Re: [ATM] Dumb Question: how to measure a...

["Donald Good" <donald.good@comcast.net>]

Make a simple turntable (e.g. plywood on a bolt) and an arm that can be
clamped or mounted firmly near the edge of the centered glass.  Tap threads
(or use a T-nut) in the arm to mount a bolt pointing radially toward the
glass edge.  Sharpen the end of the bolt to act as a crude "micrometer".
Place the glass as close as possible to the center of the turntable.  Find
the point of farthest off-center with the micrometer (just touching at one
point).  Rotate 180 degrees and move the glass 1/2 the gap.  Re-adjust the
micrometer to touch the edge (repeat if necessary).  The glass should now be
centered.  If the glass cannot be centered this way without leaving a gap at
some part of the circumference, then it is not a circle.  The manner that
the gap varies as the glass is rotated will indicate how out-of-round the
glass is.  If the micrometer touches the glass for most of the rotation, the
it is a circular glass with a "dented" side.  If the micrometer touches the
glass at 2 or more points, it is somewhat elliptical or worst: multi-lobed.
I would guess that if the variation is under 1/8" (3mm) in a 6" glass (2%),
then it probably doesn't matter.

Clear skies.
Don

-----Original Message-----
From: atm-bounces@atmlist.net [mailto:atm-bounces@atmlist.net] On Behalf Of
Dale Eason
Sent: Tuesday, May 30, 2006 5:15 PM
To: Mutalib Abdallah; atm@atmlist.net
Subject: Re: [ATM] Dumb Question: how to measure a...


The definition of a circle is all points equal radius from a point.  Thus
any diameter should be the same within a tolerance you pick and can measure.

I know of no other geometry that has that attribute. 
Can you give me one?

Dale Eason


--- Mutalib Abdallah <omegatroid@hotmail.com> wrote:

> Circle.
> 
> Given a circle of glass, how can I measure it to characterize its
> circularity?   Assume I do not know where on the
> glass is the "center", if,
> in fact a center exists.
> 
> The only way I can thing of is to measure the coordinates of a large 
> number of edge points, copy the data into a computer file, and then do 
> a curve fit to find the circle of best fit (either rectangular
> or polar coordinates).   
> Or maybe a truncated fourier series (which is a least squares best 
> fit).
> 
> Another way might be to measure the diameter at many points, but I 
> know that there are non-circular forms with constant diameter.
> 
> Suppose I was writing a specification-- how would I specify roundness?  
> Is there some way to just measure it directly?
> 
> I ask this because a friend is making round looking mirror blanks on a 
> jig that geometrically should not work.  I want to measure the error 
> of his ways.
> 
> Ideas?
> 
> 
> _______________________________________________
> ATM mailing list http://www.atmlist.net/
> 


__________________________________________________
Do You Yahoo!?
Tired of spam?  Yahoo! Mail has the best spam protection around
http://mail.yahoo.com _______________________________________________
ATM mailing list http://www.atmlist.net/

_______________________________________________
ATM mailing list http://www.atmlist.net/