Solved! Measurement accuracy - Touchdown!!

[John Piwaron]

I finally solved a measurement accuracy problem (or think I have) with the scale on my tablesaw. Since I got it, the fence's distance as measured using the scale on the guide tube and a direct measurement using a Starrett scale from the fence to the blade has not matched. So I've been using the number from the Starrett scale.

But that was bothering me. It shouldn't be. It's a Biesemeyer fence, long ago my first Biesemeyer fence, the homeshop version, had a very accurate scale, why shouldn't this one also?

I changed the scale on the guide tube. No help. The saw blade and the fence are parallel to the left miter slot. No help there. Then I started considering the geometry of how any of this works at all. It was starting to look like the fence was moving on the guide tube similar to how a car on one of those inclined railroads move. That is, I could move the fence to the 9" mark, and it did indeed move 9" but wouldn't move 9" away from the blade because the guide tube was at an angle to the front edge of the saw.

So I checked. Sure enough, that's it. That's the root of the accuracy problem. But the next issue is that there's no specific adjustment for that. Delta, when they make the fence parts must assume that the holes in the angle iron that support the guide tube must be perfectly located. The guide tube is an aluminum extrusion with a slot in the bottom for the head of the hex cap screw. The head really can't go anywhere. I had to hope there was enough play between all the relevant parts to get my own adjustment.

I made 3 blocks just under 7/8" thick to locate the guide tube evenly from the saw. Then loosened the screws, put those in the gap between the tube and the saw. Then I retightened the screws while pushing the guide tube against those blocks. Bingo! That's it! the guide tube is now parallel to the front edge of the saw, the fence itself is parallel to the miter slot, and the distance from the blade to the fence matches on the guide tube scale and the Starrett scale. Touchdown!

Then I looked in the instructions again. There's nothing in there specifically about making sure the guide tube is parallel to the saw. I think Delta (or anyone else) might want to think about instructions for that.

[Dave Richards]

So basically, you made the tube perpendicular to the blade?

I wonder how many people have the same problem.

[Brian Brightwell]

I was a little lost but you set me straight with one sentence Dave.

[Myk Rian]

How much was it off? Seems to me it would be quite a bit (very noticeable) to throw off the readings.
Did the fence have enough adjust-ability to make it up?

[pat warner]

I would add, what takes longer the sleuthing or a calibration cut?
And these saws constantly require a tweak here & there, so adjusting & calibrating a saw is just part of its use.
Now if I have the luxury of some scrap, (& I have plenty), I take my best guess at the fence position/readout and measure the cut.
Not on? Then move the fence (using the fiducials) an amount = to the error in the calibration stick.
That error should be releatively constant. So you could apply that to any fence position any time.
A calibration cut is essential for good saw health, regardless.
It often turns out that adjusting for error (empirically), rather than intense machine analysis, usually because of a machine design error, is the quickest resolution.

[John Piwaron]

Parallel

I suspect it could be a pretty good number if the user isn't critical.

[John Piwaron]

about 1/64, maybe more. I could tilt the fence itself all I wanted and it would make the slightest difference. And then it wouldn't be parallel to the blade.

The error is subtle. And took a good bit of thinking to realize what was going on.

[John Piwaron]

I would add, what takes longer the sleuthing or a calibration cut?
And these saws constantly require a tweak here & there, so adjusting & calibrating a saw is just part of its use.
Now if I have the luxury of some scrap, (& I have plenty), I take my best guess at the fence position/readout and measure the cut.
Not on? Then move the fence (using the fiducials) an amount = to the error in the calibration stick.
That error should be releatively constant. So you could apply that to any fence position any time.
A calibration cut is essential for good saw health, regardless.
It often turns out that adjusting for error (empirically), rather than intense machine analysis, usually because of a machine design error, is the quickest resolution.


I'm o.k. with anyone that doesn't care if the tube isn't parallel to the leading edge of the saw table. I know it *ought* to be, so I made it so. These errors get compounded the farther into the work one goes. One adjustment dependent on another.

I'm simply one of those guys that's going to comb through and make sure the equipment is as right as equipment of this nature can be. Now I can trust the scale on the guide tube. Before I could not. With the guide tube out of whack the way it was, the amount of error at 6 inches is one thing, It gets greater the closer to the end one gets.

[Chris Fournier]

It feels good to solve a nagging problem doesn't it? Congrats!

[Jerry Miner]

Wait a second. You're saying that since the fence was out of line with the saw, that the cursor was traveling along the scale at a slight angle, thereby traveling along the hypoteneuse of a long triangle, rather than along the leg of that triangle.

And that the "short leg" of that triangle is 1/64"!!? If you have a 50" fence, then the amount of error at the maximum 50" cut would be equal to the difference in length between the long leg and the hypoteneuse. In a 50" cut, that error is .000002 inches!! Are you seeking better precision than that from a tablesaw??!!

(with a 30" fence, a 1/64" out-of-square condition, and a 30" cut, the error would be .000004 inches). Isn't that a little overkill in precision??

[johnny means]

Wait a second. You're saying that since the fence was out of line with the saw, that the cursor was traveling along the scale at a slight angle, thereby traveling along the hypoteneuse of a long triangle, rather than along the leg of that triangle.

And that the "short leg" of that triangle is 1/64"!!? If you have a 50" fence, then the amount of error at the maximum 50" cut would be equal to the difference in length between the long leg and the hypoteneuse. In a 50" cut, that error is .000002 inches!! Are you seeking better precision than that from a tablesaw??!!

(with a 30" fence, a 1/64" out-of-square condition, and a 30" cut, the error would be .000004 inches). Isn't that a little overkill in precision??


I was thinking the same thing, but didn't have the math handy.

[John Piwaron]

Wait a second. You're saying that since the fence was out of line with the saw, that the cursor was traveling along the scale at a slight angle, thereby traveling along the hypoteneuse of a long triangle, rather than along the leg of that triangle.

And that the "short leg" of that triangle is 1/64"!!? If you have a 50" fence, then the amount of error at the maximum 50" cut would be equal to the difference in length between the long leg and the hypoteneuse. In a 50" cut, that error is .000002 inches!! Are you seeking better precision than that from a tablesaw??!!

(with a 30" fence, a 1/64" out-of-square condition, and a 30" cut, the error would be .000004 inches). Isn't that a little overkill in precision??

Not exactly what you're saying. The error was a LOT more significant than that. When I'd slide the fence over to 6", lock it down, the measure fence to blade with the Starrett rule, the distance was really 1/64 to 1/32 short. Move the fence more and get MORE error. So at 6" the error is 0.02, the the error at 36" would be 0.12" And all the while the fence is parallel to the saw blade.

I'll make a drawing if confusion persists. All I know is that now whatever distance the scale on the guide tube says the Starrett scale now agrees with.

[Kevin Guarnotta]

I've had a similar issue with my Sawstop, and had no idea what caused it. Thanks for your post, now I know what to look for...

[Clark Harbaugh]

This could be the same issue I've had for years on my Griz. Mine is not quite as pronounced as your was, though. I'm about 1/64" off at 12". I never thought about he geometry of the tube. I just became accustomed to using the rule to measure every cut.

[John Piwaron]

I've had a similar issue with my Sawstop, and had no idea what caused it. Thanks for your post, now I know what to look for...

Glad to be of help. Besides being pleased with a personal victory, however small, after clearing up my issue it occurred to me that maybe what I found might help one or two other people. Especially when I saw in the manual that there was nothing to explicitly assure that the guide tube was parallel to the front edge of the saw.

[Bill Huber]

I am trying to figure this out, my scale is not accurate at all and I can't get it to be.

Here are 2 images, the first is the fence is away from the blade but is parallel to the fence, in the second one the fence is close to the blade but is still parallel to the blade.

This is were I don't understand it, if the scale is set how does it change when you move the fence?

252581 252582

[John Piwaron]

I am trying to figure this out, my scale is not accurate at all and I can't get it to be.

Here are 2 images, the first is the fence is away from the blade but is parallel to the fence, in the second one the fence is close to the blade but is still parallel to the blade.

This is were I don't understand it, if the scale is set how does it change when you move the fence?

252581 252582

Nothing about the stick on scale on the guide tube changes. 6" is still 6". I'm using 6 just for discussion.

Your drawing *is* what I was thinking was going on. The distance the guide tube scale is measuring is the hypotenuse of a right triangle. The distance from the blade you actually move it is the long leg of that triangle. A distance that will be less than what you intended. For me, that was 1/64" + at 6". For someone else that error will be different depending on how out of parallel the guide tube is.

[Bill Huber]

Got it... thanks.

The first one is 4.27 and the second one is 4.14

252586 252587

[johnny means]

Not exactly what you're saying. The error was a LOT more significant than that. When I'd slide the fence over to 6",k it down, the measure fence to blade with the Starrett rule, the distance was really 1/64 to 1/32 short. Move the fence more and get MORE error. So at 6" the error is 0.02, the the error at 36" would be 0.12" And all the while the fence is parallel to the saw blade.

I'll make a drawing if confusion persists. All I know is that now whatever distance the scale on the guide tube says the Starrett scale now agrees with.

If you had that much of discrepancy than there was obviously some issue. A complete tear down and setup may have fixed it, but the geometrics dont work out for the rail angle to cause that large an error.

[Brian Kincaid]

When I had this problem I realized my tape measures were inaccurate. Bought starrett tape measures and all my measuring tools (Incra, etc) now match up
-Brian

[Jerry Miner]

When I had this problem I realized my tape measures were inaccurate. Bought starrett tape measures and all my measuring tools (Incra, etc) now match up
-Brian

I think this is a much more likely cause of the issue. If it were strictly an out-of-parallel fence issue, with these numbers:


So at 6" the error is 0.02, the the error at 36" would be 0.12"

then the fence would have to be out of parallel by almost 3 inches! (if the hypoteneuse is 36.12 and the long leg is 36, then the short leg is the square root of (36.12)^2 - (36)^2 , or 2.94.)

[Jerry Miner]

I don't want to beat this one to death, but let's look at the math from another angle, so to speak:

Suppose you have a 50" Biesemeyer fence. You're installing it on your saw, and the mounting holes give you 1/8" adjustability for parallel to the front of the saw. Let's say you install it as far off as you can (1/8").

Your fence would then travel along the hypoteneuse of a right triangle whose long leg is 50" and whose short leg is .125" (1/8").

You zero-out the scale at 0.0" and move the fence to the 50" mark. Instead of travelling 50", your fence will have travelled 50.000156", a little more than one ten-thousandth of an inch too far! That's why the manual doesn't address this issue. It's a level of precision that is beyond reasonable for a tablesaw.

Even if the rail is 36" long, and the installation is 1/4" out of parallel, the error at 36" is less than a thou!

[Mark Kornell]

If you had that much of discrepancy than there was obviously some issue. A complete tear down and setup may have fixed it, but the geometrics dont work out for the rail angle to cause that large an error.

Johnny is correct. Assuming that the vence was parallel with the blade, and given A^2 + B^2 = C^2, using A = 5.98 and C = 6, results in B = 0.49. In other words, the tube would need to be out of parallel with the front of the table (in actuality, since you are referencing off the blade, you mean off perpendicular to the blade as Dave Richards said) by about 1/2" over a 6" span. About 4" over a 48" span, and I'm certain you would have noticed it being that far out. Your tube rail may have needed adjustment, true, but there was also something else going on.

My saw rail can be adjusted by about 3/4". So if the rail was out of perpendicular by 3/4" over 48", and the fence had been adjusted to be parallel to the blade, the measurement error would be .006" - about 1/170 of an inch. Over 6", the error would be less than .001". So fence rail perpendicular-ness is pretty much a non-issue, which could be why the manual doesn't mention it. My saw's manual doesn't mention the issue, either. It has about 20 pages on aligning the blade, mitre slots and fence, but basically says to eyeball the rails. Having done the math, I now know why.

I did make sure my rails were within 1/16" side to side when I set it up, which puts my measurement error at something less than the day-to-day movement of wood. :-)

[Mark Kornell]

Oops, I see Jerry beat me to the punch while I was working on the math. Must've had better marks than I in high school...

[Richard Wagner]

I'm one who is very particular about alignments; especially the TS blade to fence and DP bit to table top. But when we start talking about errors in the fourth decimal position, I become somewhat uninterested.

Disinterested unless I am working at the Bridgeport. Nonetheless, this was a very interesting thread. Examples like this should have been used by my math teachers. If they had shown such "real world applications" I might have retained more of what was being taught (presented). Teaching does not occur unless there is some learning.

[Barry Wixey]

I thought maybe I would kick this up a notch since I am questioned daily about accuracy issues on table saws that use our digital readout. Most of the time the discrepancy is because the original fence scale is not accurate, however, we have tracked down large discrepancies on a few occasions to fence rails that were actually slightly bowed. If you do the math on a rail bowed on a 60 foot radius you can be way off the farther you get from the blade. This issue can be easily tracked down because the fence will be parallel to the blade at 0 inches but will swing its way out of parallel as you move away from the blade.

[John Piwaron]

Math aside, which given the numbers I posted indicate a vanishingly small error. An amount of error that's most likely not observable by the human eye.

Yet, my direct observations were not wrong and what I did to fix that after combing through how it was setup, has, in the end, apparently yielded the result I was looking for. The distance between the saw blade and the fence, as measured with a Starrett scale, now matches the number the crosshair in the head of the fence over the scale on the guide tube. I'm good. At the moment. I only checked with 2 cuts - one cut at 1" and another at 9". So I might be making an assumption that won't hold true for me at longer distances. I'll do more test cuts. FWIW - when I say the two distances match, I mean that not only do the two measurements match but also the resulting cut is what both scales said it would be. THAT is the proof in my pudding.

Is something else at play? Maybe. Probably. But I'm no longer going to worry about it. For anyone else, this remains something to consider if after everything else someone is still not getting what they want. Make no assumptions. Check everything. I spent north of 3 grand for this saw. My expectations went up with the dollar figure. And in this regard, brand is unimportant.

This is also about trust and confidence. Trusting that the scale on the tube is where it should be and when the fence is positioned at some distance, that the width being looked for is the width that results. The confidence comes in that such a thing remains so going into the future. I think for a while I'll be double checking the setting until after a while I know that doing so is redundant.

The last thing - I agree with Mr. Wagner above. This is the sort of real world example math teachers should include with course material. The practical application of the match being learned. It'd help with the "what am I ever going to use this for?" question.

[John Piwaron]

I thought maybe I would kick this up a notch since I am questioned daily about accuracy issues on table saws that use our digital readout. Most of the time the discrepancy is because the original fence scale is not accurate, however, we have tracked down large discrepancies on a few occasions to fence rails that were actually slightly bowed. If you do the math on a rail bowed on a 60 foot radius you can be way off the farther you get from the blade. This issue can be easily tracked down because the fence will be parallel to the blade at 0 inches but will swing its way out of parallel as you move away from the blade.

Wow. *More* to check. :) This is reminding me more and more of what happens when a plastic injection mold isn't giving the results it should. We now leave the world of plane geometry and enter the Twilight Zone. I mean 3D geometry. :)

[Bob Lang]

I believe that practical experience in the shop trumps sitting at a desk and running trigonometry equations. If I'm understanding the issue, the rail is actually differing in height from the saw table. I bought a Biesemeyer fence in the early 1980s, before they were a part of Delta. At the time this issue was addressed in the manual, and an L-shaped jig cut from 1/4" Masonite was included with the fence. You placed the long section of the jig on the saw table and mounted the angle iron that supported the rail so the it touched the short leg of the L. Following those instructions, as well as the advice to calibrate the cursor with dial calipers (I had to go out and buy) gave me a fence that I could trust the tape on from kissing the blade to the end of the fence.

Bob Lang

[Mike Goetzke]

I agree to disagree with many calculations here, but, maybe look at the assembly instructions for your fence. I had a Uni with a Biese fence and there was a template used to align the fence tube.

Mike

[John Piwaron]

I agree to disagree with many calculations here, but, maybe look at the assembly instructions for your fence. I had a Uni with a Biese fence and there was a template used to align the fence tube.

Mike

Yes, there's a template that gets used to set the height of the angle iron the guide tube is attached to from the saw top. Yes, I've used it. Not recently, at the time of original assembly. I won't be making fun of this idea. The template, or at least the amount it sets that iron to, will be consulted. :) It couldn't hurt to check. As long as it doesn't take too much time.

[johnny means]

I think it's relevant to note tha most Euro sliders actually adjust for fence parallelism by changing the rail angle. IME, professional level sliders are engineered to perform within tolerances that 3k cabinet saws can only dream of. I doubt any of us know more about setting up a saw than the people designing Felders, Altendorfs, Martins, etc. Apparently, they have come to the conclusion that the angle of the fence rail is not all that critical.

[John Piwaron]

It's interesting that Felder is scheduled to come to my hometown in February for the Woodworking show. I look forward to checking out their things in person. Finally.

[johnny means]

It's interesting that Felder is scheduled to come to my hometown in February for the Woodworking show. I look forward to checking out their things in person. Finally.

Be sure to wear a bib.