PID Made Simple

BY: Jim Elliott

8/20/97
 11901 W Dearbourn Ave – Wauatosa, WI 53226
(414)453-1171

PID MADE SIMPLE

This is an explanation of PID temperature control that tries to take the mystery out of temperature
control. It takes a historical point of view to give a prospective on how the concepts of PID, as it is
applied in industry, came to be.

EARLY CONTROL SYSTEMS:

In the dark ages there was ON/OFF control. This is basically the way that most ovens in
the home work. You adjust the setpoint to the desired temperature. Below this, the control
is at full output (ON). Above this setpoint, the control is at minimum output (OFF).

Diagram #1

ON/OFF control systems have only one adjustment other than the setpoint. This is called
HYSTERESIS. The primary purpose for the hysteresis adjustment is to prevent the
control device (typically a contactor or relay) from chattering when the process is at the
setpoint. While in ON/OFF control, the control turns off the output when the process is
above the setpoint. The hysteresis adjustment prevents the output from turning on again
until the process drops a certain number of degrees below the setpoint. This band
between where the output turns off on temperature rise and where it turns on again on
temperature drop is the hysteresis. Typically the units are in number of degrees.

While this control scheme is OK for simple systems like a home cooking oven and non-
critical processes, it does not control very tightly. As the temperature approaches the
setpoint, it does not see that it is getting close to the setpoint and stay full on. As a result,
when it hits the setpoint, the temperature is still rising and overshoots the setpoint. The
same thing happens as the temperature comes back down. Obviously there had to be a
better way.

2
PROPORTIONAL
Since ON/OFF control gave oscillations in temperature because it did not anticipate the
approaching setpoint, the next step was to find a way of doing this. This what
PROPORTIONAL control does.

Proportional control stretches out the ON/OFF so that the output starts to cut back before
the system ever gets to the setpoint:

Diagram #2

The width of this "stretched out ON/OFF" is what is called PROPORTIONAL BAND.
Saying it another way, the proportional band is how far below the setpoint the controller
starts to cut back, and how far above setpoint the output is completely off. Refer to the
diagram above.

Proportional band is expressed in percent of full scale of the instrument. In other words, if
your controller can display a temperature range of 0-1000°F, and you set your
proportional band to 10%, the proportional band will be 100°F wide (10% of 1000°F
=100°F).

There is another term that you may hear about: GAIN. Gain and proportional band are
really the same thing, just stated differently. Gain is the slope of the line between where
output is full ON (100%) and full off (0%). It is also the inverse of proportional band. To
find gain from the proportional band (or the other way around) use the following formula:
1
GAIN = ------------------------------------------------ .
PROPORTIONAL BAND

EFFECT OF PROPORTIONAL BAND:

The proportional term is what gives the a PID control it's stability. The wider the
proportional band, the more stable the system is. The narrower the proportional band is,
the closer it becomes to ON/OFF control, and the more unstable it becomes. So, you
might ask, why not just set the proportional band as wide as the system will allow? This
would after all, be as stable as the system could get. The reason this does not make
sense is that it also slows down the system response, making startups unreasonably
slow, and reaction to process upsets sluggish. Therefore, the ideal proportional band is
one that is as narrow as possible, without making the system unstable.

3
TYPES OF PROPORTIONING
At this point you maybe wondering: What exactly do they mean by "proportioning"? And
how are they cutting back their output? And what is 0% output, and what is 100% output
(Or more importantly, how would you get something in between, like 50% output)?

There are several way to proportion the output. These are the most common:
VALVE POSITION:
This type of proportioning is what is used to control gas valves, steam valves, and
water temperature control valves. It is pretty self explanatory: When the valve is
completely closed, it is at 0% output. When it is completely open, it is at 100%.
When it is in between, the output is in between 0% and 100%.

VOLTAGE PROPORTIONING:
This type of proportioning uses what is called a power controller to reduce the
voltage applied to the heater. It usually gets a control signal from the temperature
controller that tells it how much of the voltage into it should be applied to the
heater. This control signal is usually 4-20ma (which corresponds to 0-100% of
applied voltage).

EXAMPLE:
480Volts are applied to the power control. When the power control gets a 4ma
signal from the temperature control, the power controller puts out 0 Volts. When it
gets a 12ma signal (50% output), it puts out 240 Volts. When it gets 20ma
(100%), it puts out 480 Volts.

TIME PROPORTIONING:

This type of proportioning is by far the most common type of proportioning for
electric heaters. It cycles the heaters load full on and full off over a certain time
period called the CYCLE TIME. (This is another important term to remember).

EXAMPLE:

With the CYCLE TIME = 10 sec, a 50% output would cycle the heater ON for 5
sec and OFF for 5 sec. A 25% output would be 2.5 sec ON and 7.5 seconds
OFF.

As you can see, it is very important to choose the appropriate cycle time for the
given output device. You would not want to run a large mechanical contactor at a
1 sec cycle time. At 50% output the contact would chatter at 0.5 sec ON and 0.5
sec OFF. It would not last very long at this rate.

The best cycle time is as fast as you can set it for the final output device.

4
INTEGRAL TERM (RESET):

The next term we are going to talk about is RESET. After proportional control was
invented, it was discovered that over time the temperature would drift from the setpoint.
The reason for this can be understood with the following example and diagram.

In this imaginary process, the system is running along right at setpoint. The proportional
band is set just right to give it just the right amount of stability:

Diagram #3

Now say that the operator increases the line speed. This will start to take more heat away
from the system and the temperature will drop. But, then the proportional action kicks in.
Looking at the diagram, you will see that as the temperature drops, the output will
increase (A1).. As the output increases, this will start to put more energy into the system,
and the temperature rises.
But not enough. The system will settle out below setpoint (A2) because it now requires
more heat to maintain setpoint at the higher line speed.

The first solution to this problem was to have a "fudge factor" adjustment on the control
that would shift the proportional band so that when the temperature was at the setpoint,
the output would be higher (B). This is called MANUAL RESET. The problem with this
was that an operator would have to intercede from time to time to make sure that the
temperature hadn't drifted. If something important called the operator away for awhile,
the temperature could drift out of control and cause problems with the process.

Then somebody had a great idea of doing this automatically. They then developed what
is called AUTOMATIC RESET (OR INTEGRAL). AUTOMATIC RESET checks the
process from time to time to make sure that the temperature is at setpoint. If the
temperature is not at the setpoint, automatic reset makes the assumption that the
proportional band has had a chance to do it's thing (in other words it assumes that the
temperature is stable and the error it sees has been there for a while) and it shifts the
band in the appropriate direction. How much reset action the system has is determined
by how often the system checks for this error and makes an adjustment.

Automatic reset is expressed in sec/repeat or repeats/min. In other words how often

does the controller check to see if the temperature is at setpoint.

5
EFFECTS OF INTEGRAL TERM:

Again, the integral term (or automatic reset) is what allows a PID controller to compensate
for changing load conditions without any operator interference. And, like proportional
band, too much of it is not good. If you set the integral term too fast (that is, the integral
term tries to make adjustments too often), the assumption it makes that the proportional
band has had a chance to do it's thing is not valid. In other words, if you set the integral
term too fast it will make an adjustment to try to compensate for the temperature drifting,
and before the system has a chance to recover from that change and settle out, the
controller will make another adjustment. Pretty soon the controller will start to chase it's
tail and will never catch up. This will cause a temperature oscillation that is called
"HUNTING".

Therefore the ideal integral term is one that is just fast enough that it can compensate for
changes in the system that occur over time, yet allows the system to recover and stabilize
from these adjustments.

DIAGRAM #4

DERIVATIVE TERM (RATE):

The derivative term is just a little more difficult to visualize, but the concept is straight
forward enough. It reacts to a rate of change of temperature. In other words, it tries to
limit how fast the temperature rises and how fast it falls. The more derivative term you
have, the slower the controller will allow the process to change. Diagram "+" is what a
control system looks like on start-up with derivative enabled.

6
 OTHER CONTROL TERMS
DERIVATIVE APPROACH CONTROL:

Derivative Approach Control (DAC) allows the derivative action to act on the system on
start-up to help prevent overshoots. On most controllers both the derivative and integral
terms are disabled whenever the temperature is outside of the proportional band.
(Outside of the proportional band means that the output is either at 100% or 0% because
the temperature is either far below the setpoint or far about the setpoint. A look at the
diagrams may help.) DAC allows the derivative term to be enabled when the temperature
is outside of the proportional band. This has the effect of cutting back on the output when
the control sees the fast temperature rise that is typical on a start-up. This in turn helps to
prevent overshoots.

7
 SUMMARY OF TERMS
ON/OFF CONTROL:
The original type of temperature control, ON/OFF control turns the output full on below setpoint
(100% output) and full off above setpoint (0% output). This type of control has oscillations in the
process.

HYSTERESIS:
The only adjustment to ON/OFF control. This is the number of degrees below the setpoint at
which the output will turn on again as the process drops. It prevents the contactor or relay from
chattering.

PROPORTIONAL BAND:
The term that gives the system stability. The wider it is the more stable the system. The narrower
it is, the closer it is to ON/OFF control, and the more unstable it is. It is expressed as a % of Full
Scale of the instrument.

INTEGRAL TERM (AUTOMATIC RESET):

This is the term that compensates for temperature drifts due to load changes. It does this by
checking for errors from the setpoint from time to time and shifting the proportional band if the
temperature is not at the setpoint.

It is expressed in repeats/min or sec./repeat.

The ideal integral term is one that is fast enough to compensate for load changes, yet not so fast
that the system can not stabilize after each adjustment.

MANUAL RESET:
The predecessor of AUTOMATIC RESET, this term was used as a "fudge factor" to compensate
for temperature drifts over time due to load changes. It basically shifts the proportional band to get
the proper percent output to maintain the setpoint. It requires an operator make adjustments from
time to time to maintain setpoint.

DERIVATIVE TERM (RATE):

This term reacts to the rate of change of the temperature (speed of temperature). It has the effect
of trying to slow temperature changes. The more derivative term you have, the more it tries to
slow down the system. Too much derivative will cause oscillations. The ideal derivative term is
one that does not add instability to the system.

CYCLE TIME:
The time base used for proportioning electric heaters. (10 sec cycle time at 25% output = 2.5 sec
ON and 7.5 sec. OFF) The ideal cycle time depends on the output device. Some rules of thumb:

Mechanical Contactor: 20 - 60 seconds.

Small Amp Relay (10-20 amp): 15 - 30 seconds.
Mercury Contactor: 5 - 20 seconds.
Solid State Relays: 0.3 - 2 seconds

DERIVATIVE APPROACH CONTROL (DAC):

A percentage of the proportional band outside of which the derivative term is disabled. This term
is used to help prevent process overshoots on start-ups. This value depends on whether getting to
the setpoint quickly is of prime importance, or preventing overshoots is. The wider the DAC band,
the slower the process will get to setpoint.