Scale

I'm having some problems with objects that I cut from one drawing & paste
into another scaling up or down. Usually the scaling is minimal, objects
that should be 500mm are 500.2357 etc. I've checked the scale setting & it
seems to be ok. I thought maybe a different unit setting, but I used the
same seed file for each drawing. Is there anything else that might be
causing this?

 

New one on me. email me a file and I'll take a look

 

It's really odd. Today I encountered the same thing where most objects were
..1mm longer than they should, one or two others were .5mm longer. I'll email
a file to you tomorrow.

 

check your settings in both dgns sometimes you can have the accuracy
different in both drawings ie if i have settings readout accuracy to 0
decimal place then 500 mm not 500.123mm
also if the snaps were set to nearest and not keypoint you will also get
errors, try using the select tool and maybe use information tool (anylise
tool) to get the length
but in anycase copy paste with scale set to one as longas mu, su. psu are
set the same for both files will be correct exact copy.
BTW where are you working I'm at Yarra Valley Water in Mitcham Victoria
Where you at the Bentley V8 summit last week? I was there ...very
helpfull...
Lorys

 

Thanks for your response Lorys. I have the feeling that it might have
something to do with the dimension scale, alhtough I'm not absolutely
certain, I did have a look at it today. When I first looked at the dimension
scale setting it was set to 1.000 but the dimension was still showing with
extra digits after the decimal. When I re-entered the scale of 1 to the
dialog box however the dimension was corrected. I'm not sure this is the
problem with all the drawings since more often than not the amount the
dimension is out by changes between objects.

I work at a renewable energy company in Perth, so I wasn't able to make it
to the seminar.I have to admit that I've been strongly biased against V8
since my previous employer gave it to me to play with. I hate the thing with
a passion, admittedly things might have improved since that release which
was one of the first.

 

Hi Wanda & others,
of course there is no bug in Microstation. If the Active Scale is set to
exact "1", the size of a pasted element is the same as it was in the drawing
from where it was copied. And if the dimension scale is exact "1", the
dimensions show the right size.
But it could be that there is a very little difference to "1", let's say
1.00045, and if you have set the coordinate readout to 2 decimals you can't
see this difference in the dialog boxes.
But this is not a bug of Microstation but a "bug" in the design file since
these parameters are stored in the dgn files not in the Mirostation
environment. If this "bug" is in a seed file, of course it is in all files
that are created from this seed file. This is an error of the user who
created or changed the files, and you should make him spend a beer and beg
your pardon and correct the settings in all files (make the right settings
and then "Save settings" in the file menu).

Greetings, M.B.

 

Are you drawing in 3d- is your design file flat (clutching at straws here)


AC

 

2D

 

Wanda, I am seeing a picture here.
As I see it, you are getting different readings of distance from the same
item in different design files. Is this correct.

As you are cutting and pasting between the two, and are receiving different
numbers you are assuming that it is something in the cutting and pasting.

First question

Did both file originate from the same seed file (I ask this because you
could have
different settings between the two)

Did the same person actually draw both file

I reckon that the design file settings are different, one is set to
reporting dimensions to the
nearest millimeter, the other to 3 decimal places. The elements are going to
be exactly the same
size.

Find the element in quaetion and do an information on it. It will give you
the
absolute co-ordinates of the start point and end point of the line. Do the
math
and I bet you will find the actual lengths are 500.2357

Soory, I did not look at the post closely enough before. What you have here
is crap 2d work. Of twnty users in our office NO ONE is accurate besides
me (thats because I am sad)

Turn your grids on and set them to 1mm

Zoom in to any grid on your plans, and, so long as ther are orthoganal their
co-ordinate
(not their length) should be on a grid to the whole millimeter (unles you
are using brick sizes
in the UK when it should be to the half millimeter)

If ther are not, then it is impossible for the design to be accurate. Add to
that people who
do not snap properly and its no wonder youare getting duff numbers.

The reason you are seeing these numbers differently is probably down to a
settings
difference between files.


HTH

Anthony Cox

 

Yes. or in the unit or dimension settings... admittedly I did fiddle a bit
with the dimension settings as some drawings require more detail of
particular parts & less of others.

Ouch. I drew the drawing from scratch. I think the dimension settings might
be faulty. If I'm allowed, I might try drawing the thing from scratch using
as many decimal places as there are available, making the drawing as
accurate as possible. That might resolve some of the issues we've been
having with lengths & angles. I've always taken a lot of pride in the
accuracy of my work, but then up until recently I've only ever done
electrical & train signalling CAD work.

Because the drawings are done for a customer (Hamersley Iron) I'm not
allowed to play with the settings, although I don't think they would mind me
increasing the number of decimal places for the sake of accuracy.

 

Are you working in V8 or Pre-V8? The accuracy is very different. In V7 it
was all about Positional Units per subunit. Not so in V8

In V8, the accuracy may well be being thrown out by a low degree of accuracy
in your co-ordinate readout. But it can also be affected by where in the
workplane you are modelling. Close the origin, the data storage is more
accurate. Way out in space, not so accurate. That is the way of a floating
point database.

Dennis Barker

 

yech. definitely pre-V8.. it's J something.. 7.1 maybe? Found another issue
today, even when I drew or placed objects exactly when I went back &
measured what I had just drawn each object was out by a small amount, always
by a decimal amount, but it was surprising considering how careful I was to
draw or place it exactly... one object was placed at .0005 further than I
wanted it, I had to completely redo one area because the small amount threw
everything out enough that things were clashing.

 

I don't see your design plane / working units setup in this thread. What
are you using?
We use 2 seeds;
1) Master Units = M (metres) / Sub Units = mm (millimetre)
Resolution = 1000 mm / M - 10 Pos Units / mm
This implies that the smallest unit we can reference is 1 / 10 mm (0.0001).
Any decimals past the 4th can be ignored. The dimesions of the design plane
are 429,496 M Sqaure.
2) Master Units = M (metres) / Sub Units = cm (centimetre)
Resolution = 100 cm / M - 10 Pos Units / cm
This implies that the smallest unit we can reference is 1 / 10 cm, or
millimetres (0.001). The dimensions of the design plane are 4,294,967 M
Square.

If you are designing very small objects, and looking for 4 or 5 decimal
place accuracy, your design plane has to be small. USTN /J and prior
versions store data using 32 bit integers, and the computer does its math in
floating point real numbers, so there is always going to be discrepancies,
but you have to make it relative to what accuracy is required.

There are also readout considerations, and some of the settings are buried
in obscure places. We are not V8 yet, but soon will be. Design plane,
coordinate readout, and angle / distance measurement accuracy will have to
be tested and addressed. I do not anticipate any problem with *new jobs*,
but we won't get far without incorporating *old data*, and we'll see what
happens.

 

I just wanted to add something about the cutting and pasting
between design files. Don't rely on it. The scale will change
by varying amounts - somtimes not at all sometimes noticeably.
I worked in Bentley support a while back and ran into this
problem several times - the bottom line from the developer
was - reference the file in then copy in what you want,
MicroStation will let you cut and paste but wasn't really desinged
to work that way -

HTH

Tom F.

 

In J and before, the Integer based positional units system is fine for
horizontal and vertical whole numbers, but when you start mirroring about
mid points of odd numbers or scaling etc. it is very easy to get inaccurate
values. This is also very much the case with lines at angles.

When I first started using MS, the office was using seed files with only 10
PU per MM, and were complaining about how inaccurate their drawings became
and had chosen not to use autodimensions because of the fraction creep. This
was proven to be terrible when we started on a project that had a central
block and two wings at an angle. The Accuracy of the wing buildings was just
awful.

What you need to understand with positional units in J is that it is just
like drawing on a fine grid. While you may tell it to draw a line at 30
degrees with a length of 2135mm, what it does is draw from your start point
to the nearest addressable point. The more Positional units (PU) you are
able to have per sub unit (SU), then the more accurate this will be.

To emphasis the point I am trying to make, if you have working units set to
one PU per SU, then you will not be able to draw anything that is a fraction
of a sub unit, even though you might key that in. You can only draw to the
actual grid points in the working plane, not half-way between. So even if
you have a finer resolution setting, you still get these variations in line
lengths which is exacerbated when you draw elements at an angle. I hope this
helps in your understanding.

Dennis Barker

 

To clarify - this is in the pre-v8 versions of MicroStation.

 

If in your source file you key-in: AS=1, it will probably display as
1.0000,1.0000. If in your destination file, you key-in AS=.99999, it
will also display as 1.0000,1.0000. If you copy something in the source
file, and place in the same position in the destination file, you should
see that they are now slightly different sizes, even though by all
indications your scales are identical, when they actually are not. You
might want to ensure that the Active Scales are manually set to the same
values; however, do not trust the Scale dialog box -- which displays to
only 4 decimal places -- to display the exact scale. I think the only
way to do that is to use the tcb-> ... variables.

Geoff Houck