Straightness of a Tube

[Underspec]

I'm trying to measure the straightness of a tube. It's about 4" long. The prints tells me a .060" tolerance. What would be the best way to measure and/or dimension this type of feature?

[Wes Cisco]

Well I think that depends on a few things. You work in plastics don't you? You could measure a cylinder, level and align to it, then measure the diameter in about 10 places and compare locations. You measure a line along the length of the tube in several places around the diameter and apply straightness/flatness to each of those lines. How many hits and how far apart depends on how much variation your processes are likely to create. I often see metal tubes with a bow, but few where the deviation is more of a wave or irregular pattern, but I can see that happening with some plastic form processes. HTH

[William Johnson]

You could also put it between centers and indicate it that way, and not use the cmm. This would probably be more repeatable than the cmm.

[John Kingston]

Wes,
Couldn't he construct a line from those cylinders and dimension straightness?

Underspec,
How many do you have to measure? Does it have to be on the CMM? If it is only a few parts you'll spend more time programming than you would checking it on a surface plate. I only mention this because I've had instances were people want me to check their part (Don't go there guys) on the CMM cause it is cool. Later I find out it is a 4 piece order that I could've done a plate layout in 1/4 of the CMM programming time.

William said what I meant without whining like I did!

[kbotta]

Becareful here underspec!
There are more than 1 kind of straightness req's.
Is it str line element?
is it str axis?

Is the FCF modified at MMC???
kev

[#2#]

I was thinking along the lines of Wes's post. I was thinking measure a cylinder and align to it. Then, measure a bunch of different circles along the cylinder. Then I would try to create a line using the centers of all the circles and dimension the straightness of that. I'm not sure if workplane effects the straightness dimension though. You might have to dim. it in several workplanes and report the worst.

Putting it on centers would work.

A third way might be to put it on the surface plate and roll it looking for the high spots. Stick a gage pin under the high spot to find how high it is. Of course this way would be a little tedious if you have more than a couple of these to do.

DANG I type slow! Either that or think slow!

[Wes Cisco]

Wes,
Couldn't he construct a line from those cylinders and dimension straightness?

Underspec,
How many do you have to measure? Does it have to be on the CMM? If it is only a few parts you'll spend more time programming than you would checking it on a surface plate. I only mention this because I've had instances were people want me to check their part (Don't go there guys) on the CMM cause it is cool. Later I find out it is a 4 piece order that I could've done a plate layout in 1/4 of the CMM programming time.

William said what I meant without whining like I did!

Yep I agree entirely, for this kind of form, even if there are many of them, surface plate, centers, indicator is the way to go unless there is some reason, (like an unreasonable customer seeking the cool factor), to put it on the CMM. And yes, you could construct a line from the circles and dim straightness from it, keeping in mind what the best fit construction formulas are doing with the surface data you collect! What is most important here, the center of the tube, the outer surface, or the inner surface, at any given point? Only function can determine that.

[Underspec]

2 shots - 30pc cap study

I need a repeatable way that would eliminate as many variables as I can. I would think a CMM is best in this situation.

Kbotta,
The print does not call out any FCF. It's a "Note" that the customer wants measured. It does not specify what kind of straightness. The print does mention "straightness overall".

[#2#]

2 shots - 30pc cap study

I need a repeatable way that would eliminate as many variables as I can. I would think a CMM is best in this situation.

Kbotta,
The print does not call out any FCF. It's a "Note" that the customer wants measured. It does not specify what kind of straightness. The print does mention "straightness overall".

Oh, sounds like a good callout to me.

Any way you can get away with reporting that the tube conforms without having to actually report a number?

[Wes Cisco]

With a callout like that, I would be tempted to take a careful one eye open one eye squinted gander at arms length and say they look pretty straight over all to me, but I ain't no engineer. (Thankfully!)

[Underspec]

LOL. I guess it's a dimension semi-critical to the functionality of the part

When constructing a line through the circles, doesn't PCDMIS use bestfit? Or would constructing a line from two cylinders be best?

[craiger_ny]

How do you do a capability study on a zero based, one sided tolerance callout. Capability is a function of your process spread and its centeredness on the design. You can not have a centered process on a zero based tolerance with no minus tol. Understanding that standard deviation and the average can be calculated and that you can subtract from the average the lower spec limit (0) to divide by 3 standard deviations. However, when you calculate the upper spec limit minus the average over 3 standard deviations you could get a lower number, which would have to be the representative of your capability ratio. That number would not however, truly indicate your capability ratio because it is derived not from a process center but from a one sided process with a target that is 0. Just curious, there may be a way but I have been under the impression that this is not a good statistical tool to apply to call outs such as straightness, flatness, roundness, runout, etal. Theoretically (using a 0.005 straightness callout as an example) if you applied the formula properly a process that hovers tightly at creating parts that are out of straight 0.0025 would show a better capability than a process hovering tightly at 0.000 (perfect). Just a thought.

Craig

[Bass Assasin]

You would want to use the CPU and PPU. As you move closer to 0 you will see this number get larger. Some programs you can set 0 as a low boundry and takes this as you cannot go past it and then makes your CPK and PPK what your CPU and PPU is instead of the lower of the CPL,CPU,PPL,and PPU. Someone please jump in if I'm off the mark on this.

[craiger_ny]

You would want to use the CPU and PPU. As you move closer to 0 you will see this number get larger. Some programs you can set 0 as a low boundry and takes this as you cannot go past it and then makes your CPK and PPK what your CPU and PPU is instead of the lower of the CPL,CPU,PPL,and PPU. Someone please jump in if I'm off the mark on this.

http://i58.photobucket.com/albums/g2...er_ny/1Rep.png Rep to you, thanks I have never used this before. Then again I've never had to do cap studies on zero based tolerances.

Craig

[cmmguy]

If you construct the line and evaluate the straigthness, is it a 3D value or 2D depending on the axis viewed? Would you want to make sure that you construct the line as a 3D line?