Preferred method

[RussL]

I'm just curious as to how people generally go about checking features I believe could and should simply be dimensioned to check.
for instance, perpendicularity of a cylinder to a plane (neither is a part datum)
Most of my co-workers will go to great pains to make a new alignment using the cylinder to align to, and then inspect the plane.
They asked me to give it a try because they were not achieving the results they desired.
When I tell them I would just use a perpendicularity dimension between the cylinder and plane, that does not seem to be adequate for them.

They are just concerned with the perpendiularity here, is there a problem with simply dimensioning the features as I suggest? Is their view based perhaps on what was commonly done years ago, perhaps with TUTOR (which was used in the past here, that I have no experience with)? Or possibly simply a result of not getting the results they desired (part is probably simply OOT).

[bw_bob]

You have a perpendicularity callout with no datums. It has to be perpendicular to some kind of datum. The FCF ???

[RussL]

Customer requested check of perpendicularity between Plane and cylinder regardless of datums.

[craiger_ny]

Oh I see a drawing callout is not as strong as a customer request. Callouts are concise and matter of fact, requests are vague and amibiguous, best to go with a request (if you are the customer). Aren't customers neat? At any rate alignment should not effect the results of evaluating the orientation of one 3D feature to another. Which one you pick though will because that determines which one the datum is. However, your customer does not specify a datum so starts all over the vicious cycle.

[Jan d.]

What I do is measure the plane and level. Then measure circles all the way down the cylinder. Output the location, size and form for the individual circles. That'll teach you what happens at every level. If you have tool drift, or lobing or something, it'll immediately show.

Then construct a cylinder (make sure you press the correct button for in/or outside!!!) through all circles and output size and form. If all those look as to what you expected, evaluate the perpendicularity from the cylinder to the plane.

I find that if I measure the feature as a cylinder and I get dubious results, there is really no clear way to figure out as to what happened. If you measure cirlces, it'll be clear why you got the dubious results. Always works for me!



Jan.

[craiger_ny]

What I do is measure the plane and level. Then measure circles all the way down the cylinder. Output the location, size and form for the individual circles. That'll teach you what happens at every level. If you have tool drift, or lobing or something, it'll immediately show.

Then construct a cylinder (make sure you press the correct button for in/or outside!!!) through all circles and output size and form. If all those look as to what you expected, evaluate the perpendicularity from the cylinder to the plane.

I find that if I measure the feature as a cylinder and I get dubious results, there is really no clear way to figure out as to what happened. If you measure cirlces, it'll be clear why you got the dubious results. Always works for me!



Jan.

Me too. I am an old school skeptic of cyliders with PCDMIS. I never measure them, I always construct. Not just because I don't trust the math situation that PCDMIS uses but because like Jan you learn more from evaluating circles.

Craig

[RussL]

I construct the cylinder too.
and actually the customer was not that vague, the thought is the datum involved means less than this particular surface.
(These are old prints and there is not a chance they will be corrected).
The main question I have is why level to the plane or for that matter normalize the cylinder.
no matter what the alignment, if the features are measured correctly, shouldn't the perpendicularity dimension suffice.

[craiger_ny]

I construct the cylinder too.
and actually the customer was not that vague, the thought is the datum involved means less than this particular surface.
(These are old prints and there is not a chance they will be corrected).
The main question I have is why level to the plane or for that matter normalize the cylinder.
no matter what the alignment, if the features are measured correctly, shouldn't the perpendicularity dimension suffice.

Right, no alignment is needed. However you need to know which is the (implied) datum. The perpendicularity of the plane to the cylinder is not going to be the same as that of the cylider to the plane.

Craig

[Wes Cisco]

Craig & Jan, I use both methods quite a bit. Construct from circles and autocylinder. Which I use depends on the callouts and the depth of the cylinder. If I think two levels will be sufficent for the cylinder, (or that is all the depth will allow), and I have no reason to suspect an out of round condition I use autocylinder. If I seem to have a problem I can dim both the axis start and axis end to see if there is tool drift involved. I do this mostly for speed. I can program 1 autocylinder much faster than I can program two circles and construct a cylinder from them. However if I suspect out of round or I have a deep hole I use multiple circles and construct the cylinder. I have done several comparisons between the two methods and I have found no significant difference between a least squared two level cylinder and one constructed best fit from two least squared circles provided of course the same number of hits for each circle and same location in bore for each level. HTH

[RussL]

Is the fact that I construct the cylinder using circles defined in a work plane parallel (theoretically) to the surface being checked a possible way for error to be introduced into the measurement, if the actual surface is off, if so, I can agree that I would have to normalize the plane, before inspecting and constructing the cylinder (so the circles are on planes perfectly parallel to the measured surface plane).

[Wes Cisco]

RussL, I do not think so. To the best of my knowledge 3D elements are workplane independent at creation as well as when dimensioned. However I can say that I have any proof of this. You could experiment.

[tking]

.02

I think you need to level to the plane if you're going to probe circles first and then construct a cylinder

TK

[#2#]

RussL, I do not think so. To the best of my knowledge 3D elements are workplane independent at creation as well as when dimensioned. However I can say that I have any proof of this. You could experiment.

It is my belief also that 3D elements are workplane independent. They are, however, alignment dependent.

If you want your hit vectors to be as close to perfect as possible...

Measure the plane, align to the plane, and measure it again. Then, measure the cylinder, align to the cylinder and measure it again. This will ensure that if the cylinder isn't perp. that your hits for the second cylinder will be along the correct vector for the cylinder's angle. Then, dimension the perp of the second cylinder and second plane. Which to use as the implied datum is up to you apparently.

[cmmguy]

Wait a minute, isnt perpendicularity view DEPENDENT? ie X/Y, Y/Z, or ZX or whatever the view is. Then would not alignment have a bearing on this view????

[Paul Sarrach]

I also construct, and yes cmmguy, perrendiularity call out is independent, regardless to what ever alignment you are in, it will only only see the plane and the cylinder, actually you don't even need an alignment to do this call out on pcdmis. So I also feel it is a great waste of time to make a special alignment to get the perrendiularity results.