7.3.1 Spectral Responsivity

The TMP007 is optimized to sense IR radiation emitted by objects from approximately 250 K (–23°C) to 400 K (127°C), with maximum sensitivity from approximately 4 µm to 16 µm. The relative spectral response of the TMP007 is shown in Figure 18.

Figure 18. Relative Spectral Response vs Wavelength

7.3.2 Field of View and Angular Response

The TMP007 senses all radiation within a defined field of view (FOV). The FOV (or full-angle of θ) is defined as 2Φ. The TMP007 contains no optical elements, and thus senses all radiation within the hemisphere to the front of the device. Figure 2 shows the angular dependence of the sensor response and the relative power for a circular object that subtends a half angle of phi (Φ). Figure 19 defines the angle Φ in terms of object diameter and distance. Figure 19 assumes that the object is well approximated as a plane that is perpendicular to the sensor axis.

Figure 19. FOV Geometry Definition

In this case, the maximum contribution is from the portion of the object directly in front of the TMP007 (Φ = 0), with the sensitivity per solid angle, dR/dΦ decreases as Φ increases. Approximately 50% of the energy sensed by the TMP007 is within a FOV (θ) = 90°.

This discussion is for illustrative purposes only; in practice the angular response (dR/dΦ) of the TMP007 to the object is affected by the object orientation, the number of objects, and the precise placement relative to the TMP007. Figure 20 shows the thermopile sensor dimensions.

NOTE: Thermopile sensor is centered in the device.

Figure 20. Thermopile Sensor Dimensions

7.3.3 Thermopile Principles and Operation

The TMP007 senses radiation by absorbing the radiation on a hot junction. The thermopile then generates a voltage proportional to the temperature difference between the hot junction, T_hot, and the cold junction, T_cold.

Figure 21. Principle of Thermopile Operation

The cold junction is thermally grounded to the die, and is effectively T_DIE, the die temperature. In thermal equilibrium, the hot junction is determined by the object temperature, T_OBJ. The energy emitted by the object, E_OBJ, minus the energy radiated by the die, E_DIE, determines the temperature of the hot junction. The output voltage, V_OUT, is therefore determined by the relationship shown in Equation 2:

Equation 2.

where

• C is a constant depending on the design of the sensing element.

Note that the sensor voltage is related to both the object temperature and the die temperature. A fundamental characteristic of all thermopiles is that they measure temperature differentials, not absolute temperatures. The TMP007 contains a highly-accurate, internal temperature sensor to measure T_DIE. Knowing T_DIE and V_SENSOR enables the TMP007 to estimate T_OBJ. For each 250-ms conversion cycle, the TMP007 measures a value for V_SENSOR and for T_DIE, calculates T_OBJ, and then places the values in the respective registers.

Bits CR2 to CR0 determine the number of local and sensor analog-to-digital converter (ADC) results to average before the object temperature is calculated.

After power-on reset (POR), the TMP007 starts in four conversions per second (mode 010). In general, for a mode with N conversions, the local temperature, T_DIE, result is updated at the end of the Nth ADC conversion with the value shown in Equation 3:

Equation 3.

Similarly, the sensor voltage result is updated at the end of the Nth sensor ADC conversion with the value shown in Equation 4:

Equation 4.

These results are then used in the object temperature calculation by the math engine, which updates the object temperature result register. The total conversion time and averages per conversion can be optimized to select the best combination of update rate versus noise for an application. Additionally, low-power conversion mode is available. In CR settings 101, 110, and 111, the device inserts a standby time before the beginning of the next conversion or conversions.

The method and requirements for estimating T_OBJ are described in the next section.

7.3.4 Object Temperature Calculation

The TMP007 generates a sensor voltage, V_Sensor, in register 00h that is related to the energy radiated by the object. For an ideal situation, the Stefan-Boltzman law relates the energy radiated by an object to its temperature by the relationship shown in Equation 5:

Equation 5.

where

• σ = Stefan-Boltzman constant = 5.67 × 10^-12 W/(cm²K⁴)
• ε = Emissivity, 0 < ε < 1, an object dependent factor, ε = 1 for a perfect black body

A similar relationship holds for the sensing element itself that radiates heat at a rate determined by T_DIE. The net energy absorbed by the sensor is then given by the energy absorbed from the object minus the energy radiated by the sensor, as shown in Equation 6:

Equation 6.

In an ideal situation, the sensor voltage relates to object temperature as shown in Equation 7:

Equation 7.

Equation 8.

where

• S is a system-dependent parameter incorporating the object emissivity (ε), FOV, and sensor characteristics. The parameters S0, A1, and A2 are used in determining S.
• f(V_OBJ) is a function that compensates for heat flow other than radiation, such as convection and conduction, from nearby objects. The parameters B0, B1, and B2 are used to tune this function to a particular system and environment.

The coefficients affect object temperature measurement as described in Table 1.

7.3.5 Calibration

The TMP007 default coefficients are calibrated with a black body of emissivity, ε = 0.95, and an FOV (θ) = 110°. Use these coefficients for applications where the object emissivity and geometry satisfy these conditions. For applications with different object emissivity or geometry, calibrate the TMP007 to accurately reflect the object temperature and system geometry. Accuracy is affected by device-to-device or object-to-object variation. For the most demanding applications, calibrate each device individually.

As an overview the calibration procedure includes:

1. Defining the environmental variation range (die and object temperature range, supply voltage, temperature change speed, sampling rate and so on).
2. Making the die temperature measurements and IR sensor voltage measurements over the environmental range.
3. Generate an optimal set of coefficients based on the collected data set.
4. Load the coefficients into the TMP007 coefficients register. The object temperature register reflects the best fit from the calibration process. Perform validation measurements because accuracy may vary over the environmental range. If the object temperature measurement error is not acceptable, repeat the calibration process using more environment points, data averaging, or narrow the temperature range of T_DIE or T_OBJ.
5. After a suitable set of coefficients is obtained, they can be stored in nonvolatile memory. Each coefficient register can be programmed up to eight times. After POR, the last stored coefficient value is copied from the nonvolatile memory into the coefficient register.

The best temperature precision is available if every device is calibrated individually. Alternatively, if all the units in the application use the same coefficients, then calibrate a statistically significant number of devices, and load averaged coefficient values in nonvolatile memory.

Recalibration may be required under any or all of the following conditions:

1. Board layout changed.
2. Object or objects in the field of view changed.
3. Object distance or object surface changed.
4. Angle between device surface and direction to the object changed.
5. Object and local temperature range changed outside the environmental calibration range.
6. Object and local temperature transients significantly changed.
7. Supply voltage changed more than 1 V.
8. Air convection or conduction near the device changed.

For further information and methods for calibration, refer to SBOU142 — TMP007 Calibration Guide.

7.3.6 Sensor Voltage Format

The TMP007 provides 16 bits of data in binary twos complement format. The positive full-scale input produces an output code of 7FFFh and the negative full-scale input produces an output code of 8000h. The output clips at these codes for signals that exceed full-scale. Table 2 summarizes the ideal output codes for different input signals. Figure 22 illustrates code transitions versus input voltage. Full-scale is a 5.12-mV signal. The LSB size is 156.25 nV.

Figure 22. Code Transition Diagram

7.3.7 Temperature Format

The temperature register data format of the TMP007 is reported in a binary twos complement signed integer format, as Table 3 shows, with 1 LSB = (1 / 32)°C = 0.03125°C.

To convert the integer temperature result of the TMP007 to degrees Celsius, right-shift the result by two bits. Then perform a divide-by-32 of T_DIE and T_OBJ, the 14-bit signed integers contained in the corresponding registers. The sign of the temperature is the same as the sign of the integer read from the TMP007. In twos complement notation, the MSB is the sign bit. If the MSB is 1, the integer is negative and the absolute value can be obtained by inverting all bits and adding 1. An alternate method of calculating the absolute value of negative integers is abs(i) = i xor FFFFh + 1.

7.3.8 Serial Interface

The TMP007 operates only as a slave device on the serial bus. Connections to the bus are made using the SCL Input and open-drain I/O SDA line. The SDA and SCL pins feature integrated spike suppression filters and Schmitt triggers to minimize the effects of input spikes and bus noise. The TMP007 supports the transmission protocol for both fast and fastplus (1 kHz to 1 MHz) and high-speed (1 MHz to 2.5 MHz) mode. All data bytes are transmitted MSB first. At higher speeds, thermal dissipation affects device operation, including accuracy.

7.3.8.10 Two-Wire Timing Diagrams

Figure 23. Two-Wire Timing Diagram

1. The value of A2, A1, and A0 are determined by the ADR1 and ADR0 pins.

Figure 24. Two-Wire Timing Diagram for Write Word Format

1. The value of A0, A1, and A2 are determined by the connections of the corresponding pins.

2. Master should leave SDA high to terminate a single-byte read operation.

3. Master should leave SDA high to terminate a two-byte read operation.

Figure 25. Two-Wire Timing Diagram for Read Word Format

1. The value of A0, A1, and A2 are determined by the connections of the corresponding pins.

Figure 26. Timing Diagram for SMBus Alert

7.4.1 Temperature Transient Correction

Because the measured object temperature depends on T_DIE, transient thermal events that change the die temperature affect the measurement. To compensate for this effect, the TMP007 math engine incorporates a transient correction option for use in applications where a thermal transient is anticipated. When transient correction is turned on, a filter with programmable coefficients is used to modify the sensor voltage result before the object temperature is calculated. This function helps reduce the jump in the object temperature result when there are large transients of the local die temperature, T_DIE. The compensation incorporates the rate of change of T_DIE and of V_OBJ. The modified value for the sensor voltage used in V_SENSOR to calculate the object temperature is shown in Equation 9:

Equation 9.

where

• TC0 and TC1 are weighting coefficients programmable using the registers.
• T_{DIE_SLOPE} is the change in die temperature with time.
• V_{OBJ_SLOPE} is the change in sensor voltage with time.

As a general guideline, turn on transient correction when the local temperature is changing at a rate greater than 1.5°C/min. When transient correction is on, the function corrects transients up to approximately 20°C/min.

Turning on the transient correction also turns on the output filter shown in Equation 10:

Equation 10.

If only the use of the output filter is desired without the input transient correction arithmetic, set the TC0 and TC1 coefficient values to 0 with TC bit in configuration register set to 1. When transition correction is on, the response to a step change has a time constant of approximately five times the sampling time.

When transient correction is on, the math engine modifies the sensor voltage result based on the transient correction equations. The nonmodified sensor voltage can be recovered with TC on by setting the TC1 and TC0 coefficients to 0. The output filtering cannot be turned off with TC bit set to 1.

7.4.2 Alert Modes: Interupt (INT) and Comparator (COMP)

The INT mode maintains the alert condition until a host controller clears the alert condition by reading the status register. This mode is useful when an external microcontroller is actively monitoring TMP007 as part of a thermal management system. The COMP mode asserts the ALERT pin and flags whenever the alert condition occurs, and deasserts the ALERT pin and flags without external intervention when the alert condition is no longer present This mode is often used to notify an external agent of an alert condition.

When servicing an alert from the TMP007, in some cases it may be useful to validate the alert condition by checking the status of the nDV, MEM_CRPT, and DATA_OVF flags.

7.4.2.1 INT Mode (INT/COMP = 0)

In this mode the high and low limits form a limit window. The ALRTEN bit must be asserted if the ALERT pin functionality is desired. If the calculated temperature is above the high limit or below the low limit at the end of a conversion its respective enabled flag is asserted.

Figure 27. INT Mode

After the flag is asserted, it can only be cleared by a read of the status register, which clears the flag and the pin. The ALERT pin can also be cleared by the SMB alert response command (see the SMBus Alert Function section); however, this action does not clear the flag.

7.4.2.2 COMP Mode (INT/COMP = 1)

In COMP mode, the limits are used to form an upper limit threshold detector. If the calculated temperature is above the high limit, the high limit flag is asserted. The high limit flag is then deasserted only after the temperature goes below the low limit. The low limit register value determines the degree of hysteresis in the COMP function. In COMP mode, only the high limit enable has effect on the limit flags. The low limit enable flag does not have any effect on the low limit flags and the low limit flags always read 0. In this mode, the flags are asserted and deasserted only at the end of a conversion and cannot be cleared by a status register read or an SMB alert response.

Figure 28. COMP Mode

7.4.3 Nonvolatile Memory Description

Table 7. Internal Register Description

Table 8. Register Map

7.5.1 Sensor Voltage Result Register (address = 00h) [reset = 0000h]

7.5.2 T_DIE Local Temperature Result Register (address = 01h) [reset = 0000h]

7.5.3 Configuration Register (address = 02h) [reset = 1440h]

7.5.4 T_OBJ Object Temperature Result Register (address = 03h) [reset = 0000h]

7.5.5 Status Register (address = 04h) [reset = 0000h]

The status register flags are activated whenever their limit is violated, and latch if the INT/COMP bit is in INT mode (see configuration register). In INT mode these flags are cleared only when the status register is read. If the flag is set from a previous conversion, and at the end of the next conversion, the corresponding limit is not violating anymore, the flag is not cleared when in INT mode. In COMP mode, these flags are set whenever the corresponding limit is violated at the end of a conversion, and cleared if they are not.

7.5.6 Status Mask and Enable Register (address = 05h) [reset = 0000h]

7.5.7 T_OBJ Object Temperature High-Limit Register (address = 06h) [reset = 7FC0h]

7.5.8 T_OBJ Object Temperature Low-Limit Register (address = 07h) [reset = 8000h]

7.5.9 T_DIE Local Temperature High-Limit Register (address = 08h) [reset = 7FC0h]

7.5.10 T_DIE Local Temperature Low-Limit Register (address = 09h) [reset = 8000h]

7.5.11 Coefficient Registers

The values of the coefficient registers described above are used in the math engine. The range and resolution of the coefficients are shown in Table 9. The default coefficients, TC0 and TC1, are optimized for the default conversion mode (four averages per measurement). Different acquisition modes may require different values for the TC0 and TC1 coefficients.

7.5.12 Manufacturer ID Register (address = 1Eh) [reset = 5449h]

7.5.13 Device ID Register (address = 1Fh) [reset = 0078h]

7.5.14 Memory Access Register (address = 2Ah) [reset = 0000h]

The internal memory can be accessed through the memory access register. When the register is read, it returns the values in read name. When the register is written to, it must contain the value 6Axxh to enable the contents of the register specified by Adr4 to Adr0 to be stored in memory.