GD&T for controlling excessive "twist." 1

[whotmewory]

Anybody know of a better way - other than straightness - to specify against excessive "twist" when welding to a long, narrow flat?

I know straightness will govern "bowing along the full length, but does it also govern "twist" along an axis tangent to the long axis?

Chris in NC

 

[Heckler]

Give us a little background on this weldment. Are you using assembly fixturing to control this? What's your datum structure?

 

[whotmewory]

This piece - while simple - will have a high-visibility function on a Beta-unit that will be going around to a number of [customer's] plants for "ooohs and ahhhs."

It's a 48-inch long by 2-inch wide x 1/2-inch flat bar onto which will be welded some L-angle pieces with plugs and threads. This is a very, very simple fabrication - which means lots of room for a screw-up during welding.

While drafting this up, it occured to me that I wasn't sure how to specify this piece not to have any perceiveable twist along its full length. Straightness will cover bowing along its full length, but if you think about it, specifying straightness ACROSS the large piece doesn't do the job. The thing could "corscrew" and still incrementally fall within flatness or straightness.

 

[ewh]

If you specify flatness tight enough, there will be no perceptable "corkscrew". The problem comes in producing it.

 

[drawoh]

whotmewory,

I agree with ewh about flatness. A profile tolerance will let you control the face with respect to your mount points or some other datum surface. It depends on how much control you want.

Just don't forget this is a weldment. They cannot weld to within a profile of .001".

JHG

 

[MadMango]

Are you worried about twist or are you worried about the location of the brackets?

I think Flatness will work, but if you want to get complicated set a datum at one edge, then use a positional tolerances on the welded-on brackets. Their position should control the twist in your part if welded correctly.

[green]"I think there is a world market for maybe five computers."[/green]
Thomas Watson, chairman of IBM, 1943.
Have you read faq731-376 to make the best use of Eng-Tips Forums?

 

[ringman]

If this is for an OOOH and AHHH response, I would suggest that you not overly complicate the weldment. I would Perhaps you could forget about GD and T and verbally describe what it is you are wanting. Try to be generous with any tolerances.

 

[AusSteelMan]

Instead of (or as well as) stating flatness across the piece, state it in the drawing for along the piece. Therefore straightness will take out bowing as you suggest and flatness will take out corkscrew as ewh suggests.

Just a quick opinion
ASM

 

[AusSteelMan]

Naturally I also agree with MadMango's datum set and comments about tolerances on what seems to be some sort of prototype / display model. Constructability is a major factor in things/designs like this.

Regards
ASM

 

[whotmewory]

OK. Let's pretend this isn't a simple bracket and that - maybe I'm blowing this out of proportion - but given that this is surely something that comes up, how DO we specify for preventing Receiving or QA from accepting outsourced fabrications that are twisted or distorted but not called out opn any drawing?

Is this an oversight in GD & T?

Chris in NC

 

[ewh]

I still don't understand why you find that flatness is not an appropriate call out in this case. It will control straightness as well as flatness of a surface (if it's not straight, how can it be flat?).
In answer to your last post, if it is not specified on the drawing (or contract), then Receiving or QA has no basis for rejection. You can tell them to reject any parts that don't meet your expectations, but the supplier still has the right to be paid for manufacturing something that meets the specifications they were given (contract or drawing or both).

 

[MikeHalloran]

When you buy a shape from the mill, there are industry standard tolerances for, and industry standard ways to measure, 'camber' (deviation from flat along the length) and 'twist' (angular deviation from 'not twisted'). A single callout for flatness would be unreasonably tight for camber if made tight enough to get visually imperceptible twist.

I suggest using a flatness callout to control camber, and additionally limiting twist by means of a limit on the position of one corner relative to a line datum at the opposite end and a point datum at the adjacent corner.

Maybe I can make that more clear with a different example: Twist in automotive connecting rods is specified with extended tolerance zones implying the presence of 6" long gage bars pressed into the big and small ends, and twist is specified as a maximum deviation from planarity of one end of the smaller gage bar, relative to a plane defined by both ends of the larger gage bar and the opposite end of the small gage bar. At least that's how they did it in olden days, before GD&T left the toolroom.



Mike Halloran
Pembroke Pines, FL, USA

 

[MechNorth]

First, to drawoh, tks for mentioning Surface Profile as a possible control. Most people are timid to use it because it hasn't had appropriate emphasis/respect in most training material or in ASME Y14.5 for that matter (that's hopefully to change in the 2007 revision). It's the only way to control coplanar surfaces that are going to form a primary datum surface; it essentially becomes a flatness callout and ties the two (or more) surfaces together wrt each other. You can also use a composite surface profile to further restrict the twist, to the same effect as using a composite flatness callout. I made a suggestion for this in thread 404-139334 which whotmewory initiated on the same topic.

I'm new to this site, so I don't know the experience / training level of any of the respondents. If it helps the discussion, I am ASME Y14.5M-1994 GDTP-S certified, and I've had considerable training from a number of renowned instructors. I have used the composite flatness callout on similar applications many times; it works when the tolerances and unit area are appropriate.

Also, ewh is quite correct about the supplier's right to be paid if you don't provide them with a spec. However, you might put out a general notice to your vendors that covers such items, and cc your receivers. Much better idea to put it on the drawings though, as some parts may need tighter / looser tolerances than the norm.

 

[MadMango]

MechNorth, your contributions are appreciated. Hope to see you around the Fourms.

[green]"I think there is a world market for maybe five computers."[/green]
Thomas Watson, chairman of IBM, 1943.
Have you read faq731-376 to make the best use of Eng-Tips Forums?

 

[ewh]

Glad to have you here, MechNorth.

 

[dingy2]

Flatness is the term needed but apply a liberal tolerance. Flatness is on a plane while straightness is a componenet of flatness and is an axis.

Checking for flatness is tough on the shop floor though.

Do not use profile since this relates to datums while flatness is to itself.

Hope this helps

DD