Tim Wescott said:
OK, I'll bite -- what _is_ it good for?
True negative derivative is called (negative) velocity feedback and
is used to decrease response time at the expense of overshoot. The
frequency response of the system goes up. The system becomes stiff
but 'nervous'. Can be used to stabilize systems as the same stiffness
can be had with a lower proportional gain.
This is just what one would expect as plain-ole' derivative feedback
is used to lessen overshoot at the expense of slower response time.
Think of D as oil: it can be used to slow something down by being
thick and greasy or it can speed something up by lubricating it.
Speculation: if the controller is configurable to use derivative on
error _or_ derivative on process then the sign of the derivative
at the summing junction must change, most modern controllers do
this automatically in software but I suppose there are those that don't.
The same is true if feed-forward on SP is available.
but I _can't_ see the point in intentionally establishing an
unstable zero in your control system.
Not so simple ... it doesn't necessarily make the system unstable:
think Nyquist diagram.
I'm not sure what this has to do with software, except to know to
shout 'Hardware Error!' and go for a cup of coffee.
FWIW: dual slope and V/F are preferred conversion techniques for
process control as there are no 'bad spots'. Also work a charm
if synched to the line frequency.
Yes, this could be _very_ counter-intuitive to my target audience.
Er, maybe a new audience is needed?
The only requirements for control are monotonicity and repeatability,
everything else is icing on the cake (though some sort of linearity is
_really_ nice to have).