Need Help!- How to measure perpendicularity

[JerryBurks]

I am getting to the point of assembling the gantry of my new machine. Now I am faced with the problem of how to measure that the x and y axis are sufficiently perpendicular to each other. I do have a large 48" carpenter square but I have no idea how accurate that is and even if accurate it is difficult to discern small errors visually. An error of only 1/100 of a degree would lead to 6/1000" position error over 36 inch distance. Both axes being in different planes 8" apart does not make it easier to measure.

I can probably fine tune the y-direction to a degree using the two x-screws but since I am building a pretty stiff gantry that will literally meet some resistance.

Anyway, I am wondering if there are any good tricks or tools to establish optimum perpendicularity already while assembling without multimillion dollar measurement equipment.
Or, to say if a different way, I want to avoid making an assembly error that I cannot fix later.

Thanks, JB

[hub]

Originally Posted by JerryBurks I am getting to the point of assembling the gantry of my new machine. Now I am faced with the problem of how to measure that the x and y axis are sufficiently perpendicular to each other. I do have a large 48" carpenter square but I have no idea how accurate that is and even if accurate it is difficult to discern small errors visually. An error of only 1/100 of a degree would lead to 6/1000" position error over 36 inch distance. Both axes being in different planes 8" apart does not make it easier to measure. I can probably fine tune the y-direction to a degree using the two x-screws but since I am building a pretty stiff gantry that will literally meet some resistance. Anyway, I am wondering if there are any good tricks or tools to establish optimum perpendicularity already while assembling without multimillion dollar measurement equipment. Thanks, JB

Get a dial indicator. Fix it to X axis for example, and run it against Y to align.

[JerryBurks]

Originally Posted by hub Get a dial indicator. Fix it to X axis for example, and run it against Y to align.

I have a bunch of dial indicators. But they measure distance. I am looking for a method to precisely measure the angle.

[Michael In Cali]

Originally Posted by JerryBurks I have a bunch of dial indicators. But they measure distance. I am looking for a method to precisely measure the angle.

You don't need to measure angles. You need to measure how much you are out over a distance. Angles are horribly inaccurate.

In your example, 0.01 degrees seems like a good number until you figure you are actually out by 0.006". What matters is the 0.006.

If you care about the angle, calculate it.

[JerryBurks]

Originally Posted by Michael In Cali You don't need to measure angles. You need to measure how much you are out over a distance. Angles are horribly inaccurate. In your example, 0.01 degrees seems like a good number until you figure you are actually out by 0.006". What matters is the 0.006. If you care about the angle, calculate it.

Well, that was my point but that is where I have the problem. I need to measure that distance from two reference points along the y-axis (e.g. 0 and 36") and 8 inches up from the x-axis. That would require two long precision squares solidly connected to give me the reference point in mid-air.

One option may be to somehow measure it with a dial gage using my carpenter square and then flip the square over taking the average to eliminate the inaccuracy of the square. I should have drawn a picture to show what I want to accomplish.

I just thought there might be kind of a standard simpler method of doing that.

[RomanLini]

One-piece metal carpenters squares are usually pretty good, as they are stamped out on a steel die made on a $$$ pro CNC machine.

You can reverse the square (as you said) and if the square is good it will be the same in both directions. If your square is a bit out it will reverse the error so you will easily know if your square itself is a problem.

I don't use a dial gauge but instead slide a piece of shim (paper etc) under the square to test for gaps. It also helps to have some good flat stock to clamp to the gantry etc, I have some granite plates and some 1/4" thick glass plates which are all extremely flat.

Computer printer paper is about 0.1mm thick, and the thin sheet glossy junk mail paper is about 0.03mm thick. If you align to the point where you can't slide the 0.03mm paper under either side of a 18" square you machine will be aligned as good or better than a "pro" built router.

[louieatienza]

Originally Posted by RomanLini One-piece metal carpenters squares are usually pretty good, as they are stamped out on a steel die made on a $$$ pro CNC machine. You can reverse the square (as you said) and if the square is good it will be the same in both directions. If your square is a bit out it will reverse the error so you will easily know if your square itself is a problem. I don't use a dial gauge but instead slide a piece of shim (paper etc) under the square to test for gaps. It also helps to have some good flat stock to clamp to the gantry etc, I have some granite plates and some 1/4" thick glass plates which are all extremely flat. Computer printer paper is about 0.1mm thick, and the thin sheet glossy junk mail paper is about 0.03mm thick. If you align to the point where you can't slide the 0.03mm paper under either side of a 18" square you machine will be aligned as good or better than a "pro" built router.

An old carpenter's trick for trueing up a carpenter's (or rafter or framing) square: after checking the square, you can correct it with a center punch or nail set. If the angle is greater than 90 degrees, make a punch mark on the outside corner, and vise versa...

[hub]

Originally Posted by JerryBurks I have a bunch of dial indicators. But they measure distance. I am looking for a method to precisely measure the angle.

You can use a dial indicator to measure angle. For example, if the dial indicator says it's off 0.5mm on a 1000mm distance the angle would be 0.03deg.

[JerryBurks]

I found a solution, see below but I need to improve it to get to the required precision. More info here

[brtech]

the direct way to measure perpendicularity is to measure the diagonals of a large rectangle from the gantry extremes. Zero difference is square.

If you want to use a square, you do it with a dial indicator mounted to the gantry. You run the gantry up and down one axis, adjusting the square so that you get zero change as you move. You now have aligned the square to the axis. Then you measure the maximum difference on the other axis using the same setup. Zero would be perfectly square.

What everyone is saying to the OP is that you loosen the fasteners that the gantry is mounted to and push the gantry to get it as square as you can, then tighten. Then you use the X and A home position to get the last little bit of error out. This of course depends on how the gantry is mounted to the carriages or whatever you have that the gantry is connected to on each side. On mine, the Gantry extrusion is fastened to cncrouterparts carriages - two on each side. I loosen the bolts that hold the gantry extrusion to the carriages, move it, then tighten the fasteners.

[PaulRowntree]

It has always seemed odd (to me) that we work so hard to make the 3 axes mathematically perfect and orthogonal, when software could make corrections on-the-fly; these corrections would be best inserted into the lowest-level of software, the motion controller.
This is especially true for the X-Y axes being discussed here. It would be much simpler to simply construct a stiff system with no adjustment points, measure the skew angle, then tell Mach3 how the real tool deviates from the orthogonal ideal. In general, we can measure to a higher precision than we can build. I think that MC2 has the flexibility to redefine the motion axes, but I don't know if this feature is widely used.
Z with respect to the X-Y plane is a bit tougher because the tool axis defines a vector, not a point, but tramming Z is relatively easy.

One of the reasons that chemical instrumentation (my field) has advanced so far in the past decade is because we no longer try to build perfect instruments : we now build them to be stable and predictable, and use software to correct for any non-linearities or residual problems.

[JerryBurks]

Originally Posted by brtech the direct way to measure perpendicularity is to measure the diagonals of a large rectangle from the gantry extremes. Zero difference is square. If you want to use a square, you do it with a dial indicator mounted to the gantry. You run the gantry up and down one axis, adjusting the square so that you get zero change as you move. You now have aligned the square to the axis. Then you measure the maximum difference on the other axis using the same setup. Zero would be perfectly square.

That works when the whole thing is assembled. But I was trying to get the gantry posts in the proper position BEFORE assembly and there was nothing to run along the y-axis. That is why I had to make do with rules and squares. Now after I have everything together I can do what you say and verify that it is still good.

Originally Posted by brtech What everyone is saying to the OP is that you loosen the fasteners that the gantry is mounted to and push the gantry to get it as square as you can, then tighten. Then you use the X and A home position to get the last little bit of error out. This of course depends on how the gantry is mounted to the carriages or whatever you have that the gantry is connected to on each side. On mine, the Gantry extrusion is fastened to cncrouterparts carriages - two on each side. I loosen the bolts that hold the gantry extrusion to the carriages, move it, then tighten the fasteners.

I suppose this one of the disadvantages of my build method....Once assembled (i.e. glued and screwed) it is a block. I have to get it right on the first attempt. On the other hand, nothing is going to wiggle loose here