User manual

PID Loop Operation
(DL450 Only)
Maintenance
8--33
PID Loop Operation (DL450 only)
DL405 User Manual, 4th Edition, Rev. A
NOTE: You may elect to leave the tuning dialog blank and enter the tuning
parameters in the DirectSOFT PID View.
Derivative Gain Limiting
The derivative gain (rate) has an optional gain--limiting feature. This is provided
because the derivative gain reacts badly to PV signal noise or other causes of
sudden PV fluctuations. The function of the gain--limiting is shown in the diagram
below.
Process Variable
Σ
Error Term
+
--
Control
Output
Setpoint
PID Mode 1 Setting V+00
013456789101112131415 2Bit
Derivative gain limit select
Σ
+
P
I
D
Loop Calculation
+
+
Derivative
Derivative,
gain-limited
0
1
Integral
Proportional
Loop Table
V+25 Derivative Gain Limit00XX
The gain limit can be particularly useful during loop tuning. Most loops can tolerate
only a little derivative gain without going into uncontrolled oscillations.
If Derivative Gain Limiting is selected, a unit of 0--20 for Limit must also be entered.
NOTE: When first configuring a loop, it’s best to use the standard error term until
after the loop is tuned. Once the loop is tuned, you will be able to tell if these functions
will enhance control. The Error Squared and/or Enable Deadband can be selected
later in the PID setup. Also, values are not required to be entered in the Tuning
dialog, but they can set later in the DirectSOFT PID View.
Error Term Selection
The error term is internal to the CPUs PID loop controller, and is generated again in
each PID calculation. Although its data is not directly accessible, you c an easily
calculate it by subtracting: Error = (SP -- PV). The PID calculation operates on this
value linearly to give the result. However, a few applications can benefit from
non--linear control. The Error--squared method of non--linear control exaggerates
large errors and diminishes small error
Error Squared y When selected, the squared error function simply squares the
error term (but preserves the original algebraic sign), which is used in the
calculation. This affects the Control Output by diminishing its response to smaller
error values, but maintaining its response to larger errors. Some situations in which
the error squared term might be useful:
S Noisy PV signal -- using a squared error term can reduce the effect of
low--frequency electrical noise on the PV, which will make the control
system jittery. A squared error maintains the response to larger errors.
S Non--linear process -- s ome processes (such as chemical pH control)
require non--linear controllers for best results. Another application is
surge tank control, where the Control Output signal must be smooth.