What's new
What's new

Machinist square

There's crude and there's not so crude. The crude way is to pein the acute side of the beam near the edge (but not at it. The metal's thinning in the vertical direction expands it in the horizontal thus forceing the beam back into square but this repair is not too stable.

If the beam is soldered in the stock (as it once was almost universal) you can gently heat the square up to 400 degrees F or so and tap the beam back onto square. This is tricky because the heat if not uniform can distort the beam or the stock. Newer squares may be assembled with any one of dozens of high strength adhesives and no movement is possible without major surgery.
 
Stan:

Quote: "Sure. Mount it on a good angle plate and grind it."

I can think of how to do either the blade or the stock parallel if one inside edge is in good shape, along with one outside edge square, on say my 6 x 12 machinist square. Can you detail your response a little bit as to how you would finish the other inside edge? (or both) I have angle plates, magneticcubes, compound sine chucks, and tilting magnetic chucks, so creativity it ok. But my biggest grinder only has a 10" wheel so I've never resolved a good solution in my mind. A cup wheell sounds risky?

BTW I've scraped a few CI squares, but as a grinding job on a typical machinist square, I'm interested in hearing more.

Thanks!
smt
 
What size square?

How out of square?

How are your measuring it?

How did it end up out of square?

Most importantly- square to what level of accuracy -and what is your time worth?
 
J R asks some astute questions.

How do you know if a square - um - aint? Scribing two lines etc works for a framing square but it's useless for a hard square whose error is measured in parts per ten thousand as a slope.

A square having a parallel stock (the shorter thick part) and a beam (the blade looking part) is a self-checking tool. That is you need no reference standard to determine the direction and degree of error. The square when reversed serves as its own comparator.

Proving a square is fairly straight forward but for some reason the how-to is not wide-spread. All you need are common shop tools: a dial test indicator, a surface plate, a pair of 1-2-3 blocks, a surface gage, and the square under test. With careful technique you can use this equipment to determine the square's error to 1 part in 10,000.

First clean up the square and glide a stone over it to detect and remove any burrs or raised metal. Mike the stock and beam to ensure parallelism. Check the stock and beam for straightness against the plane of a surface plate.

Invert the mast of the surface gage and position the ball end so it's just off the surface plate. Slant the mast back a ways, mount the DTI in the snug and arrange it so the ball of the mast and the DTI are aligned on a perpendicular from the surface plate.

You need to raise the surface gage so you can take readings from both sides of the beam without disturbing the surface gage. For that, you need precision 1-2-3 blocks to raise it.

Place the 1-2-3 blocks on the surface plate (1 x 3 edges up) and parallel so the stock of the square can be straddled by them. The 1-2-3 blocks straddles the stock raising the surface gage and keeping it parallel to the reference surface. Plant the surface gage on the 1-2-3 blocks.

Allow the square to reach thermal equalibrium. Hereafter, handle the square so your hands put no heat into it. A pair of clean white cotton gloves is recommended.

Place the stock of the square on the surface plate with the beam vertical. Gently bring the beam into contact with the surface gage ball and the DTI. Adjust the surface gage so the ball is in contact with the square while the DTI registers zero. The ball and the DTI tangencies should contact the beam in the center. Approach the surface gage from several slight angles to ensure the vertical plane of the mast ball and DTI contact setting is correct.

Reverse the square and take a second reading. Reverse back into the original position and push the square into contact again for a repeat zero. Repeat this test several time to ensure you get consistant readings.

This is delicate and sensitive work. The heat of your hand can throw off your most careful readings and drive you nuts. However, once you've done it a few times the task becomes less ornerous.

The readings and the repeat zero mean nothing unless they are both present in the test. Record the sign of the error and the amount. Prepare a calibration sticker or certificate containing the date, the method of test (in this case "self check by reveral") your name, the amount of error over the tested distance, the span of the tested distance from mast ball to DTI contact), and an exaggerated sketch of the square showing whether the interior angle is acute or obtuse. Don't forget: the error will be HALF of the actual reading.

Place the certificate in your calibration records (I keep my calibration records an accordian envelope) and apply the sticker to the stock where it will not interfer with its use.

The beauty of this method is it quantitative, that is expresses the error as an actual slope or angle to the degree of accuracy your apparatus affords. Non-quantitaive methods are pass/faill only; that is one square is compared to another which may show an error but you will not know how much nor unless the other square is calibrated. Nor can you determine whether the interior angle is acute, on the money or obtuse. Three squares will mutually prove each other only if they were progressively reworked to eliminate error. At the end of the calibration you will have proven squares to the extent a bright line will demonstrate (damn close by the way) but no numerical figure or calibration accuracy can assigned to them. With suitable gage heads and a differential gaging amplifier plus some metrology refinements in technique, errors in a hard square can be determined to a few millionths of an inch slope per inch. To be practical a surface gage and a 0.0005" resolution DTI is good enough and attainable whith common shop equipment.

If you choose to express the error in angular terms divide half the raw reading by the span between the ball and the indicator contact. Determine the angular error by taking the ARC SINE from the calculated slope making the appropiate conversions from decimal angles to degrees minutes and seconds.

If you know the square's error it doesn't have to be perfect. Knowing the error allows you to compensate for it. By calibrating your square (and to the extent possible your other measuring tools) you can be confident the work you check with a less-than-perfect square is as reliable as if accomplished with a perfect square.
 
Last edited:
Guy Lautard's Second Bedside Reader explains how to make a precision square. I have half a notion to use the method to improve the accuracy of some of my cheapies.

The technique he explains requires the use of three squares. Fine tuning is done by lapping.

I'm not advocating his technique over any others. I'm just pointing out another way of doing things and am not saying one is better than another.

Regards,

Orrin
 
What size square? I have a few

How out of square? Not much

How are your measuring it? on a granite block using a Starret butted up againist it.

How did it end up out of square? Don't know. I got several in a package deal

Most importantly- square to what level of accuracy -and what is your time worth? In my hobby shop my time isn't worth much!

Thanks for the help!
 








 
Back
Top