John Kasunich wrote:
> Stuart Stevenson wrote:
>> John,
>>    There are two pictures on www.mpm1.com:8080/cinci
>> one is the positive capture - the one that jumps                   -
>> positive-jumps.png
>>
>> one is both the positive and negative capture                       -
>> pos-neg.png
>> the positive on the left and the negative on the right
>> I moved the axis .0001 positive and then .0001 negative
>>
>> they both look the same
>>
>> it must be the machine - SCR amp?
> 
> I only now got a chance to see the scope pics.  Something doesn't make
> sense.
> 
> In BOTH sets of traces, the backlash-corr signal jumps (as expected)
> both directions.  But backlash-filt is only filtered when going
> negative, when going positive it jumps too.
> 
> Both edges should be filtered.
> 

OK, we've been looking at this a bit on IRC, and we found that there is
indeed a problem, and identified the cases when it happens.  The bug is
in the filtering code.

For normal backlash compensation, the unfiltered values are always above
and below nominal by one half of the total lash.  For a typical screw
comp curve, the values may vary from place to place, but again, the one
value is usually above the nominal, and the other is below nominal.  The
code works correctly in both of those cases.

But in your file:

> -0.5 -0.502 -0.4972
> 0.0 0.000 0.0043
> 0.5 0.5005 0.5052

The third line describes an area of the screw where both values are
above the nominal value.  The filtering code doesn't work right in that
case - it only filters in one direction, the other direction is
completely unfiltered.  The same thing happens if both values are below
nominal.

We will be looking into a fix (or maybe two fixes, one for CVS trunk and
one for version 2.2.x), but that might take a bit of time.

In the meantime, you can avoid the problem by adding or subtracting a
small constant offset to all of your correction values.  The goal is to
have one column of numbers always be less than nominal, and the other
column always be greater than nominal.  Hopefully your data will let you
do that.

Thanks for spotting and reporting this - we'll get it fixed.

Regards,

John Kasunich




-------------------------------------------------------------------------
This SF.Net email is sponsored by the Moblin Your Move Developer's challenge
Build the coolest Linux based applications with Moblin SDK & win great prizes
Grand prize is a trip for two to an Open Source event anywhere in the world
http://moblin-contest.org/redirect.php?banner_id=100&url=/
_______________________________________________
Emc-users mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/emc-users

Reply via email to