TMP006 Customer Training
Applications Team
Sensing Products
What is a Thermocouple / Thermopile
Thermocouple
Thermopile
+
+
Tcold
Thot
-
Vout
Thot
Thot
Vout
Tcold
Thot
Heat
Absorber
Heat Sink
Thot
Heat Absorber
-
Cold Junction
Vout = G x (Thot – Tcold)
Vout = n x G x (Thot – Tcold)
G: Seebeck Constant
G: Seebeck Constant
Heat Transfer (1): Conduction
• Conduction: Heat is transferred through direct touch.
– Example: Touching a hot cup of coffee
Heat Transfer (2): Convection
• Convection: Heat is transferred by a moving liquid or gas.
– Example: Fan cooling
Heat Transfer (3): Radiation
• Radiation: Heat is transferred by waves (photons)
– Example: The warmth you feel from the sun
– The amount of energy radiated is dependent on the absolute (Kelvin)
temperature of the object.
Device Principle of Operation
Object
Infrared Radiation
TMP006
PCB
TMP006 at its core is a heat sensor. It converts the heat being
transferred to a voltage.
PCB Recommended Layout
• The sensor at it’s core is a heat sensor and is sensitive to conduction
and convection in addition to radiation.
• Recommended PCB layout isolates the sensor from the rest of the PCB
GND
Via
Connecting
to Circuit
Thermal
Break
Copper Dot
TMP006 Conduction & Convection Offsets
•
The sensor, is a heat sensor at its core. It will detect heat from all
three sources: conduction, convection, and radiation.
•
The signal created by heat conduction and convection is relatively
small in a setup with the recommended PCB layout but it is not zero.
•
Calibration is needed to cancel out remaining offsets
Field of View Consideration (1)
• For best performance it is recommended that the diameter of object is 4
times the distance to the TMP006 sensor.
• d = h – 250µm
• object size ≥ 4×d
Case
object size ≥ 4d
distance = d
h
TMP006
PCB
Field of View Consideration (2)
• For best performance it is recommended that the diameter of object /
opening is 4 times the distance to the TMP006 sensor.
• d = h – 250µm
• Opening size ≥ 4×d
Case/Lens
opening size ≥ 4d
heat sink
distance = d
h
TMP006
PCB
Emissivity Considerations
• Not all objects emit the same amount of energy at the same
temperature.
• Emissivity is the ratio of energy radiated by an emitter to the theoretical
ideal emitter. Emissivity is always less than one.
• Since the emissivity may change for different systems, then the system
gain will need to be calibrated.
TMP006 Equation
Needs calibration. Depends on field
of view and emissivity of object
S  S0
1  a T
1
die
 T ref
  a T
2
die
System independent: No
need to calibrate
 T ref
V os  b 0  b1 T die  T ref   b 2 T die  T ref
f V obj   V obj  V os   c 2 V obj  V os 
2

2

2
Models voltage offset due to
conduction & convection.
System dependent & needs
calibration
System independent: No
need to calibrate
T obj 
4
T
4
die
 f (V obj ) 

 

S


Transient Correction Basics
• The hot junction is isolated from the rest of the sensor by the sensor’s thermal
resistivity
• When the sensor’s cold junction temperature is changed through heat conduction
from the outside, the hot junction follows, but is delayed by a time constant
• This effect can be modeled with a first order RC filter that delays the hot junction
relative to the sensor substrate temperature
Thot
Cth
Thermopile
Tcold
+
Rth
Tcold
Thot
• Rth Thermal resistance of thermopile
• Cth Thermal capacitance of thermopile
Heat
Absorber
Heat Sink
-
Vout
Tcold
Thot
time
Temperature
Temperature
Transient Error Example
Tcold
Thot
time
Transient Correction Basics (2)
V Err  Seebeck  
V Err  
V Err  
 T Cold
t
 T Cold
t
 T Cold
t
A temperature gradient is
created on the thermopile
during transients that is
proportional the slope of the
transient itself.
This gradient generates a
voltage error through the
Seebeck effect.
 R th C th
  Seebeck  R th C th 

α   Seebeck  R th C th   2 .96  10
-4
(V  sec / C)
Alpha is a constant which corresponds to the
multiplication of Seebeck constant and the thermal
time constant
With transient correction
Without transient correction
Effect of Transient Correction
Coefficient Calibration
• Coefficients that get calibrated are: S0, b0, b1 & b2
• Before calibrating the coefficients, transient correction should be
applied to the sensor’s output voltage to cancel out the transient
effects.
• At this point in time it is recommended that the customer send us the
data of their system and we will provided a custom set of “b”
coefficients for optimum accuracy.
• An automatic calibration software is being developed at this point and
should be available for customer use in the near future
Setting Up The System For Calibration
Object
Reference Temperature
Sensor (Example: TMP112)
Glued to object using thermal
epoxy
TMP006
PCB
• For calibration: the data from the reference: Tobj as well as the TMP006
data: Tdie & Vout need to be collected simultaneously.
Calibrating S0

S  S 0 1  a 1 T die  T REF   a 2 T die  T REF
V os  b 0  b1 T die  T REF   b 2 T die  T REF
 V obj 1  V os 1
4
4
T obj 1  T die 1  
S1

T
4
obj 2
T
4
die 2




 V obj 2  V os 2
 
S2

2 
2
To be able to accurately calibrate S0
you need at least two points with the
same Tdie values but different Tobj
values.
More than one measurement is
recommended to reduce the effect of
noise




if T die 1  T die 2 
 V os 1  V os 2  V os
 S1  S 2  S


 T obj 2  T obj 1 
4
 S0 
4
T
4
obj 2
V obj 2  V obj 1
S
 


V obj 2  V obj 1
S 0 1  a 1 T die  T REF   a 2 T die  T REF
V obj 2  V obj 1
 T obj 1  1  a 1 T die  T REF   a 2 T die  T REF
4
2 
2 
Calibrating “b” Coefficients
T
4
die
 V obj  V os
 
S






 V os  V obj  S T obj  T die
4
4

V os  b 0  b1 T die  T REF   b 2 T die  T REF
• The residual offset based on the
calibrated S0 value is calculated
according to the equation above.
• The calculated offset is drawn
versus (Tdie – Tref) and a second
order curve fit for the data is done
to calculate the b0, b1 and b2
coefficients.
2
Vos = Vobj – S(Tobj4 – Tdie4)
T
4
obj
2nd order curve fit
(use excel)
Coefficients of
curve fit are b0, b1
and b2
Tdie – Tref
Importance of the Number and Range of Points
y-axis
Error at the extreme points
Calibrated curve based
on only middle three
points
Actual device behavior
Points collected
don’t exactly match
the actual curve
due to noise
x-axis
• The larger the number of points the lower the effect of noise.
• The further apart the points are the better the correction.
Calibrated coefficients
Default coefficients
Default Versus Calibrated Coefficients