On Tolerances and Daftness in Woodworking...

[ian maybury]

To throw one out for discussion. I've just got off the telephone from a retired toolmaker friend, having sat through ten minutes of his berating me to the effect that I'm mad for talking in terms of tight tolerances in woodworking, and in woodworking machines and equipment. Hand plane sole out of flat? No problem - just push harder on it. Cabinet not square - just clamp it up and nobody will notice.

The same argument breaks out regularly on forums too - where the fact that wood moves considerably with changes in humidity is pointed out, and the argument put forward to the effect that it's daft to try to set machines up to fine tolerances. Machine manufacturer under pressure because his tables are not flat? for sure that one will be trotted out. (along with the claims of high precision in the advertising blurb)

I can't claim any great expertise in this regard, but my personal take on this sort of thing is that it all depends. If your hand plane is not properly flat and otherwise sharpened and set up it's going to drive you nuts. No, it normally doesn't matter if the first piece you cut for your cabinet is 3mm longer than intended, but you'd better get the other three dead nuts the same. Better not be 3mm too long either if it's got to fit in a space exactly. Back of your chisel not flat - no problem if you're using it to open paint tins, but if you want to pare accurately?

Ditto your jointer. 0.005in isn't a lot for a cast iron table to be out of flat, but find that in the form of a hump in the wrong place in the outfeed table and try to get it jointing straight. Then try clamping up the result maybe 5x 1mm gaps at the ends of those 1.5m boards you are gluing together to make a table top.

0.005in cup in your fence? Tough luck (see below) - don't tell me that 0.005in gap in the joint when a cross member butted against a leg (say a mortice and tenon) won't be obvious. Or that forcing the pieces together with clamps won't cause other problems.

Ditto if your rip fence springs out of alignment, next you know the back teeth on the blade are chewing up your work.

Cross cut fence out 1/4 degree. No problem? Think again - that's an error of 2.5mm or roughly 0.1in in the width of a 24in panel, more in a wider one. Even 1/10 deg (the limit of most commonly available electronic angle measurement equipment short of fancy metrology room stuff) delivers errors of almost half of that. The five sided cut is way ahead in terms of its potential accuracy.

Precision fitting a wood to wood drawer? The method suggests for good reason that you don't try to achieve the fit by dead reckoning (fitting tother pieces made to specific dimensions), that you build in a little oversize and creep up on the required fit using carefully planned methods. But getting the eventual fit right hinges on achieving a gap correct to a small part of a thou.

Of course the wood moves afterwards by substantial amounts with moisture changes, and you'd better have got it right from a design point of view or this movement is going to take your piece apart. Get it right and the whole lots moves in unison.

So what do you guys think? My sense is that we're a bit schizo on the topic - that sometimes we argue one way on this, and then the other. Trouble is that there's times when precision isn't that critical, and equally lots of places where the utmost precision is critical.

Or is there some great woodworking secret divulged only to the initiates that allows you to work to 0.1 in all the way and have stuff come together perfectly?

??

[Mark Ashmeade]

Like any other engineering situation, it is subjective. Some things matter, other don't. One variable is price. I wouldn't expect a $2000 piece to have any flaws in the materials or workmanship. A $200 version of the same thing can be forgiven some.

In order to simplify things, society has developed "rules of thumb", and people go by those, sometimes to the utter detriment of the end result, and very often to their wallet, because either they didn't know how to engineer it right in the first place, or couldn't be bothered to do so. The "rule of thumb" has a lot to answer for, once you've got past "it's convenient".

It never ceases to amaze me when magazine and website contributors say "use a rare earth magnet" when a normal one would do just fine. "Use 3/4 Baltic Birch ply" when 1/2" C/D would suffice. "Make sure you buy a USA-made extension cord". Why? Are European ones no good then? Electrics are another one. "These 15A routers are better than those 12A routers". Current has less to do with it than power - which is measured in Watts.

So much for daftness!

[mike holden]

Oh Boy! Religious wars!
First, wood CAN be worked to a 0.005" total tolerance across the entire piece. This is normal patternmaking tolerance for wood models. (and although I have seen many pictures of superbly equipped shops, I have never seen one that had a CMM, Coordinate Measuring Machine, for checking projects)
But SHOULD it be worked that closely for an end table? Probably not.

And lets not confuse tolerances for tools with tolerances for parts. My metal tools are tuned to MUCH closer tolerances than the parts I build to make an end table (and I am not picking on end tables here, insert whatever woodworking project you like).

For most of us, this is a HOBBY. That means that as long as YOU are enjoying yourself, the rest dont matter.

Of course the significant other would like to see something come out of the shop once in a while, but not many will measure those things with a micrometer (grin)

Relax,
enjoy
Mike

[Paul Murphy]

Sometimes we work to high tolerance without actually measuring that tolerance, such as when fitting a mortise and tenon joint. Too loose, and the gappy joint will fail if stressed, too tight and it will split on assembly. The acceptable fit range for a reliable M&T joint is fairly small, and unless you want to fit each and every joint in a project you end up working to a fairly tight tolerance consistantly across the entire project.

Sometimes we work around inaccuracy, such as when a jointer fence is not perfectly square and we edge glue the complimentary edges of boards together so that they come together flat despite the jointer inaccuracy.

I favor accuracy like you do Ian, and over the years I've corrected my machines where I can.

[glenn bradley]

To throw one out for discussion. I've just got off the telephone from a retired toolmaker friend, having sat through ten minutes of his berating me to the effect that I'm mad for talking in terms of tight tolerances in woodworking, and in woodworking machines and equipment. Hand plane sole out of flat? No problem - just push harder on it. Cabinet not square - just clamp it up and nobody will notice.

Wait, I know that guy. He and his ilk built all the ill fitting, force-fit kitchens in my neighborhood :D. All kidding aside, there are many types of woodworking that will survive house-framing tolerances. I lean toward the exaggerated joinery style of Greene&Greene and some Arts & Crafts pieces. These styles draw the eye to the joinery. Having something out 1/4 of a degree or a gap of 1/64" (or even less IMHO) would stick out like a red flag waving at a bull. To each their own and each of us should do what makes us happy. Some of us build houses, some of us build kitchens, some of us build tables, chests and chairs. What is acceptable for one does not always translate well to the other. I'll shoot for a bit better than your friend advises. That's just how I am ;).

[Carl Beckett]

I'm with Mark.... It's 'subjective'

If I'm off a 32nd on a stretcher length dimension, nobody will ever notice. But they do need to be consistent to each other. A 32nd on a M&T or dovetail would be catastrophic.

So it depends on design purpose.

Is my table saw and jointer 'flat'? I have no idea, and don't have a CMM to find out. I can lay a straightedge across them but this in no way represents flat.

More and more I use story poles. No dimensions needed. Repeatable yes ( to the degree I can transfer the marks, which with a 0.5mm mechanical pencil is likely not within .005). But no real absolute dimension.

So my answer is along the lines of, you need to be 'accurate enough'

[David Helm]

Your friend is OK when talking about house framing. 1/8 inch tolerance is as accurate as you need because it will change with the next rainfall. Rough framing is just that. Woodworking, on the other hand, needs to look well made. To do that it must be done with much tighter tolerances. Yes wood moves. Proper (whatever that means) design is the answer.

[Jim Underwood]

Oh you know what they say...

"Measure with a micrometer, mark it with a piece of chalk, cut it with a hatchet!"

[ian maybury]

Ta guys, good to see the religious war hasn't materialised! (yet) :) Pardon my cutting loose on that one, but it's about the third time in recent weeks that I've got lectured on the topic by people who when it suits them often adopt the role of the arch exponent of precision.

I think you've hit it spot on. That there's no room for gross generalisations. Compulsively doing everything to tight tolerances makes no sense, but there are circumstances where we end up working to very tight tolerances indeed.

There's some good stuff in there (pardon my not mentioning more):

Paul: 'Sometimes we work to high tolerance without actually measuring that tolerance'.

Mike: 'lets not confuse tolerances for tools with tolerances for parts'

Mark: The "rule of thumb" has a lot to answer for, once you've got past "it's convenient".

Carl: 'you need to be 'accurate enough'

Paul again: 'Sometimes we work around inaccuracy'.

Mike again: 'For most of us, this is a HOBBY. That means that as long as YOU are enjoying yourself, the rest dont matter.'

It's a bit like life really - the moment you take your eye off the issue it tends to bite you up the posterior. What's peculiar to my mind (as a guy with a background in engineering and manufacturing) about woodworking is how often we end up second guessing/finessing equipment that isn't routinely capable of high levels of accuracy into delivering.

That said I have to admit to enjoying to the Zen like ('caring but not caring'/'always being required to connect with reality') nature of the discipline....

ian

[Steve Rozmiarek]

Measure once then use the stops to make all the parts exactly the same. Done. Careful planning eliminates the need to measure, and makes the gazillionth of an inch idea irrelevant.

[Dave Zellers]

Oh you know what they say...

"Measure with a micrometer, mark it with a piece of chalk, cut it with a hatchet!"

Isn't that supposed to be, measure twice with a micrometer, cut once with a hatchet?

:)

[Mel Fulks]

Pattern making was mentioned as a place for close tolerances, and that is certainly true .But it's also true that the pattern accuracy and the casting accuracy are different things. Many years ago ,as the grunt in a pattern shop,I had to hand chisel as much as a heavy sixteenth off off some match plates .Because the metal shrinkage did not follow the rules and the manufacturer customer did not want to have to machine so much metal on thousands of castings. Accuracy is not always accurate.

[paul cottingham]

Measurements and tolerances don't matter a wit. How does it look? In proportion? Joints look tight? I have several tables that aren't perfectly square, they are out by at least 1/16".
-gasp-

As the boatbuilders say....if it looks fair, it is fair.

Then again, I may be just excusing my lack of patience and skill.

[Peter Aeschliman]

I think people often respond negatively when people post tirades against a manufacturer for not having tight tolerances on this or that when they purchased a budget hobby machine.

The thing is, if you spend $1,500 for a cabinet saw, it's a lot of money for one hobby machine. It feels like a huge purchase to a hobbiest, so you expect high quality. But as far as cabinet saws go, that's a budget machine. That budget machine will cut accurately enough for the vast majority of peoples' needs. But the engineers among us get out the feeler gauges and dial indicators and obsess over details that really don't matter.

My "rule of thumb" is this... if you can see a gap in your joinery or have fitment issues, your machines or processes aren't dialed in. If you see burn marks, your fence isn't aligned. If you don't see symptoms, there isn't a problem. Doesn't hurt to check your settings when you do routine maintenance. But sometimes people obsess over things that have no impact in reality.

['Jacques Malan']

In our workshop the importance of accuracy increased when the CNC router came in.

The other comment on tolerances I have to make is that it is a very good guideline to see how experienced a woodworker is. Anybody can work accurately, for some it just takes a little bit longer. The experienced woodworker know where and when precision is important and where not, and concentrate his/her effort on the important stuff.

[Rich Engelhardt]

For most of us, this is a HOBBY. That means that as long as YOU are enjoying yourself, the rest dont matter.
Very true....
However....

I get a whole lot of satisfaction and enjoyment out of running my finger over a glass smooth joint :).
Just the other day i went to great pains to measure all the cuts I was making using my track saw. I had to cut 6 pieces of 3/4" ply to as close to the same size as i could get for a closet unit I made.
When I was done. I stacked them on top of each other and ran my hand over the edges....{{sigh}}} what a wonderful feeling it was to find each side to be as smooth as glass.

[Keith Christopher]

I would say why not true up your tools. It will save alot of time that would be spent on tuning joints in your project if you simply popped it out of the box and started cutting. but how far down to tune it, well that is a personal taste matter. if you plan your project properly then wood movement will not impact the joints or project so cutting parts to a very tight tolerance is not a problem. So it's like chicken soup for a cold IMHO.....doesn't hurt.

[ian maybury]

That's a big part of it for me too Rich, and it illustrates just how sensitive the eye and the finger are so far as detecting tolerances is concerned. Bet if you measured those pieces that they were within a very small tolerance of each other...

Trouble is it then becomes frustrating if a machine is unpredictable - it puts out accurate cuts most of the time, but every now and then throws a wobbly. Say you have a planer with a hump over five thou high 3/4 way down the outfeed table. Presuming the machine is otherwise set up right then every time you joint a short piece it'll come out straight, but joint a longer piece and the hump will kick it up when it reaches it - creating an angle in the jointed face which will be continued to the end of the cut.

It's very tempting to put the time in to sort a problem like this if the means is available, but there's a limit too beyond which it starts to become obsessive...

ian

[Jeff Duncan]

Ian, here's my take for what it's worth. If you look at the piece that you just built and are happy with it.....your accuracy is fine, end of discussion. Somebody wants to berate you for your accuracy or lack thereof MUST also be criticizing the quality of your work. For without examining your work they have no leg to stand on. I have been working wood since I was old enough to pick up tools and have been doing it for a living for well over a decade. In all that time I have never had someone look at my work and question what tolerances I work to. For most of that time I never owned a dial indicator or even a basic caliper! Only recently have I started using these items help in my machinery maintenance.

So in short, we can discuss our opinions of accuracy just for conversations sake and alls well and good....but at the end of the day your work is the final say and everything else is talk;)

good luck,
JeffD

[Mike Cogswell]

Measurements and tolerances don't matter a wit. How does it look? In proportion? Joints look tight? I have several tables that aren't perfectly square, they are out by at least 1/16".
-gasp-



It depends on the circumstances. You have a table that's out of square by a sixteenth and it's unlikely anyone but you will ever know or care. But I recently made a full set of replacement window sashes for my daughter's 1880 house. That took hundreds of pieces of SYP that had to be mortised or tenoned together and I couldn't have achieved nice tight joints and coplanar faces without some accuracy in my work. But the kind of accuracy required didn't involve micrometers. It was far more important that all the parts be the same thickness than that they be a specific thickness to a few thou. So, all the final thickness planing was done at the same time with the same machine setting. The same for lengths. If you have multiple muntins, they better all be the same length. Even a slight gap at the end of a short one would stick out like a sore thumb. So, use a stop block or jig to make them all the same. Again, it's the difference (tolerance) that matters more than the absolute length. But having said that, the actual values still matter, else the windows will not function smoothly.

Accuracy is not the end-all, be-all in woodworking that it is in some metal work, but in my opinion the key to good wood working is knowing when it really matters and when not. I've built many a wood project without using a ruler, just using an eyeball design, a story stick and jigs and stop blocks to make similar parts the same length or shape.

[paul cottingham]

It depends on the circumstances. You have a table that's out of square by a sixteenth and it's unlikely anyone but you will ever know or care. But I recently made a full set of replacement window sashes for my daughter's 1880 house. That took hundreds of pieces of SYP that had to be mortised or tenoned together and I couldn't have achieved nice tight joints and coplanar faces without some accuracy in my work. But the kind of accuracy required didn't involve micrometers. It was far more important that all the parts be the same thickness than that they be a specific thickness to a few thou. So, all the final thickness planing was done at the same time with the same machine setting. The same for lengths. If you have multiple muntins, they better all be the same length. Even a slight gap at the end of a short one would stick out like a sore thumb. So, use a stop block or jig to make them all the same. Again, it's the difference (tolerance) that matters more than the absolute length. But having said that, the actual values still matter, else the windows will not function smoothly.

Accuracy is not the end-all, be-all in woodworking that it is in some metal work, but in my opinion the key to good wood working is knowing when it really matters and when not. I've built many a wood project without using a ruler, just using an eyeball design, a story stick and jigs and stop blocks to make similar parts the same length or shape.
I must admit, I was not thinking about windows when I made my heretical statements. There is a reason I don't make doors or windows. :-)

[ian maybury]

It gets interesting to bounce this stuff around Paul - there's such a wide variety of situations we can end up woodworking in.

Lots of our level of equipment Mike is I guess once set up not bad in terms of repeatability, but rather harder to hit or to return an absolute dimension with. Which makes experience/knowing how to correctly sequence work an important part of the skill set.

To be fair to engineers it's just as clearly the case when working metal that overdoing the tightness of tolerances is a no no - one that leads to runaway costs. A decent designer needs to know when it's needed, and when not - and what it takes to achieve it. It's just that there are a lot of common design elements in engineering that by their nature need tight tolerances to function properly - a journal bearing for example. Structural steelwork though is a very different matter to building press tools or moulds.

Another major difference in engineering is that it often relates to the design of products that are manufactured in quantity, and to the building of the systems to make them. To rigorous quality standards.

Back in the day engineered parts were hand made and individually fitted as we do in the craft activity that is woodworking. (a mechanic over here is still called a 'fitter') Trouble is the advent of mass production led to the need for interchangeable parts - that's parts that aren't just just fitted together one to one, but parts which will always fit because they are manufactured to absolute dimensions and degrees of surface finish. (the bit we're not so good at) Which has tended to drive the requirement for reliably repeatable precision in machining/manufacturing processes.

More recently lean manufacturing and the cost driven preference to use less skilled labour has added to this requirement for precision the need for machines/processes that don't require much skill to set up. We by virtue of a lot of care and by knowing our machines well can pull off remarkable stuff, but that's not a reliable or a cost effective way of doing it in volume manufacturing. It's often more cost effective to spend relatively large amounts of money on machines/processes that are capable to a very high degree of reliability delivering as dialled up on the controls from first to last - with no messing about, and no black arts involved in getting to this.

I guess that's the case in the sort of volume production of wood products that goes on in lots of places too - it's the distinction in fact between our craft based approach and volume manufacturing of commercial wooden products. There's batch manufacturing environments you come across in wood manufacturing too - like the volume production of windows in your example Paul. Lots of much more expensive tooling, much more powerful and probably more precise machines (high end shapers and the like) than what we typically run.

Which I guess brings us around to the fact that many have pointed out - there's an awful lot of subsidiary worlds in woodworking, and each with their own needs. That it's not really possible to generalise. Solutions always have to fit the needs of the situation......

ian

[Rich Engelhardt]

Ian, all I can really say is that when I started to pay more attention to how accuruate I did things, the better they started to go together and they started to actually look like something other than a botched attempt at doing something.

The other thing I found out is that it's a lot easier and quicker putting things together that fit right than it is to make "filed engineering changes". ;)

I'll gladly trade a little "obsessive" for the benefits ;)

[ian maybury]

That tends to be my experience too Rich, and it's definitely the path I've followed. On the basis that it tends in the end to create less hassle, frustration pain. ( :) racing motorcycles and high performance model aircraft tend to extract quite a high price for overlooking even minor engineering and related issues) That said there's clearly those that see it differently....

[Ben Hatcher]

What I've learned so far in my woodworking is that there's a huge benefit to setting up machines as accurately as I possibly can and assembling parts immediately after machining them. Yes, things will move, but probably not before I finish gluing them up. I can maybe close a gap with a clamp, but doing so puts the joint at a greater risk of opening later on. I can't guarantee anything by using these practices, but I can certainly improve my odds of success.

I think that, ultimately, the accuracy required on a finished piece is determined by eye and the cooresponding level of tolerance required depends on where the piece will be viewed/used. Without some kind of reference, it is difficult for a casual observer to see a flaw like a table being slightly out of square. Your kitchen wall might look perfectly strait until you put a cabinet up against it.

[Mike Cogswell]

<snip>
There's batch manufacturing environments you come across in wood manufacturing too - like the volume production of windows in your example Paul <actually, Paul was quoting my msg>. Lots of much more expensive tooling, much more powerful and probably more precise machines (high end shapers and the like) than what we typically run.

<snip>

ian

:LOL: adjectives like expensive, powerful and precise are not what I would use in describing my 35 or 40 year old Craftsman contractor saw that cut all those pieces to length and also cut all the tenons. It's been well maintained and I have carefully set it up (and check it periodically) to be sure the blade, fence and miter slot are all parallel to one another. That and a very carefully made crosscut sled work pretty well. You're right that it was mostly about the process. My annual Christmas project is to make a batch of identical family gifts for an extended family and a few close friends, so I'm used to production runs of about 30 objects. The key to success is to plan the build and assembly so that all operations that use the same machine setting are done at the same time. Changing a setting and then trying to go back and duplicate it later never seems to work well for me.

[Alan Schaffter]

I agree the answer is "it depends." A lot of this discussion has focused on thickness, width, and length accuracy/precision, but there are a number of cases in woodworking where +/- 1/8", 1/16", or even 1/64" is not good enough! Here are some examples:

How many of you have labored to set the proper bit height so your dovetails fit properly and look nice? I would guess the setting tolerance is pretty narrow.

Box joints must be cut within close tolerances, too- too tight and they won't fit and too loose they will look like crap. I made a series of joints to demonstrate this for an I-BOX demo at the recent Klingspore WW Show. I used the I-BOX to make what I considered a perfect joint- then made a few more sets of joints with varying amounts of micro adjustment in each direction. The joint made with .006" of micro in one direction (from center) could barely be assembled, while .006" in the other direction resulted in a joint that was too loose.

I've been working on a set-up jig for lock miter bits- bit height and fence position need to be set precisely here too- within a sharp pencil's width!