On Thu, Feb 3, 2011 at 1:33 PM, andy pugh <bodge...@gmail.com> wrote:

> On 3 February 2011 19:09, Igor Chudov <ichu...@gmail.com> wrote:
>
> > It would appear that friction compensation (let me call it FFF for now)
> is
> > just one more term,
>
> As somebody else said, inverse-deadband ought to do what you want.
> (I have seen a comp somewhere to do that).
> Assuming that the input to the PID is position, then I have a feeling
> you could get the same effect by using a very large gain on FF1 with a
> limit (maxcmdD) set to ever so slightly less than the current required
> to move the table.
>
>
Andy, I really like your last suggestion, you are truly on to something.

Since you apparently know a lot about how PID loop is implemented (I just
looked into pid.c myself after seeing your older message), would you think
that adding a friction compensation is relatively straightforward?

In line 375 of pid.c, ff0, ff1 and ff2 are used.

All I need to do is after that, add a statement to use FFF as friction
compensation. FFF would be held in fffgain (new attribute):

tmp1 += *(pid->fffgain) * sign( command - feedback )

where sign( ... ) is a function that returns +1 if its argument is above 0,
0 if its argument is 0, and -1 if argument is < 0.

The above is SIMPLIFIED because command-feedback needs to be corrected for
deadband. I was just trying to convey the idea.

This will work well if the axis can stop within the deadband area, with the
dricving force removed, due to friction or something. Otherwise oscillations
may result.

Am I making any sense at all?

i
------------------------------------------------------------------------------
The modern datacenter depends on network connectivity to access resources
and provide services. The best practices for maximizing a physical server's
connectivity to a physical network are well understood - see how these
rules translate into the virtual world? 
http://p.sf.net/sfu/oracle-sfdevnlfb
_______________________________________________
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Reply via email to