Accuracy of Square

1. Aluminum
Join Date
Sep 2005
Location
Posts
120
Post Thanks / Like
Likes (Given)
2
6

## Accuracy of Square

I am just finishing up a CNC router and want to square it up. I do have a machinist square, but I am suspect if it is very accurate. How can I test?

thanks
d1

2. Cast Iron
Join Date
Feb 2007
Location
wilson nc
Posts
441
Post Thanks / Like
Likes (Given)
0
4
The three, four, five rule will work if you can measure accurately enough. Three squared plus four squared equals five squared. (Pythagoras) Five being the hypotenuse of the triangle and three and four the legs. multiples of the above also work. 6,8,10

3. Best IMHO would be to indicate a granite master with your ways.

Another possibility would be to make up a rigid test bar using Homemade's principle. I'd add some plugs to the end of the test bar (could be just a flat steel bar).

All you'd have to do is drill 4 holes at the outer limits of travel and check how well the plugs drop into your drilled holes. (IOW, checking diagonals of the superimposed rectangle).

4. Carpenters lay theirs against a straightedge, draw a line, flip the square over so the leg's facing the other way, and draw another line. Then, they look for non-parallelism, usually at the extreme end. No divergence of lines=square enough for woodwork.

With metalworking measurement tools, you can use the same general method. A straightedge and dial indicators would get you in the ballpark, unless you have some fancier stuff. Make sure to test the outside edge against itself, rather than against the inside edge. If you do the latter, you're just measuring the parallelism of the blade, not the "squareness of the square."

Testing your router (wood?) for square could be as straightforward as making the largest triangle you can, and then testing it as if it were a carpenter's square. You could 'register' the material against the side of the frame, if that's important to you, or you could just cut x-y then diagonal, which would measure the accuracy of the router location system without regard to what it's mounted in or on.

Hope this helps...
Chip

5. Depends on the size of your table. If it is 12" X 12" a 6" X 4" or 8" X 6" precision square should do fine. If it is larger, even much larger, than that I would use the 'equal-diagonals' method.

You need to have accurately machined reference surfaces or points at the corners.

Assuming your CNC router bed is a square or rectangle.
It helps to set everything up on as flat a surface as you can find.
Use a trammel, which can be just about any piece of metal (pref steel) with some carefully mounted points.
First determine that two opposite ways are as exactly parallel as you can make or adjust them. Then do the same for the other set of ways.
Then make or mark something as carefully as you can at each corner.
Determine that the length of each diagonal is as equal as you can make them. The actual length in inches or mm or whatever doesn't matter. What matters is that they are equal in length.

If you can get those measurements listed above as close as you can... your table will be as square as you can make it.

It isn't rocket science. The Egyptians (and probably others) figured it all out using sticks and strings about seven thousand years ago.

-DU-

6. Stainless
Join Date
Feb 2007
Location
Angleton, Texas
Posts
1,714
Post Thanks / Like
Likes (Given)
939
536

How are you going to adjust it?

When I used to square the travel on our water-jets I'd cut a 12" x 12" square and measured across the corners.

The rest was just math and knowing which number to put where.
Last edited by KilrB; 04-29-2009 at 04:09 PM. Reason: Not wordy enough.

7. Cast Iron
Join Date
Jul 2006
Location
Cumberland, Maine
Posts
453
Post Thanks / Like
Likes (Given)
44
95
Originally Posted by d1camero
I am just finishing up a CNC router and want to square it up. I do have a machinist square, but I am suspect if it is very accurate. How can I test?

thanks
d1
Others have suggested some geometric methods, which are fine if you can measure accurately enough.

I'd test the square to see if it is accurate. You need a flat surface - a granite surface plate is ideal, but a tablesaw top or jointer bed may be OK for your purposes - and two other squares. Rest the beams of two squares on the surface plate and bring the blades together. A light source behind them will let you see how well they match.

You do need to test with three squares. Two squares can match up perfectly, but can be off by the same amount in opposite directions. If the third square matches each of the first two, you know they are all good.

8. Aluminum
Join Date
Sep 2005
Location
Posts
120
Post Thanks / Like
Likes (Given)
2
6
Thanks everyone for the great suggestions. I tried a number of the above methods and the machinist square that I thought was good is way off. About .003" over 6". Not good enough for helping to set up my large CNC router.

So... KBC has a few options:

(1) a 8 1/2" TMX brand for \$123CDN (accuracy class "H" - whatever that is?!?!)

(2) a Starrett #55 6" for \$283CDN (no accuracy specification)

(3) a Mitutoyo 6" for \$238CDN (.00014")

(4) also there is a SOWA STM 9 1/2" square (accuracy class H) for about \$150

Starrett might be a nice name, but no accuracy specification bugs me. What do you guys suggest?

thanks
d1

BTW I use my machinists square for my other machine work (milling and turning). So having a good quality square will be a bonus.

9. ## Granite Master Angle

Mine is accurate, i.e. flat, square and/or parallel as appropriate, to 25 millionths of an inch per foot.

Use this with a tenths-reading dial indicator to check squareness and you'll be pretty close.

- Leigh

10. Mac
Aluminum
Join Date
Sep 2004
Location
Carrollton, Georgia
Posts
128
Post Thanks / Like
Likes (Given)
0
2
If you have a lathe and a mike, it is fairly simple to make a cylinder square to compare your square to.

11. Aluminum
Join Date
Sep 2005
Location
Posts
120
Post Thanks / Like
Likes (Given)
2
6
Originally Posted by The real Leigh
Use this with a tenths-reading dial indicator to check squareness and you'll be pretty close.
Leigh, I am all ears and have the gear, just not the know how.

12. chip chester has the right idea. A square is self checking with very little work.

I don't know what you are checking, but say its x to y. Make a good surface on the table where you can put the beam of the square. Put the square in several times and tram the blade, and average the readings. Lets just say the square reads .006 acute in 12" travel. Now flip the square over and repeat, keeping the beam against the same stops in the same location. Tram several times and average. Lets say your readings are dead nuts square now. This tells you that the square is the average of the two (.006 and 0), or .003 acute. And your machine is also out .003.

This assumes you have no straightness errors in your ways, but I would expect you have already checked that.

Using this method you can build and calibrate your own square for machine checking. The absolute squareness is not as important as the calibration.

13. This comes up from time to time. As was noted in earlier posts a square is self checking. A machinist square is a far more accurate tool than a framing square so the scribes line method will not be satisfactor except for rejecting squares far enough out of calibration that sucj a test reveals error.

In a machine shop you need to quanitfy error not merely detect it. If you know the magnitude of the error and it's direction you can compensate and work to greater accuracy that the toolig woukld otherwise allow.

Checking a square by miking its diagonal contans a built in error: you don't know the amount the corners are rounded from perfect sharpness. If you could determine the error (is it a 0.010" R or a 0.015" R - or a wear hyperbola) you could compensate for it obtaining a valid figue for the hypoteneuse.

Here is a general instruction for checking square to a high degree of accuracy using simple shop tools.

------- --------- --------- ---------- ---------

Part of being a machinist is knowing your tools and equipment and whether they are accurate enough for a particular set of circumstances.

Those of you having no need for calibrated squares or making verifiable squareness determinations should read no further nor need they scoff at the following text's applicability to their particular line of work. It's intended to illustrate calibration technique for those having an interest or a need for the topic.

Checking square against square only ensures that their stocks are parallel when their beams are butted together. Scribing a perpendicular, then reversing a square to eyeball the error is adequate for a framing square but laughably crude when applied to even rudimentary squareness determinations used in the machine shop. OTH, you need three squares to check in rotation. While this method can be used to verify absolute squareness it lacks the ability to quantify the squareness of any particular item needed to be square.

Quantify means ",,,express as a number or measure or quantity..." My definition of quantify as applied to squareness error means a measure of the error and its direction (ie acute or obtuse stated in terms of the interior or exterior angle. Also it means that the square's calibration can be verified by third party inspection.

A square is self checking, that is no standard is required to assess a particular square’s accuracy: it can be used to check itself.

Quantitatively checking a square to very close tolerances is bone simple with a minimum of equipment. All you need is a granite flat, a surface gage, a pair of good 1-2-3 blocks, and a dial test indicator.

First and most important: clean and de-bur the square to be tested. Verify the parallelism and straightness of the stock and beam independently. Verify parallelism and equal size of the 1-2-3 blocks in the 2" dimension.

Reverse the surface gage mast so its ball end extends down through the notch of the surface gage base. Use the surface plate as a flat reference:

Mount the dial test indicator on the upper part of the mast.

Set the pair of 1-2-3 blocks on the 2" edge or on end so the stock of the square can pass between them

Set the surface gage base on the 1-2-3 blocks. Ensure the stock of the square can pass under the base and between the 1-2-3 blocks.

Position the square so the ball of the mast ball contacts the beam close to the stock.

Adjust the DTI so its ball contacts the beam towards the upper end. Beware of cosine error and adjust the indicator;s contact so it is withint 20 degrees of parallel to the beam. Note: work carefully and gently. minimizing heat input to the square and the surface gage. The DTI contact has to be on a perpendicular over the tangency of the ball end on the square. This is necessarily a finicky adjustment. The DTI cannot be properly nulled unless this adjustment is within close limits.

Slide the square back and gently re-contact the ball of the mast rotating the square slightly in the vertical plane to null the reading. Zero the indicator. Re-test a few times to ensure consistency.

Slide the square out and reverse it 180 degrees on the surface plate. Contact the ball and the indicator again. Note the reading.

Reverse the square and run the stock back under the surface gage base. Note the REPEAT ZERO READING. If the reading reading is not zero adjust and tweak until it is.

The indicator reading is DOUBLE the actual square error. Note the algebraic sign of the error. Calculate the raw readings to convert them to a slope. Apply a calibration sticker to the square or note in the calibration records (you meticulously keep). Place a certificate of calibration in the square's box. I suggest you use wording like "Error 0.000X per ft - interior angle ACUTE (or OBTUSE)." This provides enduring notice of the amount of square's error and its direction. Anyone using it can use the error information to compensate its known error. Thus it's possible to make squareness determinations that are more accurate than the square itself.

I discussed this procedure in some detail in my Home Shop Machinist article in Jan/Feb 2005 and there are some pictures to see. I didn't use a surface gage for the article. Instead I used a gadget I built. A surface gage is quite tricky to adjust for this evolution and, once set, its adjustments can easily slip.

The same procedure may be used to check other items. Angle plates may be checked by using adapting this technique but a pair of precision parallels and a couple of clamps must be added to the equipment.

Wish I was smart enough to post pictures.

14. Originally Posted by d1camero
Thanks everyone for the great suggestions. I tried a number of the above methods and the machinist square that I thought was good is way off. About .003" over 6". Not good enough for helping to set up my large CNC router.
Actually... that is fine. You now know exactly (depending on the method) how 'square' your square is. A bit more of a hassle to use but still usable. You just need to fit a 0.003" feeler gage at the end that is 'out'. IOW what you now have is a 'calibrated' square.

Starrett might be a nice name, but no accuracy specification bugs me. What do you guys suggest?
Starrett is a nice name and all my Starrett No. 55, 20, and 21 squares are at least as accurate as Starrett claims. Starrett claims an accuracy of 0.0001" per 6" for their No. 55, 20 and 21 model squares. The Starrett No. 55 is the same as their No. 20 except with beveled edges. I bought all my Starretts used off of eBay, never payed more than \$50, and I suppose I have been lucky because they were all in excellent condition and square. Nicest square I have is a Mauser 7.5" beveled edge square that was made in Germany for Scherr Tumico. \$70 off of eBay. Looks to be brand new. As far as I can measure it appears to be dead accurate for it's full length.

As mentioned earlier you can make your own cylindrical square on the lathe. Turn a piece of large diameter (3-4 inches) by as long as you need on the lathe so that you get a consistent diameter over the entire length. Face one end (or both) undercut the faces about 0.050" leaving a 1/4" wide 'belt' at the ends.

You can sometimes get them even cheaper from Enco sometimes they go on sale for as little as \$100 for the 12"X4" model.

As far as 'squareness checkers' are concerned that has been discussed a number of times on the forum. Here, here, and here (to name a few).

BTW I use my machinists square for my other machine work (milling and turning). So having a good quality square will be a bonus.
Good idea to have several. I even find the little 1-1/2" or 2" Starrett No. 20s to be pretty handy. You must be very careful using precision steel squares for setting up on the mill. Very easy to ding the edge of the blade which may raise a burr several thousandths. If that happens carefully stone off the burr.

-DU-

15. Stainless
Join Date
Jan 2007
Location
Florida
Posts
981
Post Thanks / Like
Likes (Given)
3
69

16. Thank you Glenn. That's the procedure I was trying to describe exactly. BTW I learned theis technique from "Two Teths" Tommy Thompson, my mentor when I rotated through lay-out and inspection when I was an apprentice in 1963.

What's illustrated differs slightly because a straight edge is used to negate the need to null the indicator but this carries repeatability risks. Stuff moves quite easily on a clean granite plate.

17. Diamond
Join Date
Feb 2004
Location
peekskill, NY
Posts
24,506
Post Thanks / Like
Likes (Given)
0
4597
Unless I miss my guess that photo is from the moore book, foundations of mechanical
accuracy. The surface plate is a four foot square hand-scraped cast iron one! (not
granite....)

Jim

18. Hot Rolled
Join Date
Mar 2005
Location
Canberra, Australia
Posts
592
Post Thanks / Like
Likes (Given)
0
0
Making a cylindrical square isn't quite as easy as David describes. Because the spindle of a lathe can pump in and out swashplate-fashion if the thrust bearing is imperfectly perpendicular to the spindle, facing cuts can be slightly out of truth. To avoid this the final facing cut should be done between centers using "dead" centers. If there is any runout at this stage due to the centers drilled in the end of the cylindrical square not being perfectly concentric with the outside of the cylinder, of course that will create an error.
A cylindrical square is perfectly easy to check in a similar fashion to described by Forrest, but all you need is to rotate the square, it should always give the same reading against the DTI.

- Mike -

19. Originally Posted by HelEx
Making a cylindrical square isn't quite as easy as David describes. Because the spindle of a lathe can pump in and out swashplate-fashion if the thrust bearing is imperfectly perpendicular to the spindle, facing cuts can be slightly out of truth. To avoid this the final facing cut should be done between centers using "dead" centers.
Mike,

Good point. Thanks for the correction.

If there is any runout at this stage due to the centers drilled in the end of the cylindrical square not being perfectly concentric with the outside of the cylinder, of course that will create an error.
Perhaps best to do all cuts between centers then. My commercial (SPI) cylindrical square is, I believe, ground between centers. It has a very fine finish so it may even be lapped. The centers are still quite obvious and there is also an offset hole that may have been used for a driving pin.

A cylindrical square is perfectly easy to check in a similar fashion to described by Forrest, but all you need is to rotate the square, it should always give the same reading against the DTI.
My cylindrical square weighs about 45 lbs. I prefer to move the my right angle gage around the cylinder

-DU-

20. I bought a granite square from CDCO for not much money,for the shop I worked at. I also have 3 other American made granite squares. With high spot blue as a guide,the CDCO square was quite accurate compared to my high quality granite squares.

#### Posting Permissions

• You may not post new threads
• You may not post replies
• You may not post attachments
• You may not edit your posts
•