Inaccuracy / Precision drawing
Lloyd Rubidge
are fractionally out of position, or rotated very slightly. Has anyone also
come across this, or have any ideas how to prevent it?
I suspect that this is due to small mistakes in the drafting process (like
snapping to an endpoint in a hatch pattern) or the use of blocks converted
between metric and imperial. It could also be the result of rotating drawing
elements or mirroring them across a non-orthogonal line.
These small errors are then propagated through the whole drawing, becoming
worse as slightly skew lines are used as mirror lines or for offsets. This
is a problem that I have experienced in a number of drawing offices.
Is this common?
D.R.A.
correcting the actual object. Of course they could be picking the incorrect
points.
As for the skewed line, I have found them mostly in drawings from LT users
who have tried to change the angle of a drawing and do not have the "align"
command available
"Lloyd Rubidge" <nospam_...@yahoo.com> wrote in message
news:3b40a...@news1.mweb.co.za...
Dennis Shinn
wrote:
>..... running object snaps. What many do not realise is that when
>you have running object snaps on, if a particular point of geometry is
>not available, a point at the crosshairs is used.
Gawd I wish more people would realize this! The snap markers only
exaccerbate the situation although I do find them handy.
Too few people understand that at some point some obscure intersection
or line enpoint may be a candiate for the next person's snap request.
But so long as the drawing prints correctly, who the hell cares?
That's the attitude that seems to prevail.
To the original poster - keep up your good work and by all means, if
the occasion allows, point out these inaccuracies to whomever is in
charge. Unless he's the one that produced them (hehe)
Well, perhaps even then.
olddog
Endemic, in my experience
Using architect's files as (obviously) the reference source for
producing workshop fabrication data, I can't trust their accuracy.
Architects, in general, see their function as purely creative- fair &
reasonable you may think.
Not when supplying electronic data, it is'nt. Not paying attention to
detail is just not on where CAD's concerned; its not hard to get it
right.
Before going further, a caveat- not all architects generate sloppy
files; but a surprising proportion do! (Also, this seems to apply
mostly to the small to medium commercial projects, those of limited
budgets & tight schedules)
The problem originates in the way users relate to CAD; seeing it as an
electronic pencil results in exactly the faults raised by Lloyd.
Hasn't everybody worked in files where intersecting lines are
'approximated', & don't actually meet. The closer the mismatch, the
harder to detect- painful!
I maintain the best way to avoid this crap is to become
'object-oriented'.
Use polylines, closed shapes, grouped elements, blocks, etc. instead
of lines/arcs/individual elements. Easier to ensure orthogonal
relationships, fully conjoined elements & thus accuracy when
mirroring, rotating etc. Sounds so obvious I feel a bit silly raising
it. (Perhaps the greatest gain in 3D modelling is object generation &
manipulation)
Conclusion- I have to rework every file to ensure it can be used as a
manufacturing template. Anything else is too risky
Thoughts, anyone?
OD
PS Apologies if my terminology is faulty, I'm not an ACAD user
Dennis Shinn
>....
>Thoughts, anyone?
I would simply agree with your observations, for the most part. But it
doesn't seem to matter (in our area) what the size or complexity of
either the project or the office as to the quality of the electronic
files.
You're right in a sense that the architect's role in the process is
one of design and creativity. But as one contactor told me, "If you
can draw it, we can build it, but we can't read your mind".
Michael Bulatovich
> are fractionally out of position, or rotated very slightly. Has anyone
also
> come across this, or have any ideas how to prevent it?
>
> I suspect that this is due to small mistakes in the drafting process (like
> snapping to an endpoint in a hatch pattern) or the use of blocks converted
> between metric and imperial. It could also be the result of rotating
drawing
> elements or mirroring them across a non-orthogonal line.
One other opportunity exists for screwing things up at a very small scale
that the other posters have not mentioned, and that is unit accuracy. If a
drawing is created with acad is set to, say, half inch accuracy, and then
you insert it into a drawing set to 1/256 inch accuracy you will find
dimensional discrepancies. The same is true with angular accuracy.
I used to work in an office that thought that just because they never
dimensioned anything less than half an inch that setting unit accuracy to
that number would be fine. Of course, barring any severe hardware
limitations in the early eighties, they should be drawing with unit accuracy
as high as they could manage and have dimensional accuracy limited to half
an inch if they want to prevent insignificant fractions appearing in their
dimensions. Personally, I have always found that those fractions were good
indicators of trouble with the model and things that needed my attention.
--
MichaelB
Lloyd Rubidge
Really?
I thought that Autocad used full depth of data storage no matter what the
unit and precision settings. So I would not have thought that this would
have been the source of such discrepancies. What I can imagine is the
introduction of strange floating point errors (well, not errors as such)
when rotating elements. Perhaps the point rotated to can not be fully
described by Double precision (8 byte) data. Then, when the point is rotated
back there is very small discrepancy.
Have I got this right?
Any thoughts?
Lloyd Rubidge
"Michael Bulatovich" <don'tspambu...@ica.net> wrote in message
news:RVi27.261106$Z2.31...@nnrp1.uunet.ca...
Jim Patrick
>"Michael Bulatovich" wrote:
>> "Lloyd Rubidge" wrote:
>> > I have often discovered drawings with varying levels of inaccuracy......
>> One other opportunity exists for screwing things up at a very small scale
>> that the other posters have not mentioned, and that is unit accuracy. ......
>I thought that Autocad used full depth of data storage no matter what the
>unit and precision settings. So I would not have thought that this would
>have been the source of such discrepancies......
You are both right and wrong. The program keeps full accuracy, so
you're right there. But a coarse units setting allows the operator to
miss osnaps, and a subsequent check (with coarse settings) won't show
the inaccuracy.
If you can stand the mental conversions, I've found that decimal units
with fractional dimensions are good for preventing that problem.
This is a problem in many CAD packages; one solved it by dis-allowing
fractional dimensions unless they were within .001 of the dimension.
It's a royal pain, but did make for good drawings.
Lloyd Rubidge
From: "Jim Patrick" <jpat...@shentel.net>
Newsgroups: comp.cad.autocad
Sent: Wednesday, July 11, 2001 11:33 PM
Subject: Re: Inaccuracy / Precision drawing
> This is a problem in many CAD packages; one solved it by dis-allowing
> fractional dimensions unless they were within .001 of the dimension.
> It's a royal pain, but did make for good drawings.
What do you mean by "dis-allowing fractional dimensions"?
We work in Metric, to a millimeter scale, so nothing should be done in
fractions. This should avoid the kind of error you are describing.
Lloyd
Tom
Michael Bulatovich <don'tspambu...@ica.net> wrote in message
news:RVi27.261106$Z2.31...@nnrp1.uunet.ca...
>