Rounding "Problem" on Dims?

[engAlright]

We have two parts, that are mirrors of each other, that each have dimensions that measure 33.96875. Our standard is 16th fractional dims for these parts - on one drawing the above dimension shows as 34", on the other drawing the equivalent dim shows 33 15/16". When you set each dim to decimal, they are equal - 33.96875. The dim is taken from a model edge to another model edge in each case.

33.96785 is exactly halfway between 33 15/16" and 34" so it seems like SW is using different ways to round the dim on each drawing, though I can't figure out what's driving that.

Has anyone seen anything similar?

Thanks!

 

[rockguy]

Yes this happens to us all the time. It just seems to depend on which way SolidWorks feels like going.

Rob Rodriguez CSWP

http://www.axiscadsolutions.com

http://www.robrodriguez.com

http://designsmarter.typepad.com/rob_rodriguez/

Eastern Region SWUGN Representative

http://www.swugn.org

http://www.nvtswug.com

SW 2007 SP 2.0

 

[ctopher]

I don't it is rounding them different, just showing different dims, ie decimal, fraction or whole number, etc.
If you want the part to be 34", then it would be created as 34.00000, not 33.96785.
If you want the part to be 33-15/16", then create it at 33.93750

Chris
SolidWorks 07 3.0/PDMWorks 07
AutoCAD 06
ctopher's home (updated 04-08-07)

 

[CorBlimeyLimey]

What may be happening is that the actual dimensions are 33.963750000000001 and 33.96374999999999999 or similar.

I agree with Chris. Make the part the size it is supposed to be. Why design a part at one size and dimension it as another?

It's one thing to express 33.96875 as an exact fraction (33-31/32), it's quite another to express it as rounded to another denominator (33-15/16 or 34). To make it more extreme, consider 33.5" rounded it to the nearest 1" ... or the nearest foot?

If it's a purchased part, then an exact reference dimension should be used.


SW07-SP3.1
SW06-SP5.1

 

[engAlright]

Thanks for the feedback!

Long story short these parts are formed sheet metal. We take the flat pattern and use that for laser-cutting - laser cuts to 1/32" tolerance so we usually design parts to "nominal" dimensions whatever those might be. On the drawing we also show the "formed" shape of the sheet metal part, on which the tolerances are 1/16" - which is where the SolidWorks rounding gets involved.

I was just curious as to why SW interprets the same dim 2 different ways on 2 different drawings.

Thanks again!

 

[rockguy]

It's not that the parts are modeled to an incorrect dimension. They are modeled correctly and happen to be of a dimension that falls on the 1/32" increment but the dimension settings are set to a 1/16" accuracy. The real issue is SW rounds the exact dimension to its 1/16" accuracy equivalent in differeent directions for no reason. I can measure the same part ten times and it will round differently (to the nearest 1/16" every time. It's not consistant. Of course if we used decimals this would be a non issue since we could give the exact dimesnion.

Rob Rodriguez CSWP

http://www.axiscadsolutions.com

http://www.robrodriguez.com

http://designsmarter.typepad.com/rob_rodriguez/

Eastern Region SWUGN Representative

http://www.swugn.org

http://www.nvtswug.com

SW 2007 SP 2.0

 

[ctopher]

thread559-129217


Chris
SolidWorks 07 3.0/PDMWorks 07
AutoCAD 06
ctopher's home (updated 04-08-07)

 

[engAlright]

Interesting, I never really understood that odd/even rule either!

However, in this case the "true" dimension is exactly the same on both drawings, yet SW is displaying them 2 different ways.

What's even more funky...this morning we tried creating a virtual sharp to a corner, rather than measuring off edges (another way of taking the same dimension) and the dim displayed the same way on both drawings: 34" (which I suppose also fits with the odd/even rule).

So when dimensioning to the virtual sharp the display dim is consistent between drawings, however when dimensioning to model edges it is not (at least in this case)...

The plot thickens...

 

[macduff]

Ok, now this is real scary....Is SW changing this on-the-fly on software releases without notice? What happens to legacy data that's brought into a new release? Does it feel like rounding up or down? How about a supplier that has an old worn copy of a drawing and requests a re-print of that same document with no ECO change? It sounds like the dimensional values will change automatically and now you’re screwed because your part is out-of-print.

Colin


Macduff
Colin Fitzpatrick
Mechanical Design Engineer
Solidworks 2007 SP 2.2
Dell 390 XP Pro SP 2
nVida Quadro FX 3450/4000

 

[macduff]

One more thing....
Our CATIA friends at Dassault were infamous at doing this. It cost my last company thousands of dollars making ECO changes and re-checking of drawings.

Macduff
Colin Fitzpatrick
Mechanical Design Engineer
Solidworks 2007 SP 2.2
Dell 390 XP Pro SP 2
nVida Quadro FX 3450/4000

 

[CorBlimeyLimey]

That's another good reason the model should be exactly the size you want the dimension to show.

This is SolidWorks not ACAD.


SW07-SP3.1
SW06-SP5.1

 

[macduff]

CBL,
Our standard back at my old company in Cleveland was draw to nominal no mattter how many decimal places it was. Most of time it was 5 places and show 4 places on the drawing. (The tolerance was a text value/not link to the dimension) Example: .5123+.0003-.0002 with the nominal being .51235. What was happening was the drawing value changed from: .5123 to: .5124 between software releases and was never told by Dassault nor documented. The older release was taking the 5 digit which was the value of 5 and rounded it down. the new release of CATIA was taking the 5 digit which was the value of 5 and rounded it up, bumping the dimension automatically.

Macduff
Colin Fitzpatrick
Mechanical Design Engineer
Solidworks 2007 SP 2.2
Dell 390 XP Pro SP 2
nVida Quadro FX 3450/4000

 

[EEnd]

Discussions like this are why I do not trust reprinting an old version of a drawing by using the as built revisions from PDMWorks. There just seems to be too may ways that the new print might be different than the original.

Eric

 

[ctopher]

From PDMW, you will only see dims changed that were changed. It's not going to auto change rounding of dims in the as bult rev or newest rev ... unless something related to that geometry has updated.

Chris
SolidWorks 07 3.0/PDMWorks 07
AutoCAD 06
ctopher's home (updated 04-08-07)

 

[engAlright]

Looks like I may just have to send this in to our VAR...I can't reproduce the problem on new parts so it may be a quirky thing with this specific drawing...

Thanks for the feedback

 

[looslib]

You said that one of the parts was a mirror. The problem may be out further in the precision of the model/assembly. A difference of as little as .000000000001 can cause the dimension to round up or down. Mirrored designs are known to have very small inconsistencies like this.


"Wildfires are dangerous, hard to control, and economically catastrophic."

Ben Loosli
Sr IS Technologist
L-3 Communications

 

[ctopher]

That's why I hate looking in the mirror.

Chris
SolidWorks 07 3.0/PDMWorks 07
AutoCAD 06
ctopher's home (updated 04-08-07)