Skip site navigation (1)Skip section navigation (2)
Date:      Fri, 2 Mar 2018 15:14:08 +0100
From:      Pieter de Goeje <pieter@degoeje.nl>
To:        Jeff Roberson <jeff@FreeBSD.org>, src-committers@freebsd.org, svn-src-all@freebsd.org, svn-src-head@freebsd.org
Subject:   Re: svn commit: r329882 - in head/sys: conf kern sys vm
Message-ID:  <1782e475-06d8-c806-0d0c-ae5a5893baf5@degoeje.nl>
In-Reply-To: <201802232251.w1NMpqn4025101@repo.freebsd.org>
References:  <201802232251.w1NMpqn4025101@repo.freebsd.org>

next in thread | previous in thread | raw e-mail | index | archive | help
Hi,

I'm curious, it looks like Kd is quite significant by default, I assume 
to limit the slew rate of the output.

Given that the error is an integer, and pidctrl_daemon() is called at 
irregular intervals, it seems to me that the derivative will be very noisy.

In my experience with PID controllers controlling physical processes, 
one would need some kind of filtering on the input to make Kd useful, 
otherwise it would just contribute more instability to the system. In 
practice even a simple exponential (single pole) filter can help a lot 
at the cost of increased response time.

Is this something that was considered?

- Pieter de Goeje



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?1782e475-06d8-c806-0d0c-ae5a5893baf5>