The Thermocouple Amplifier by sufianabrar

VIEWS: 764 PAGES: 25

									Instrumentation and control
      Project file
  Farrukh Ali khan -1909
  Sufian Abrar        - 1934
          19 (2007-2008)

         “ENGR. Abdul wAhAb”

The world cannot rely on the fact that
temperature measurements using devices
are easily operated manually. The
industrial process cannot continue with
manual temperature measurements by
human because in industrial process
temperatures are so high that cannot be
measured by human. It requires a vast
technology for industrial process. The aim
of this project is to demonstrate how high
temperatures measuring devices are
design and how these devices work in the
industrial process.
               USING ??
 The course incharge of instrumentation
  and control decides to give an object to
  his students in which the every group of
  students design a circuit related to his
  subject and the students gave a full
  demonstration about the circuit which
  they design to the course incharge
 After this our group decided to design a
  circuit which is totally uncommon to
  others circuits
 After the couple of days we decided to
  design a circuit related to thermo couple
  because it is the most important topic in
  our course
    There was much much problem
     about the designation of circuit
     because we only knew the theory
     not thermo couple practically.

    Before designing thermocouple we
     should have to know about its function,
     its types, its application and much
A thermocouple is formed when two electrical conductors composed of dissimilar materials,
come in contact with each other. Thermocouples are able to generate voltages through a
process called the thermoelectric effect. The ability to generate voltage as a function of
temperature makes them suitable as a temperature measuring device.

The underlying principle behind thermocouples, the thermoelectric effect, was discovered by
Thomas Johann Seebec. The thermoelectric effect, sometimes called the Seebec effect, occurs
when a thermal gradient is applied to a conductor and produces an Electromotive Force
generated across the gradient (See Figure 1). The magnitude of the Electromotive Force is a
function of the temperature differential applied, and the property of the material being used.
(Each material type has its own unique temperature characteristics.)

                                           Figure 1
In Figure 1, if the temperature T1 is greater that T2 then the Electromotive force causes a
potential to develop at T1 as compared to T2. The reverse is true if T2 is greater than T1.

While interesting, this is hardly useful, since measuring the voltage at T1 and T2 requires
additional conductors (e.g., voltmeter probes) which pick up not only the voltage at T1 and
T2 but also the temperature at T1 and T2. The resultant thermal gradients in these new
conductors counteract the Electromotive force and so the voltage reading is always zero.

A second conductor of a different type can be introduced to create a circuit for current to
flow. In Figure 2, two conductors of differing materials are joined at each end. (The
connection points between the conductors are thermocouples, also called junctions, or
thermocouple junctions).

                                             Figure 2

Since the two conductors are of differing materials, they have different Electromotive forces
for any given temperature gradient applied. Remember the Electromotive Force is dependent
on the conductor material (each material has unique thermoelectric properties and so can
produce greater or lesser Electromotive Forces). In Figure 2, with differing materials and a
temperature gradient between T1 and T2, the Electromotive Force will cause a current will
flow around the loop. Assuming Material A generates a stronger EMF for a positive thermal
gradient between T1 and T2, current will flow around the loop in a counterclockwise fashion
(A wins over B). Conversely, if T2 is hotter than T1 the current flows in a clockwise fashion
(A wins over B again, but in the opposite direction). The magnitude of current is a function of
the temperature difference and controlled by the resistance of the conductors. (i.e., higher
current will be generated given heaver wire, or higher temperature gradient. So the
temperature differential between T1 and T2 can be determined by knowing the materials and
their properties, the conductor resistances and by measuring the current.

Needless to say the circuit is difficult to use and a more complex circuit is necessary if the
Seebec Effect is to be realized into a useful tool.

Before launching into discussions on more complex circuits, a point of interest is in order.
Any two dissimilar conductors can be used for the circuit in Figure 2 and will create
electromotive forces characteristic of those two conductors. Over many years, thermocouple
manufacturers have painstakingly tested and measured the characteristics of various material
pairs and developed a thorough understanding of the properties for these pairs. This has
resulted in a standardization of a number of useful wire pairs with well known characteristics.
Figure 3 depicts a practical circuit for measuring the temperature difference between T1 and
T2. The circuit introduces a third conductor, Copper (CU), and three thermocouples instead of
just two. A meaningful Temperature/Voltage measurement on the Copper wires depends
entirely upon holding the same temperature on two junctions at T2. With these two
thermocouples at the same temperature, the Copper material’s Electromotive Force
contribution cancel away, and the circuit can be treated as having just two material (similar to
the circuit in Figure 2).

                                            Figure 3

Furthermore, since the voltmeter at Vt is high impedance the wire resistance is no longer a
factor. This circuit is quite functional and is quite repeatable given any particular T1 and T2.

Unfortunately the voltage produced at Vt is only a function of the difference between T1 and
T2 and not an absolute measurement of the T1 temperature. Historically, this has been solved
by holding T2 at a precisely known temperature. Commonly the junctions at T2 were
submersed into an ice bath, as is shown in Figure 4. Holding temperature T2 to exactly 32°F
finally allows the temperature at T1 to be accurately calculated from the voltage read at Vt.
As a side note, the practice of holding T2 in an ice bath earned it the name the cold junction.
And as one would expect, T1 is referred to as the hot junction. (The term cold junction is still
used even though ice baths are hardly ever used today.)

                                            Figure 4

Earlier we discussed that various material pairs had been painstakingly tested and their
thermoelectric properties evaluated. Through this process, standard pairs have been
established. As it turns out, regardless of the thermocouple type, the voltage produced is
nonlinear with respect to the temperature at T1, and T1 cannot be accurately determined by a
simple slope and offset conversion from Vt.

The relationship between the voltage at Vt and temperature can be described fairly accurately
as a polynomial equation of the form:
Where the order of the polynomial (n) is between 5 and 8, depending on the thermocouple
type. The coefficient values ( have been calculated for each thermocouple type and are
readily available for reference. Calculations such as there are somewhat difficult to
implement and a simpler approach is available.

Fortunately, data tables have been produces that very accurately define the relationship
between voltage and temperature for each thermocouple type. These data tables have
excellent resolution with voltage values for each degree of temperature over the whole
working temperature range of the thermocouple. As expected these data tables are referenced
to an ice bath (32°F or 0°C).

Standard thermocouple types are identified by their letter code, which define the materials
being used. Common thermocouple types include:

Type B, R and S     Platinum / Rhodium   Stable, lower sensitivity, good at high
Type C              Tungsten / Rhenium   Used at extremely high temperatures
Type E              Chromel / Constantan High output, suitable for cryogenic used
Type J              Iron / Constantan    Limited range, medium sensitivity, problematic
                                         due to corrosion
Type K              Chromel / Alumel     General purpose, wide range, inexpensive,
Type N              Nicrosil / Nisil     High Temperature, good stability
Type T              Copper / Constantan  Non-magnetic (good for measurements in
                                         strong magnetic fields

                                            Table 1

Looking at any of the thermocouple tables, the zero point for volts is at 32F (or 0C) with
positive voltages above freezing and negative below. This makes sense since the
thermocouple tables were developed with T2 in an ice bath, 32F. (i.e., no gradient, no

While the ice bath approach produces accurate results, it is difficult to use and limited in
practicality for most applications. A better approach is shown in Figure 5 where an absolute
temperature sensor is used to measure the temperature of T2. While the temperature T1 may
be at an extreme temperature (e.g., inside a furnace, or cryogenic chamber) T2 is likely to be
in a rather fairly benign temperature environment where an accurate absolute temperature
sensor can be used. A typical silicon sensor, such as Analog Devices AD590, has a fairly
wide working temperature range of -55°C to 150°C which supports most environments for
                                           Figure 5

To implement this approach, and use the zero-reference thermocouple tables, the voltage read
at Vt must be corrected for the temperature at T2. One way of thinking about the voltage at Vt
is that it has two contributors, Thermocouple T1 and Thermocouple T2. The zero-reference
table must be used twice to perform the compensation. To compensate, read the absolute
temperature sensor and then using the zero reference table determine the voltage contribution
of T2. Subtract this value from the voltage reading Vt. And next use the zero-reference table
again to determine the temperature at T1.
 Then we take referance from our
  course incharge who suggests us to
  use the k type of thermo couple in
  our circuit which is easy to
  understand and assembled in the
Datasheet of Thermo couple
Thermocouples are the most popular
temperature sensors. They are cheap,
interchangeable, have standard connectors
and can measure a wide range of
temperatures. The main limitation is
accuracy, system errors of less than 1°C
can be difficult to achieve.
  Then from our research about the
   circuit from the internet, library and
   other resources we first compile a
   simple circuit which is:
  The above circuit is the thermo
   couple amplifier
  In this circuit the Operational
   Amplifier is used which amplify the
   voltage generation produce by
   thermo couple
  Also we use the IC LM35 which is
   also a temperature sensor
   integerated circuit
  After few days we compile a circuit
   which named as

The schematic of simple circuit is:
Now let's create a practical circuit using this popular sensor. Because the actual voltage is
small, you need a lot of gain to make this signal big enough to pass to an ADC. The other
tricky bit is that the measurement depends on the temperature both at point of interest
(measure junction) AND the where the thermocouple connects to your circuit (reference
junction). Because the reference junction will change with room temperature, you need to
compensate for these changes (cold-end compensation).

Armed with a thermocouple, a local temp sensor and a summing amplifier you can design a
practical preamplifier. As a project goal, let's design a circuit for a k-Type thermocouple that
produces a 0 to 100 mV output for a 0 to 100 deg C temperature change - effectively, an
output of 1 mV / C.


What circuit can juggle the multiple functions required by the thermocouple preamp? The
summing amplifier is perfect for accomplishing these three functions.

1. Amplify the thermocouple signal.
2. Subtract the effect of the reference junction at room temperature.
3. Add an offset adjustment.

The summing amplifier is nice because you can adjust each input independently using input
resistors R1, R2 and R3.

Vo = (-RF/R1) · V1 + (-RF/R2) · V2 + (-RF/R3) · V3

Let's design a circuit to accomplish these three functions above using three simple steps.

How much gain is needed to generate a 100mV for 100 deg C temp change? First, find how
much voltage Vtc the thermocouple produces. Although it's somewhat nonlinear over a large
temp range, calculate the average temperature coefficient for a k-Type thermocouple over
100 deg C.

dV/dT = [Vtc(100) - Vtc(0)] / [100 deg C - 0 deg C ]

     = [4.096mV - 0 mV ] / 100 deg C

     = 40.96 uV / deg C

Given this coefficient and a target output of dVo/dT = 1 mV / deg C, we calculate the gain
K1 needed.

K1 = (dVo/dT) / (dVtc/dT) = (1mV / C) / (52.69uV / C) = 24.414 V/V

All that remains is finding R1 to achieve the gain. First choose RF to be 100k (you can
always choose a different value if the other resistors are not practical or available.) Then,
calculate R1

R1 = RF / K1 = 100k / 24.414 = 4096 ohms.

Remember the summing amp is inverting! To get a positive output, connect the neg side of
the thermocouple to the amplifier input.


A major thorn in the side of a thermocouple measurement is that they don't measure the
absolute temperature of the junction. However, they do measure the temperature
difference between the measure junction Tjunc and the reference junction Tref. So what
happens if Tjunc = 100 C and Tref (room temperature) increases from 22 C to 23 C. The
difference changed from 100 C - 22C = 78 deg C to only 100 C - 23 C = 77 deg C!
Accordingly, the thermocouple voltage changed proportionately by -1 deg C even though the
measure junction did not change! Correspondingly, the preamp's output will change by -1

To compensate for this decease in voltage, you need to add a voltage to the output that's
proportional to +1 deg C = +1 mV. To accomplish this feat, simply place a local
temperature sensor close to the reference junction. Then, use the summing amp to get +1mV
output for every deg C of local temp change.

For this design, we've chosen an LM335 IC. This sensor produces a voltage change of
+10mV / deg C. (Actually the sensor operates in Kelvin, 0 deg C = 273 K, generating an
output at 22 C of Vsensor = (273 + 22) C * 10mV/C = 2.95 V.) So how much gain do we
need to transform 10mV / C into 1 mV/deg C?

K2 = (dVo/dT) / (dVsensor/dT) = (+1mV / C) / (+10 mV/ C) = 0.1 V/V

Then, calculate R2

R2 = RF / K2 = 100k / 0.1 = 1000k ohms.
Again, because the summing amp is inverting and we need a +1mV/C at the output, connect
the sensor's negative terminal to V2. The upshot of this is V2 = -10mV/C and Vo2 =
+1mV/C. The only challenge left is subtracting the sensor offset of 2.95V.


For the final step, we need to subtract the large offset of the local temp (-2.95V) and
accurately adjust the output for the correct temperature. This fairly easy step involves picking
a specific measurement and reference temperature, then using our summing amp equation to
find the offset voltage Voff required at V3. (Voff is typically implemented with a

Our grand equation looks like

Vo = (-RF/R1) · V1    + (-RF/R2) · V2      + (-RF/R3) · V3
    (-RF/R1) · Vtc    + (-RF/R2) · Vsensor + (-RF/R3) · Voff

All that remains is choosing R3 and rearranging the equation to find Voff.

    Voff = ( Vo + (RF/R1)·Vtc + (RF/R2)·Vsensor ) / ( -RF/R3 )

To make our calculation easier, let's choose the Tjunc = 22 and Tref = 22 C. Why is this
easier? With the Tjunc == Tref, the thermocouple output is zero, Vtc = 0 mV. Let's substitute
what we know and choose R3 = 1000k. Given our design goal of 1mV/C, we expect 22mV at
the output.

22mV = (-100k/4096) · 0mV + (-100k/1000k) · -2.95V + (-100k/1000k) · Voff
     =                    (-100k/1000k) · -2.95V + (-100k/1000k) · Voff

Voff = ( Vo     + (RF/R2) · Vsensor )    / (-RF/R3)
          = ( 22mV + (100k/1000k) · -2.95V ) / (-100/1000k)
          = +2.73 V

Question: can you easily generate Voff with an available reference voltage and a
potentiomemter? Imagine for this design we have a Vref = +5V and a 5k ohm pot connected
from 5V to ground - piece of cake! The wiper of the pot can be adjusted from 0 to 5V.
However, if your Voff was greater than Vref, you could choose a different R3 such that Voff
< Vref.

Here's the SPICE circuit of the thermocouple preamp.
To simulate the measure junction temperature Tjunc at V(10) and reference junction
temperatures Tref at V(11), we'll use two voltage sources where V really represents T in deg
C. The Piece-Wise Linear (PWL) sources will linearly ramp the temperatures over a 100ms

V_TJUNC 10 0 PWL(0MS 0 100MS 100)
V_TREF 11 0 PWL(0MS 22 100MS 22)

Initially, we'll sweep Tjunc from 0 to 100 deg C while keeping Tref fixed at 22 deg C. These
voltages (temperatures) feed the thermocouple defined by a subcircuit . The negative terminal
connects to R1.

X_TC1 10 11 0 1 TC_J_1

The cold-end compensation is achieved by a local temperature sensor (LM335 IC) modeled
simply as a Voltage Controlled Voltage Source (VCVS) at V(11). An equation describes it's
V vs. T behavior of +10 mV / K where 273 K = 0 deg C. Similar to the thermocouple, the
negative terminal connects to R2.

E_VSENSOR 0 2 VALUE = { 2.73 + V(11)*0.010 }

A simple voltage source creates the offset

VOFF 3 0 DC 2.73

Finally, our op amp SPICE model is a simple one capturing it's open-loop gain and gain-
bandwidth product.

XOP1 0 4 5 OPAMP1

Let's see if our calculations actually mean something. We'll simulate operation from 0 to 100
deg C with the reference junction held at 22 C (room temperature).

 CIRCUIT INSIGHT Run a simulation of TC_PREAMP1.CIR. Plot Tjunc at V(10) and Tref
at V(11). Open a second window and plot the thermocouple output at V(1). Open a third
window and plot the preamp output at V(5). Does the output match the expected 0 to 100
mV? The preamp's output should be within a mV of the target value.

Actually, because we calculated Voff at conditions Tjunc = Tref = 22 deg C, the output
should be spot on at this temperature. Move your cursor to around 22 deg C. How accurate is
the output in mV? Because the thermocouple is a non-linear device, (requiring an 8th order
polynomial to approximate it's output), the accuracy falls off as you move away from the 22
deg C adjustment point.

Finally, let's put our cold-end compensation to the test! Keep Tjunc fixed at 100 deg C and
sweep Tref (room temperature) from 10 to 40 deg C. How well does the local temp sensor
and circuitry keep the measurement steady? Rerun the simulation and check V(5). Did the
output hover near 100 mV as the room temperature swung through it's expected range?
Seeing the cold-end compensation was the big payoff for me. You can see the thermocouple
V(1) changing with room temperature while the local sensor (open a new window and plot
V(2) ) was moving in the opposite direction to cancel the effect!


This was a marathon topic - high fives for staying the course! Just a few closing notes. To get
a highly accurate reading over a large temperature swing, many systems use calibration and
software to linearize the highly non-linear output of a thermocouple. These T vs. V equations
are typically published with the V vs. T equations used to create the SPICE model.

The gain resistor R1=5269 will not likely be a standard resistor value. But you can always
use a potentiometer or a parallel/series combination to get very close.

Also, you'll find the local sensor for cold-end compensation comes in many forms - IC
sensor, thermistor, diode, Vbe junction. Which ever is used, simply choose the polarity and
apply a gain such that it's output cancels thermocouples change due to local junction
temperature change. In a real circuit, the local sensor LM335 needs a bias resistor not shown
in the simulation. We didn't need one because our simplified SPICE model used a voltage
source defined by an equation.
A friend of mine designed and bread-boarded a preamp circuit for the K thermocouple.

He added a gain adjust (potentiometer) for R2 as well as one for the offset adjust voltage
Voff. For calibration, he used two handy temperature references inside every house - a
container of ice cubes and a pot of boiling water. Thank you Mike for the picture of the home

V_TJUNC 10      0      PWL(0MS 0       100MS 100)
RD10    10      0      1MEG
V_TREF 11       0      PWL(0MS 22      100MS 22)
RD11    11      0      1MEG
X_TC1   10 11 0 1      TC_J_1
R1      1       4      5269
E_VSENSOR 0     2      VALUE = { 2.73 + V(11)*0.010 }
R2      2       4      1000k
VOFF 3 0        DC     2.73
R3      3       4      1000K
RF      4       5      100K
XOP1    0 4     5      OPAMP1
*** J THERMOCOUPLE SUBCIRCUIT ************************************
*        T_JUNC(C) - 1      T_REF(C) - 2    V_TC+(V) - 3 V_TC-(V) - 4
.SUBCKT TC_J_1 1 2 3 4
*     V = B1*T + B2*T^2 + ...
E_TJ 5 0 VALUE = {
+    0.503811878150E-01*V(1)       +   0.304758369300E-04*V(1)**2 +
+ -0.856810657200E-07*V(1)**3 +        0.132281952950E-09*V(1)**4 +
+ -0.170529583370E-12*V(1)**5 +        0.209480906970E-15*V(1)**6 +
+ -0.125383953360E-18*V(1)**7 +        0.156317256970E-22*V(1)**8   }
E_TR 6 0 VALUE = {
+    0.503811878150E-01*V(2)       +   0.304758369300E-04*V(2)**2 +
+ -0.856810657200E-07*V(2)**3 +        0.132281952950E-09*V(2)**4 +
+ -0.170529583370E-12*V(2)**5 +        0.209480906970E-15*V(2)**6 +
+ -0.125383953360E-18*V(2)**7 +        0.156317256970E-22*V(2)**8   }
* V_TC = V_TJ - V_TR
E_TC     3       4       VALUE = { (V(5)-V(6))/1000 }
* connections:        non-inverting input
*                     |    inverting input
*                     |    |    output
*                     |    |    |
.SUBCKT OPAMP1       1   2     6
RIN      1       2       10MEG
* DC GAIN (100K) AND POLE 1 (100HZ)
EGAIN    3 0     1 2     100K
RP1      3       4       1K
CP1      4       0       1.5915UF
EBUFFER 5 0      4 0     1
ROUT     5       6       10
.TRAN    1MS 100MS
 Taking digital output
 When the circuit is fully operated by
  resulting analog output we want to take
  output of it is in digital way
 But the main problem is to convert analog
  signal into digital signal and for this
  problem what should we have to use?
 After some research we use ICL7107
  which is high performance, low power
  A/D converter
 For digital output we use 31/2 digital
  segment which gives digital output
 The data sheet for ICL7107 is :
 From more research we also compile the
  analog to digital converter circuit:

 Figure-ICL7107 Test circuit and typical application with LED
 display components selected for 200mv full scale
    The components selected for it is

    The final circuit which we design by hard
     struggle is



ICL7107                        LM324

      Figure- Digital temperature measurement using thermocouple,
      the main components in above figure are shown in small text
  The circuit is so cheap to design

  The design is assembled easy for
   any industrial purpose

  All reading can br known easily
  The whole circuit cannot be placed
   to the high energy area because
   except thermocouple the all
   components are so delicate or
   cruicial to high temperature
I would like to express deepest
appreciation to our course incharge
ENGR. ABDUL WAHAB, who has the
attitude and the substance of a genius:
he continually and convincingly
conveyed a spirit of adventure in
regard to guide us in every step when
we tired of thinking so much about
project. Without his guidence and
persistent help this dissertation would
not have been possible.
I would like to thanks my group leader,
FARRUKH ALI KHAN (1909), who played
a vital role for designing this project.

To top