hassel was worth it. It just seemed that a bigger table needed bracing in places I would not be able to put it since my general planning called for a gantry that had a part that engaged the lead screw and passed back and forth under the table.
I did, however, come up with the thought of having the lengths of X and Y axes reversed. That would have that part of the gantry travelling the short distance (about 48") where I could get away with no extra bracing.
But now I wonder if such a gantry would be so wide it wouldn't roll back and forth without a tendency to rack. I could just try it, I guess, and if it doesn't work out do something else. That's got to be one of the greatest things about 80/20.
I read somewhere elso on this site that some recommend a drive motor on each side of wide gantries They say the software can easily keep the two motors synchronized. But I don't know about my board handling it. Seems the current would be a problem.
I bought a kit from Hobby CNC that included a 4 axis control board and 305 oz. steppers. Maybe those people have an opinion about the double motor approach.
By the way, one thing that never occurred to me, since I'm pretty much a novice, is the accuracy factors of the different lead screw types. If ACME threads claim .009"/ft I will really try hard to get ball screws, though they will most likely cost me much more. I just picked up a Z axis actuator off EBay that has ACME (or something else besides ball screw). But the overal travel on it is 8". And on any one project I expect the actual travel to be considerably less than that. The other axes (about 36" and 48" of travel each) will require a bigger chunk of change; if I can even get them somewhere. I have seen some on EBay, but they are all more than 2" Dia. I want 1/2" to 1", or so, I think. Anyway, thank you all for your advice.

Mike M

[BobF]

As Gerry said repeatability is more important than ultimate accuracy. .009 per foot may seem like a lot but its not. ACME srews like this do fine on many machines and cut lots of usable parts for guitars and lots of other things. Of course if you can afford ball screws and want them, go for it.
Look at Joes 4 x 4 hybrid to see an example of a machine with dual motors on the X axis.
Most machines I have seen have the long axis along the table and the shorter across the gantry. There must be a reason.

[ger21]

Keep in mind that .009 is worst case, and most likely they will be better. They can also vary from ft to ft. You might be +.003 at one ft and back to perfect the next foot. As long as everything is tight and you can get it consistent, I wouldn't be too concerned about it. Ballscrews have similar tolerances, but usually slighly better, around .005. You can get acme with better tolerances as well, but it'll cost as much as cheap ballscrews.

[spoiledbrat]

Maybe it is worth mentioning that sag is not the only enemy to the worktable. A helical cutter (upcut), while cutting, will pull the table up. I learned this because although my table was very rigid, my parts were getting strange tool marks in them until I enhanced my fixturing system.

[Geof]

Originally Posted by mhmslm ......a bigger table needed bracing in places I would not be able to put it since my general planning called for a gantry that had a part that engaged the lead screw and passed back and forth under the table. I did, however, come up with the thought of having the lengths of X and Y axes reversed. That would have that part of the gantry travelling the short distance (about 48") where I could get away with no extra bracing. But now I wonder if such a gantry would be so wide it wouldn't roll back and forth without a tendency to rack.... .....the accuracy factors of the different lead screw types. If ACME threads claim .009"/ft.... Mike M

Couple of comments:

Making the gantry long will make it prone to racking, and it will require extra bracing so it will be heavy; and remember it is the part that is moving.

Your idea of driving the gantry from a screw under the table is feasible, but it does need a lot of stiffening to the table. Actually if you think about it with this system the table has to be a stationary 'gantry' that spans the length of the ball screw. It will be much easier to build this than a long heavy moving gantry.

Regarding your queries about accuracy. Make sure you know the difference between accuracy and precision.

Precision is what you want. The ball screw you mention may be inaccurate to .009" per foot but if it has no backlash then that does not matter if all the mating parts are made on the same machine; they all have precisely the same inaccuracy.

Accuracy is very important if you are making parts that have to be assembled with parts made on another machine; then both machines need to be precise and accurate.

[mhmslm]

This forum serves to remind me that I'm not the clever person I used to think I was in my younger days. I have only been coming here a little while and already have learned a wealth of lessons, answers to problems and just plain better ways of doing things. And I would never have considered a cutting tool pulling the table up as it did its work. I would have been looking at my loused up engraving, or whatever, scratching my head wondering, "What just happened here?" Thanks for the warning.

BTW I went to a new surplus place a friend had shown me recently. I was looking for a ball screw and rails. The first thing I saw when I walked in the door was a matching pair of ball screws with 60" threads for very cheap. I laid claim to both of them immediately and now plan to drive my gantry with a screw on each side. Hopefully that will prevent racking. I guess the only concern there might be if one screw's motor should suddenly stop working and the other doesn't. Or maybe there's a software setup that can can shut things down in a hurry if that happens. I plan on going with Mack 3. Anybody using that program know if it will do that?

Mike