Precision of the program

[jessica.miller]

How does one adjust the precision at which the program reads data?  We've been bitten more than once when the precision of one entity is 0.0000001 off from what it should be.  I'd like to set the working precision for the program to a lesser value.  In other words increase the tolerances of data input.  Please advise.  

[jessica.miller]

In relation to the above:

We tend to key in stations b/c we've been burned numerous times for picking (with the mouse in model space) a station rather then keying it in.  Reason being the "picked" point is accounted for out to .00000001 for accuracy.  When our data & all annotation round to .01.  This will result in a bad cross section every time!!  I want to know how to make the "picked" point only recognize .01.  Please advise.

[Joe-Bouza]

Hi Jessica

Sounds like microstation.

If I understand you correctly there is no working precision in AutoCAD only display precision everything internally calculates out 16 decimals ( I thinks is the number bandied about).

UNITS will give you the display precision settings.

[Joe-Bouza]

If you are picking points the transparent command tool bar offers more precision by allowing you to key in stations and offsets to the precision needed

[jessica.miller]

...And this we do all the time to get around the 16 or 8 decimal place precision; to ensure accuracy.  You add in the human factor & the tight precision of the program is a killer!  It would be nice to have this adjustment within the programming/settings.

[Joe-Bouza]

Not sure if I follow you correctly. Instead of precise key entry via transparent commands are you suggestion the program force precision to a randomly picked point? 

Can you elaborate on what exactly you are doing with these cross sections? What points are you picking points?

[dgorsman]

Coordinates are double-precision floating point values, so extreme coordinates will lose some precision.  Although I have to admit the original question could benefit from more context.  16 decimal places for common units e.g. meters, feet, even kilometers surpasses all engineering tools and most scientific ones:

0.0000001 meters = 0.0001 MILLIMETERS

0.0000001 kilometers = 0.0001 meters = 0.1 MILLIMETERS

0.0000001 feet = 0.0000012 INCHES

Being "off" by that much shouldn't be a factor.

[jessica.miller]

I'm referring mainly to the relation of sample line stations & corridor frequency stations.  Currently if I want a section view to display correctly at 200+00 I have to have:

1) sample line @ 200+00.00

2) section view @ 200+00.00 (generated from a sample line)

3) corridor frequency @ 200+00.00

When adding sample lines & or corridor frequencies, there is an option to either key in the station or to physically "pick" the point in model space.  

If I key in the station for both sample line & frequency, everything is kosher.  

If I "pick" the point in model space instead, the program sees that station identification out to 200+00.00000001 for the sample line & 200+00.0001 for the frequency, causing a discrepancy in the station value; resulting in the section not looking like it should because the sample line & the frequency are technically not the same station.  What I want the program to do is to recognize the point "picked" in models space at 2 decimal place precision.  OR to recognize any station >199+99.5 & <200+00.4 as 200+00, so that if my frequency & sample line are only off by .000001 it will still recognize the break lines from inserting the assembly at said station.

I have not found where to change this precision...

[jessica.miller]

My point exactly!! but as we are using the program today, it recognizes that difference.  Causing us negative results in production.  

How do I set the program to recognize a max. difference?

[Joe-Bouza]

If you are going to pick and not snap to an exact location then like I said earlier use the transparent tool bar, and key it in.

[jessica.miller]

Right.  This is our current process, but as a cad manger I can't make/insure that each individual is doing so.  Hence wanting the option for precision.

[Joe-Bouza]

I cant think of anything to help that except a SR to the deskers.

Forgive me for dragging this out, but how does 0.0000001 adversely effect your section? Your topo shots are probably plus minus 1.25-2 inches using total station an GPS 

[jessica.miller]

I'm sorry, a what?

Thanks for all of your feedback!

[Cadguru42]

How do you even know it's picking such a small number when you're selecting a station from the model space? I just don't see what the problem actually is as others have pointed out. The base data can't be that accurate and neither can something be constructed to that kind of tolerance. I've had problems with Bubba being able to get within 0.2' of a foot for accuracy when actually constructing, let alone 0.000001'. 

[jessica.miller]

We use code sets to annotate our cross sections.  Corridors are either xref'd or data ref'd into production .dwgs.  In order for the program to show this annotation, or even earthwork shapes correctly, in section view at the desired station, a corridor frequency has to be at that EXACT station.  The program otherwise triangulates across the station; either not displaying the annotation or showing cut/fill outside our toc/tos.  I know this because I've had to move a sample line .00001 before to match a frequency & vice versa.

[rl_jackson]

I'm trying to wrap my head around why your concerned with a number that exceed PPM when your labeling to the tenth or hundredth of a foot. Is the Back of Curb a highway to Tulsa or Mars. The program itself is going to have a certain degree of error no matter what you do. Rounding is rounding, weather it happens in the tenths or hundreds, or 16 places behind the decimal, it just happens based on normal scientific procedures. Maybe adjust the precision of your units to show only 2 or 3 places when you list a line or inquire on a station.

[jessica.miller]

It has zero to do with smart labels...  The precision of the label is irrelevant, other than to identify the problem.

The program recognizes a difference in stationing out to that precision.  I don't want it to.  I want it to only recognize about 4 or 2.

Here's a similar situation, exact same concept"

Profile starts at 20+00.00000000

corridor starts at 199+99.99999999

The corridor will shoot down to zero for that 0.00000001.  The surface will also triangulate for the 0.00000001 difference.  Not cool.

I'm merely suggesting that the program disregard the human error of 0.00000001.  That the user be able to give the program a tolerance of some number, say 0.0001 instead of 0.00000001.  That it consider the two stations as the same.  

[Cadguru42]

---

@jessica.miller wrote:

We use code sets to annotate our cross sections.  Corridors are either xref'd or data ref'd into production .dwgs.  In order for the program to show this annotation, or even earthwork shapes correctly, in section view at the desired station, a corridor frequency has to be at that EXACT station.  The program otherwise triangulates across the station; either not displaying the annotation or showing cut/fill outside our toc/tos.  I know this because I've had to move a sample line .00001 before to match a frequency & vice versa.

---

I'm still not understanding what the problem actually is. When you create sample lines you can tell it to make them on even stations. If you need a section on other geometry points, such as a PC or something, then it's easy as snapping to the alignment when picking. I've never seen a section label not work due to the sample line not being exactly on a corridor frequency station. Even if the sample line is 0.00001 off from the station and it's averaging the data between the stations, a 0.00001 difference won't even show up to make a difference in any labeling nor any earthwork quantities, let alone surface creation. 

If your concern is due to things like the corridor not reading the beginning profile then just extend your profile slightly before the corridor region starts and/or start your corridor slightly past the beginning of the profile. This has been standard practice since C3D came out almost 12 years ago. 

[jessica.miller]

---

@Cadguru42 wrote:

---

@jessica.miller wrote:

We use code sets to annotate our cross sections.  Corridors are either xref'd or data ref'd into production .dwgs.  In order for the program to show this annotation, or even earthwork shapes correctly, in section view at the desired station, a corridor frequency has to be at that EXACT station.  The program otherwise triangulates across the station; either not displaying the annotation or showing cut/fill outside our toc/tos.  I know this because I've had to move a sample line .00001 before to match a frequency & vice versa.

---

I'm still not understanding what the problem actually is. When you create sample lines you can tell it to make them on even stations. If you need a section on other geometry points, such as a PC or something, then it's easy as snapping to the alignment when picking. I've never seen a section label not work due to the sample line not being exactly on a corridor frequency station. Even if the sample line is 0.00001 off from the station and it's averaging the data between the stations, a 0.00001 difference won't even show up to make a difference in any labeling nor any earthwork quantities, let alone surface creation. 

If your concern is due to things like the corridor not reading the beginning profile then just extend your profile slightly before the corridor region starts and/or start your corridor slightly past the beginning of the profile. This has been standard practice since C3D came out almost 12 years ago. 

The highlighted portion hits the point exactly.  It's NOT as easy as snapping, & it should be!  When one snaps, even with osnaps on, the user may actually pick station 200+00.00000001 unknowingly.   Then because of that, the corridor doesn't see the profile at 20+00.00 & the sample line doesn't see the corridor at station 200+00.00000000 because, to the program, it doesn't exist...such a negligible difference away...I don't know how to make this more clear.  I know what to do to make it work for what I need, but I want the program to fix it for me.

Also, let me clarify that we do not use section labels for all annotation.  (Maybe we should be...)  We are using code sets for repetitive annotation.  Even if we changed our labeling system, we'd still want the variances of precision option in the program, mainly to lessen the human error factor.  

I hope this helps.

[doni49]

I can see what you're saying -- I've run into it myself.  Alignment ends at 200+00.  Corridor ends at 200+00.  Profile ends at 199+99.99999.  So because there was no profile at station 200.00, the corridor dropped to 0.00.  I usually alter my profile to make sure it ends at 200.00.