Physics 221 - Electronics

Temple University, Fall 2005-6

C. J. Martoff, Instructor

On/Off Temperature Control; Controlling Wall Current with an Op-Amp

1 Objectives
• Introduce the method of closed loop control with a sensor and an ac-
tuator

• Apply this method to building a rudimentary on/off heater controller

that will maintain a sample at a desired temperature

Components used:

• thermistor

• Zener diode

• LED

• relay

• op-amp (as comparator only)

• bipolar transistor

2 Thermistor + Op-Amp + Relay = Sensor + Ac-

tuator
We are all familiar with using control circuits like a heating thermostat in
the home. In general, a control circuit contains two essential elements:

• A sensor that responds electrically to a variable of interest (tempera-

ture, pressure, humidity, magnetic field, light level, etc.)

1
Figure 1: Basic Thermistor sensing circuit, using an op-amp without feed-
back as a comparator

• An actuator (heater, valve, humidifier, current source, light bulb, etc.)

that the circuit causes to do something to influence the variable of
interest.

In this lab we will control temperature. The sensor to be used is a “ther-

mistor”, which is just a resistor that has a resistance which varies strongly
and reproducibly with the temperature. We will use so-called “NTC ther-
mistors”, which means the resistance decreases if the thermistor is heated
up. Resistance changes are sensed by “biasing” the thermistor (running cur-
rent through it) and comparing the voltage across the thermistor to another
voltage (the “set point”) representing the temperature we want the sample
to stay at.
The actuator to be used is an “AC-powered heater”, in other words
something that plugs into the wall. These heaters are rather high-powered
devices (50-1000 Watts). Part of the goal of the lab is to see how easily low-
powered electronics on our breadboard can be made to control such high
powered actuators, using an electromechanical switch called a “relay”.

3 Basic Sensor/Actuator Circuit

3.1 Circuit Explanation
Consider the circuit in Figure 1.This looks complicated but is really is not.
Start by looking at the op-amp inputs. The inverting input is connected
to the wiper of a 10K potentiometer, and the ends of the potentiometer
are connected to Vcc = +9 V and Vee = ground, respectively. (The op-
amp we are using is specified to work with V ee = ground, sometimes called
“single-supply operation”) This input setup is of course a voltage divider
with variable divider ratio. By adjusting the potentiometer setting we can
vary the voltage V− at the inverting input from 0 to 9 Volts. We will use

2
the potentiometer adjustment to set the temperature we want the thing to
maintain, called the “set point”.
The non-inverting input is connected to another voltage divider, formed
by the thermistor T and a fixed 10 K resistor to ground. In use, the ther-
mistor would be in good thermal contact with 03 sample whose temperature
we want to control. Remember that the thermistor resistance varies with
temperature. So the divider ratio in this divider, and hence the voltage at
the non-inverting input, will also vary with temperature.
Since we use an NTC thermistor, the voltage V + at the non-inverting
input will increase as the temperature increases. (Better think this through.)
When V+ is bigger than the set point V− , the op-amp output will go high.
We will use this output to turn off the heater (the sample is already too
hot). Whenever V+ is smaller than the set point V− , the op-amp output
will go low. We will use this output to turn on the heater (the sample is
still too cold).
To play with this circuit we start out with no sample, just the thermistor
on the breadboard (we can change its temperature by touching it with a
finger). Instead of a heater we use an LED (light emitting diode) to show
whether the heater would be on or off.
Notice that there is no negative feedback applied to the op-amp. This is
unusual and in a way is a 17-use of the op-amp. There is another special IC
called a “comparator” which should really be used for this. If we wanted to
do more sophisticated temperature control (for example reduce the heater
current as the set point is approached) we could introduce feedback and get
a signal out of the op-amp that was proportional to the difference between
actual and set point temperatures.
With no feedback, the high gain of the op amp makes its output just
go to one or the other of the power supply voltages in response to any tiny
difference (V+ -V− ). This is called “railing” or “hitting the rails”. The power
supply voltages Vcc and Vee are called the “rails”.
The LED shown in the circuit is there to indicate the state of the op-
amp output without having to measure it on a meter all the time. An
LED lights when it is “forward biased”, which means conventional positive
current flows in the direction of the arrow in the diode symbol. Therefore it
will light when the op-amp output goes low. When the op-amp output goes
high, the LED is reverse biased. That doesn’t hurt anything, but it doesn’t
light the thing up either.
So, putting this altogether, when the temperature sensed by the ther-
mistor is less than the set point (controlled by the potentiometer), V out at
pin 6 of the op-amp is near Vee = ground, and the LED lights up. When
the temperature is higher than the set point, the output is close to
Vcc = +9V the LED has no forward bias voltage across it and does not
light. To reverse this behavior, we could switch the inputs to pins 2 and 3
of the op-amp.

3
Figure 2: Thermistor sensing circuit with 4 and output driver stage added

3.2 Build It!

1. Build the circuit of Figure 1 on your breadboard. To get the LED
oriented properly, the longer lead must go to the positive voltage.

2. Rotate the potentiometer until LED just goes on.

3. Warm the thermistor with a finger – the LED will go off.

4. Wait till thermistor cools – the LED will go on again.

5. Try reversing inputs to op-amp and verify that the operating cycle
is also reversed. This could be used to control temperature by cooling
rather than heating– or for other purposes (see below).

4 Controlling A Real Actuator with a Relay

4.1 Circuit Explanation
We wish to use the thermistor to control a ”real” heater, not just an LED.
A suitable heater requires 1 - 10 Amps, while the maximum output power
the op-amp can directly supply is less than 250 mA. To control such a high-
current element using the puny op-amp output, we need two additional
components. The “relay” is shown at the extreme lower right of Figure 2. A
relay is an electrically operated switch. A spring-loaded contact is mechan-
ically “pulled in” (opened or closed) by the magnetic force of a coil. The

4
coil requires only a low current, low voltage input to operate the switch, but
the switch has enough power-handling capability to open or close the 120V
circuit. The relay used here actually has two switches in it, one “normally”
(with no current in the coil) open (normally open is abbreviated NO) and
one normally closed (NC). Of course there are also solid-state relays which
have no mechanical moving parts.
For safety reasons, the 120VAC connections don’t go on your breadboard.
They are built into an aluminum “chassis” box with connectors for standard
“IEC line cords” to supply the 120 VAC input from the wall and output to
the heater.
The circuit in Figure 2 also has another complication. A “ bipolar tran-
sistor” has been added to increase the current available to drive the relay.
The bare op-amp output would be marginal for this unless a special low
current relay were used. The transistor is denoted by:

. This writeup is not the place for an introduction to

bipolar transistors. Suffice it to say that in this circuit the transistor is
used basically as a voltage-controlled resistor. When the voltage at B is
higher than that at E, terminals C and E are basically shorted together.
Current then flows 1 from the 9 V supply through the transistor and into
the relay coil and the LED. When the voltage at B is lower than that at E,
the transistor acts like a very high resistance (more than a megohm). No
current flows through the transistor. The advantage of using a transistor in
this way is that very little current (microamps) is required from the op-amp
input to drive the B input, while the EC side of the transistor easily handles
several hundred milliamps of drive current needed for the relay coil.
Note also that the input connections to the op-amp have been reversed
in Figure 2. This allows the “normally open” contacts of the relay to be
used to run the relay.

4.2 Build It!

1. Build the circuit of Figure 2, using a 2N222 transistor. The data-sheet
shows the pin-out. Use the relay boxes provided but do not connect
any 120 VAC for now.
2) Test in the same manner as for the previous version, warming and
cooling around the set point. You can hear the relay clicking as it
is turned on and off. The LED should also light when the relay is
energized.
3) Check the state of the relay outputs (NO and NC) when the mea-
sured temperature is above set point. Which set is doing the right
thing to power a heater?

5
 +10V

10K
pot RS
47
R1
3 7
T
6 500
2 4
40K
R NC
2
NO

.
Figure 3: Thermistor sensing circuit with relay and output driver stage
added

5 Final Circuit
5.1 Improving the Stability with Schmitt Trigger
Figure 3 shows the final circuit. Notice that the setpoint circuit has been
changed. There is now a fixed resistor R 2 to ground. The 10K pot is now
being used as a single variable resistor. Only two of its three terminals are
connected to anything– the “lower” end and the wiper. The upper end is
not connected to anything.
Also new is the positive feedback resistor R S . The reason for putting
this in is explained below.
During testing you may have noticed the relay “buzzing” - chattering
open and closed very rapidly as the temperature fluctuates very slightly
around the set point. Even if your setup did not exhibit this behavior, it is
generally a problem with simple on/off controllers like this. You can imagine
what would happen if this behavior occurred in a circuit controlling a 100
KW gas furnace in a house.
To eliminate the chattering behavior we use an electronic fix called a
Schmitt Trigger. Basically what this does is change the setpoint a bit de-
pending on whether the output state is on or off. Say the temperature is
low and the heater is on. After awhile the sample warms up, the setpoint
is exceeded, and the heater turns off. With the previous circuit (Figure 2,
as soon as the heater goes off and the sample barely cools off, we will again
be below the set point temperature and the heater will be turned on again.
Hence the chattering. What we would like is to slightly lower the setpoint

6
Figure 4: Temperature vs. time plot for circuit of Figure 2, no Schmitt
Trigger. Temperature is tightly controlled but heater is cycled on and off
very rapidly.

every time the heater turns off because the old setpoint was exceeded. That
way it will be awhile before the sample cools below the new setpoint and
calls for heat again. This is amazingly easy to do with a Schmitt Trigger.
We trade a little looser control of the actual temperature for not wearing
out our expensive gas furnace ignitors.
The Schmitt trigger consists simply of resistor R S connected as positive
feedback. Say we initially start adjusting the set point with the op-amp
output low (output at ground; set point temperature below sample temper-
ature). Then RS is connected to ground in parallel with R 3 . The set point
voltage we get will be slightly smaller than 10V * R 3 /(R3 +R4 ). This does
not really matter, we just adjust it to the desired temperature.
But the magic occurs when the sample heats up and the op-amp output
goes high (10 Volts), RS is then connected to 10 V in parallel with R 4 . This
changes the setpoint. The setpoint voltage divider ratio is now higher than
it was before, due to the extra resistor R S now in parallel with R4 . A higher
setpoint voltage corresponds to a lower set point temperature because we
are using an NTC thermistor.
The effect on sample temperature control can be seen from Figures 4
and 5.
To determine the proper value for RS some design decisions and calcu-
lations are required.

5.2 Design of Schmitt Trigger

First off, notice a mathematical fact about the equation for resistors in par-
allel. Say we want to figure out what resistor R p to put in parallel with some
existing resistor R, to get a parallel combination with 1% lower resistance
than R alone. Some algebra or just trial and error should convince you that
Rp = 100 R will accomplish this. In general to get a parallel resistance
(1-δ)R, (for small δ) we need to put in parallel R p = R/δ. Conversely, we
can see that δ = R/ Rp . The per cent effect of adding a given resistor in
parallel is less, the larger the resistor to which it is added in parallel.

7
Figure 5: Temperature vs. time plot for circuit of Figure 3, including
Schmitt Trigger. Note the “dead zone between upper and lower set points,
which reduces heater cycling.

Secondly, consider the effect on the voltage divider ratio if we change

one of the resistors. If we change R1 by a fraction δ the ratio becomes:
R1 (1 + δ)
Vout = Vin (1)
R1 (1 + δ) + R2
Applying the binomial theorem to expand the denominator and keeping
only terms up to first order in the small parameter δ we obtain:
R1 R1
Vo ut = Vi n (1 + δ − δ ) (2)
R1 + R 2 R1 + R 2
What this says is the following. If R 1 (the one we changed) is by far the
smaller of the two resistors, the first δ term dominates and so the divider
ratio also changes by a fraction δ. If R 1 is by far the larger, then the two
δ terms nearly cancel. The fractional change in the divider ratio is smaller
than δ by a factor of the (small) divider ratio itself. Finally, if R 1 is about
equal to R2 , the divider ratio changes by a fraction ∼ δ/2.
In Figure 3 let’s choose R2 = 40K, the same as the resistor forming a di-
vider with the thermistor. Since the thermistor resistance ranges from about
0.5KΩ to 2.5KΩ depending on temperature, the variable resistor formed out
of the 10K pot will cover that same resistance range depending on the de-
sired temperature. So we have R1  R2 . Now consider the case RS 2C 
R2 . According to the above discussion of parallel resistors and voltage di-
viders, the biggest change in the divider ratio for the setpoint divider will
take place when RS is switched into parallel with R2 , changing by a fraction
δ ∼ R2 /RS . So to pick the value of RS to use, we should decide how much
we want the set point to change when the op-amp switches.
Referring to Figures 4 and 5 we see that the set point change leads to a
range of temperature fluctuations. For a household thermostat, fluctuations
of say 2◦ C would be acceptable. For scientific applications, more precise
control might be necessary. The choice here also depends on how much
heater cycling we can tolerate. For this example we will choose a setpoint
change equivalent to 1◦ C. Using the thermistor calibration data provided,

8
calculate the change of setpoint divider ratio that is equivalent to this tem-
perature fluctuation, and from there get the value of R S . (hint: Show that
the thermistor resistance changes by about 3.5% per ◦ C near 60 ◦ C. Since
the ratios for the thermistor divider and the setpoint divider are similar, we
want R2 to change by about this much when RS is switched in.)

5.3 Run It!

1. Build the circuit of Figure 3. For initial testing without the heater,
the relay leads can be left un-connected and the thermistor can be left
on the breadboard.

2. Verify that the Schmitt trigger provides separate heater-on and heater-
off setpoints. Explain exactly what you measured to verify this.

3. Choose a set point temperature in the range 30 - 60 ◦ C and calculate

the corresponding value of R1 using the thermistor calibration data
provided. Show your calculation to the instructor.
Next, prepare for testing with the relay and heating appliance.

4. Use half-a-meter of hook up wire to connect the relay leads to the relay
box. Verify operation of the relay box before applying AC power and
have the instructor check it.

5. When the instructor gives the OK, connect the IEC line cords to the
wall plug and the heater appliance.

6. Move the thermistor off the breadboard (use hook up wire to extend
its leads), and put it in good thermal contact with the sample. For
water samples it is sufficient to immerse the thermistor in the water.
If using a hotplate as a heater, turn the hotplate control knob all the
way to “hot”, to avoid having dueling controllers. Have the instructor
verify your setup.

7. Operate your controller, monitoring the temperature of the sample us-

ing a passive, visually read thermometer provided. Watch and record
the temperature and the amplitude of its fluctuations. Compare with
expected values.

5.4 Extra Credit–Chart It!

It is interesting to chart the temperature vs. time to compare with Figures 4
and 5. An old-fashioned ink-on-paper X-Y recorder is available in the lab
for this. Why not hook the recorder inputs directly to pin 2 and ground?
To run the recorder, put the thermistor voltage through an op-amp buffer
and then to the chart recorder input. (The chart recorder input requires a
signal range of 0 - 5 V and draws several mA of current). See if you can get
curves like Figures 4 and 5 by keeping and removing R S .