Re: [foxboro] PIDA 'dead-lock' ?

Patrick,

Your example does illustrate some unexpected behavior. I think the BIAS
function may be working OK, but the problem appears to be in the way the
value of internal feedback is computed during output limiting.  Repeat =
your
example but use a scan period of 2 or 5 seconds on all blocks.  Now when =
you
trend them I think you will see the BIAS does have an effect, but it is
immediately undone on the next scan by the action of the internal =
feedback
balancing with slightly different behavior depending on the setting of
LIMOPT.  I made the following simple test for various PID types (On a
version 7.0 system):

PID
PIDE
PIDX
PIDA

In each case make the block with a PERIOD of 2 sec, a MEAS of 50.0, a =
SPT of
55, PB of 100%, INT of 1.0, FBK to it's own output, MODOPT of 4, and the
LOLIM of 10%.  For the PIDA you can try each version of the LIMOPT, but
maybe start a 1.  All other parameters at default, no other blocks =
needed.

Trend the controller SPT, MEAS, and OUT.  Put the controller in AUTO and
observe it will integrate up.  Now put the HOLIM down to 0.0 for a =
second or
two and then release it back to 100%.  Only the PIDA jumps to 5% above =
the
LOLIM and then integrates.  The other controller outputs go to 10% and
integrate at the reset rate from 10%.  None of the LIMOPT options 1,2,or =
3
(or 0 or 4 for that matter) will give the more desirable (my opinion)
behavior of the earlier PID block types.  LIMOPT of 2 does not solve =
this
problem.  For example, put the PIDA to manual and set the BIAS to 50%, =
and
then put the PIDA back to AUTO.  Now set the BIAS to 0 to force the =
output
into the LOLIM and then back to 50% and wham the output jumps up much =
higher
than it's original value when the BIAS started at 50% - this is not
acceptable for a "feedforward" signal that has a brief transient. =20

The FOXDOC LIMOPT description is not too clear to me. Even I-only
controllers (MODOPT=3D2) behave in this unexpected manner.

Applications that use the HOLIM and LOLIM to position the output of a
controller in AUTO or applications using BIAS or MULTIN behave =
considerably
differently than the same application using previous PID block types.  =
While
the PIDA has some very nice features, like the setpoint ramping, NLNBLK, =
and
especially the PRETUNE/FBTUNE functions, the LIMIT behavior is a bit
puzzling.  The PIDX would seem better suited. Another undesirable =
feature is
changing the LIMOPT parameter on a PIDA in AUTO will cause the =
controller
output to jump.  I think LIMOPT needs an option 4 or 5 that makes it
backwards compatible to the earlier PID types.

Rick Rys P.E.
www.R2Controls.com

=20



 
 
_______________________________________________________________________
This mailing list is neither sponsored nor endorsed by Invensys Process
Systems (formerly The Foxboro Company). Use the info you obtain here at
your own risks. Read http://www.thecassandraproject.org/disclaimer.html
 
foxboro mailing list:             http://www.freelists.org/list/foxboro
to subscribe:         mailto:foxboro-request@xxxxxxxxxxxxx?subject=join
to unsubscribe:      mailto:foxboro-request@xxxxxxxxxxxxx?subject=leave

Other related posts: