JP2011086742A - Semiconductor device - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To flexibly adapt to design change etc., by improving precision of correction on a temperature measured by a temperature sensor mounted on a semiconductor device. <P>SOLUTION: The semiconductor device (1) generates a correction table (33) holding computed temperature data corresponding to each temperature by computing the temperature data for every temperature using an inverse function of a characteristic correction function generated by correcting a characteristic function, showing ideal correspondence relation between temperature data measured by a temperature sensor portion (31) and temperatures, with a correction coefficient showing an error between the temperatures measured by the temperature sensor portion and actual temperatures. When temperature measurement is performed, the temperature data measured by the temperature sensor portion is retrieved from the correction table, and data (34) of a temperature made to correspond to the retrieved temperature data is output. <P>COPYRIGHT: (C)2011,JPO&INPIT

Description

  The present invention relates to a technique for correcting a measured temperature in a semiconductor device having a built-in temperature sensor, for example, a technique effective when applied to a microcomputer that uses measured temperature data.

  In recent years, microcomputers for automobile engine control have been increasingly required to operate at higher temperatures. In accordance with this, there are many demands for mounting a temperature sensor on a microcomputer. Since the engine of an automobile becomes very hot, it is necessary to make temperature detection on the high temperature side highly accurate especially for the temperature sensor to be mounted.

  As a method of temperature detection by an on-chip temperature sensor mounted on a conventional microcomputer, for example, a voltage having a positive temperature characteristic or a negative temperature characteristic and a reference voltage having no temperature dependency generated by any method are used. There is a method of detecting the temperature by comparing these voltages by an analog or digital method in the temperature sensor. As the voltage having the positive temperature characteristic or the negative temperature characteristic, for example, the following voltages are used.

  As a voltage having a negative temperature characteristic, that is, a voltage that decreases with increasing temperature, a parasitic diode or a parasitic bipolar transistor between wells formed on silicon by a CMOS process is used as a diode. The voltage across the diode generated by the current flowing in is used. In addition, a voltage having a positive temperature characteristic, that is, a voltage that increases in proportion to an increase in temperature, is the value of the current that flows into the voltage at both ends generated by flowing a current into the diode as in the above method. The difference voltage obtained by taking the difference between the respective voltages when changing is used. Another example of a voltage having a positive temperature characteristic is a base that appears when two parasitic transistors having different emitter areas are used and the same current flows into each of the two bipolar transistors.・ In some cases, the voltage difference between the emitters is taken and the difference voltage is used.

  These voltages are relatively linear with respect to temperature. However, actually, it has a second-order or higher-order nonlinear component, and this is a factor that degrades the accuracy of temperature detection. In addition, devices (MOS transistors, resistors, capacitors, bipolar transistors, diodes, etc.) formed on silicon by the CMOS process have variations in absolute values and relative values of element characteristics due to process variations and power supply voltage variations. However, these are also factors that degrade the accuracy of temperature detection. Therefore, in order to reduce deterioration in temperature detection accuracy due to process variations or the like, a technique such as trimming that corrects variations or the like for each chip is often used. Also, there is a method of improving the detection accuracy by performing a calculation for correcting the temperature by a hardware or software method at the time of temperature detection. For example, Patent Documents 1 to 3 listed below disclose conventional methods related to the calculation for correcting the temperature.

  The semiconductor device having the temperature sensor circuit of Patent Document 1 further includes an actual temperature measurement circuit, and the temperature measured by the actual temperature measurement circuit is regarded as a true temperature, and the correspondence between the temperature and the output value of the temperature sensor circuit Based on the above, the temperature measured by the temperature sensor circuit is corrected and output. From the ideal calculation formula showing the correspondence between the output value of the temperature sensor circuit and the temperature obtained by experience in advance, the output value of the theoretical temperature sensor circuit at a predetermined temperature is obtained, and the output value and the actual value at that temperature are calculated. A difference from the output value of the temperature sensor circuit is obtained, and the difference is used as correction data. In temperature measurement, a value obtained by adding / subtracting the correction data to / from the output value of the temperature sensor circuit is output as a measured value.

  The semiconductor device having the temperature sensor circuit of Patent Document 2 corrects the measured temperature by the temperature sensor circuit using a correction function representing the relationship between the voltage across the diode used as the temperature sensor circuit and the temperature. The temperature is measured in advance by a temperature measuring device installed separately from the semiconductor device, and the measured temperature is set as a true temperature, and the voltage across the diode at that time is measured. The voltage value is measured at two temperatures, and a linear function of the voltage value and temperature is derived from the measured data and used as a correction function. In the temperature measurement, the voltage value of the temperature sensor circuit is converted into a temperature using the correction function, and the temperature is output as a measured value.

  The semiconductor device having the temperature sensor circuit of Patent Document 3 creates correction data corresponding to the output value of the temperature sensor circuit in advance outside the chip, and outputs the correction data addressed with the output value as a measured temperature in actual operation. To do. That is, the temperature is measured by a temperature measuring device installed separately from the semiconductor device, and the power supply voltage and output value data of the temperature sensor circuit at that time are acquired in advance as the true temperature. The correction data is created based on the acquired data. At the time of temperature measurement, correction data addressed by the output value of the temperature sensor circuit is output as a measured temperature. Note that the accuracy of temperature correction is determined by the number of acquired data corresponding to the set number of temperatures and power supply voltages set when generating correction data.

JP 2004-134472 A JP 2001-298160 A JP-A-4-225250

  However, in the conventional method described above, the inventors consider the following problems.

  First, in the temperature correction method described in Patent Document 1, the temperature measured by the actual temperature measurement circuit on the same silicon as the temperature sensor circuit is set as a true temperature. The measured temperature lacks reliability as a true temperature. Further, although a method of adding / subtracting correction data to / from the output value of the temperature sensor circuit is adopted as a temperature correction method, sufficient accuracy with respect to temperature correction cannot be obtained by calculation only by addition / subtraction.

  Secondly, in the temperature correction method described in Patent Document 2, the relationship between the voltage output by the temperature sensor and the temperature for correction is expressed by a linear function. Since there is a non-linear component, sufficient accuracy cannot be obtained with respect to temperature correction.

  Third, in the temperature correction method described in Patent Document 3, in order to increase the accuracy of temperature correction, a plurality of temperatures and power supply voltages are set using a tester without performing a calculation using a function when generating correction data. Since it is necessary to set a pattern and acquire data individually for each LSI, it is not realistic.

  In order to solve these problems, a method of correcting the temperature by a secondary or higher-order correction function or a calculation by multiplication or the like can be considered. However, when the method is mounted on a semiconductor device, hardware such as a multiplier is used. Need to be added, increasing the amount of hardware implementation. In addition, these hardware needs to be redesigned when changing the calculation method of the correction function, etc., changing the accuracy of the detected temperature according to the customer's request, and changing the process, etc. Can not do it.

  An object of the present invention is to provide a semiconductor device that can improve the accuracy of correction of a measured temperature by a mounted temperature sensor and can flexibly cope with a design change or the like.

  The above and other objects and novel features of the present invention will be apparent from the description of this specification and the accompanying drawings.

  The following is a brief description of an outline of typical inventions disclosed in the present application.

  That is, the semiconductor device corrects the characteristic function indicating the ideal correspondence between the temperature data measured by the temperature sensor unit and the temperature with a correction coefficient indicating an error between the temperature measured by the temperature sensor unit and the actual temperature. Using the inverse function of the characteristic correction function, the temperature data is calculated for each temperature, and a correction table for holding the calculated temperature data corresponding to each temperature is generated. At the time of temperature measurement, the temperature data measured by the temperature sensor unit is searched from the correction table, and temperature data corresponding to the searched temperature data is output.

  The effects obtained by the representative ones of the inventions disclosed in the present application will be briefly described as follows.

  That is, it is possible to improve the accuracy of correction of the measured temperature by the temperature sensor and flexibly cope with a design change.

FIG. 1 is a block diagram of a semiconductor device 1 according to the present invention. FIG. 2 is an example of a temperature-voltage conversion circuit using a bandgap reference voltage. FIG. 3 shows temperature characteristics of the reference voltages VPTAT 312 and VREF 313. FIG. 4 is a graph showing the relationship between the voltage ratio VREF / VPTAT and the AD conversion result t. FIG. 5 is a graph showing an ideal curve 200 showing the correspondence between the AD conversion result t and the chip temperature T. FIG. 6 is a graph showing the ideal curve 200 and the characteristic function 201. FIG. 7 is an explanatory diagram showing the influence of process variation on the function of the chip temperature τ and the AD conversion result t. FIG. 8 is an explanatory diagram showing the variation range of the characteristic function 201. FIG. 9 is an example of a graph representing the characteristic correction function. FIG. 10 is a schematic diagram of the correction table. FIG. 11 is a flowchart when creating a correction table. FIG. 12 is an example of a circuit configuration at the time of a product pre-shipment test for the semiconductor device 1. FIG. 13 is a flowchart showing a flow from acquisition of the correction coefficient to temperature detection for the semiconductor device 1. FIG. 14 is a block diagram of the semiconductor device 2 according to the present invention. FIG. 15 is an example of a circuit configuration at the time of a product pre-shipment test for the semiconductor device 2. FIG. 16 is a flowchart when the chip temperature is measured by the thermal diode 60. FIG. 17 is a flowchart when calculating the correction coefficient using the temperature measured by the thermal diode 60. FIG. 18 is an explanatory diagram showing diode connection of a bipolar transistor. FIG. 19 is an example of a layout diagram in the chip of the parasitic bipolar transistors TB_NPN_1 and TB_NPN_8 and the thermal diode 60 in the temperature-voltage conversion circuit 310. FIG. 20 is a block diagram of the semiconductor device 3 according to the present invention. FIG. 21 is an example of the layout of the resistors 70 and the resistors in the temperature-voltage conversion circuit 310. FIG. 22 is an example of a circuit configuration at the time of a product pre-shipment test for the semiconductor device 3. Figure 23 is an example of a correspondence between the possible voltage range and temperature conversion constant _a n ~a ₀ voltage VR10. FIG. 24 is a flowchart when the temperature conversion constant is selected. FIG. 25 is a flowchart when calculating the correction coefficient in the semiconductor device 3. FIG. 26 is a block diagram of the semiconductor device 4 according to the present invention. FIG. 27 is an explanatory diagram showing a connection relationship between the sensor unit 39 and the diodes 80_1 to 80_4.

1. First, an outline of a typical embodiment of the invention disclosed in the present application will be described. Reference numerals in the drawings referred to in parentheses in the outline description of the representative embodiments merely exemplify what are included in the concept of the components to which the reference numerals are attached.

  [1] A semiconductor device (1 to 4) according to a typical embodiment of the present invention includes a temperature sensor unit (31, 39), a temperature correction control unit (32), and a data processing unit (13) on a semiconductor substrate. And a memory unit (20), wherein the memory unit has a characteristic function indicating an ideal correspondence between temperature data measured by the temperature sensor unit and temperature, and is measured by the temperature sensor unit. A storage area for storing a correction coefficient indicating an error between the measured temperature and the actual temperature, and the data processing unit uses an inverse function of a characteristic correction function obtained by correcting the characteristic function with a correction coefficient, The temperature data is calculated for each temperature, a correction table (33) for holding the calculated temperature data corresponding to each temperature is generated, and the temperature correction controller reads the temperature data read from the correction table as the temperature data. In the temperature sensor When matching the boss was temperature data, and outputs the data of the temperature corresponding to the temperature data. According to this, if only the correction coefficient is calculated and stored in advance, the characteristic correction function for each LSI can be generated, and by using the inverse function of the characteristic correction function, the semiconductor device itself, The correction table can be easily created. In addition, since the correction using the correction table is performed, the measurement temperature can be corrected with higher accuracy.

  [2] In the semiconductor device of [1], the temperature correction control unit reads temperature data addressed by candidate data from the correction table, and the temperature data matches the temperature data measured by the temperature sensor unit. The candidate data is updated until addressing is repeated, and if they match, the temperature data (34) corresponding to the candidate data at that time is output.

  [3] In the semiconductor device of [1] or [2], the correction coefficient indicates an error between a temperature measured by the temperature sensor unit and an actual temperature in a predetermined temperature range. According to this, it is possible to obtain a characteristic correction function specialized in the temperature range for which temperature correction is desired with high accuracy.

  [4] In the semiconductor device of any one of [1] to [3], the data processing unit creates the correction table in an initialization process instructed by a reset signal or at the first temperature measurement. According to this, it is possible to reduce the burden of the data processing unit creating the correction table after the initialization process using the reset signal.

  [5] In the semiconductor device according to any one of [1] to [4], the temperature correction control unit outputs an interrupt signal based on the temperature data. According to this, even if the data processing unit or the like performs a thermal runaway, the data processing unit or the like can be initialized.

  [6] The semiconductor device according to any one of items 1 to 5, further including a transfer control unit (40) that performs data transfer control in response to a data transfer request from the data processing unit and the temperature correction control unit. The temperature correction control unit temporarily stores the temperature data and gives a data transfer request (12) to the transfer control unit at a predetermined timing. The transfer control unit Control is performed to transfer the temperature data stored in the temperature correction control unit to the memory unit. According to this, the history of measured temperature can be recorded.

  [7] The semiconductor device according to any one of [1] to [6], further including a diode (60) for measuring temperature, wherein the actual temperature is a voltage when a current is supplied from the outside to the diode. And the temperature calculated using the difference between the current and the voltage when n times the current (n is greater than 1) is supplied. According to this, since the actual temperature is not the ambient temperature of the semiconductor device but the temperature inside the semiconductor device, the measurement accuracy is further increased.

  [8] The semiconductor device as described in any one of [1] to [7], further including a resistor (70) for measuring a variation in resistance value, wherein a predetermined current is supplied to the resistor at a predetermined temperature. The characteristic function corresponding to the resistance variation width calculated from the voltage values at both ends of the resistor is stored in the memory unit. According to this, since the correction table is created based on the characteristic function corresponding to the variation in resistance, temperature correction with higher accuracy is possible.

  [9] In the semiconductor device according to any one of items 1 to 8, the semiconductor device further includes a plurality of diodes (80_1 to 80_4) for measuring temperatures at a plurality of locations in the semiconductor device, and the temperature correction control unit includes: For the temperature data of each location measured by the diode, calculate relative temperature data for the measured temperature data measured by the temperature sensor unit, and based on the relative temperature data, the temperature of each location measured by the diode Output data. According to this, it becomes possible to measure the temperature at a plurality of locations in the chip without incorporating a plurality of temperature sensor sections, and it is possible to know the temperature gradient and temperature distribution.

  [10] A semiconductor device (1 to 4) according to a typical embodiment of the present invention is a semiconductor device including a temperature sensor (31, 39) and a data processing unit (13) on a semiconductor substrate. The data processing unit includes a characteristic function indicating an ideal correspondence between the temperature data measured by the temperature sensor unit and the temperature, and a correction indicating an error between the temperature measured by the temperature sensor unit and the actual temperature. A correction is performed by inputting a coefficient, calculating the temperature data for each temperature using an inverse function of the characteristic correction function obtained by correcting the characteristic function with a correction coefficient, and holding the calculated temperature data corresponding to each temperature. A table (33) is generated, and the correction table enables output of temperature data before correction searched using candidate data as an index. According to this, there exists an effect | action similar to claim | item 1.

2. Details of Embodiments Embodiments will be further described in detail.

<< Outline of microcomputer according to the present invention >>
FIG. 1 is a detailed explanatory diagram of a microcomputer for carrying out automobile data processing and the like, incorporating a temperature sensor, as an embodiment of a semiconductor device according to the present invention. The semiconductor device 1 in FIG. 1 is not particularly limited, but is formed on a single semiconductor substrate such as single crystal silicon by a known CMOS integrated circuit manufacturing technique.

  The semiconductor device 1 shown in FIG. 1 includes a CPU 13, a temperature sensor circuit 30, and other circuits to be described later. Based on a correction table 33, the temperature measured in the internal temperature sensor circuit 30 at the time of temperature detection. Correct and output. Although the details will be described later, the correction table 33 stores temperature data before correction in a searchable manner using temperature data after correction as an index. That is, when temperature detection is performed in the semiconductor device 1, the temperature sensor circuit 30 performs temperature data before correction addressed by the corrected temperature data as candidates from the correction table 33 created in advance by the CPU 13. Is read. The temperature sensor circuit 30 updates the candidate corrected temperature data until the temperature data before correction and the temperature data measured by the sensor unit 31 in the temperature sensor circuit 30 match, and repeats addressing. If they match, the measured temperature is corrected by outputting temperature data corresponding to the corrected temperature data of the candidate at that time.

<< Configuration of Microcomputer According to the Present Invention >>
In addition to the CPU 13 and the temperature sensor circuit 30 described above, the semiconductor device 1 includes an interrupt controller unit 14, a transfer control unit 40, a memory unit 20, and a bus 50, and further includes other peripheral circuits not shown.

  The CPU 13 performs processing of data used for engine control and the like of the automobile. Further, the CPU 13 creates the correction table 33 as described above and stores it in the table storage memory 38 in the temperature sensor circuit 30.

  The memory unit 20 includes a nonvolatile memory (hereinafter referred to as “NVM (Non Volatile Memory)”) 22 and a volatile memory (hereinafter referred to as RAM (Random Access Memory)) 21.

  The NVM 22 stores a program for instructing the CPU 13 to perform data processing and the like. Further, a temperature conversion constant and a correction coefficient of a characteristic function, which are parameters necessary for creating the correction table 33 in temperature detection, are also written. The NVM 22 is, for example, a flash memory.

  The characteristic function is an approximate function representing an ideal correspondence between temperature data measured by the sensor unit 31 in the temperature sensor circuit 30 and temperature, and the temperature conversion constant is a parameter of the approximate function. One. The correction coefficient is a parameter representing an error between the temperature measured by the sensor unit 31 and the actual temperature. Details of these parameters will be described later.

  The RAM 21 is used for temporary storage of data in data processing or the like. The RAM 21 is, for example, an SDRAM.

  The interrupt controller 14 interrupts the CPU 13 in response to an interrupt request from another block.

  The transfer control unit 40 sets transfer conditions and the like by the CPU 13, and transfers data to the memory unit 20 in accordance with transfer requests from the CPU 13, the temperature sensor circuit 30, or other peripheral circuits not shown. Do. The transfer control unit 40 is, for example, a DMAC (Direct Memory Access Controller).

  The temperature sensor circuit 30 includes a sensor unit 31, a temperature correction control unit 32, a bus interface unit 37, and a table storage memory 38.

  The temperature sensor circuit 30 uses the correction table 33 stored in the table storage memory 38 by the temperature correction control unit 32 based on the temperature measured by the sensor unit 31 in response to a temperature detection request from the CPU 13 or the like. It is corrected and output via the bus interface unit 37.

  The sensor unit 31 includes a temperature / voltage conversion circuit 310 and an AD conversion circuit 311.

  The temperature voltage conversion circuit 310 outputs two reference voltages having temperature dependency.

  The temperature-voltage conversion circuit 310 is configured by, for example, a circuit shown in FIG.

  FIG. 2 shows an example of the temperature-voltage conversion circuit 310 using the bandgap reference voltage.

  The temperature voltage conversion circuit 310 shown in FIG. 2 is configured using bipolar transistors, and outputs a reference voltage VPTAT 312 having a positive temperature dependence on temperature and a reference voltage VREF 313 having a low temperature dependence. The control signal 314 from the temperature correction control unit 32 performs control such as switching of enable / disable of the temperature voltage conversion circuit 310, adjustment of the temperature dependence of the reference voltages 312 and 313, and the like.

  As the bipolar transistor, a parasitic bipolar transistor in a CMOS process is mainly used. In FIG. 2, a vertical NPN type parasitic bipolar transistor is used.

  The operation principle of the temperature-voltage conversion circuit 310 shown in FIG. 2 will be described.

  In FIG. 2, the output voltage VREF 313 of the OP amplifier is applied to the base (B1) of one parasitic bipolar transistor (TB_NPN_1) and the base (Bn) of n parasitic bipolar transistors (TB_NPN_n) (n is an integer of 1 or more). As a result, a difference occurs in the current density of the current flowing through the emitter (E1) of the parasitic bipolar transistor (TB_NPN_1) and the emitter (En) of (TB_NPN_n). Due to this difference in current density, the difference voltage (ΔVBE_n) between the base-emitter voltage (VBE_1) of the parasitic bipolar transistor (TB_NPN_1) and the base-emitter voltage (VBE_n) of the parasitic bipolar transistor (TB_NPN_n) is generated across the resistor R3. And the difference voltage has a positive temperature characteristic. Using the difference voltage ΔVBE_n, a reference voltage VREF 313 and a reference voltage VPTAT 312 having a positive temperature characteristic are generated, and these reference voltages are expressed by Expression (1) and Expression (2). Here, k is Boltzmann's constant, q is the charge amount of electrons, and T is temperature. Note that the value of VREF 313 shown in (Equation 1) is about 1.2V.

  These functions are represented in a graph as shown in FIG. FIG. 3 shows the temperature characteristics of the reference voltages VPTAT 312 and VREF 313.

  The AD conversion circuit 311 receives analog signal reference voltages VPTAT 312 and VREF 313 output from the temperature voltage conversion circuit 310, and the ratio of the reference voltages VPTAT 312 and VREF 313 at a predetermined temperature is j bits (j is an integer of 1 or more). ) And an AD conversion result 316 is output. This AD conversion result 316 becomes temperature measurement data by the sensor unit 31. Specifically, in FIG. 3, when the temperatures τ1, τ2, τ3, and τ4 are taken at equal intervals, and the values of the reference voltages VPTAT 312 and VREF 313 for each temperature are V21 to V24 and V11 to V14, for example, at the temperature τ1. Voltage ratio V11 / V21 is converted into a digital value, and voltage ratio V12 / V22 is converted into a digital value at temperature τ2.

  FIG. 4 shows the relationship between the voltage ratio VREF / VPTAT and the AD conversion result t, that is, the input / output relationship of the AD conversion circuit 311. In FIG. 4, the voltage ratio VREF / VPTAT is non-linear with respect to the AD conversion result t. This is because the reference voltage VPTAT 312 has a linear characteristic with respect to temperature, whereas the reference voltage VREF 313 has a non-linear characteristic with respect to temperature.

  The AD conversion circuit 311 controls the AD conversion circuit 311 such as enabling / disabling and adjusting the error due to the offset of the AD conversion circuit 311 according to the control signal 315 from the temperature correction control unit 32. Etc. are performed.

  The table storage memory 38 is a storage device such as an SDRAM or a register, for example, and stores the correction table 33 created by the CPU 13 as described above.

  The temperature correction control unit 32 corrects the AD conversion result 316 based on the correction table 33 and outputs the result.

  The temperature correction control unit 32 includes a control unit 329, a comparator 324, an AD conversion result register 320, a comparison source register 321, a temperature register 326, and a selection circuit 327.

  The AD conversion result register 320 stores the AD conversion result 316 output from the AD conversion circuit 311.

  The temperature register 326 stores candidate temperature data as the corrected temperature of the AD conversion result 316.

  The selection unit 327 addresses the correction table 33 using the corrected temperature data corresponding to the temperature data stored in the temperature register 326 as an index, and the temperature data before correction corresponding to the corrected temperature data. Is stored in the comparison source register 321.

  The control unit 329 includes a control logic circuit 330, a register 332 for temporarily storing a correction coefficient and a temperature conversion constant, which will be described later, and a register buffer 331 for storing corrected temperature data corresponding to the measured temperature. Have. The control logic circuit 330 controls the overall operation related to the correction of the measured temperature.

≪Specific method of temperature correction in temperature detection≫
Here, the correction of the measured temperature, that is, the correction of the AD conversion result 316 by the control logic circuit 330 will be described in detail.

  First, after the semiconductor device 1 is activated, for example, the CPU 13 creates the correction table 33 in the initialization process instructed by a reset signal and stores it in the table storage memory 38. Then, the CPU 13 sets an initial value in the temperature register 326. As the initial value, any corrected temperature data stored in the correction table 33 is set. For example, if the correction table 33 stores temperature data after correction from a temperature of −40 ° C. to 160 ° C. for each step of 4 ° C., the temperature data corresponding to the temperature of −40 ° C. is the initial value. Set as a value.

  Next, when the CPU 13 requests the temperature sensor circuit 30 to detect the temperature, the control logic circuit 330 causes the temperature data after correction stored in the temperature register 326 to be input to the selection unit 327. The selection unit 327 addresses the correction table 33 stored in the RAM 21 using the received corrected temperature data as an index, selects the temperature data 317 before correction corresponding to the index, and selects the comparison source. Store in the register 321. At this time, the AD conversion result register 320 stores the AD conversion result 316 output from the AD conversion circuit 311. Then, the comparator 324 compares the AD conversion result 316 stored in the AD conversion result register 320 with the temperature data 317 before correction stored in the comparison source register 321, and whether or not they match. A comparison result 325 indicating whether or not is output to the control logic circuit 330. If the comparison result 325 indicates a mismatch, the control logic circuit 330 increments the candidate temperature data set in the temperature register 326 to newly add the candidate temperature data. The corrected temperature data corresponding to the temperature data is set and given to the selection unit 327 as a new index. For example, the control logic circuit 330 increments temperature data representing −40 ° C. set as an initial value in the temperature register 326, and newly adds temperature data representing −36 ° C. to the temperature register 326. Set. The selection unit 327 addresses the correction table 33 using the corrected temperature data corresponding to the temperature data of −36 ° C. as an index, and compares the temperature data before correction corresponding to the corrected temperature data with the comparison. The original register 321 is set. When the value of the comparison source register 321 is set, the comparator 324 performs the comparison operation again and outputs the comparison result 325 to the control logic circuit 330. Here, when the comparison result 325 indicates a mismatch, the series of operations from the update of the value of the temperature register 326 to the comparison operation are repeated until the comparison result 325 indicating a match is output. Done. When the comparison result 325 indicating the coincidence is output from the comparator 324, the control logic circuit 330 stops updating candidate temperature data set in the temperature register 326, and the temperature register 326 The currently set temperature data 34 is stored in the register buffer 331 and output to the CPU 13 via the bus interface unit 37.

  By the above method, the result measured by the sensor unit 31 is corrected and output.

  The control logic circuit 330 performs not only the above-described temperature correction but also an interrupt process based on the temperature data value stored in the register buffer 331. For example, when the temperature data represents a temperature exceeding a predetermined temperature, the temperature sensor circuit 30 outputs an interrupt request 11 to the interrupt controller 14, and the interrupt controller 14 receives an interrupt signal. By outputting 15, the CPU 13 is initialized and the operating frequency is adjusted. Further, the control logic circuit 330 gives a data transfer request 12 to the transfer control unit 40 at a predetermined timing, and transfers the temperature data stored in the register buffer 331 to the memory unit 20. A history of the measured temperature is recorded, and control for saving the data stored in the RAM 21 to the NVM 22 is also performed.

«How to create a correction table»
Next, a method for creating the correction table 33 will be described in detail.

  In the semiconductor device 1 according to the present embodiment, the creation of the correction table 33 requires a characteristic correction function based on two parameters, the temperature conversion constant of the characteristic function and the correction coefficient. First, the parameters and the characteristic correction function will be described.

  As described above, the temperature characteristics of the two reference voltages in the temperature-voltage conversion circuit 310 are shown in FIG. 3, and the relationship between the ratio value of the reference voltages and the AD conversion result is expressed as shown in FIG. . From these relationships, the relationship between the AD conversion result t and the chip temperature T is expressed as shown in FIG.

  FIG. 5 shows an ideal curve 200 showing the correspondence between the AD conversion result t and the chip temperature T. Note that the chip temperatures T1 to T4 in FIG. 5 are values corresponding to the chip temperatures τ1 to τ4 in FIG.

  FIG. 6 shows the ideal curve 200 and the characteristic function 201. The characteristic function 201 is a function that mathematically approximates the ideal curve 200 with a quadratic or higher function.

  The characteristic function 201 is expressed by (Equation 3).

Here, T ′ represents the temperature, and t represents the AD conversion result. Further, a _{n to} a ₀ are set as the temperature conversion constant, and are set to values common to all chips. The temperature conversion constants a _{n to} a ₀ are previously written in the NVM 22 of the memory unit 20.

  As shown in FIG. 6, if the characteristic function 201 is a quadratic or higher approximation function, the error between the temperature T′1 to T′4 of the characteristic function 201 and the temperature T1 to T4 in the ideal curve is reduced. Is possible. At the time of temperature detection, if the AD conversion result 316 is converted into temperature based on the characteristic function 201, the measured temperature can be known with high accuracy.

  Next, the variation regarding the characteristic function will be described.

  As described above, the AD conversion result can be converted into temperature by using the characteristic function of (Equation 3). However, the ideal curve 200, which is the basis of the characteristic function, ignores process variations during chip manufacturing, and is actually affected by process variations.

  FIG. 7 shows the influence of process variation on the function of the chip temperature τ and the AD conversion result t. As shown in FIG. 7, due to the influence of process variations, the ideal curve 202 between the AD conversion result t and the chip temperature τ has a characteristic that varies within the range indicated by reference numeral 203. This is because the values of the reference voltages VPTAT 312 and VREF 313 vary due to the influence of process variations, and the AD conversion result by the AD conversion circuit 311 varies. Therefore, for example, in FIG. 7, the AD conversion result t at the chip temperature τ1 varies in the range of (t1−δt1) to (t1 + δt1) with t1 as the center value. Based on this result, FIG. 8 shows a variation range of the characteristic function 201 with the temperature T ′ on the vertical axis and the AD conversion result t on the horizontal axis. In FIG. 8, the characteristic function 201 has a characteristic that varies within a range indicated by reference numeral 204. For example, the AD conversion result t output from the sensor unit 31 at the temperature τ1 takes any value within the range from (t1−δt1) to (t1 + δt1) with t1 as the central value. When these values are converted into temperature, T′1 is the central value, and the value is in the range of (T′1−δT′1) to (T′1 + δT′1), and the temperature measured by the sensor unit 31 is in this range. The value varies. Therefore, in order to reduce the variation in the measured temperature, the semiconductor device 1 uses a function corresponding to the process variation of each chip. Specifically, the characteristic function is corrected by the correction coefficient calculated for each chip, and the corrected function is used. This corrected function is a characteristic correction function.

  As described above, the correction coefficient is a parameter representing an error between the temperature measured by the sensor unit 31 and the actual temperature. For example, the correction coefficient is a temperature difference between the measured temperature and the actual temperature at a predetermined temperature. is there. The correction coefficient is calculated for each chip by a method to be described later at the time of a product shipment test, and is written in the NVM 22 of the memory unit 20 together with the temperature conversion constant.

  The characteristic correction function is obtained by correcting an ideal characteristic function expressed by (Equation 3) using the correction coefficient for each chip. Specifically, the characteristic correction function is expressed by (Equation 4), with temperature data Y after correction, temperature data before correction X, and correction coefficient A.

  The AD conversion result t represents the temperature data X before correction.

  FIG. 9 is an example of the graph of the characteristic correction function, and shows the characteristic correction function 205 and the characteristic correction function 206 of two chips that output different AD conversion results t under the same temperature condition.

  At the time of product shipment test, the chip whose output of the sensor unit 31 at the temperature T2 is (t2-δt2) uses the characteristic correction function 205 based on the correction coefficient at the temperature T2, and performs AD conversion of the sensor unit 31 The measured temperature is output after correcting the result t. Further, the chip whose output of the sensor unit 31 at the temperature T2 at the time of the test is (t2 + δt2) uses the characteristic correction function 206 based on the correction coefficient at the temperature T2, and the AD conversion result t of the sensor unit 31 Is corrected and output. Thus, if the AD conversion result is converted into temperature by the characteristic correction function for each chip, as shown in FIG. 9, even if the AD conversion results output from the sensor unit 31 of each chip differ under the same temperature condition. In addition, variation in measurement temperature as a chip can be suppressed, and measurement accuracy can be increased. In FIG. 9, the correction coefficient calculated at the temperature T2 is used. However, when the measurement accuracy near the temperature T1 is particularly desired to be increased, the correction coefficient at the temperature T1 is calculated during the test. The characteristic correction function based on the correction coefficient may be used.

  As described above, if the characteristic correction function is used for each chip, the temperature measurement accuracy can be improved. However, each time the temperature is measured, the CPU 13 performs a correction calculation of the measured temperature using the characteristic correction function. As a result, the load on the CPU 13 increases. Further, when the CPU 13 runs out of heat, the correction calculation cannot be performed and interrupt processing or the like cannot be performed. Therefore, the semiconductor device 1 according to the present invention does not perform the temperature correction calculation every time the temperature is measured, but creates the correction table 33 based on the characteristic correction function in advance, Using the correction table 33, the measured temperature is corrected and output.

  FIG. 10 shows a schematic diagram of the correction table 33. The correction table 33 represents the correspondence between the temperature data X before correction and the temperature data Y after correction as described above, and uses the inverse function of the characteristic correction function expressed by (Equation 4). Thus, the temperature data Y after correction is generated by calculation for calculating the temperature data X before correction corresponding to the temperature data Y after correction. In FIG. 10, the corrected temperature Y is expressed as Y = y [i], and the temperature data X before correction is expressed as X = x [i]. However, i = 0 to n. For example, when it is desired to measure the measurement temperature range from −40 ° C. to 160 ° C. in units of 4 ° C., the corrected temperature data Y is set to −40 ° C. to 160 ° C., and the temperature range is corrected corresponding to every 4 ° C. The previous temperature data X is calculated and the correction table 33 is created. If the inverse function is not used, it is difficult to accurately create the correction table 33 for each target step temperature. In addition, if the above two parameters are stored in the memory unit 20 when the correction table 33 is created, it is not necessary to actually measure the temperature.

  A procedure for creating the correction table 33 will be described with reference to FIG.

  FIG. 11 is a flowchart when the correction table 33 is created.

  In FIG. 11, first, in the initialization process after the semiconductor device 1 is activated, the register buffer 331 and the register 332 in the control unit 329 are initialized (S101). Thereafter, the temperature conversion constant and the correction coefficient are read from the NVM 22 and stored in the register 332 (S102). The CPU 13 uses the inverse function of the characteristic correction function based on the temperature conversion constant and the correction coefficient stored in the register 332 to convert the temperature data X before correction corresponding to the temperature data Y after correction to the temperature. Each time it is calculated and stored in the table storage memory 38 or the RAM 21 as the correction table 33 (S104). When the calculation is completed in the entire corrected temperature range required, the creation of the correction table 33 is terminated (S105).

≪How to obtain correction coefficient≫
As described above, the correction coefficient is calculated for each chip at the time of a product shipment test, and written in the NVM 22. A specific method for obtaining the correction coefficient will be described.

  FIG. 12 shows an example of a circuit configuration at the time of a test before product shipment.

  In FIG. 12, the semiconductor device 1 is installed on a stage 401 for installing a measurement target in a tester 400, and the tester 400 measures the temperature measured by the sensor unit 31 of the semiconductor device 1 and the semiconductor The correction coefficient is calculated from the temperature measured by the high-accuracy temperature sensor 402 installed separately from the device 1 and written in the NVM 22 of the semiconductor device 1.

  The semiconductor device 1 is the same as the semiconductor device 1 shown in FIG. 1, and components other than those necessary for explanation are omitted.

  The semiconductor device 1 is connected to the tester 400 via I / O interfaces 407A and 407B, and the measurement result measured by the sensor unit 31 is output from a PAD 408A which is one of the I / O interfaces 407A. The

  The high-accuracy temperature sensor 402 is installed on the stage 401 and measures the temperature of the stage 401.

  The tester 400 applies arbitrary DC or AC voltage and current to the semiconductor device 1 to be measured via the I / O interfaces 407A and 407B, and adjusts the temperature of the stage 401. A test apparatus for examining the characteristics of the semiconductor device 1 under various temperature conditions.

  The tester 400 includes a heat generation / cooling device 406 serving as a heat source, a storage unit 404, a system control unit 403, and a monitor 405.

  The storage unit 404 has a storage area for storing the measurement result of the sensor unit 31 and the measurement result of the high-precision temperature sensor 402, and the correction coefficient calculated by the system control unit 403, the temperature conversion constant, and the like. Also store.

  The system control unit 403 controls the heat / cooling device 406 to adjust the temperature of the stage 401.

  Further, the system control unit 403 controls the calculation of the correction coefficient, the storage of the correction coefficient and the measurement result in the storage unit 404, and the writing of the correction coefficient and the like to the NVM 22. Note that the writing control of the correction coefficient and the like to the NVM 22 can be performed by the semiconductor device 1 using the measurement result of the sensor unit 31, the measurement result of the high-precision temperature sensor 402, and the like. .

  The system control unit 403 calculates the correction coefficient A by taking the temperature measured by the high-accuracy temperature sensor 402 as a true temperature and taking the ratio between the chip temperature measured by the sensor 31 and the true temperature. To do. Moreover, although the formula (4) is shown as a specific example of the characteristic correction function, when the function shown in the following formula (5) is used as another example of the characteristic correction function, the high-precision temperature sensor 402 is used. The correction coefficient B is calculated by taking the difference between the temperature of the chip measured by the sensor 31 and the true temperature as the true temperature.

  The correction coefficient A, the correction coefficient B, or other correction coefficient suitable for the applied characteristic correction function is referred to as a correction coefficient indicating an error between the temperature measured by the temperature sensor unit and the actual temperature.

  The monitor 405 displays a measurement result, a control state of the tester 400, and the like.

  The flow of obtaining the correction coefficient by the apparatus shown in FIG. 12 will be described with reference to FIG.

  FIG. 13 is a flowchart showing a flow from acquisition of the correction coefficient to temperature detection for the semiconductor device 1.

  First, the semiconductor device 1 is set on the stage 401 in the tester 400 (S201). The system control unit 403 controls the heat generation / cooling device 406 to set a desired temperature. The desired temperature is a predetermined temperature in a temperature range in which the measurement accuracy of the temperature sensor circuit 30 is particularly desired to be improved. Then, after a lapse of time during which the high-precision temperature sensor 402 on the stage 401 and the semiconductor device 1 are in a uniform temperature state, the tester 400 displays the measurement result of the temperature measured by the high-precision temperature sensor 402. While storing in the memory | storage part 404 (S202_1), the measurement result of the temperature measured by the sensor part 31 in the said semiconductor device 1 is stored in the said memory | storage part 404 (S202_2). The system control unit 403 calculates the correction coefficient based on the measurement result, writes the correction coefficient in the NVN 22 in the semiconductor device 1 (S203), and ends the test (S204).

≪Operation flow at temperature detection≫
Next, the flow of operation during temperature detection will be described with reference to FIG.

  In FIG. 1, when there is a temperature detection request 36 from the CPU 13 or the like during the operation of the semiconductor device 1, the temperature sensor circuit 30 starts measuring the temperature (S205). At this time, the control unit 329 checks whether or not the correction table 33 has been created and stored in the table storage memory 38 (S206). When the correction table 33 has not been created, the control unit 329 makes a request 35 for creating the correction table to the CPU 13, and the CPU 13 creates the correction table 33 and stores the table storage memory 38. (S100 to S105), and an initial value of candidate temperature data is set in the temperature register 326. In the initialization process at the time of power-on reset of the semiconductor device 1, if the CPU 13 has created the correction table 33, it is not necessary to create the correction table 33 when performing temperature detection.

  Next, the sensor unit 31 performs AD conversion on the ratio of the reference voltage by the AD conversion circuit 311, stores the AD conversion result in the AD conversion result register 320, and the selection unit 327 stores in the temperature register 326. The corrected temperature data corresponding to the initially set temperature data is addressed as an index, and the corresponding temperature data before correction is stored in the comparison source register 321 (S207). The comparison operation is performed by the comparator 324, and the control logic circuit 330 repeatedly updates the temperature register 326 until the AD conversion result 316 and the temperature data 317 before correction match (S208). When the AD conversion result 316 matches the temperature data 317 before correction, the control logic circuit 330 stores the temperature data stored in the temperature register 326 in the register 332 and outputs it to the CPU 13. (S209).

  As described above, according to the semiconductor device 1, if the correction coefficient and the temperature conversion constant are stored at the time of a shipping test, the semiconductor device 1 can easily create the correction table 33 by itself. By performing the correction using the correction table 33, the measurement temperature can be corrected with higher accuracy. In addition, when the correction table 33 is created, the built-in CPU 13 performs an operation using the inverse function of the characteristic correction function expressed by a function of second or higher order, and therefore dedicated hardware such as a multiplier is provided. This can be realized without adding, and it is possible to flexibly cope with design changes and the like. Furthermore, since the correction table 33 is created in the initialization process at the time of startup or at the start of temperature detection, the burden on the CPU 13 can be reduced, and the CPU 13 or the like has run out of heat. In addition, the CPU 13 and the like can be initialized based on the measured temperature corrected by the control unit 329.

≪Built-in thermal diode≫
FIG. 14 is a detailed explanatory diagram of a microcomputer for carrying out data processing and the like for an automobile, incorporating a temperature sensor, as another embodiment of the semiconductor device according to the present invention. The semiconductor device 2 in FIG. 14 is not particularly limited, but is formed on a single semiconductor substrate such as single crystal silicon by a known CMOS integrated circuit manufacturing technique.

  In the semiconductor device 1 described above, the temperature on the stage 401 measured by the high-accuracy temperature sensor 402 installed outside the semiconductor device 1 is set as a true temperature when the correction coefficient is acquired in the pre-shipment test. The semiconductor device 2 shown in FIG. 14 further includes a thermal diode 60 for measuring the temperature, and the correction coefficient is calculated using the temperature inside the chip measured by the thermal diode 60 as a true temperature.

  Regarding the semiconductor device 2, the same components as those of the semiconductor device 1 are denoted by the same reference numerals, and the detailed description thereof is omitted. In FIG. 14, only the components necessary for the description are displayed.

  The thermal diode 60 is disposed in the vicinity of the temperature / voltage conversion circuit 310 in the temperature sensor circuit 41. As the thermal diode 60, a diode-connected parasitic bipolar transistor is used. The base terminal and collector terminal of the thermal diode 60 are short-circuited and connected to the PAD 409A, and the emitter terminal is connected to the PAD 409B.

≪Method of measuring temperature with thermal diode≫
A method for measuring the temperature by the thermal diode 60 will be described.

In the semiconductor device 2, the base-emitter voltage VBE corresponding to each current when the value of the flowing current is changed using the base-emitter voltage VBE generated by flowing a current through the thermal diode 60. The temperature is measured by using the difference voltage. That is, at a temperature Td1, the base-emitter voltage when pouring the current _{I 0} to the thermal diode 60 and VBE0, the base-emitter voltage when pouring the n times the current nI0 current I0 was VBEn Then, the difference voltage ΔVBE between these voltages is expressed by equation (6), and the temperature Td1 is expressed by equation (7). In the semiconductor device 2, the temperature T is calculated using (Equation 7) by measuring the ratio of the current flowing through the thermal diode 60 and the differential voltage ΔVBE.

  A specific method of the measurement will be described with reference to FIG.

  FIG. 15 shows an example of a circuit configuration for performing a pre-product shipment test on the semiconductor device 2.

  In FIG. 15, the semiconductor device 2 is installed on a stage 401 for installing a measurement target in the tester 500, and the tester 500 measures the temperature measured by the sensor unit 31 of the semiconductor device 2 and the thermal The correction coefficient is calculated from the temperature measured by the diode 60 and written to the NVM 22.

  In the tester 500 shown in FIG. 15, the same components as those of the tester 400 are denoted by the same reference numerals, and detailed description thereof is omitted.

  The tester 500 further includes a current source circuit 501 and a multimeter 502.

  In the current source circuit 501, the switch SW1 and the switch SW2 are switched by the system control unit 504, and the current I0 or the current nI0 is supplied to the base terminal of the thermal diode 60 via the PAD 409A.

  The multimeter 502 measures the current values of the current I0 and the current nI0 flowing from the current source circuit 501, and the potential difference between the PAD 409A and the PAD 409B when the current flows, that is, the base of the thermal diode 60. The emitter voltage VBE is measured.

  The storage unit 503 stores the current I0 and current nI0 of the thermal diode 60 measured by the multimeter 502 and the base-emitter voltages VBE0 and VBEn of the thermal diode at the respective current values.

  The system control unit 504 calculates the temperature using (Equation 7) from the current value and the base-emitter voltage value stored in the storage unit 503 and stores the temperature in the storage unit 503. The other functions are the same as those of the system control unit 403.

  A flow of temperature measurement by the thermal diode 60 will be described with reference to FIG.

  FIG. 16 is a flowchart for measuring the chip temperature by the thermal diode 60.

  First, the system controller 504 in the tester 500 controls the heat generation / cooling device 406 to set a desired temperature (S301). The setting of the desired temperature is the same as that of the tester 400. After a sufficient time for the temperature state of the semiconductor device 1 to stabilize, the system control unit 504 controls the current source circuit 501 to flow the current I0 into the thermal diode 60, and the multimeter 502 The base-emitter voltage VBE0 at that time is measured, and the values of the base-emitter voltage VBE0 and the current I0 are stored in the storage unit 503 (S302). At this time, the tester 500 determines whether or not the base-emitter voltage VBE0 is within the process management range (S303). That is, it is checked whether or not the base-emitter voltage VBE0 of the bipolar transistor in the semiconductor device 2 is within the variation width allowed for the process. If the base-emitter voltage VBE0 is out of the process management range (S304), the semiconductor device 2 is excluded from the sample (S305). If it is within the process management range, the system control unit 504 controls the current source circuit 501 to flow a current nI0 into the thermal diode 60, and the multimeter 502 is connected between the base and emitter at that time. The voltage VBEn is measured, and the base-emitter voltage VBEn and the value of the current nI0 are stored in the storage unit 503 (S306). Next, the system control unit 504 calculates a difference voltage ΔVBE based on the base-emitter voltages VBE0 and VBEn stored in the storage unit 503 (S307), and calculates a temperature Td1 based on the difference voltage ΔVBE. At the same time, the temperature is stored in the storage unit 503 (S308), and the temperature measurement is terminated (S309).

  With the above method, it is possible to accurately measure the temperature of the chip of the semiconductor device 2 by using the thermal diode 60.

≪Calculation method of correction coefficient using temperature measured by thermal diode≫
A method for calculating the correction coefficient using the temperature measured by the thermal diode 60 in the semiconductor device 2 will be described with reference to FIG.

  FIG. 17 is a flowchart when calculating the correction coefficient using the temperature measured by the thermal diode 60.

First, similarly to FIG. 16, the system control unit 504 in the tester 500 controls the heat generation / cooling device 406 to set a desired temperature (S301). And by the above-mentioned method, temperature is measured using the said thermal diode 60 (S302-S309), and temperature Td1 in a chip | tip is obtained (S401). In addition, the system control unit 504 reads the temperature conversion constant of the characteristic function from the storage unit 503 (S402). Here, the case where the characteristic function is a quadratic function is taken as an example, and the temperature conversion constants a _{2 to} a ₀ are read. Next, the AD conversion result t of the temperature measured by the sensor unit 31 in the semiconductor device 2 is output via the PAD 408A and stored in the storage unit 503 (S403). The system control unit 504 calculates the temperature Td0 from the AD conversion result t stored in the storage unit 503 by using the characteristic correction function, writes the temperature Td0 in the storage unit 503 (S404), and obtains the measurement result Td0. (S405). The characteristic correction function at this time sets the correction coefficient A to “1” and uses the temperature conversion constants a _{2 to} a ₀ .

  When the temperature Td1 measured by the thermal diode 60 and the temperature Td0 measured by the sensor unit 31 are obtained, the system control unit 504 compares the values (S406). In the comparison, it is determined whether or not the temperature Td1 and the temperature Td0 are equal (S407). If they are equal, it is determined that no correction is required (S408), the correction coefficient A is determined to be “1” (S412), and the test is performed. Is finished (S413). In step 407, if the comparison results are not equal, it is determined whether or not the temperature Td1 is higher than the temperature Td0 (S409). When the temperature Td1 is higher than the temperature Td0, it is determined that a value larger than “1” is necessary for the correction coefficient A (S410), and the correction coefficient A is determined to be a value larger than “1” (S412). Is finished (S413). In step 409, if the temperature Td1 is lower than the temperature Td0, it is determined that a value smaller than “1” is necessary for the correction coefficient A (S411), and the correction coefficient A is smaller than “1”. (S412), and the test ends (S413).

  In the flow chart of FIG. 17, when the correction coefficient A is determined, the calculation of the temperature Td0 and the comparison between the temperature Td0 and the temperature Td1 are described as an example. When it is desired to improve the accuracy, the following method may be employed. For example, after step 410 or step 411, the value of the correction coefficient A is changed, the value of the temperature Td0 is calculated again, and the operation of comparing the calculated temperature Td0 and the temperature Td1 is performed a plurality of times to perform the correction. A method for determining the value of the coefficient A may be adopted.

«Placement of the thermal diode»
As described above, the thermal diode 60 in the semiconductor device 2 uses a parasitic bipolar transistor.

  FIG. 18 shows diode connection of a bipolar transistor.

  Bipolar transistors generally have two types of device structures, PNP type and NPN type. In a parasitic bipolar transistor formed in a CMOS process, a P-type well and an N-type well are formed vertically in the same direction as the direction of impurity diffusion or implantation in the process of well formation in the CMOS process. There are straight bipolar transistors and lateral bipolar transistors formed perpendicular to the direction of impurity diffusion or implantation. The thermal diode 60 uses the PNP type or NPN type of the above-mentioned parasitic bipolar transistor, and short-circuits the base terminal and collector terminal of the bipolar transistor to form a diode connection, so that the P side is the anode and the N side Is used as the cathode.

  FIG. 19 is an example of a layout diagram of the parasitic bipolar transistors TB_NPN_1 and TB_NPN_n and the thermal diode 60 in the temperature-voltage conversion circuit 310 in FIG. Here, the number n of the parasitic bipolar transistors TB_NPN_n is eight.

  The thermal diode 60 is disposed in the vicinity of the parasitic bipolar transistors TB_NPN_1 and TB_NPN_8 that detect the temperature in the temperature-voltage conversion circuit 310 in order to correct the measurement result by the sensor unit 31 as accurately as possible. When the thermal diode 60 is disposed, the anode side, that is, the collector region 551 needs to be spaced from the parasitic bipolar transistors TB_NPN_1 and TB_NPN_8 of the voltage conversion circuit 310 to separate wells. This is because the parasitic bipolar transistors TB_NPN_1 and TB_NPN_8 in the temperature-voltage conversion circuit 310 and the parasitic bipolar transistor forming the diode 60 are constituted by a vertical NPN type parasitic bipolar transistor, and the collector region of the bipolar transistor is This is because it is an N type well region formed deep in the silicon substrate, so that it is necessary to provide a space for well separation defined by the process, that is, a space indicated by reference numeral 550. The parasitic bipolar transistor TB_NPN_1 is laid out together with dummy parasitic bipolar transistors arranged so as to have the same collector size as that of the parasitic bipolar transistor TB_NPN_8. The base terminals of the elements are connected in common and connected to the ground. As a result, the parasitic capacitance between the N well used as the collector region of the parasitic bipolar transistor and the P-type silicon substrate is reduced, and the AC characteristics of the temperature-voltage conversion circuit 310 can be improved. In addition, a leak current flowing from the N well toward the P-type silicon substrate at a high temperature can be suppressed, and errors in the reference voltages VREF 313 and VPTAT 312 can be reduced.

  As described above, according to the semiconductor device 2, the temperature in the vicinity of the sensor unit 31 can be measured using the thermal diode 60 when the correction coefficient is acquired, and the temperature can be corrected with higher accuracy. It becomes possible.

≪Built-in correction resistor≫
FIG. 20 is a detailed explanatory diagram of a microcomputer for carrying out data processing for automobiles, which incorporates a temperature sensor, as another embodiment of the semiconductor device according to the present invention. The semiconductor device 3 in FIG. 20 is not particularly limited, but is formed on a single semiconductor substrate such as single crystal silicon by a known CMOS integrated circuit manufacturing technique.

As described above, the characteristic function also varies due to the process variation of the chip. The variation in the characteristic function is largely influenced by the variation in resistance used in the sensor unit 31. Therefore, the semiconductor device 3 incorporates a resistor 70 of the same type as the resistor used in the sensor unit 31 and selects an optimum characteristic function by knowing the variation of the resistor 70 to obtain a higher accuracy. the correction of the temperature. That is, in the semiconductor devices 1 and 2 according to the above-described embodiment, the temperature conversion constants a _{n to} a ₀ that are common to all chips are used. The temperature conversion constants a _{n to} a ₀ are used.

  In the semiconductor device 3, the same components as those in the semiconductor device 2 are denoted by the same reference numerals, and detailed description thereof is omitted. In FIG. 20, only components necessary for the description are displayed.

  The semiconductor device 3 further includes a correction resistor 70 with respect to the semiconductor device 2.

  Terminals at both ends of the resistor 70 are connected to the PAD 410A and the PAD 410B, respectively.

  FIG. 21 shows an example of the layout of the resistor 70 and the resistor Rerr 730 used in the temperature-voltage conversion circuit 310.

  The correction resistor 70 is disposed in the vicinity of the resistor Rerr 730 used in the temperature-voltage conversion circuit 310 in the temperature sensor circuit 42. The resistor 70 is the same type of resistor as the resistor 730 and has the same unit resistor shape. The resistor Rarr 730 is disposed on both sides together with a dummy resistor 731 of the same type as that of the resistor and having the same unit resistor shape, and is fixed to a predetermined potential in order to prevent the relative accuracy of the resistor element from deteriorating. It is arranged in a guard ring 732 formed by a well or the like. The same applies to the resistor 70. However, the sum of the unit resistances does not have to be equal to the resistance Rerr730.

≪Method for selecting temperature conversion constant using correction resistor≫
Selection of the temperature conversion constants a _{n to} a ₀ using the resistor 70 will be described.

  FIG. 22 is an example of a circuit configuration at the time of a product pre-shipment test for the semiconductor device 3.

  In the tester 600 shown in FIG. 22, the same components as those of the tester 500 are denoted by the same reference numerals, and detailed description thereof is omitted.

  In FIG. 22, the current source circuit 601 of the tester 600 supplies a current I10 to the resistor 70 via the PAD 410A under the control of the system control unit 604. The system control unit 604 measures the voltage VR10 appearing at both ends of the resistor 70 at that time using the multimeter 602 and stores the voltage VR10 in the storage unit 603.

The storage unit 603 stores the temperature conversion constant a corresponding to the allowable variation width of the resistance 70 based on the process management width of the resistance value per unit resistance, that is, the allowable variation width of the resistance value. _{n to} a ₀ are stored in advance. For example, when the variation width of the resistor 70 is in the range from the minimum value Rmin to the maximum value Rmax, the center value of the resistor 70 is Rtyp, and the voltage when the current I10 is passed through the resistor 70 is VR10. The range that the voltage VR10 can take is from VRmin to VRmax. Note that VRmin = Rmin · I10 and VRmax = Rmax · I10. The possible range of the voltage VR10 is divided into a plurality of voltage ranges, and the temperature conversion constants a _{n to} a ₀ corresponding to the divided voltage ranges are determined in advance. For example, a correspondence relationship as shown in FIG. 23 is stored in the storage unit 603. FIG. 23 shows the temperature conversion constants a _{n to} a ₀ corresponding to each voltage range by dividing the range that the voltage VR10 can take into three voltage ranges.

  The system control unit 604 determines which voltage range the measured voltage VR10 is based on the correspondence relationship between the voltage range stored in the storage unit 603 and the temperature conversion constant, and the voltage range Is selected and written to the NVM 22 of the semiconductor device 3.

  The flow when selecting the temperature conversion constant using the resistor R70 will be described with reference to FIG.

FIG. 24 is a flowchart when the temperature conversion constant is selected using the resistor R70. Here, the case where the characteristic function is a quadratic function, that is, the case where the temperature conversion constants a _{2 to} a _{0 are used} is taken as an example.

First, as described above, the system control unit 604 in the tester 600 controls the heating / cooling device 406 to set a desired temperature (S301), and measures the chip temperature using the thermal diode 60. (S302 to S309). The storage unit 603 stores the correspondence between the voltage range and the temperature conversion constant as shown in FIG. 23 (S501). The system control unit 604 controls the current source circuit 601 to apply the current I10 to the resistor 70, measures the voltage VR10 across the resistor 70 at that time (S502), and stores it in the storage unit 603. Store the value of voltage VR10 (S503). Then, the system control unit 604 determines which voltage range of the voltage range indicated in step 501 is the voltage VR10 (S504). The system control unit 604 initially determines whether the value of the voltage range which the voltage VR10 represented by V _X (S505), if a value in the range shown in the voltage range, the voltage range The temperature conversion constants a _{X2 to} a _X0 corresponding to are selected (S506), and the temperature conversion constant is determined (S513). Further, in step 505, if not a value of the range shown in the voltage range, it is determined whether the value of the voltage range which the voltage VR10 represented by V _Y (S507), indicated in the voltage range If the value is within the range, temperature conversion constants a _{Y2 to} a _Y0 corresponding to the voltage range are selected (S508), and the temperature conversion constant is determined (S513). Further, in step 507, if not a value of the range shown in the voltage range, it is determined whether the value of the voltage range which the voltage VR10 represented by V _Z (S509), indicated in the voltage range If the value is within the range, the temperature conversion constants a _{Z2 to} a _Z0 corresponding to the voltage range are selected (S510), and the temperature conversion constant is determined (S513). In step 511, when the value is not in the range indicated by the voltage range, it is determined that the resistance is out of the process management range (S511), and the sample is excluded (S512).

≪Calculation method of correction coefficient using correction resistor≫
The flow of calculating the correction coefficient in the semiconductor device 3 will be described with reference to FIG.

  FIG. 25 is a flowchart when calculating the correction coefficient in the semiconductor device 3.

  25, first, the system control unit 604 of the tester 600 selects the temperature conversion constant according to the procedure shown in FIG. 24 (S601). The system control unit 604 controls the heat generation / cooling device 406 to set a desired temperature (S301), and reads the selected temperature conversion constant from the storage unit 603 (S402). Subsequent procedures are the same as those in the flowchart of FIG. 17, so that the correction coefficient can be determined.

According to the semiconductor device 3 above, since selecting the temperature conversion constant a ₀ ~a _n in accordance with the process variations of each chip by using the resistor 70 with a built, allows more accurate correction of the measured temperature to become. In the above description, the case where the resistor 70 is built in the semiconductor device 2 is shown as an example, but the resistor 70 may be built in the semiconductor device 1.

≪Use of correction resistor 70 as a heating element≫
A method for performing a high temperature test using the resistor 70 as a heating element will be described.

  When the semiconductor device is placed in a high temperature environment during the pre-shipment test, as described above, there is a method of adjusting the temperature of the stage by a tester to bring it into a high temperature state. However, the semiconductor device 3 has a built-in resistor 70. By flowing a large current, the resistor 70 can generate heat, and the temperature in the vicinity of the sensor unit 31 can be changed to a high temperature state. According to this method, heat can be generated at a position close to the temperature-voltage conversion circuit 310 that detects the temperature of the semiconductor device 3, so that heat can be transferred efficiently.

  When the resistor 70 is used as a heating element, it is necessary to complete the selection of the temperature conversion constant described above. The reason is that, due to the large current that flows when the resistor 70 is used as a heating element, the characteristics of the resistor 70 and the metal wiring between the resistor 70 and the PAD 410A and PAD 410B deteriorate. This is because there is a possibility of disconnection.

≪Measurement of temperature gradient in the chip≫
FIG. 26 is a detailed explanatory diagram of a microcomputer for carrying out data processing and the like for an automobile incorporating a temperature sensor as another embodiment of the semiconductor device according to the present invention. The semiconductor device 4 in FIG. 26 is not particularly limited, but is formed on a single semiconductor substrate such as single crystal silicon by a known CMOS integrated circuit manufacturing technique.

  In order to measure the temperature gradient inside the chip with high accuracy, the semiconductor device 4 has temperature measurement diodes 80_1 to 80_4 that enable temperature measurement relative to the temperature-voltage conversion circuit 310 at a plurality of locations. .

  In the semiconductor device 4, the same components as those in the semiconductor device 1 are denoted by the same reference numerals and detailed description thereof is omitted. In FIG. 26, only the components necessary for the description are displayed. A PLL (Phase-Locked Loop) 90 is displayed as a representative example of the peripheral circuit in the semiconductor device 4.

  The temperature measuring diodes 80 </ b> _ <b> 1 to 80 </ b> _ <b> 4 are arranged in the vicinity of components that tend to generate heat in the chip and components that easily affect the performance of the microcomputer.

  The temperature sensor circuit 43 includes the temperature correction control unit 32, the table storage memory 38, and a sensor unit 39.

  The sensor unit 39 includes the temperature-voltage conversion circuit 310, the AD conversion circuit 350, and the bias circuit 351, and is connected to the diodes 80_1 to 80_4.

  FIG. 27 shows a connection relationship between the sensor unit 39 and the diodes 80_1 to 80_4.

  In FIG. 27, the diodes 80_1 to 80_4 are connected to the switches SW1b to SW4b of the bias circuit 351, respectively.

  The bias circuit 351 is controlled by the control unit 329 in the temperature correction control unit 32 and switches and outputs a current I0 or a current nI0 that is n times the current I0 at a certain timing, and outputs the diodes 80_1 to 80_4. Either is selected and a current is passed.

  The AD converter circuit 350 is an example of a ΔΣ type AD converter circuit that generally has high AD conversion accuracy.

  A method for measuring the temperature gradient using the temperature measuring diodes 80_1 to 80_4 will be described in detail.

  A case where the temperature is measured by the diode 80_1 during the operation of the semiconductor device 4 will be described as an example.

  First, the control unit 329 controls the switch SW1b in FIG. 27 to connect the diode 80_1 and the bias circuit 351. The control unit 329 controls the switch SW1a of the bias circuit 351 to flow the current I0 into the diode 80_1, and controls the switch SW2 with the base-emitter voltage VBE0 of the diode 80_1 generated by the current. To the AD conversion circuit 350. The AD conversion circuit 350 taking in the voltage VBE0 performs AD conversion on the value and outputs the value. The control unit 329 stores the AD converted VBE0 value in the register buffer 331. Next, the control unit 329 turns off the switch SW1a of the bias circuit 351, turns on the switch SW2a, flows a current nI0 into the diode 80_1, and controls the switch SW2, thereby generating the diode generated by the current The AD conversion circuit 350 takes in the base-emitter voltage VBEn of 80_1. Then, the value of VBEn subjected to AD conversion in the same manner as described above is stored in the register buffer 331. The controller 329 calculates a difference voltage ΔVBEtd between the values of the voltages VBEn and VBE0 stored in the register 311. This difference voltage ΔVBEtd is a temperature-dependent voltage and is expressed by the following (formula 8).

  Further, since the VPTAT is expressed by the above-described (Equation 2), the relative temperature with respect to the temperature measured by the temperature-voltage conversion circuit 310 can be measured by comparing the ΔVBEtd with the reference voltage VPTAT. it can. For example, when the temperature measured by the temperature-voltage conversion circuit 310 is T2, and the temperature measured by the diode 80_1 is T1, the relationship between the reference voltage VPTAT and the difference voltage ΔVBEtd is (Equation 2) and (Equation 8). ) And is represented by the following (formula 9).

  Therefore, it is possible to know the relative temperature T1 with respect to the temperature T2 measured by the temperature-voltage conversion circuit 310. By performing this measurement on each of the diodes 80_1 to 80_4 in a time-sharing manner, the temperature gradient in the chip can be obtained. Can be measured.

  (Equation 9) is an ideal equation in the case where there is no error due to variation between the diode 80_1 and the parasitic bipolar transistor in the temperature-voltage conversion circuit 310. Therefore, if there is such an error, the difference between the reference voltage VPTAT 312 and the difference voltage ΔVBEtd is measured under the same temperature condition during the pre-shipment test, and the difference is written into the NVM 22 so that the temperature gradient is obtained during actual operation. When measuring, the relative temperature by the diode 80_1 may be calculated in consideration of the difference. As another method, a method of adjusting the value of the current I0 or the current nI0 flowing into each of the diodes 80_1 to 80_4 so as to cancel the error may be used.

  As described above, according to the semiconductor device 4, it is possible to know the temperature gradient and temperature distribution in the chip without significantly increasing the chip area by disposing the diodes 80_1 to 80_4 in the chip. It becomes. The means for measuring the temperature gradient or the like is not limited to being applied to the semiconductor device 1 but can be applied to the semiconductor device 2 and the semiconductor device 3.

  Although the invention made by the present inventor has been specifically described based on the embodiments, it is needless to say that the present invention is not limited thereto and can be variously modified without departing from the gist thereof.

  For example, the measurement temperature correction means according to the present invention is applied to a microcomputer for an automobile, but the present invention is not limited to this, and can be applied to a microcomputer having a temperature measurement function. The thermal diode 60, the diodes 80_1 to 80_4, etc. in any of the semiconductor devices 1 to 4 use parasitic bipolar transistors in the CMOS process, but are not limited to this. For example, bipolar transistors and diodes in the BiCMOS process may be used. . Further, the number of diodes 80_1 to 80_4 for measuring the temperature of each part in the chip in the semiconductor device 4 is not particularly limited.

  In the semiconductor device 3, the method of using the resistor 70 as the method of selecting the temperature conversion constant has been described. However, the method of using the variation width of the base-emitter voltage VBE of the thermal diode 60 instead of the resistor 70. Can also be adopted. In this case, in the flowchart of FIG. 24, instead of measuring the voltage of the resistor 70, the base-emitter voltage VBE0 when the current I0 is passed through the thermal diode 60 is measured. Then, the temperature conversion constant corresponding to a voltage range obtained by dividing the variation width of the base-emitter voltage into a plurality of parts is determined in advance, and the temperature conversion constant is selected according to the measured value of the base-emitter voltage VBE0. . According to this, it is possible to correct the measurement temperature with higher accuracy as in the method using the resistor 70.

  As another method for correcting the measured temperature in any one of the semiconductor devices 1 to 4, the correction table 33 created by the CPU 13 is stored in the memory unit 20, and the CPU 13 detects the temperature when detecting the temperature. The temperature data measured by the sensor unit 31 or the sensor unit 39 is received, the coincidence between the temperature data and the temperature data before correction in the correction table is determined, and temperature data corresponding to the matched temperature data is obtained. It is also possible to adopt a method of correcting the measured temperature by outputting.

1-4 Semiconductor device 11 Interrupt request 12 Transfer request 13 CPU
14 Interrupt controller 15 Interrupt signal 20 Memory 22 NVM
21 RAM
40 Transfer control unit 50 Bus 30, 41, 42, 43 Temperature sensor circuit 31, 39 Sensor unit 32 Temperature correction control unit 33 Correction table 34 Temperature data 35 Request for creation of correction table 36 Request for temperature detection 37 Bus interface unit 38 Table Storage memory 311, 350 AD conversion circuit 310 Temperature voltage conversion circuit 312 Reference voltage VPTAT
313 reference voltage VREF
314, 315 Control signal 316 AD conversion result 320 AD conversion result register 321 Comparison source register 317 Temperature data before correction 324 Comparator 325 Comparison result 326 Temperature register 327 Selection circuit 329 Control unit 330 Control logic circuit 331 Register buffer 332 Register 200, 202 Ideal curve 201 Characteristic function 203, 204 Variation range 205, 206 Character correction function 400, 500, 600 Tester 401 Stage 402 High-precision temperature sensor 404, 503, 603 Storage unit 403, 504, 604 System control unit 405 Monitor 406 Heat generation / Cooling device 407A, 407B I / O interface 408A, 408B, 409A, 409B, 410A, 410B PAD
60 Thermal diode 501, 601 Current source circuit 502, 602 Multimeter 550 Minimum spacing for well separation defined in the process 551 Collector region of thermal diode 60 70 Correction resistor 730 Resistance Rar
731 Dummy resistor 732 Guard ring 80_1 to 80_4 Diode 90 PLL
351 bias circuit

Claims (10)

A semiconductor device provided with a temperature sensor unit, a temperature correction control unit, a data processing unit, and a memory unit on a semiconductor substrate,
The memory unit includes a characteristic function indicating an ideal correspondence between temperature data measured by the temperature sensor unit and temperature, and a correction coefficient indicating an error between the temperature measured by the temperature sensor unit and the actual temperature. A storage area for storing,
The data processing unit calculates the temperature data for each temperature using an inverse function of the characteristic correction function obtained by correcting the characteristic function with a correction coefficient, and corrects the calculated temperature data in correspondence with each temperature. to generate a table,
The temperature correction control unit outputs temperature data corresponding to the temperature data when the temperature data read from the correction table matches the temperature data measured by the temperature sensor unit.   The temperature correction control unit reads temperature data addressed by candidate data from the correction table, updates candidate data until the temperature data matches the temperature data measured by the temperature sensor unit, and performs addressing. 2. The semiconductor device according to claim 1, wherein when the data coincides with each other, temperature data corresponding to the candidate data at that time is output.   The semiconductor device according to claim 2, wherein the correction coefficient indicates an error between a temperature measured by the temperature sensor unit and an actual temperature in a predetermined temperature range.   The semiconductor device according to claim 3, wherein the data processing unit creates the correction table in an initialization process instructed by a reset signal or at the time of first temperature measurement.   The semiconductor device according to claim 4, wherein the temperature correction control unit outputs an interrupt signal based on the temperature data. A transfer control unit that performs data transfer control in response to a data transfer request from the data processing unit and the temperature correction control unit;
The temperature correction control unit temporarily accumulates the temperature data, and gives a data transfer request to the transfer control unit at a predetermined timing,
The semiconductor device according to claim 4, wherein the transfer control unit performs control to transfer the temperature data stored in the control unit to the memory unit in accordance with the request. A diode for measuring the temperature;
The actual temperature is a temperature calculated using a difference between a voltage when a current is supplied from the outside to the diode and a voltage when a current n times as large as the current (n is a number greater than 1) is supplied. The semiconductor device according to claim 4, wherein It further has a resistance for measuring variation in resistance value,
5. The characteristic function corresponding to a resistance variation width calculated from a voltage value at both ends of the resistor when a predetermined current is supplied to the resistor at a predetermined temperature is stored in the memory unit. semiconductor device. A plurality of diodes for measuring temperatures at a plurality of locations in the semiconductor device;
The temperature correction control unit calculates relative temperature data with respect to the measured temperature data measured by the temperature sensor unit for the temperature data of each location measured by the diode, and measured by the diode based on the relative temperature data The semiconductor device according to claim 4, wherein the temperature data of each of the locations is output. A semiconductor device having a temperature sensor unit and a data processing unit on a semiconductor substrate,
The data processing unit includes a characteristic function indicating an ideal correspondence between temperature data measured by the temperature sensor unit and a temperature, and a correction coefficient indicating an error between the temperature measured by the temperature sensor unit and the actual temperature. A correction table for calculating the temperature data for each temperature using the inverse function of the characteristic correction function obtained by correcting the characteristic function with a correction coefficient, and holding the calculated temperature data corresponding to each temperature. Generate
The said correction table is a semiconductor device which makes it possible to output the temperature data before correction | amendment searched by using candidate data as an index.

Images