Checking precision squares

[Pete.]

I wanted to make sure that my box square and Chinese granite square were telling the truth in terms of squareness measurement. The box square I got used from somewhere and it's obviously good quality but very old. The granite square I bought new and it did come with a certificate of squareness but we all know what they are worth.

First the box square. I already knew that it had a very tiny error in one face and suspected that it was in the bottom face which is the one that it would be normally standing on.

I already had the gauge set up from yesterday so I used that to check the two pairs of faces on the box square for parallel. The two side faces were spot on, the top and bottom showed out of parallel by 0.00015" (just under 4 microns). The two vertical faces read perfectly parallel (to the limits I could measure them) so that gave me a starting point.

We know that the top and bottom faces were out of parallel but what we don't know is which one is out of square with the sides (or if both of them are). To check for this I set up the following gauge:



If you ever wondered what the ball is for on a surface gauge - this is it (or at least, this is one use). I set the dial point as close to directly above the ball end as I could. It doesn't need to be perfect for these comparative measurements. This dial gauge is VERY sensitive 1 micron resolution (0.00004") and it's very important to make sure all readings are set up exactly identical. To do this I put the square and the gauge alongside the Mercer comparator stand base and took some readings.

I put the square on one good known parallel face and brought it up to the dial gauge then zeroed the dial. This is measuring the top face which is the one I thought would be OK.



Next I carefully backed the square away, flipped it over and brought it up to the dial again.


I got the same reading exactly. The top face must be dead square to both of the two sides. That means it's the bottom face that's got some wear. It's not surprising given the age of the square and the fact that the face has only two narrow lands and a vee in between for measuring cylindrical objects.

So I mixed some 5 micron diamond paste with light oil to make a thin slurry and spread it on one of my lapping plates. I put the square on the slurry mix, held the 'low' end and hinged the 'high' end to and fro in the slurry puddle. The idea is that the end that needs lapping down more has the most movement in the slurry and the low end is just a pivot. I stopped a couple of times when I saw the slurry turning grey (indicating that it was removing metal) and in no time flat I had the faces reading parallel again.

Here's the result from parallel measurement. I moved the square, not the gauge stand. The results were more consistant that way.



To check the granite square I simply set up the stand and zeroed it against the now-known-true box square, then brought the granite square up to the dial.



2 microns out over 8" That's less than a ten-thousandth of an inch so I'm leaving it well alone. Better than I expected TBH.

 

[Parm]

That’s awesome, but why would you ever need that level of accuracy?

Or is this just one of those projects because you know how, and you can?

 

[Hopefuldave]

That’s awesome, but why would you ever need that level of accuracy?

Or is this just one of those projects because you know how, and you can?

If you want to make anything else square, Good Practise is to check it with something considerably more accurate, or you can wind up with cumulative errors - example is checking a mill's spindle or quill's square to the table - DTI mounted to the spindle, against square on the table, run it up and down a few times.

Dave H. (the other one)

 

[Pete.]

That’s awesome, but why would you ever need that level of accuracy?

Or is this just one of those projects because you know how, and you can?

As Dave says these are checking references so they should be at least one order of value more accurate than what you're checking. The square is about 8" square, but the longest way on my mill is 30". If I used the 8" square to check the 30" way for squareness to another I will multiply any error in the instrument by nearly 4 (or rather, I would only be able to detect 1/4 of any error in the way).

Everything about rebuilding is about averaging-out errors. The smaller the errors in my references, the better the result, and those errors might be stacking up or cancelling out there's no way for me to tell. For example a while ago I did a thread about lapping my surface plates flat. This is why - any reading I get also depends not only on how square the box square is but also on how flat the plate is and how large the base is on my surface gauge (hence the use of that huge comparator stand. If I attempted to do this work with the plates all full of dips and bumps I would have been chasing my tail because the erorrs in the plates would have been greater than the errors in my parts.

The damn things also move according to time and temperature. Some of my straight edges are quite new so I have to check them regularly and adjust them sometimes as any redsidual stresses in the casting relieve over time. 0.0002" bend in a straight edge and it doesn't blue up all along, it either pivots on the middle or touches only at the ends. It's a tiny amount and takes only minutes to correct but it does show up in the work. In summer and autumn the sun shines through the door and warms up whatever isinside. If I'm scraping and the sun warms up my straight edge, it bends upwars in the middle quite severely.

 

[mtt.tr]

Here's and example of why and where you would want an accurate square @Parm

 

[Windy Miller]

Here's and example of why and where you would want an accurate square @Parm

View attachment 340247

So what exactly is going on here? Trueing up the cross slide?

 

[Stupoty]

Have you seen the radius'd foot on the indicator holders in oxtools vid on checking squares ? (reminded me of setting up square stuff in a four jaw chuck a bit.

Stu

 

[mtt.tr]

So what exactly is going on here? Trueing up the cross slide?

Yes that indeed, cross slide was scraped into correct alignment.

One side parallel to the bed, though the cross slide is not scraped square
but you need a reliable reference

 

[addjunkie]

That’s awesome, but why would you ever need that level of accuracy?

Or is this just one of those projects because you know how, and you can?

Some like to do things properly to the best of their ability. Others use the near enough for pit work or the Dewhurst Butchers methodology.

 

[Parm]

As Dave says these are checking references so they should be at least one order of value more accurate than what you're checking. The square is about 8" square, but the longest way on my mill is 30". If I used the 8" square to check the 30" way for squareness to another I will multiply any error in the instrument by nearly 4 (or rather, I would only be able to detect 1/4 of any error in the way).

Everything about rebuilding is about averaging-out errors. The smaller the errors in my references, the better the result, and those errors might be stacking up or cancelling out there's no way for me to tell. For example a while ago I did a thread about lapping my surface plates flat. This is why - any reading I get also depends not only on how square the box square is but also on how flat the plate is and how large the base is on my surface gauge (hence the use of that huge comparator stand. If I attempted to do this work with the plates all full of dips and bumps I would have been chasing my tail because the erorrs in the plates would have been greater than the errors in my parts.

The damn things also move according to time and temperature. Some of my straight edges are quite new so I have to check them regularly and adjust them sometimes as any redsidual stresses in the casting relieve over time. 0.0002" bend in a straight edge and it doesn't blue up all along, it either pivots on the middle or touches only at the ends. It's a tiny amount and takes only minutes to correct but it does show up in the work. In summer and autumn the sun shines through the door and warms up whatever isinside. If I'm scraping and the sun warms up my straight edge, it bends upwars in the middle quite severely.

Thanks for that, I’ve learnt something

 

[Parm]

Here's and example of why and where you would want an accurate square @Parm

View attachment 340247

I don’t really understand what going on there

 

[Pete.]

I don’t really understand what going on there

The cross slide way opposite the gib (almost always the side nearest the chuck) controls the direction that the cross slide moves. It has to be almost perfectly square to the bed ways (almost square because the cross slide should always put a tiny amount of dish in the end of a part when facing). If you put a square on the bed as in the picture and indicate the long side by putting the indicator on the saddle then you know that once that long side is clocked in, the short side is at right angles/parallel to the cross slide way. If you want to put some dish into the way you simply indicate a shallow angle into the long way then clamp it in place, then you don't have to think about anything when working on the cross slide, just scrape it zero-zero.

Those small angled ways are a bit of a cowson to do. Take just as long as a large way for me.

 

[Windy Miller]

Just shows how little I know about all this stuff. I had always assumed that the cross slide would automatically be square without giving any though as to WHY or HOW it was square.
The more I learn about this, the more I realise how little I know, and the more I worry that my lathe and mill are knackered!

 

[AKN]

It has to be almost perfectly square to the bed ways (almost square because the cross slide should always put a tiny amount of dish in the end of a part when facing)

Can I ask why?

 

[mtt.tr]

Can I ask why?

From what I understand allows for wear and 2x concave faces will fit together but 2x convexs won't

 

[Parm]

Thanks folks, never ceases to amaze me what you know and how little I know about any of this

 

[Windy Miller]

Thanks folks, never ceases to amaze me what you know and how little I know about any of this

^^what he said!^^