There are at least two ways of making parts that fit together exactly. The first way is the Cartesian way, and the machinists way. Imagine that you could specify the size of both the hole and the peg that you’d like to put into it. Just make sure your tolerances are tight enough, and call out a slightly wider hole. Heck, you can look up the type of fit you’d like in a table, and just specify that. The rest is a simple matter of machining the parts accurately to the right tolerances, and you’re done.
The machinist’s approach lives and dies on that last step — making the parts accurately fit the measure. Contrast the traditional woodworker’s method, or at least as it was taught to me, of just making the parts fit each other in the first place. This is the empirical way, the Aristotelian way if you will. You don’t really have to care if the two parts are exactly 30.000 mm wide, as long as they’re precisely the same length. And woodworkers have all sorts of clever tricks to make things the same, or make them fit, without measuring at all. Their methods are heavy on the jigs and the clever set-ups, and extraordinarily light on the calipers. To me, coming from a “measure carefully, and cut everything to measure” background, these ways of working were a revelation.
This ends up expressing perfectly the distinction between accuracy and precision. Sometimes you need to hit the numbers right on, and other times, you just need to get the parts to fit. And it’s useful to know which of these situations you’re actually in.
Of course, none of this is exclusive to metal or wood, and I’m actually mentioning it because I find myself using ideas that I learned in one context and applying them in the other. For instance, if you need sets of holes that match each other perfectly, whether in metal or wood, you get that precision for free by drilling through two sheets at one time, or by making a template — no measuring needed. Instead of measuring an exact distance from a feature, if all you care about is two offsets being the same, you can find a block of scrap with just about the right width, and use that to mark both distances. Is it exactly 1.000″ wide? Nope. But can you use this to mark identical locations? Yup.
You can make surprisingly round objects in wood by starting with a square, and then precisely marking the centers of the straight faces, and then cutting off the corners to get an octagon. Repeat with the centers and cutting until you can’t see the facets any more. Then hit it with sandpaper and you’re set. While this won’t make as controlled a diameter as would come off a metal lathe, you’d be surprised how well this works for making round sheet-aluminum circles when you don’t care so much about the diameter. And the file is really nothing other than the machinist’s sandpaper (or chisel?).
I’m not advocating one way of working over the other, but recognizing that there are two mindsets, and taking advantage of both. There’s a certain freedom that comes from the machinist’s method: if both parts are exactly 25.4 mm long, they’re both an accurate inch, and they’ll match each other. But if all you care about is precise matching, put them in the vise and cut them at the same time. Why do you bother with the calipers at all? Cut out the middle-man!
precision is fast, accuracy is portable
Origins of Precision, from the Machine Thinking channel
https://www.youtube.com/watch?v=gNRnrn5DE58
:o)
Measure twice, cut once, and sand one thousand times :)
This is true. :) Assuming that the other party is playing with the same accuracy benchmarks.
Noice!!
But what about lumberjacks?
They’re OK
They sleep all night and work all day.
I cut down trees, I eat my lunch
I go to the lavatory
On Wednesdays I go shoppin’
And have buttered scones for tea.
Wood is not a stable material. There’s not much room for precision. If the parts dont fit you hammer them down so they do. That’s why ikea furniture is a pita to assembe even though they are using precise and accurate cnc equipment.
Wonder if that depends upon the wood, not to mention any kind of treatments.
Somewhat, but changes in humidity and temperature will throw your measures off in any case.
Even when treated there’s some expansion and contraction in wood. It’s never perfectly stable.
Density and grain interlock will affect that although that can also be negative. More likely to split than swell with the weather if it’s too interlocked.
Metal has the benefit of mostly being affected only by temperature when it comes to fit.
Wood shrinks/swells depending on the humidity. Engineered wood is a bit more stable.
A good paint job helps.
Tell that to Japanese puzzlebox makers
The trick is to know that wood expands across the grain. If you try to bind something down by crossing the grain with two pieces, it either works the joints loose or cracks the board you’re trying to pin down. The puzzle boxes are built with careful attention to the grain direction, so the mechanism doesn’t bind.
QUOTE ” Wood is not a stable material. There’s not much room for precision.”
In 1944 I went to “nursery school” at the Gesell Institute at Yale. (https://www.gesell-yale.org/) They had great wooden toys. Sometimes they would sit down with you and ask you to do something like put different blocks through different shaped holes in a board. You had to get them lined up perfectly.
14 years later I had figured out what that was all about. And I met this guy who ran a precision machine operation in New Haven, drilling DEEP holes. Like aircraft machine gun barrels. In his home workshop he was making wooden blocks. Rock Maple, finished with beeswax. He trued each block up in a jig on a very fine sandpaper belt. Then he tested them under a dial indicator. He wanted them to be within a thousandth of an inch.
We talked about Yale/Gesell and he told me he sometimes made identical tests for some other universities. I said, “How do you make the holes in the board??” “”Ahh..”. he said, “THAT”s the hard part…”
Interesting. I attended something similar at the University of Saskatchewan when I was little. I can say it did influence my life path positively.
I remember talking about this to another machinist. He pointed out that making unique parts that fit each other is fine if you are making one-off prototypes. But if you are making multiple parts that need to match things made by other people, or if you are delivering parts that will be used by others in an assembly, you really have to work to dimensions — so it is good to discipline yourself to do so. The whole business of specifying tolerances for parts is a science in itself.
Metalworkers didn’t worry so much about accuracy and tolerances before the invention of mass production. If a skilled clockmaker is machining all the parts of a clock and assembling them one at a time, she only needs to make them fit each other. If you have a production line assembling a thousand clocks a day, the parts need to be interchangeable because there is no time for doing multiple trial assemblies and making tiny adjustments to every piece.
I’m all for precision and keeping in practice, but only when its practical to do so. The most important thing in any build/repair is that it works and you could make the part in whatever time frame you needed at a cost you can afford. To do real precision work often means custom fixtureing, and that both takes time, and costs in materials, where if you can tap it in well enough and just use your normal vice, the part might have taken longer in setup, but its done well enough and cheap…
Really doesn’t work so well in wood either, which is partly why you think of working to measures and tolerances as machinist not wood working, as well as why most ikea furniture isn’t really wood. In wood you will have to adapt to that annoying placed knot, or shuffle around to avoid a void or warped section in a key spot (also remember it changes volume considerably with moisture content)..
Its a natural product, full of ‘errors’ that make the 100% precision so any part from any run fits with any other in any order much harder to do well – Its not impossible, but suddenly you have to be very selective in species and cut of wood. Why lots of old wooden structures and benches etc use tapered pins or wedges (or metal mating surfaces), it lets you cut the important features as well as you can in the right sort of place and the wedge (or metal parts) will tighten up the imprecisions.
When I worked at Lawrence Livermore Labs, the on site craft folks were known to say:
“Machine to spec. Bang to fit.”
When doing carpentry, my Grandfather would say, “Dang! I cut it off twice, and it’s still too short!”
He cut of the wrong end.
Next time cut of the short end and not the long end.
Beat to fit, paint to match, and call it a day.
Plan to measure, hammer to fit, paint to cover.
And file and paint to cover.
Sounds like how the space orcs do it
You actually do see some of the “equal by shared setup” in normal precise machining; this is why good 1-2-3 blocks and parallels come as matched pairs. The dimensions are made to hopefully quite good tolerances already, but the pair was made together, grinding them both in the same setup, so that they are _equal_ to a much tighter degree than their overall dimensions. You then lean on those equal-by-construction dimensions when you use them to have less error buildup.
Wood has a tendency to be a little flexible, so for example when building wooden frames for walls for my hackspace with my father. We’d notice some of the straight pieces would be a bit bent, but it didn’t matter as once everything was screwed together it would then straighten / pull out the bent bits and make everything straight as the final result.
Metal on the other hand is very rigid by comparison, cast iron much more so that steel, which is why getting a perfect fit with metal is more important than wood. With wood you can just file a bit off more easily and it will shift a little bit over the year anyway as the humidity changes (sometimes noticeable with door frames if the gap around them isn’t big enough)
Cast iron is actually more flexible, but it’s more brittle so you don’t get to experience the effect. The Young’s modulus of cast iron falls between 1/2 to 3/4 that of carbon steel.
People often confuse brittleness with rigidity. Concrete for example is about as flexible as stiff wood, such as oak. It just breaks at any tiny amount of tension, which is why it has to be built with a large enough cross section that there is negligible bending – which creates the illusion that concrete itself is stiff, when it’s the structure that makes it.
Wow, that’s changed my view of concrete!
Indeed, I remember coming in my classroom one day in college, with all the tables and chairs put against the walls, and distinctly saw the floor lowering as we were walking in.
That is part of the difference. The other is that wood is much more affected by its environment. There is no point to working in thousandths with wood since it is moving constantly. It is also difficult to find two pieces of wood that would remain exactly in sync with each other in movement due to the variability of the material. Metals on the other hand remain more stable in size and in comparison to each other. Metal parts are very hard on each other if clearances are not correct. Another point is that metal parts are often made to be interchangeable with replacement parts, wood parts are generally built to only mate with each other and replacements would be custom made to fit. Interchangeable parts naturally require more careful consistent manufacturing processes.
Lol you talk about not having to “measure “ something and then in the next paragraph about creating a circle you say you “precisely marking the centers of the straight faces”
Dude just learn to measure
“precisely marking the centers of the straight faces”
Totally possible without quantifying the space. Center can be found with a compass.
Measuring is measuring, even though you don’t measure by numbers.
Finding the center of a face is at the same time measuring half its width.
Not necessarily. If you do it with something like a compass or gauge, yeah you have to set it to half the width before you mark it. But it’s common in woodworking to use a jig that has pegs on either side of the face of the workpiece, with the awl or pencil in the center. No measuring or setting required; just hold it at an angle with the pegs snug against either side, and the mark will be the center of the face.
To Kerry’s point, it is a perfectly reasonable technique and easier to get a round part than with other methods. Some of which are hard to do without a hole in the center (like using a pencil on a string or similar). Honestly making a circular part out of flat stock seems hard to do with measurements, at least if you care about roundness.
I think its a spectrum…..wood working, precise (*very* generally) means different things than to say, people making air bearings, means something different to those in metrology. By the same token, techniques vary depending on your tolerance. Drill a hole through multiple parts? How perpendicular is your spindle? How much flex does the bit have? It may not matter depending on your tolerances, but then again, maybe it does. So I really think this is a problem of knowing your tolerances, knowing techniques, and knowing which techniques fit which tolerances. If your tolerance is 0.005″ you can be pretty sloppy and still get 0.005″ tolerances. I know, I’ve experimented on a machine I can maintain 0.0002″ tolerance with some effort and a relatively low rejection rate. So with a 0.005″ tolerance, its pointless to slow down, clean everything to exacting precision between parts, make sure as many surfaces are machined in the same setup as possible, measure every finished surface and adjust offset for thermal growth, etc….But when your tolerance is 0.0002″, you may not be able to get away with drilling multiple parts at once (how perpendicular is your spindle?, drill bit flex/walk), using a vise stop (how much space is between your part and the stop? How square is the surface touching the part, how square is your stop, etc). Make your tolerances even smaller, and you may not be able to to grind in the same operation (wheel wear), etc. So it really is a spectrum. The tighter the tolerance, the more exacting you have to be. The looser the tolerances, the more you can use woodworking techniques.
Often many fail to catch the difference between accuracy and precision – many MCUs incorporate an internal temperature sensor. They are often very precise, easily reading to a tenth of a degree. But, simultaneously, if not calibrated, the temperature can be several degrees off. So, precisely inaccurate.
Make it accurate or make it adjustable. I make a lot of one-off machines using a bridgeport mill, metal lathes and woodworking tools. Digital readouts on everything so I have a shot of getting what I expect for parts. Still, adjustability is good when I have to change the machine for other use. Sometimes its just because the final use isn’t well defined. Currently making a Rose Engine Lathe using 6 steppers and a servo motor. Everything is adjustable.
Would love to see that here when possible!
Seconded- I’ve done work in metal on one, would like to make my own since the real ones are so expensive and rare, please post that project when you’re done.
Betting you already know of the Lindow Rose Engine Newsletter- if not, look it up.
I think we all want to see it! tips@hackaday.com when you get something down on paper. Or whatever passes for paper on the internets…
Are you making it compatible with the Holtzapffel books for those lathes? It’d be neat to be able to use the data from those. Dunno why Charles Babbage didn’t team up with Holtzapffel. They made precise and accurate interchangeable parts for their fancy turning lathes. Calculating engine parts should have been easily within their abilities.
I suggest you acquaint yourself with the 19th century American invention of interchangeable parts. Even in 1900 a *lot* of metalwork was done with spring calipers and machines without calibrated dials.
Parts were made to fit. That was especially the case in job shops where the majority of the work was repairs without the benefit of dimensional data other than the actual object.
Yep, if your making one offs precision is getting things to fit right, but if your making millions of things precision is making every part the same. In the 7 Years War if a Brown Bess broke it had to be rebuilt by a gunsmith, while in World War II an Indian Soldier could replace the bolt in an SMLE build in Britain with a bolt built in Australia and fire rounds made in Canada, both precision, both different.
I’m a life-long wood worker, CNCer and aspiring machinist. I think this missed a very important point – fit. Tolerances are a part of this but not the whole story. A lot of projects have parts that fit together. If you machine a 25.4 mm hole and a 25.4 mm peg, no matter what the material, they won’t fit together. At least not without a lot of force and maybe some cracking.
So how much smaller than the hole do you make the peg? Well, it depends on the material and the kind of fit you want. Take a look at recommended hole sizes for screws and bolts and you will see “tight fit” and “loose fit” recommendations which are dependent on the materials involved. Just about everything you make needs fit factored in – by hand, CNC machined, laser cut, 3D printed. I find this issue far bigger than precision or accuracy. You can make something very accurately but if you want it to be perfect, you need to get the correct fit.
I think that was covered in the linked article on fits and tolerances.
If you are referring to the writeup here and thr link to Wikipedia then his points still stand because the summary given does not capture the essence.
I was referring to “Know Your Fits and Tolerances,” Lewin Day, February 25, 2019. The link is in the first paragraph of this article. The video that article is about is very informative, has demonstrations, and even points you to industry standards for various types of fits.
Oops. I replied to the wrong post. You are right. This was supposed be about the accuracy and precision topic.
“You can make surprisingly round objects in wood by starting with a square, and then precisely marking the centers of the straight faces, and then cutting off the corners to get an octagon. Repeat with the centers and cutting until you can’t see the facets any more. Then hit it with sandpaper and you’re set. While this won’t make as controlled a diameter as would come off a metal lathe, you’d be surprised how well this works for making round sheet-aluminum circles when you don’t care so much about the diameter.”
A nice way of doing that is to rig up a shear for thinner or softer materials, so you can set a center pivot at the right radius from the blade, and just chop material off as you rotate the piece. Works for some kinds of saw too, and against a disk sander.
I’m afraid the real definition of ‘precision’ doesn’t fit the theme of this article.
To start, a ‘reading’ is a single test of some dimension or property. A ‘measurement’ is the average of a series of readings. One reading is only a ‘measurement’ in the trivial sense. A measuring system is ‘controlled’ if the average of the readings converges (always gets closer) to some value.
Using that framework, ‘accuracy’ describes how well the measurement converges to the real value of the thing being measured. It’s also called ‘trueness’.
‘Precision’ describes how quickly the measurement converges. You can also see it as a measure of variation from one reading to the next.. the measurement converges slower when the difference between readings is a large fraction of the average, and converges faster when the difference between readings is small fraction of the average. It’s also called ‘repeatability’.
So ‘precision’ only applies when you do the same thing multiple times.
The trim-to-fit process implies a noticeable variation from one fit to the next. That’s the definition of low precision. Trim-to-fit is also called ‘truing up’ or ‘making true’, which fits the alternate name for accuracy.
What we really have is a difference between consensus standards and ad-hoc standards.
Consensus standards are maintained internationally by organizations like NIST in the US. Their goal is to make sure a 1.000″ hole fits in a 1.002″ hole with the same amount of clearance, regardless of whether they’re made in the same shop or in two shops with no communication other than calibration trails to their national standards organizations.
Standards come in two general forms: artifact standards and process standards. Consensus artifact standards are things like the platinum-iridium bars that used to define the international meter, and the ones that still defines the international kilogram. Process standards are things that can be reproduced accurately, like the triple-point of water and the wavelength of a krypton-86 laser.
Ad-hoc artifact standards are objects used as a reference for comparison within a given job or shop: a carpenter’s story stick is a great example. If you watch Adam Savage work in his cave, you’ll see him create and use ad-hoc standards over and over.
The woodworker’s cut-and-fit process uses one part as the ad-hoc standard for the mating part. The part used as the standard doesn’t have to obey any conditions for size or uniformity, it just has to exist. The process of adjusting the mating part to fit the standard is one of improving accuracy: bringing the mating part into closer agreement with the standard.
Thanks for getting to the essence: the difference between precision and accuracy. To me, it’s understandable as shooting 100 times at a target. If 12 shots are to the core, and 88 scattered around the periphery, of which 40 are clustered very tightly in one edge.
Although 12 shots were accurately in the desired place, the shooter or the weapon need to recalibrate because the majority of shots were precisely in the wrong place.
Respectfully you are using a specific definition of “precision” from metrology. But the word has been around for hundreds of years (therefore longer than metrology), and that’s not the only valid meaning. Traditionally, in English it means “exactly or sharply defined.”
Outside of metrology, in the context of this article, it means that a machinist can make his parts accurately (with each part relatively close to the specified dimension—in other words, on average the machining process will be reasonably true to the specified dimension, which is necessary for interchangeable parts), and a woodworker can make his parts precisely (whether he even knows the dimensions or not, they fit precisely). The machinist probably also needs precision, depending on the required fit, but that’s not the point of the article.
If you’ve ever seen a hand cut dovetail you will hopefully understand why “precision” applies to it even though nobody is going to pretend it’s NIST traceable in any way. “Sharply defined” doesn’t even do it justice.
The dovetail is not made by first cutting the parts and then fitting them together, but first cutting one part, then using it as a standard for the matching part, then making the matching part a bit oversize and whittling it down until you can bang it in. Then you trim off the bits that stick out and plane or sand the remaining surfaces flush, so you’re actually forming both parts at the same time.
And, if you leave a tiny gap, you take a pinch of the sawdust and a dollop of glue, and mash it in the gap to hide your crime.
That’s also a good way to fix a squeaky floor… Carpenter’s glue, sawdust, and a thin wedge and hammer to work it deep into the crack.
As a long time – and pretty decent – woodworker turned machinist, I can confidently say that thinking like a machinist will make you better at making that dovetail drawer front. And a lot of other things too.
We, collectively, had to learn a lot to make modern machines and the parts we make on them. And that knowledge is applicable even if you’re using wood as your material. You spend a lot less time fussing and squaring and massaging if you machine your (wooden) parts accurately. And even if you can’t hold tight tolerances in your pine or mahogany like you can in a piece of brass, thinking about the deflection and rigidity and tool geometry you need to make a precise AND accurate part on your metal lathe or knee mill means you spend a lot less time sanding and shaving your wood parts to match each other.
Of course you also eventually lose the sense of shame that comes from sticking a piece of metal in your project instead of keeping it traditional, or clamping up a board in the mill vice to cut a mortise…. -shrug-
I think that flat pack furniture is rather relevant. One cannot produce that economically if the parts are not interchangeable. Except for the tolerances, there is *no* difference between metal and wood work in a production environment. Comparing artisan practice to mass production is specious. The material is irrelevant. The significant difference is the number of pieces to be made. James Krenov and Ikea are worlds apart. Both are good, but the methods, products and prices are *very* different
Meanwhile the olde tyme machinists are still fitting things with dykem and doing strange things with chisels and iron.
Ray’s rule of precision: Measure with a micrometer, mark with chalk, cut with an axe.
I’m installing kitchen cabinets. To drill holes for the handles I made a simple jig from a piece of wood that’s about 1/2″ by 3/4″ and two tapered shims glued and brad nailed on at a right angle.
Two of the small drawers had their fronts held on by a pair of screws exactly 3″ apart and halfway down the front. I drilled those through the drawer front and through the shims. With a center mark on the piece of wood for alignment I put a piece of painters tape on the top of a drawer front then measure and mark the center. Align the mark on the jig to the mark on the tape, then run the drill through the jig to mark the holes. Remove jig and finish the holes. The screws holding the handles reinforce the drawer front’s mounting to the drawer body.
For the door handles, which have to be in a corner, and vertical, I measured where I wanted the holes on one door. Next, I aligned the edge of one shim with a horizontal feature at the top of one left base cabinet door, then measured out where the holes would be. I drilled those with a tiny bit so they just leave a small mark. To finish I flipped the jig over to a right door, aligning the other shim to the same feature. Then I finished the jig by making holes for the right door handles.
Took me around 10 minutes to make the jig from scraps, and it’ll save a lot of time drilling all rest of the holes, plus ensure all the handles are properly positioned.
Cool story, bro.
This article caught my attention and speaking as both a master wooden puzzlemaker and a master CNC machinist I can say that my own take on this is that precision is how many places past the decimal point you can make a measurement whereas accuracy is how close to the nominal intended dimension you make a part. So for example being able to take very precise measurements can help you make a part very accurately yet you can also make accurate parts that “do the job” just fine without needing to be precise about it, as this author has demonstrated with examples. Here is another way to look at it: just because something was made to a precise dimension does not mean it is accurate, for example I could precision grind a rod to 1.0000 inch diameter +\- 0.0001 which would be a very precise grinding job, yet if the customer wanted 1.1250 inch diameter then the part would be totally inaccurate. Also another example to consider is that when machining threads (of almost any sort) the standard way to measure the threads is by using calibrated GO / NO-GO gages and “feeling” how well the go gage fits, as opposed to directly measuring the threads in some sort quantitative way, like for example by using the 3 gage pin and micrometer technique.
Key point – metal parts are often built to be interchangeable, wood parts are often not. Metal parts are often in much harsher wear environments where fit matters more. Metal parts are often used in motion applications, wood much less so. It is that simple.
Not necessarily. Lignum vitae has been used in industrial applications like hydroelectric plants as bearings and even in a nuclear submarine (USS Nautilus).
It’s a tropical hardwood with good wear resistance, high strength, extremely high density, and is naturally oily which provides a degree of self lubrication. There’s even bearings still in use today from back in the 70’s on turbines.
You can machine it to within a few tenths and will remain relatively dimensionally stable. That said it is a bastard to machine and avoid tearout. Since it’s been logged to all hell and back it’s CITES appx II restricted so don’t expect to find much on the market.
Good point.. but… just because those are the common applications does not mean that there are outlying scenarios such as for example making wooden puzzles which must have pieces which slide across each other with high degree of accuracy, and which must also be interchangeable. It is possible to “machine wood” the same way one works with metal, and it is possible to hold tolerances with (certain types of) wood that would meet “basic” precision specifications for many machined metal parts (i.e. +/- 0.005″)
Verawood is similar to Lignum vitae, but more readily available. It machines beautifully.
Padauk is also incredibly dimensionally stable.
I once worked on a GPS project with bad requirements. I knew your location to within the width of a hair, plus or minus a quarter mile.
:)
A friend of mine, fresh out of school, was hired to work in a metal shop
His job was to mill out some blocks of metal according to a blueprint he was given.
They needed 1000 of that part, in a given timeframe.
The mill was and old worn out thing, and he did his best, but sometimes a part was out of spec and had to be scrapped.
He had 2 pallets, one with scrap parts and one with good parts, one day he walked over to the next building where they mounted the parts he made, and discovered that none of the measurements were critical within 1mm or so, but the blueprint specifyed all tolerances to 1/10th of a mm.
He then walked over to the constructor (in house) and asked him why the tight tolerances, the answer was “it’s the default in the program I design in, I never change that”