Precision

[johnsonri]

When I first started in my shop, I followed the "measure twice, cut once" practice.

While "measure twice, cut once" worked out very well, I am happy to say that I have long since moved on to simply "lay and cut".

Almost all of my tools have either 1/32" or 1/64" scales attached so that I can simply lay a board down at the tool along the scale and cut the needed length.

My question though is with the level of precision achieved using either a 1/32" scale or a 1/64" scale. If accurately using a 1/32" scale, would that mean that you have accuracy to +- 1/32" or accuracy to  +- 1/64"?

It seems to me that whatever scale that is used would give accuracy to +- of the next increment...eg 1/32" scale is +- 1/64" and a 1/64" scale is +- 1/128".

Thoughts?

 

[Steve Jones]

I think you're confusing unit of measure with accuracy.

If you use a 1/32" scale - that's your unit of measure, accuracy is how well you use that scale.

For example: set up and cut a small sample piece (small enough that you can measure it very accurately by some other method, preferably a micrometer or digital caliper) How close your sample actually is to it's intended size is a measurement of your accuracy.

One of the major advantages to the Incra jig (Available for router tables and table saws) is that they lock in in 1/32" increments, once you have them set up they will always be that accurate at any measurement. (they lock into the nearest 32nd" and then you can fine adjust to 1/1000")

I had one set up one my cutting table for a while - on testing I could consistently get accuracy of about 4/1000" (1/250") on cutting sheet goods for cabinets (which is a little too intense for that purpose, I put it back on the router table where it belongs.

I believe they also make a fence that you can add to miter saw stations and pretty much any other tool to get the same level of accuracy.

One of these days I'll go crazy and add those incra fences and jigs to everything in the shop, they're hard to find in retail stores but do have a good web site. Woodcraft carries some of their products but not the really fine (and expensive) stuff for some reason.

 

[Notorious T.O.D.]

Measuring in woodworking is inherently dangerious and I avoid it as much as possible.  Instead use measurements to make your initial layout and cuts in your projects first pieces.  Work off those pieces as you cut and add more pieces to the project.  If you try to cut all the project parts to size and then assemble it you will have a disaster on your hands.  Measurement errors tend to accumulate in a project too, another reason not to measure too much.

And if you are going to measure invest in a good set of steel rules with etched markings and bea aware that not all tape measures measure the same distance as the printing can vary especially between brands.  I do most of my measuring with a 6, 24 and 36 inch Bridge City Tool Works steel rulers.  And don't underestimate the value of a story stick for marking out multiple pieces of the same part.

I think it was Frank Klaus in his dovetail video who said something like "....the board knows how big it is..." meaning that each and any board has a width, length and thickness which we can reference to and from without ever measuring anything.

Best,

Notorious

 

[johnsonri]

If you use a 1/32" scale - that's your unit of measure, accuracy is how well you use that scale.

This one statement is hinting on what I talking about. If you use a 1/32" scale and measure out perfectly (atleast perfect in your mind  ) to a 32nds of an inch, then would you say that your piece is accurate to a 32nd of an inch, or +- 1/64th?

I am not questioning how to measure, or what devices to use. Personally, I mostly use Incra rules with 1/64th scales and almost all of my bench and major tools have 1/32" scales attached. Story sticks are great too. I use and have demonstrated them a bit in my Festool MFT Retrospective.

One of these days I'll go crazy and add those incra fences and jigs to everything in the shop...

I am one of those guys who have gone crazy and added scales and jigs to everything in the shop  ;D  and to be absolutely sure, calipers, protractors and other fine measuring devices are at the ready for a confidence check when needed  ;D

Notorious, I do agree with you...been there done that, in my earlier days. I haven't used any of the BCTW steel rulers. Once I started down the path with Incra, I was happy.

 

[dirtydeeds]

a mate of mine is a hobby machinist, he works to thou's if he wants better accuracy he switches to microns

this is what he calls precision ( apparantly microns are more smaller than thou's )

we are doing carpentry
our tools do not allow us to work to these tollerances

precision for a carpenter is achieved in different ways

RULE 1.............. no carpenter can cut better that 1/64th of an inch  (that is about 1/2 of a millimeter)
RULE 2.............. any carpenter / carpentress who can, is wasting their time ( there is no need )

your eyes are far more accurate than 1/64th, and you can achieve this accuracy VERY easily

your block plane or chisels (if sharp) can remove as little as 1 or 2 thou

high end trim carpentry, cabinet work and production trim carpentry have different quality requirements so the degree of precision required is achieved in different ways

case 1, high end trim carpentry (or single "one offs")

precision in trim carpentry has nothing to do with measurement

it is all to do with tricking the eye to see straight lines and right angles where there are none

it is pointless spending time cutting to 1/64th if it LOOKS wrong

to get precision you mark the piece in place, then cut, then if necessary use your block plane

{loading and unloading all your production power tools out for a one off wastes time, use hand tools instead}

case 2 cabinet work

not the humidor boxes ive seen here (which i GUESS have everything to do with the eye)

accuracy in cabinet work is achieved by cutting all similar pieces to the SAME dimension

clearly you have to cut straight and square and in measuring from a SINGLE point

the best expanation of measuring from a single point is setting out ballustrades for a deck handrail

if you measure the position of each ballustrade from the STARTING point any minor errors get lost and cant be seen

HOWEVER

if you meausure each ballustrade from the previous you WILL end up with a culumulative error

case 3, production trim carpentry

this type of trim carpentry has slighty lower quality requirements

so you mix the rules to get the speed required for contract work.

Cut the pieces as you do for cabinet work (the same lengths for similar pieces) then make the sub assembly

Then......    you "bend" the sub assembly to get the best fit by eye

 

[nickao]

Precision is repeatability so in the terms you are speaking you can not "work to"precision only accuracy. You can work with precision. It semantics only I think your premise is correct.

In case 1 if you were precise everything would be identical in measurement even if it looked wrong. I believe in case one you are talking about accuracy which is the amount the precise measurement is off by. Your statement should say to get Accuracy(not precision). You can not change and use the block plane etc making everything slightly different and call it precision becasue precision means repeatability, even if something looks wrong. What you are stating is accuracy by staying within a certain amount of the precise measurement.

"Accuracy indicates proximity to the true value, precision to the repeatability or reproducibility of the measurement"

http://en.wikipedia.org/wiki/Accuracy_and_precision

As you say a carpenter does a dis service to the client by being to precise because reveals would look like crap! You need to be accurate making everything "look" good. My incra stuff is precision. But when I put the stuff together I work to be accurate I could never assemble everything identical every time, no one can except maybe a computerized robot.

Try to imagine something being accurate, but not precise or being precise and not accurate, which are both possible. It is a struggle to visualize , isn't it 

This is the toughest thing I came across in college physics becasue I took them to be the same(precision and accuracy) and they are not, I think most people do think of them as the same and in most cases when laymen are talking about something they are. Until you get in a scientific setting of course.

 

[dirtydeeds]

forgot to say tollerance affects precison and accuracy  :

 

[slb]

Nick, you are right on the money about the technical meanings of precision vs accuracy, how the words are generally (mis)used interchangeably, and how they matter for woodworking.  Accuracy is correctness on average, precision is the amount of scatter among measurements that were supposed to be the same.  Engineering teaches that you can always calibrate a precise thing to make it accurate, but you can't use accuracy to achieve precision.

For woodworking we always need precision because we want parts to match each other closely when we assemble them, even if they aren't quite the intended size (aren't accurate).  When parts are made independently (e.g. by separate measurements rather than a jig, by separate setups of a fence, or by different people) we also need them to be accurate.  That is, if I make a 10 cm part and you make another, we could each be precise to 0.1mm but if your 10 cm is 0.5 mm larger than mine we won't get a good fit.

Returning to the original poster's question, I think accuracy depends mostly on the quality of the measuring tool.  If your tape measure has a bent hook or has stretched, you will never get accuracy.  Precision depends on what you are doing and how you do it, not on the mark spacing on the ruler.  For example, with care you can easily get better than 1/64 using a rule with markings at 1/8 increments - provided you are trying to measure something that is a multiple of 1/8.  If you are sloppy (or if your eyes are like mine and have trouble seeing very close markings any more) you can get worse than 1/32 using a scale with 1/64 markings  .

Steve

 

[Jesse Cloud]

Interesting thread!  I learned the lingo very slightly differently... precision is the degree of fineness of the measurement, e.g., 1/1000th or 1/2; reliability is the likelihood that if you take the same measurement more than once, you will get the same result; accuracy is how correct the measurement is, taking into account both precision and reliability.

Ironically enough, true 'accuracy' often doesn't matter in woodworking.  It doesn't really matter exactly how long the side of that box is, just that its the same as the other side.

Which brings me to a common mistake I confess to: measuring with different instruments.  Probably the worst thing you can do is to measure with whatever is close at hand.  Pick a tool for all your final measurements then hide all the rest - amazing what this will do for the quality of your work!

 

[johnsonri]

Good points all the way around.

So, I am after a two pieces each with a length of 5 1/32".
>> Piece one is cut to 5 1/32" plus a hair
>> Piece two is cut to 5 1/32" minus a hair

Using a digital caliper that measures in 64ths, the readout for both pieces is 5 1/32".

It seems to me that as long as the plus/minus a hair is less than 1/64th, then the caliper will indicate the length is 5 1/32" for both pieces. With that it seems like the precision accuracy of this scenario is +- 1/64. Is that right?

 

[slb]

Interesting thread!  I learned the lingo very slightly differently... precision is the degree of fineness of the measurement, e.g., 1/1000th or 1/2; reliability is the likelihood that if you take the same measurement more than once, you will get the same result; accuracy is how correct the measurement is, taking into account both precision and reliability.

Ironically enough, true 'accuracy' often doesn't matter in woodworking.  It doesn't really matter exactly how long the side of that box is, just that its the same as the other side.

Which brings me to a common mistake I confess to: measuring with different instruments.  Probably the worst thing you can do is to measure with whatever is close at hand.  Pick a tool for all your final measurements then hide all the rest - amazing what this will do for the quality of your work!

Agree with everything except that I was taught to call the "fineness of measurement" resolution.  Ironically, some instruments, especially cheap digital ones, resolve things smaller than their accuracy or precision!

The point about using the same instrument for all measurements is especially apt.  My tape measures vary from one another by as much as 1/32 inch!

Good points all the way around.

So, I am after a two pieces each with a length of 5 1/32".
>> Piece one is cut to 5 1/32" plus a hair
>> Piece two is cut to 5 1/32" minus a hair

Using a digital caliper that measures in 64ths, the readout for both pieces is 5 1/32".

It seems to me that as long as the plus/minus a hair is less than 1/64th, then the caliper will indicate the length is 5 1/32" for both pieces. With that it seems like the precision accuracy of this scenario is +- 1/64. Is that right?

Well, kinda...the accuracy is still unproven - they could both actually be 5 1/4 for all you can tell (not likely unless your caliper is a real piece of junk, but possible  )!  And the resolution of the caliper's fraction mode is obscuring its potential precision.  These digital gadgets have internal logic to decide when to display a particular fraction.  Some round to the nearest, others show the fraction when the actual reading is within .005, and doubtless there are other variations.  If you put it in pure digital mode (if it allows) you may find that its actual precision is .001 or better, though at that level it depends a lot on your technique.  For example, I can cause the readings on my caliper to vary by several thousandths based on how hard I crank down on the thumb wheel.

 

[nickao]

One thing about using the same tape, etc for every measurement. I used to follow this, but with the Woodpecker or Pinnacle measuring items of which I have over 20 every single one measures exactly the same!

I think they may not make a tape measure because they know it would be tough to keep the same level of quality in making a tape always read the same as the rest of their stuff, squares etc.

 

[slb]

One thing about using the same tape, etc for every measurement. I used to follow this, but with the Woodpecker or Pinnacle measuring items of which I have over 20 every single one measures exactly the same!

I think they may not make a tape measure because they know it would be tough to keep the same level of quality in making a tape always read the same as the rest of their stuff, squares etc.

That's accuracy, which you get when you pay for it ;D.  Same with Starrett, though they are even more expensive.  I also like Woodpeckers (who make Pinnacle for Woodcraft).  In the end, tape measures should be limited to carpentry.  Trouble is, really accurate long rules are very expensive.

 

[dirtydeeds]

really accurate rules tend to be steel or aluminium

so.....................

their accuracy is affected by heat

you wanna' get REAL ACCURATE??  then you get a REAL headache

rocket science type stuff

at a certain point you need to take the curvature of the earth, altitude and the earths LOCAL MAGNETIC signature into consideration

(aparantly the last bit is really difficult because it alters over a period of time)

then........... it gets stupid

different timber, heartwood and sapwood, plastics and metals react and alter dimensions due to humidity as well as all the rest above

 

[AlexThePalex]

at a certain point you need to take the curvature of the earth, altitude and the earths LOCAL MAGNETIC signature into consideration

Just or the sake of getting the best result possible I'd like to add that you also should NEVER forget to compensate for fluctuations in the space-time continuum. 

 

[dirtydeeds]

now you are getting a bit too tricky for me........... im a limey carpenter

im not a time lord or a vulcan

 

[johnsonri]

Tom and Alex,

I like the macro approach that you've taken here. I was able to compensate for the curvature of the earth and altitude. Although, my caliper still read the same afterward. But, I had peace of mind...so all is right in the world. Well, my little world anyway.

 

[tbellemare]

Tom, Alex, and Rey:

A while back, we instituted a policy at Tool Home that we would use the XML formula put out by UPS to calculate our space/time continuum. If it doesn't work right, it's best to call me personally (512-428-9140).

Tom

 

[dirtydeeds]

tom, there is the answer

UPS arrived on our shores a few years ago, with "A class" american size vans..............

they found out that english roads dont have the same space time continum as north america

now they use sprinters.......................  smaller and MUCH MUCH quicker

the uniform is still a naff "crap" color