2015: dynamic block increment stretch off by <1/256"?!

[p0026376]

I have a block with a segment of base length 1-81/256". I want to have an incremental stretch so it can be a multiple of that length. However, when I set it up to stretch, with an increment of 1-81/256" and then stretch it, it does not line up with a line copied at that same segment distance. In even more odd outcome, the distance it is off measures as 0" (units set to 1/256"), but if I stretch it far enough I start to get increments of 1/256" offset (3*~1/256", 6*~2/256", etc).

 

Has anyone seen this behavior or know how to "fix" it?

 

thanks!

[Dadgad]

Have you tried doing REGEN? I suspect that the line and the stretch will then line up.

 

The rest of it sounds pretty odd.

Pardon my asking, just curious, 256ths of an inch seems rather precise, coming from a carpenter's & steel detailer's perspective.

What sort of work are you doing, or does the block represent?

[Dana W]

You are right Dadgad, there is no such thing as 1/8" out in the field. Heck, muh saw blade's thicker'n that.

 

I am guessing NANO-COMBAT ROBOTS.

 

I am not sure you can measure distances less than your precision level. (UNITS >> Precision) AutoCad rounds less than half to zero. Anything greater than half your precision level is rounded up to it, 1/256. You just told AutoCad you don't care about anything smaller.

 

I keep my units precision at least double my dimstyle precision, sometimes I use an even wider spread, for arrays and such. That probably is tilting with windmills, though. I have been told that AutoCad cannot be inaccurate. Bhhuuuuwwhaahaahaahhaaaaahhhaaaa ha.

 

I have seen simple arrays get a little off too, especially without specifying a separation distance. It is pretty much always less than the precision level off.

 

Yeah, I am not sure AutoCad can get that accurate all the time. I am pretty convinced it draws at exactly the precision level you give it. In other words 1/256 will be a fudge factor, of sorts.

 

I've even seen objects mirrored with a line placed orthogonally, be twisted off of perfectly perp. Only way to know is to see that one little zigged pixel in a ten foot line.

[p0026376]

the drawing was one that was handed to me, so the initial conditions were out of my control. I normally don't think anything of the chosen units in cases like this. That said, I'm using the drawing elements as a basis for creating a standard set of blocks for producing this type of work in the future.

 

In doing so is when I noticed this oddity.

That said, I began to redraw the elements in more sane units and this issue did not arise.

 

That said, it seems the issue is similar to a drawing done in, say 1/256, but then the units are changed to 1/16, and that lost precision appears in objects showing as a certain distance actually being "in between" that distance. I just thought that was odd since I can't choose a more granular measurement than 1/256".

 

thanks!

[Dana W]

The incremental miss error may be in the very first object, and either builds or falls off with the stretching, maybe?

[Dadgad]

Alwys a good idea to start from scratch, then you know what you've got.

Rounding fractions is bound to cause problems, so glad to have given up the Imperial system and have moved on to decimal based metric.