Axis of a square shaft

[CheckerHater]

I have a simple question which unfortutely doesn't have simple answer (or I couldn't find it).

Imagine a shaft having two sections - round and square
To simulate working conditions of the shaft I have to put square end into collet, rotate it and measure Run-out of the round end - something perfectly do-able in the real physical world.
Now to document this process I have to specify Run-out in relation to the axis ... of a square, and ASME Y14.5 is not very clear on that matter.

If you look into definition of the Coaxiality, it basically sais that everything can have an axis, as long as it has a cross-section, but I REALLY don't want to invoke Coaxiality on that part.

Any ideas?

 

[KENAT]

Well, your choice of handle is unfortunate, but I suppose I'll try and help anyway.

I'd look along the lines of making the widths of the square the datum features (both directions). Then tying in the diameter portion to those datums.

What do you mean by invoke coaxiality? I'm not familiar with a separate coaxiality control. There are several controls which relate to/can control coaxiatily, position, runout, concentricity... but not an explicit coaxiality control that I'm aware of.

Posting guidelines faq731-376

http://eng-tips.com/market.cfm?

(probably not aimed specifically at you)
What is Engineering anyway: faq1088-1484​

 

[CheckerHater]

Thank you KENAT,

In definition of Coaxialty they want you to slice multiple cross-sections, find centers of those cross-sections, draw the line as close as possible to the centers, etc., etc., etc. This way you can find axis of any irregular shape. Nevertheless Coaxiality is way too complicated control to use for square stick with round end. That's what I meant.

About handle: if you ever had your design butchered by illiterate hack who didn't even understand how the machine works, you wouldn't judge so harshly. I guess I've had disproportional number of those in the past; kind of traumatic childhood experience

 

[dingy2]

Approximately, how long is the shaft? How long is the square section on the length of the shaft? I will assume that one end is square while the other end is cylindrical??

Dave D.

http://www.qmsi.ca

 

[CheckerHater]

to dingy2:

You are correct. Lengths are different (multiple parts) but definately enough to get a grip.

To develop KENAT's idea, please see the picture.

Datum A is a plane, datum B is a plane, 2 planes have common line, so you MIGHT say "common axis of A and B", but I've never seen anything like that before.

 

[CheckerHater]

Sorry.

 

[drawoh]

CheckerHater,

How accurate do you need to be?

The two sets of opposing faces of your shaft make perfectly good primary and secondary datums if they are accurate. Apply the datum symbols to your width specifications. You will have to decide whether to call up the two sets of faces at MMC or not. It is easier to design the fixture at MMC.

You can use a positional tolerance to locate the round section of your shaft, in combination with the diameter specification. Run-out is meaningful if your faces are significantly more accurate than the run-out specification.

The whole point of datums is that these are your fixturing features. You are meant to clamp to the two sets of opposing faces for inspection and for fabrication. Your two flat faces are good fixture points if your cantilevered round section is short and rigid. If you are supporting your round section at the end, you need to re-think your datums, and your inspection process, and your drawings.

JHG

 

[KENAT]

I'm not sure about "A-B", otherwise that's kind of what I was thinking.

I realize A and B are equally important in this case, not primary then secondary which I think is what you're trying to say but I'm not sure that's how it would be interpreted & don't have time to look it up right now.

As to the handle, funny, most of the people I know that hate checkers it's because of their own failings which they didn't like the checker pointing out. Although I've heard a few horror stories.

Posting guidelines faq731-376

http://eng-tips.com/market.cfm?

(probably not aimed specifically at you)
What is Engineering anyway: faq1088-1484​

 

[dingy2]

One can create an axis by using the square feature as mentioned by Kenat.

I would suggest using a leader line to the square feature and then use the square symbol followed by the size of the square feature with tolerance. Have a drop down datum symbol from the leader line with an "A". The axis is now developed by the square feature.

The square feature is indeed long enough to become a stable datum. If the length of this feature was extremely short, then one couldn't use it as a datum by itself.

One can either have a circular runout or, as you have shown, a total runout relative to datum A. Total runout is quite severe since it includes off center, roundness and size distortion along the length of the feature.

If you chose to use a circular runout, you can dictate where it is to be checked by using chain lines otherwise it probably would be confirmed in the center of the feature's length. It includes off center and roundness combined.

Hope this helps.

Dave D.

http://www.qmsi.ca

 

[CheckerHater]

Thank you all!

To dingy2: agree that total runout may be overkill, but I was just trying to illustrate the idea. Will adjust.

About applying one single datum A to entire square; I never realized that applying datum to square symbol works the same way as applying datum to diameter. Probably missed it. Could you provide a citation (point to text or picture in 14.5)?

 

[KENAT]

Just had a quick flick through 14.5M-1994 and nothing jumped out at me. I was trying to extrapolate what Dingy says from first principles, or by comparison to what you can do with a pattern of features, but it escapes me.

Posting guidelines faq731-376

http://eng-tips.com/market.cfm?

(probably not aimed specifically at you)
What is Engineering anyway: faq1088-1484​

 

[dingy2]

Searching through the 2009 version, I found fig. 4-46 on page 84 reflects a square of 12 mm as datum B. I does qualify with a profile tolerance all around but it is an example of a square feature of size as a datum.

Fig. 4-3 page 50 has a slot shown as a datum. Here, again, the non-round feature of size qualified or had to meet a profile requirement.

Fig. 3-18 page 43 has an example of the square symbol for a square feature of size.

I didn't find that much in the 94 standard except many examples of widths as datums.

Here is how I extrapolated this concept from the 94 standard. We apply positional on features of size and those features could be square rather than round. If we have 1 square hole in a plate where a carriage bolt assembles, could we designate that hole as datum B? Does it need to be round hole to become a datum? It can be qualified with perpendicularity from datum A if warranted the same as a round hole. We show the shape of the hole (square symbol rather than round) with its size tolerance.

We may need another hole for anti-rotation so we pick another hole and qualify with a positional tolerance from datum A and B. The tertiary hole could also be square or it could be round.

Does this makes sense?

Dave D.

http://www.qmsi.ca

 

[KENAT]

It certainly doesn't sound unreasonable, but I wonder if it would be understood by others - or hold up in court if necessary.

Posting guidelines faq731-376

http://eng-tips.com/market.cfm?

(probably not aimed specifically at you)
What is Engineering anyway: faq1088-1484​

 

[pmarc]

I would just like to add something shortly. Paragraph 9.3.1 of Y14.5-2009 std. entitled "Datum Features for Runout Tolerances" says:

"The datum axis for a runout tolerance may be established by a cylindrical datum feature of sufficient length, two or more cylindrical datum features having sufficient axial separation, or a cylindrical datum feature and a face at right angles to it."

There is nothing here about datum features which have square shape. Of course one can say that standard only says "may be established" so this should not be treated as a strict rule.

 

[dingy2]

pmar:

You are correct. In this case, we only have an option of positional in RFS.

Dave D.

http://www.qmsi.ca

 

[MechNorth]

I like the idea of using the square as an irregular datum feature, but I think there is a problem with using linearly-toleranced width dimensions; how do the center planes of each opposed pair of faces relate to each other? If you use two separate datums (A & B) as opposed to multiple datum features (A-B), then the secondary can be related to the primary. Using the profile all-around (i.e. Fig. 4-46)relates the four faces to each other (and to the primary datum), and therefore the centerplanes as well; the intersection of those centerplanes, of course, is the datum axis.

As an alternative, use linear tolerances on the widths, designate areas on all four faces as datum target areas-A1 thru A4. It will work if the relationship between the four faces is non-critical. Plus, the geometries of the areas can reflect the actual in-use contacts.

Jim Sykes, P.Eng, GDTP-S
Profile Services

http://www.profileservices.ca

TecEase, Inc.

http://www.tec-ease.com

 

[flash3780]

I think you can establish the tolerance zone you're looking for by using a composite position (to A, B, & C at RFS), profile (without datum references), and cylindricity applied to the round surface.

A and B would be the centerplanes of the opposing square faces and C would be the end of the shaft.

 

[TheTick]

Giving OP a star because this is one of the best questions I have seen here. A real puzzler, and not the spawn of ignorance of standards or "folk drafting".

 

[KENAT]

Just thinking out loud here, but how about reversing the situation, depending obviously on function. How about making the cylinder your datum feature and locating the 'square' to it?

This eliminates the issue of the width one way of the square being primary and the other way secondary.

Posting guidelines faq731-376

http://eng-tips.com/market.cfm?

(probably not aimed specifically at you)
What is Engineering anyway: faq1088-1484​

 

[CheckerHater]

To dingy2: This was exactly my concern: When it comes to non-round shape profile is understood, but axis is kind of murky.

To KENAT: Sorry to confuse you with "coaxiality" in earlier post, it was definately "concentricity"; just happen to think os something else at the same time.

To TheTick: Thank you. I was going thru the pile of old drawings and found this check (square in collet, run-out of the round end) placed on the drawing as a verbal description. I thought: "Stupid people, why not to use GD&T?" Now I know why.

To everybody: Thank you again. I guess establishing that there is no universally accepted solution (at least today) is also an answer. Maybe definition of run-out (or axis) still has long way to go.