tool and table tolerance?

[keith ouellette]

I just recently failed in my attempt to make a large router table. It didn't end up being as flat as I wanted and it got me to wondering about tolerances of things.

When I was worried about it being .003 out of flat across its width a number of people told me it wasn't enough to worry about. Then I discovered I had accidently built a high ridge along the width of the table which made it about 1/32 out of flat on its length.

Using the router table as an example:

Even if the table is only out by .003 then the fence and the bit would always end up a certain amount out of perpendicular in relation to the work piece. the problem could double (or possibly quadruple) if the fence was off a bit also. I a world of magical possibilities if the fence was off in a way to cancel out the unevenness of the table then everything would be fine, But I doubt that is much of a possibility.


A distance of .003 sounds small; just about the thickness of a piece of paper.

But my middle age eyes can see the edge of a piece of paper without any strain so wouldn't they be able to see a gap in a joint that was that size?

Do I have this thought right?

How much of a gap could .003, .006, or .012 make in a joint made at the router table? Am I really splitting hairs?

[Dan Friedrichs]

A human hair is generally around 0.004", so yes, you are LITERALLY splitting hairs :)

[Brian Kent]

Since it is made, try it with a few projects, including something pretty small and sensitive to accuracy and see if it works for you.

[pat warner]

A 3 mil error is not serious but if the work rocks (not well milled) it can be.
If a 4' stick has a 3 mil fulcrum, it may well sit far higher than 3 mils on one end or the other.
Either forgetabboutid or be consistant with your chase for precision and accuracy.
Routs to .001. (http://www.patwarner.com)

[Bruce Page]

Keith, the 1/32 ridge would be a show stopper but .003 is in the noise.
I'd be willing to bet that there isn't a commercial router table made that has a TIR of .003

[keith ouellette]

How about the combination of .003 out of flat on a table along with a fence that was .003 out along with a router bit that was tilted to the side so it was .003 to the left or right of perpendicular to the table and fence.

How do all these little errors combine. Are they simply added together or are they multiplied?

[Rod Sheridan]

Hi Keith, at the risk of sounding like a smart, you can't possibly be serious about worrying about 0.003" in router table can you?

I don't know how your router is held into the table, what kind of router it is, or what the stiffness of your router table and fence are.

It's possible that your router will deflect more than 0.003" under the force of cutting.

You're not using an industrial type shaper, you're using a hand held power tool designed for light work.

A router simply isn't that accurate to start with, coupled with the fact that you will be hand feeding the work means that you'll introduce more variables. The bit will deflect, the router will move in the table, the table will flex, the fence will flex etc.

If you're lucky some of the 0.003" errors will be opposite in direction so they will cancel.

Most machinery won't have a top that's flat to 0.003", however machinery will normally have tight specifications for arbour runout and verticality, although I've had to shim shapers for that.

The 1/32" ridge however is a problem, it's aproximately 10 times your 0.003" error.

Regards, Rod.

[Tom Esh]

A human hair is generally around 0.004", so yes, you are LITERALLY splitting hairs :)

You left out a couple zeros there Dan. A piece of 20# paper is around .004". An average hair is more like .00004, or 100 times thinner.

[Bill White]

I sure don't want to sound like a pain in the a&&, cranky old fool, but the growing obsession with aircraft/space vechicle accuracy in woodworking is startin' to make me even more astonished. I would bet ya that none of the old masters measured in .000/inch. I am also quick to add that the "close enough for gummint work" syndrom is a shift in the opposite direction.
.003? Get over it and go to work. "Paralysis by analysis" strikes again.
Now I feel better.
Bill :rolleyes:

[Matt Benton]

Aren't you assuming that the jointer gets the stock to within .003 to begin with. If not, I'd worry about that first...

[Jack Ellis]

Your stock's going to move more than .003".

If you can repeatably work to 1/64th, I'd say you're doing fine.

Yes, errors can accumulate. They can also cancel each other out.

[Scott Loven]

You left out a couple zeros there Dan. A piece of 20# paper is around .004". An average hair is more like .00004, or 100 times thinner.
I have very thin hair and it measures .0015, thats 15/10,000 of an inch.

[David DeCristoforo]

FWIW, I have been doing this kind of work for more years than I care to mention and I have never worried about these kinds of "errors". My "bottom line" for all but the most critical woodwork (exposed joinery for example) has been 1/64th of an inch.

For machine setups, that would be too much error but a few thousandths of an inch is simply not significant. You will undoubtedly have more variance than that in the piece of wood you are milling.

[Chris Tsutsui]

I agree with the above posters about how 0.003 is overkill... even 1/32nd is accurate enough for some applications. Being accurate to 1/64 is great. haha

Trying to be more accurate than that setting up machinery is going to take a long time and probably cause more frustration.

I wonder what's your table made out of? Mony home made router table tops are MDF which is pretty stable. To flatten a mistake you can run it through a shop's large drum sander.

[Andrew Joiner]

Keith,
Materials move around over time. Your top may be flatter in a month or sag a hair in a year. Don't worry! Yours is fine.

[Chip Lindley]

You left out a couple zeros there Dan... ...An average hair is more like .00004, or 100 times thinner.

TOM TOM TOM! "An average hair"??? Are you referring to a human hair? I think NOT! Perhaps the *hair on a gnats ? .00004" is 4/100,000ths! Gimme a Break Please!

As reference, a Class V interference fit (press fit) --such a bearing pressed on a typical 5/8" shaft, clearance between the shaft and pulley is, at minimum, .0003" (three ten-thousandths)

[Chip Lindley]

Keith, be *Practical*! Not OCD! You ask what tolerance *joints* should have, made on a router table?? What kind of joints are you referring to? I am scratching my head here! Those Incra dovetail or finger joints? IMO Incra was the first to induce this paranoia among woodworkers that anything less than *their* tolerances just won't DO!

My router table is used primarily for routing profiles on edges, pattern shaping, or producing small trim mouldings. IMO, consistency is needed much more than high precision! A gap of .003" can be clamped closed!!

Remember that most stock run thru a router table has some flexibility! .003" is nothing to be concerned about, with featherboards, hold-downs or the *armstrong* approach!

A straight fence, square to the table AND the router bit being square with the tabletop are quite other issues!

Find the cause of the 1/32" problem at table's-edge and deal with that. LIVE with .003" and ENJOY!

[keith ouellette]

I understand no one seems to worry over .003 of an inch. I guess it started to concern me because of my table saw tun up. Before that I didn't think of tolerances much at all. When I noticed not only the difference in the way a rip cut looked, even one with a cheep blade, along with the change in the sound of the blade made as it was cutting

thats when I really started to think of .003 being an issue. It made me want to go back and recheck the saw and re align so I could try and get rid of the tiny bit of run out left in the blade.

Now that I am moving on to the router table I have started to think of the relation of the bit to the table.

I never really thought about the bit flexing as Rod pointed out
or that the stock will flex as Jack and chip pointed out.

Now I'm more bothered.

I'm sure some (or all) think I'm being anal about it but when you put the error in the table together with the error in the fence along with the bit flex and the stock flex and the jointer probably being off a bit also then you can end up with a pretty poor joint.

Any one of them alone doesn't bother me much ( it does bother me a little) but since I saw my TS cut improve so much I just want to strive a little harder for excellence. Just not sure If I can.

[Tom Esh]

TOM TOM TOM! "An average hair"??? Are you referring to a human hair? I think NOT! Perhaps the *hair on a gnats ass??* .00004" is 4/100,000ths! Gimme a Break Please!

Yeah, sheesh. Must have got fat-fingered with the calculator;)
Wikipedia has the range as 17 - 181um or .0007 - .007"

[Rod Sheridan]

Hi Keith, to put it in perspective, I recently helped a friend make some raised panel doors.

He used a set of router bits in my shaper to make the cope and stick cuts.

The cope cuts were first, made by clamping the rails in my home made coping sled. The sled is made from 3/4" BB ply, with DeStaco toggle clamps and a fence that's glued to the sled.

A piece of masking tape, 6" from the end of the fence calibrates it to as square as I can measure.

Nothing fancy obviously, yet it produces joints that fit well.

This would be the sort of accuracy you would get with a router table, fine for woodworking.

Normally I use shaper cutters, which can accept very thin metal shims to produce perfect fitting joints.

If your router table is halfway flat and square you'll produce nice work with it.

The glue itself in a joint will take care of a few thou......Rod.

[Pete Bradley]

I understand no one seems to worry over .003 of an inch. I guess it started to concern me because of my table saw tun up. Before that I didn't think of tolerances much at all. When I noticed not only the difference in the way a rip cut looked, even one with a cheep blade, along with the change in the sound of the blade made as it was cutting

thats when I really started to think of .003 being an issue. It made me want to go back and recheck the saw and re align so I could try and get rid of the tiny bit of run out left in the blade.

Now that I am moving on to the router table I have started to think of the relation of the bit to the table.

I never really thought about the bit flexing as Rod pointed out
or that the stock will flex as Jack and chip pointed out.

Now I'm more bothered.

I'm sure some (or all) think I'm being anal about it but when you put the error in the table together with the error in the fence along with the bit flex and the stock flex and the jointer probably being off a bit also then you can end up with a pretty poor joint.

Any one of them alone doesn't bother me much ( it does bother me a little) but since I saw my TS cut improve so much I just want to strive a little harder for excellence. Just not sure If I can.

If you saw a really major improvement, I'm betting you did more to your TS than take out .003 of runout. Perhaps it was pinching at the back of the blade before?

I think the thing that has really changed woodworking forums (not necessarily woodworking) is the availability of low-cost precision measuring equipment. People with no machinery or manufacturing experience can buy a straightedge or a dial indicator and get a reading. It looks so precise, it must mean something, right? Magazines like FWW compound this with ridiculous "precision" comparisons that are just looking at noise inside of reasonable tolerances. Much of it wastes time that would be better spent getting experience.

Pete

[Andrew Joiner]

I'm sure some (or all) think I'm being anal about it



Glad you said it and not me, but I'm sure some here might be thinking some thoughts along those lines.

Just move on, do some test cuts. You'll see .003 is nothing.

[Bruce Page]

Keith, you'd make a good tool maker. ;)

[Dan Friedrichs]

You left out a couple zeros there Dan. A piece of 20# paper is around .004". An average hair is more like .00004, or 100 times thinner.


Tom,

I don't think I did. I worked in a clean room that had a scanning electron microscope in it for a few years, and we'd routinely stick a piece of human hair into it for a "whiz-bang" demo for high school tours, etc. The digital diameter measurement usually read ~100um, or 0.004". Obviously there are huge variations between samples, but I'm not off by two orders of magnitude....

Just for fun, I attached an image showing what the microscope outputs. The attached image IS NOT A HAIR (naturally, the one image I was trying to find isn't there...) - but rather, a gold line in a microprocessor. Note the 75nm (or 0.000003") measurement :). Now THAT may be too precise for woodworking!

[Tom Henderson2]

Keith, be *Practical*! Not OCD! You ask what tolerance *joints* should have, made on a router table?? What kind of joints are you referring to? I am scratching my head here! Those Incra dovetail or finger joints? IMO Incra was the first to induce this paranoia among woodworkers that anything less than *their* tolerances just won't DO!

My router table is used primarily for routing profiles on edges, pattern shaping, or producing small trim mouldings. IMO, consistency is needed much more than high precision! A gap of .003" can be clamped closed!!

Remember that most stock run thru a router table has some flexibility! .003" is nothing to be concerned about, with featherboards, hold-downs or the *armstrong* approach!

A straight fence, square to the table AND the router bit being square with the tabletop are quite other issues!

Find the cause of the 1/32" problem at table's-edge and deal with that. LIVE with .003" and ENJOY!

I think the OP is having a bit of fun with us.

Unless you have rather sophisticated measuring equipment, you can't even measure table top flatness to anywhere near a few thousandths.

So let's cut him some slack...

-TH

[keith ouellette]

If you saw a really major improvement, I'm betting you did more to your TS than take out .003 of runout. Perhaps it was pinching at the back of the blade before?

I think the thing that has really changed woodworking forums (not necessarily woodworking) is the availability of low-cost precision measuring equipment. People with no machinery or manufacturing experience can buy a straightedge or a dial indicator and get a reading. It looks so precise, it must mean something, right? Magazines like FWW compound this with ridiculous "precision" comparisons that are just looking at noise inside of reasonable tolerances. Much of it wastes time that would be better spent getting experience.

Pete

I forgot to mention the fence tapered in towards the saw by about .002 and that got fixed also. It made a remarkable difference. I was a little shocked that it sounded different. other than the sound of the motor the blade went from a whine to a whisper.

Now that I have been thinking about it, just in case anyone is a little interested, if the table is out of flat then the router bit will double in its out of perpendicular measurement on the x and or the y axis because it can not sit flat on the table ( the table would be the z axis in a 3 dimensional field if I remember my geometry).

So if the table tilts down towards the bit position by .003 along either its length or its width then the bit would end up being tilted towards the table by .003 if the router plate sat flush with the table on all sides like they must to keep the plate out of the way. now the cut will be off by .006.

I had more to say but I'm sure anyone who read this far is already board with me (and thinks I'm super anal. I'm usually not.

[M Toupin]

the thing that has really changed woodworking forums (not necessarily woodworking) is the availability of low-cost precision measuring equipment. People with no machinery or manufacturing experience can buy a straightedge or a dial indicator and get a reading. It looks so precise, it must mean something, right? Magazines like FWW compound this with ridiculous "precision" comparisons that are just looking at noise inside of reasonable tolerances. Much of it wastes time that would be better spent getting experience.

Now that's the truth!

Unfortunately "low-cost" and "precision" never go hand in hand... cheap or accurate, choose one.

Keith - .003" is about 2.5 swipes with 220g sandpaper or 2 swipes with a scraper. Your joints will compress more than .003" whren you clamp them, it's not all that.

Put down the micrometers and feeler gages and cut some wood...

Mike

[keith ouellette]

I think the OP is having a bit of fun with us.

Unless you have rather sophisticated measuring equipment, you can't even measure table top flatness to anywhere near a few thousandths.

So let's cut him some slack...

-TH

NOT so at all. I'm not making any kind of joke here, I didn't realize I was getting so ridiculous.

I am using a straight edge and a marked feeler gage. may not me super accurate but it will be off th same amount each time.