Re: test results

[Ray Henry <rehenry-at-up.net>]

Wow Hassan, this is some great stuff.  Thanks for all your effort.  

The first thing that jumps out at me is that repeatablilty is much better 
for the EMC than it is for the spectraCAM.  I wonder if these plots are 
typical?  If they were, I'd be willing to bet that a test for 
significance would show it.

The second thing that jumps out is that there is some scaling difference 
between x and y within the machine itself.  This shows up as an egg 
shaped circle on both plots.  This is fixable with the emc as I'll 
describe later.

I notice that the bit of egg shaped circle for the EMC appears to be 
rotated a few degrees clockwise from the xy of the plot.  If you draw a 
rectangle over this plot made of lines drawn tangent to the flattened 
sides of the egg you will see that they do not lay perfectly along the x 
axis.  I have heard a complaint about this before from someone who was 
measuring a milled circle.  

A lot of the egg shape of it could be taken out by changing scale.  I 
don't really know how the circularity number is computed by the ballbar 
software but it looks like you could take half that value as a percent 
and add to y and subtract from x.  For example if your input_scale was 
2000, I'd try setting y to 2020 and x 1980. See how the test comes out 
with that. If it looks good, do some long moves and measure the distances 
traveled for both.  Based on the linear measures, you may want to shift a 
few pulses from one axis to the other.

Ray

BTW -- Will you be able to make it to NAMES this year?  I'd drive over 
and get you if you can.  We could get an extra cot if you need a place to 
stay in Southfield.  Any chance to show the machine?  If not at least a 
poster session would be fantastic!  Perhaps you could include the show 
and comments from people in your report on the project.  I'd be pleased 
to speak with your professor if it would help.

BTW2 -- My tkemcex is in the linuxcnc.org dropbox but I'm nearly finished 
with a revision that works somewhat better.



On Saturday 15 February 2003 01:16 pm, you wrote:
> Hello List,
> I know that some of you were interested in the results of the renishaw
> ballbar test running from EMC. I have few comments to make about the
> test:
> * the test was conducted on the machine that i designed and
> manufactured. * when i ran the g-code that will run the test on EMC,
> while stepmode.o is on I noticed that there is a fraction of time stop
> between each of the arcs in a 2 circle 8 arcs g-code program. but when
> freqmode.o is on, that stop dissapeared.
> * maybe I need to further tune emc.ini, (freqmode.o) in order to get
> butter results. *since there is no way to view the results of the
> renishaw ballbar except using the software that runs on windows
> systems, so I included a picture of the main results that I obtained.
> the first pic. is for the test that was conducted using SpectraCAM, a
> win. CAM software running under win98, and the second pic. is for the
> same machine on the same conditions but while EMC is running the test.
>
> any comments regarding the tests??
> where can i find the updated file of the Raytkemce GUI??
> is there anyone who knows of a freeware that accepts an image and
> converts it into a G-code??
>
> Best Regards
> Hassan