How straight can you cut?

[promark747]

An unstraight line is an arc, or a more complex set of curves.

Yes, but I don't see mention of the error being due to an unstraight line.  I can understand that with aluminum extrusions, errors could be due to a curve/warp...in which case the extended error could actually be lower over 10' than they are over 1'.

 

[Svar]

An unstraight line is an arc, or a more complex set of curves.

Yes, but I don't see mention of the error being due to an unstraight line.  I can understand that with aluminum extrusions, errors could be due to a curve/warp...in which case the extended error could actually be lower over 10' than they are over 1'.

See the OP: "I can't seem to get a dead straight cut they are all slightly less than .25mm bowed towards the rail over a 2.4m sheet"

 

[promark747]

An unstraight line is an arc, or a more complex set of curves.

Yes, but I don't see mention of the error being due to an unstraight line.  I can understand that with aluminum extrusions, errors could be due to a curve/warp...in which case the extended error could actually be lower over 10' than they are over 1'.

See the OP: "I can't seem to get a dead straight cut they are all slightly less than .25mm bowed towards the rail over a 2.4m sheet"

Gotcha, thanks

 

[Svar]

If your 10' long rail only deviates by .010" or less, then you're straightness is .001" per foot.

It is actually 0.00016" per foot. This is arc (circular) where height to length relationship is non linear.

First, did you mean 0.0016?
Second, I don't agree that it's not linear, at least as far as how Cheese phrased it.  If you are out exactly 0.001 per foot, you would be out exactly 0.010 over 10 feet.  There was no mention of an angle.

No, I meant 0.0001".
I mentioned that I assume circular arc (which is a fair assumption, also see OP).
The height of the segment is h=R(1-cos(alpha/2)). Non linear.
Look at the difference in segment height H and h in this greatly exaggerated picture. In shallow arc the difference is even more pronounced.
If the shape of the curve is unknown, than we can not downscale deviation from 10' to 1' and there is nothing to talk about (as in Cheese's comment).
[attachimg=1]
Forgive my monologue, I'm obsessive-compulsive, which is what we have in common with the original poster.  [big grin]

 

[charley1968]

@ Michael : i stand corrected..