I'm wondering if anyone can give me the inside scoop on commodity- grade CNC machines, such as the CarveWright, CompuCarve, Rockler CNC Shark, and so on. What I'm interested in is precision joinery. The promotional videos for CarveWright have a lot to say about carving text into wood, and making cute angel designs and whatnot, but what I want to know is if it can cut a mortise to a depth of exactly 3 mm and a width of exactly 10 mm, exactly the same way on 10 identical pieces of wood.
You'll want to investigate accuracy/repeatability and programming capabilities for each machine you consider.
What does "exactly" mean? I bought a ShopBot that's good to about +/-
1/1000" and wasn't satisfied with that for joinery, so I built my own that's good to within 1/4800" (photos at link below), which seems adequate for the kinds of stuff I wanted to do.
What???? You're working with wood, right? Are you sure you didn't add a couple extra zeroes for effect? BTW: How exactly did you measure that? Did you use your optical comarator or your CMM?? : )
FYI - 1/4800" is .000208333 of an inch. If a standard sheet of notebook paper is approx. .003-4" thick, then you're claiming accuracy to within 1/20 the thickness of a sheet of paper!
Fairly tight tolerances, even in the aerospace industry...
I don't have the capability of accurately measuring it. It's pure calculation using 400 steps/revolution of the motors and 12 revolutions/inch for the leadscrew (with no detected/measurable backlash).
That's what my calculator says too. The calipers that I measure with give me readings in increments of 0.0005" and if I cut out a piece, for example, 1" long and measure, it consistantly comes out 1.0000" on the calipers - which only tells me that the actual part is between 1.0000" and 1.0004".
If I assume that the last step produced a complete increment of movement, then the maximum error is fairly close to being in the range [-0",+0.0004"). This isn't as good as the calculated value, but the results (whatever the measurement might really be) are likely to be at least some amount better than that maximum error and - bottom line - have been adequate for my purposes.
Even if the error range were to be the worst case [-0,+0.0004), it's still not too shabby for a wooden shop-built woodworking tool.
The aerospace guys have bigger budgets, more accurate tooling, and a lot more power available - but I doubt they have more fun with their stuff than I do with mine... :)
On Oct 23, 1:04=A0am, Morris Dovey wrote: [snipped context]
It is always about fun with you, isn't it...? (* Envy indicator LED's feathering*)
...and all *I* get to do, is build infra-structure components, such as air-lines, DC duct-work, coffee machines, JUST so I can hunker down for the winter and MAYbe enjoy myself. *S*
I'm still looking for a source for quality E25 collets. I found a few collections, but I really don't know much about their quality...
I'm not doubting your *mad* skills. And yes, I agree that this is "not too shabby for a wooden, shop-built woodworking tool." Neither am I questioning the veracity of your claim, or whether you've achieved this level of accuracy (I have no doubt you have). I'm simply wondering why you feel it's necessary to do so -- in light of the fact that we're all working in wood.
Let's not lose perspective here. We are talking about wood. WOOD! You, me, the OP, and anyone else we know -- we're NOT going to notice any appreciable difference between a joint's that's close to within .001 and one that's .000208. Now, if we were talking about metalworking.... well, that's a different story. But fact is, .005-.010 for a woodworking application, that's more than adequate -- even for the most discerning of woodworkers.
I try to make it so, even when the task at hand is distasteful, boring, or loaded with adversity. It's a decision, closely related to Roger Miller's observation that "you can be happy if you've a mind to."
Ok, since you called me out on this fun business - how about allowing yourself to anticipate the very real satisfaction of air-lines that don't deliver running water to your tools, a shop where sneezing and runny noses don't become an automatic response when you walk in, and a coffee spot that's so easy to clean up that the coffee even /tastes/ better...
I bought a set (#7-171-295) from
formatting link
near Detroit and added a 3/4" (emergency use only) collet from my spindle vendor. The set comes from Poland and is branded "Bison" - and I haven't had any problems with any of them so far.
OK, I don't know Morris personally. Never had a cup of coffee with the man. But I recognize some of the symptoms he has from the things people nag and tease me about.
I think he does it because he can. I don't know that this kind of accuracy was the actual intended end product, but his normal way or problem solving and personal standards led him to it.
His postings over the years have led me to believe woodworking is just one of the things he fusses over.
Why not wood as well? If that's his standard, or if it is just the unintended consequence of his personal quest to build a piece of woodworking machinery, why not? Notice, he didn't advocate this standard for everyone.
Absolutely no doubt. Remember, he is stating that is the result of his machinery output. He didn't say that was the mitered tolerances of a complimentary miter angle using crown molding the someone's living room.
On Wed, 22 Oct 2008 23:04:39 -0600, Morris Dovey wrote (in article ):
Stepper motors cannot be assumed accurate to the steps/rev level unless there is no load applied. Typically the error is a function of the shaft load, coil current accuracy, and mechanical accuracy of the motor construction.
Set a stepper motor to a given position/step, increase the load, and eventually it will 'jump' to the next stable pole position which depending on the motor design can be multiple steps away (7.5 degrees of shaft rotation is typical). Encoder feedback can help, but often the reason for choosing steppers in the first place is to eliminate the need for a servo system; the encoders only provide positional information.
That aside, your machine is way cool! You certainly know how to put together a system which does a fantastic job at minimal expense.
I wish I had the time to build something like that....
One of the things that I've enjoyed has been the design of 3D joints that don't require glue. I was upset to discover that +/- 1/1000" wasn't good enough in some instances, and that was what motivated me to build a machine just for joinery. The little JBot was a first attempt to meet that challenge (and I learned enough to build what, in my mind, would be a much better machine).
Baltic birch wasn't my first choice for the tool structure. It was simply what I could afford to buy and what I had tooling to work. My thought at the time was that if I could sell 10 of 'em at $2500, I'd be able to afford the tooling to build better (metal) machines. That didn't happen, but the wooden JBot does well enough for my explorations.
Ever tried stuffing a tenon that's 0.005" too big into a mortise that's
0.005" too small? Even with glue as a lubricant, it's problematic (and prone to splitting). Now consider the somewhat friendlier case where the tenon is 0.005" undersize and the mortise is 0.005" oversize. The latter works until you, as I did, decide that you want the joint to work without glue - it's prone to wobbling and racking.
It's less about being a discerning woodworker (which I'm really not) and more about achieving some fundamental mechanical objectives - and, as Robatoy kidded me about, having a bit of fun meeting the challenges. :)
Full agreement. Fortunately, I started out ignorant and (by accident MOL) bought controller and steppers that handle the loads I've handed 'em quite well.
It could be better. The minimal expense part was an inflexible requirement. What turned out to be most interesting was production of a design with a minimum number of critical dimensions - other than relative hole placements, there are /very/ few places where tolerances are tight - and yet the machine is (almost) unreasonably accurate.
If it looks like fun to you, then let me encourage you to make the time. I have to admit that I felt pretty good about myself when I put this one to work. :)
That could be fun, and (maybe?) within the shrinkage achievable with a fractional percentage change in moisture content. I wonder if they can be made to fit with a few passes of a hair dryer. Sorta like heating up a chunk of steel to allow insertion for an interference fit. Or the dried and compressed chips we call biscuits.
I have a fellow ShopBotter friend who kids me about "one hour" and "same day" joints because, if they're not assembled within that time window, they can't _ever_ be assembled.
One of the joints is designed so that /any/ change in moisture content of either or both members results in a tightening of the joint - and a heat gun would just aggravate the problem.
It probably sounds like a worthless preoccupation with precision until you play with the joined construction - and discover that the joint is the point of greatest strength _and_ that it's inherently rack-proof. :)
Not at all; I'm fascinated. Do you have examples posted somewhere? I have in mind the intricate joints in Japanese timber framing as a starting point. CNC guided cutters opens doors for shapes other than that of the cutting tool.
I did for a couple of years, but deleted most of my CNC-related pages a while back. I posted drawings (including those for the JBot) and most of the photos to ABPW - they may still be available on one archive or another.
I'll look around tonight to see if I have copies stashed anywhere...
from short-grain weakness. Easier to simply rout an X and fit splines, if one doesn't have the CNC capability (OK, splines are old fashioned, and there are fiddly bits of loose wood involved... no solution is perfect!)
HomeOwnersHub website is not affiliated with any of the manufacturers or service providers discussed here.
All logos and trade names are the property of their respective owners.