Accuracy of a square

[Cochese]

Don't have one, besides assembly squares from Rockler. So while I was at HD I picked up their cheapest steel square to see how accurate it was, and if it wasn't to practice bringing one into square.

I measured just now, and got about the following: as best I can measure, I had between a .7 to 1mm delta over a length of about 196mm. That's a .0035-.0051 variance. I give a range, as I measured the graphite lines as best I could.

Would you consider that square enough? I know wood is going to move, but the more dead on I am in the beginning, the better off I'll be I reckon.

Posted this a few places so I can get a larger response value.

[cabinetman]

How did you perform the test? Was it a left line from an edge, and then flip the square over and draw another line. Then measure any gap between the lines?

.

[LCHIEN]

if ~1 mm is the error over 196 mm from vertical then you have a .29 degree error.
I'm assuming what you did was to use the square one way then flip it and mark the other way. That also assumes your reference edge was perfectly straight....? if you used a table edge did you check the edge for straight along the whole length??? If the table edge has a 1 mm curve variance over the length of the twice your baseline then all bets are off.

And, I assume if you measured 2mm between the lines you divided by 2 to get 1mm error since the true vertical lies halfway in between.

if these are all correct then the .29 degrees I would guess is fair... probably good for one of those one piece steel or aluminum "L" squares 24" x 16. But if you are talking 196mm, that's short, about 7 inches. What square do you have?

[Cochese]

This is the square I bought. By no means was I expecting it to be able to do machine work when I bought it, I was probably just curious.

http://www.homedepot.com/Tools-Hardw...&storeId=10051

I used a cutoff sheet of plywood that I can assume to be reasonably straight, and did the flip test. I saw some discrepancy, so I did the flip test again but didn't try to put the second line directly on the first. I used my digital calipers to try and take as accurate of a reading as I could. I measured the bottom and the top of the run, the total distance between the lines at the outside of the graphite. I subtracted the two and came up with .7-1mm (hard to be more precise with a mechanical pencil line). I divided that by the approximate run of the lines (don't have anything that measures mm that long, so I used the step process with the calipers) and came up with my .0035-.0051 reading.

[cabinetman]

You could get a more accurate line test with using an x-Acto knife with a #11 blade, instead a pencil of some sorts.


.

[LCHIEN]

thats a small square, 8" x 12". How'd you align the bottom edge withthe edge of the plywood as that square has no lip... lots of error can get in there.

Using calipers to measure between two pencil lines is also very inaccurate for small distances, calipers are best suited for measuring between physical surfaces and or features like surface to bottom or surface to surface where the jaws can be used as accurate references.

I wouldn't give a whole lot of credence to your values... and that's not the kind of square I'd use for real accurate measurements anyway for lack of a lip.

why are we using mm at all? Isn't the square marked with distance in inches so you know the run length... Small errors in run length don't matter for figuring very small angles.

Measurement techniques are everything if you are trying to measure small errors. You should use a very flat, straight item like a 24" level along the bottom of your wood piece, then rest the edge of the square along the same surface of the level, for example, to ensure the base of the square is perfectly parallel tot he edge of the wood. a factory cut on plywood is probably relatively straight, but maybe not to the amount you are looking for (a few mils - .001" - or less). Make sure the level is straight (test against another level edge to edge and see if you can see light between them, then flip one to make sure they don't have complementary errors). Then push the level against the wood and against put a light behind it and see if you can see light of get a feeler gauge between.

Then you'll know the wood is straight... we have all these assumptions to prove...

If you really want to tune up a square, take a 12" combination square. It has a lip to reference against the edge. Now draw a line and flip the square and draw another line as nearly coincident as you can. If you can see the line diverges, then you can grind a hair off the spot where the rule hits the holder to change the angle. Repeat until you get the lines as parallel as possible. THen your square will be at 90 degrees. this test is pretty good - the error is always doubled by flipping the squares. You should easily be able to see nonparallelism by naked eye of around .005" or better over 10" (250mm) of the blade you'll have extended. That's about 1/8th of a mm. So that should get you down to .014 degrees error from 90.

[sailor55330]

After reading this thread, it has again become completely clear that my mind does not function in the same manner as that of a trained engineer.

At .29 degrees (which I would have never thought to even calculate), I would have simply pulled it tight with the clamp and moved on.

[sailor55330]

After reading this thread, it has again become completely clear that my mind does not function in the same manner as that of a trained engineer.

At .29 degrees, I would have simply pulled it tight with the clamp and moved on.

[LCHIEN]

I took the mental challenge as one of

how close did the OP actually get? (the answer was .29 degrees IF he measured correctly.)
How close is achievable with the correct methods?

now .3 degrees or so, might have been more or less, is actually a lot. Cut a 2x4x8' board in half with a miter saw at an actual cut angle of 89.7 degrees (i.e. .3 degree error) and stand them on the cut end back to back, top face to top face. The tops will be .5" apart and you can surely see that.

OTOH, if you can find a perfectly straight 2x4 to do this with, let me know...

[durango dude]

My $15 Empire combination square (HD) is reasonably accurate. No - it's not precise --- but neither is my measuring and cutting ability. The imprecision is less than the width of a pencil point at 18". I would note that when I made a mantle clock out of walnut, I started appreciating the value of precise cuts. Even so - I likely introduce more error with my sawing techniques than I do with my combination square. For me, I'm nowhere near needing a Starrett combination square.

[Cochese]

FWIW, I did forsee the problem with not having a lip and used a similarly straight piece of 3/4" plywood, ripped for a French cleat, as a registering base. Those are the results I posted. It's possible neither piece of plywood is straight, but it was my assumption. I can repeat the process with the level.

I chose metric because it's much easier to do calculations with. I also have an orange triangle with a lip that I need to check out as well. And an engineer square. Not having a known 90º, I've been attempting to find a reference tool for my shop.

[jdon]

Assuming you have a straight edge and a trammel, you can create your own reference right angle: http://www.mathopenref.com/constangle90.html

Ah, geometry

[greenacres2]

My $15 Empire combination square (HD) is reasonably accurate. No - it's not precise --- but neither is my measuring and cutting ability. The imprecision is less than the width of a pencil point at 18". I would note that when I made a mantle clock out of walnut, I started appreciating the value of precise cuts. Even so - I likely introduce more error with my sawing techniques than I do with my combination square. For me, I'm nowhere near needing a Starrett combination square.

Not meaning to inject levity into a serious discussion, but Durango's comments would apply to me as well...

So, if i can find a square that is off by the inverse of my most commone error--i'd have a chance at perfect resluts???

earl

[gsmittle]

Assuming you have a straight edge and a trammel, you can create your own reference right angle: http://www.mathopenref.com/constangle90.html

Ah, geometry

That made my head hurt. Not to mention I have no clue what the video was doing.

g.

[Cochese]

Not meaning to inject levity into a serious discussion, but Durango's comments would apply to me as well...

So, if i can find a square that is off by the inverse of my most commone error--i'd have a chance at perfect resluts???

earl

Resluts? I'm not sure you want those the second time around.