FX Series Programmable Controlers Applied Instructions 5
5-106
Effective use of the input filter
α
S
3
+2
To prevent the PID instruction from reactin
immediatel
and wildl
to an
errors on the
Current Value, there is a filterin
mechanism which allows the PID instruction to observe and
account for an
si
nificant fluctuations over three samples.
The quantitative effect of the input filter is to calculate a filtered Input Value to the PID
instruction taken from a defined percenta
e of the Current Value and the previous two filtered
Input Values.
This t
pe of filterin
is often called first-order la
filter. It is particularl
useful for removin
the
effects of hi
h frequenc
noise which ma
appear on input si
nals received from sensors.
The
reater the filter percenta
e is set the lon
er the la
time. When the input filter is set to
zero, this effectivel
removes all filterin
and allows the Current Value to be used directl
as
the Input Value.
Initial values for PID loops
The PID instruction has man
parameters which can be set and confi
ured to the user’s
needs. The difficult
is to find a
ood point from which to start the fine tunin
of the PID loop to
the s
stem requirements. The followin
su
estions will not be ideal for all situations and
applications but will at least
ive users of the PID instruction a reasonable points from which to
start.
A value should be
iven to all the variables listed below before turnin
the PID instruction ON.
Values should be chosen so that the Output Manipulated Value does not exceed ± 32767.
Recommended initial settin
s:
T
S
= Should be equal to the total pro
ram scan time or a multiple of that scan time, i.e. 2 times,
5 times, etc.
α
= 50%
K
P
= This should be ad
usted to a value dependent upon the maximum corrective action
to reach the set point - values should be experimented with from an arbitrar
75%
T
I
= This should ideall
be 4 to 10 times
reater than the
T
D
time
K
D
= 50%
T
D
= This is set dependent upon the total s
stem response, i.e. not onl
how fast the
pro
rammable controller reacts but also an
valves, pumps or motors.
For a fast s
stem reaction T
D
will be set to a quick or small time, this should however
never be
less than
T
S
. A slower reactin
system will require the
T
D
duration to be lon
er. A
be
innin
value can be
T
D
twice the value of T
S
.
Care should be taken when ad
ustin
PID variables to ensure the safet
of the operator and
avoid dama
e to the equipment.
With ALL PID values there is a degree of experimentation required to tune the PID loop
to the exact local conditions. A sensible approach to this is to adjust one parameter at a
time by fixed percentages, i.e. say increasing (or decreasing) the K
P
value in steps of
10%. Selecting PID parameters without due consideration will result in a badly
configured system which does not perform as required and will cause the user to
become frustrated. Please remember the PID process is a purely mathematical
calculation and as such has no regard for the ‘quality’ of the variable data supplied by
the user/system - the PID will always process its PID mathematical function with the
data available.
On FX
2N
MPUs pre-tunin
feature is available that can quickl
provide initial values for the
PID process. Refer to pa
e 10-28 for more details.
