INFICON SQM-242 Thin Film Deposition Controller Card Operating Manual - Manual (Page 97)

9 - 11
IP
N 07
4-
54
9-
P1
A
SQM-242 Operating Manual
ITAE is insensitive to the initial and somewhat unavoidable errors, but it will weight
heavily any errors occurring late in time. Optimum responses defined by ITAE will
consequently show short total response times and larger overshoots than with
either of the other criteria. It has been found that this criteria is generally most
useful for deposition process control.
The most satisfactory performance criterion for deposition controllers is the ITAE.
There will be overshoot, but the response time is quick, and the settling time is
short. For all of the above integral performance criteria, controller tuning relations
have been developed to minimize the associated errors. Using manually entered
or experimentally determined process response coefficients, ideal PID controller
coefficients can be readily calculated for the ITAE criteria as shown below.
[15]
[16]
[17]
For slow systems, in order to help avoid controller windup (windup is the rapid
increase in control signal before the system has the chance to respond to the
changed signal), the time period between manipulated variable (control voltage)
changes is lengthened. This allows the system to respond to the previous controller
setting change, and aggressive controller settings can be used. A secondary
advantage is that immunity to process noise is increased since the data used for
control is now comprised of multiple readings instead of a single rate
measurement, taking advantage of the mass integrating nature of the quartz
crystal.
With process systems that respond quickly (short time constant) and with little to
no measurable dead time, the PID controller often has difficulty with the deposition
process noise (beam sweep, fast thermal shorts of melt to crucible, etc.). In these
situations a control algorithm used successfully is an integral/reset type of
controller. This type of controller will always integrate the error, driving the system
towards zero error. This technique works well when there is little or no dead time.
If this technique is used on a process with measurable lag or dead time, then the
control loop will tend to be oscillatory due to the control loop over-compensating
the control signal before the system has a chance to respond.
K
c
1.36 K
p
   L T
1
  
0.947 –
=
T
i
1.19T
1
  L T
1
  
0.738
=
T
d
0.381T
1
  L T
1
  
0.995
=