Enhanced temperature sensor

Abstract

Embodiments of this disclosure relate to an enhanced temperature sensor. A calibration method for a temperature sensor is provided. The temperature sensor has a current source and a ring oscillator that generates a square pulse signal having a square pulse frequency related to temperature. A first square pulse frequency measurement obtained from the square pulse signal at a first temperature forms a first measurement point. A second square pulse frequency measurement obtained from a second square pulse signal at a second temperature forms a second measurement point. The determination of relational data represents an affine relationship between the square pulse frequency measurement and the temperature. The affine relationship is defined by a scaling factor used to modify the measurement scaling factor relative to the affine relationship linking the first and second measurement points.

Description

[0001] Cross-reference to related applications

[0002] This application claims priority to French patent application No. 2012056, filed on November 24, 2020, which is incorporated herein by reference in its entirety. Technical Field

[0003] This disclosure relates to the field of electronic circuits, and more specifically to temperature measurement integrated into electronic circuits via a temperature sensor including a ring oscillator. Background Technology

[0004] For electronic circuits (especially microcontrollers), accurate knowledge of circuit temperature is advantageous. In fact, the characteristics of many electronic components are affected by the temperature at which they are exposed. Furthermore, temperature monitoring can be used to detect attacks designed to compromise the security of electronic circuits by placing them under extreme temperature conditions. Therefore, accurate temperature knowledge is beneficial for a variety of reasons. Summary of the Invention

[0005] Having a temperature sensor calibration method, device, or both is advantageous.

[0006] To address this, a method for calibrating a temperature sensor is provided. The temperature sensor includes a current source configured to transmit an excitation signal proportional to a temperature-based current. A ring oscillator, including a series inverter, is configured to be powered by the excitation signal and to generate a square pulse signal having a square pulse frequency dependent on the excitation signal. The temperature is determined from relational data established between the square pulse frequency and the temperature.

[0007] The calibration method includes a first measurement of the frequency of a square pulse from a square pulse signal of a ring oscillator when a temperature sensor is placed at a first temperature. The first square pulse frequency measurement and the first temperature form a first measurement point for the ring oscillator. The method further includes a second measurement of the frequency of a square pulse from a square pulse signal of the ring oscillator when the temperature sensor is placed at a second temperature—a second temperature higher than the first temperature. The second square pulse frequency measurement and the second temperature form a second measurement point for the ring oscillator. The method further includes determining affine relationship data representing the square pulse frequency measurement and temperature from both the first and second measurement points of the ring oscillator. The affine relationship is defined by a scaling factor used, which is modified relative to the measurement scaling factor of the measured affine relationship linking the first and second measurement points of the ring oscillator.

[0008] The methods accomplished by the following different features, individually or in different possible combinations of them, offer several advantages.

[0009] The square pulse frequency increases with temperature, the measurement proportionality coefficient is positive, and the proportionality coefficient used is smaller than the measurement proportionality coefficient.

[0010] The measurement scale factor and the scale factor used have the same sign.

[0011] The scale factor used is determined by subtracting a certain amount from the measurement scale factor, and this amount varies depending on the measurement scale factor.

[0012] Without using any measurement points other than the first and second measurement points, determine the relationship data between the square pulse frequency and temperature.

[0013] The second temperature is at least 50°C higher than the first temperature.

[0014] The relational data includes the scaling factor used, or the relational data includes a square pulse frequency table, where the temperature pairs are composed of the scaling factor used.

[0015] The temperature sensor may include multiple ring oscillators, each ring oscillator having an inverter connected in series, the inverter being configured to be supplied with an excitation signal and configured to generate a square pulse signal having a square pulse frequency that depends on the excitation signal.

[0016] The ring oscillators differ from one another in the channel width / length ratio of the transistor forming the inverter of each ring oscillator. A first square pulse frequency measurement is obtained at a first temperature for each ring oscillator. A second square pulse frequency measurement is also obtained at a second temperature for each ring oscillator. Each ring oscillator has a first measurement point and a second measurement point.

[0017] The method may then further include the following steps: for each ring oscillator in the ring oscillator, determining a measurement scaling factor for the ring oscillator that represents the affine relationship between the measurements of a first measurement point of the ring oscillator and a second measurement point of the ring oscillator. Further, before determining the relational data representing the affine relationship between the square pulse frequency measurement and temperature, the measurement scaling factor is selected based on its proximity to a previously defined target scaling factor.

[0018] This disclosure also relates to a temperature sensor including a current source configured to transmit an excitation signal proportional to a temperature-based current. The ring oscillator includes: an inverter connected in series, configured to be supplied with the excitation signal and configured to generate a square pulse signal having a square pulse frequency dependent on the excitation signal; and a memory storing relational data representing an affine relationship between a square pulse frequency measurement and temperature, wherein the temperature sensor is calibrated according to a calibration method of this disclosure, and the relational data establishes an affine relationship defined by a scaling factor used, the scaling factor being modified relative to a measured scaling factor of the measured affine relationship, the measured affine relationship linking a first measurement point and a second measurement point of the calibration method.

[0019] A temperature sensor may include multiple ring oscillators; each ring oscillator has an inverter connected in series, the inverter being configured to be supplied with an excitation signal and configured to generate a square pulse signal having a square pulse frequency that depends on the excitation signal. The ring oscillators are different from each other by the channel width-to-length ratio of the transistors forming the inverters of each ring oscillator. Attached Figure Description

[0020] To gain a more complete understanding of this disclosure and its advantages, reference is now made to the following description in conjunction with the accompanying drawings, wherein:

[0021] Figure 1a This is a simplified view of components of a temperature sensor with a cascade-free current mirror according to a possible embodiment of the present disclosure;

[0022] Figure 1b This is a simplified view of a temperature sensor with cascaded current mirrors according to a possible embodiment of this disclosure;

[0023] Figure 2 A graph is shown that reveals the temperature error of the sensor for different power supply voltages, depending on the temperature, for non-cascaded (Figure A) and cascaded (Figure B);

[0024] Figure 3 This is a simplified diagram illustrating the steps of a calibration method according to a possible embodiment of the present disclosure;

[0025] Figure 4 This is a simplified view of a temperature sensor according to a possible embodiment of the present disclosure, the temperature sensor having a plurality of ring oscillators arranged therein;

[0026] Figure 5 This is a simplified diagram illustrating the steps of a calibration method according to a possible embodiment of the present disclosure when multiple ring oscillators are present;

[0027] Figure 6The diagram illustrates the relationship between square pulse frequency measurement and temperature for both single oscillator (Figure A) and multiple oscillators (Figure B); and

[0028] Figure 7 The graph shows the dispersion of the error on the temperature measured by the temperature sensor based on the temperature before calibration compensation (Figure A) and after calibration compensation (Figure B). Detailed Implementation

[0029] U.S. Patent Application No. 14 / 760,762 describes a temperature sensor with an oscillator adapted to alternately generate a temperature-dependent output frequency in a detection mode and a temperature-independent output frequency in a calibration mode. A controller is adapted to provide at least one first input signal (VREF) to the oscillator and to vary the first input signal (VREF) to switch the oscillator between the generation of the temperature-dependent output frequency and the generation of the temperature-independent output frequency. However, this temperature sensor has several drawbacks; in particular, it causes non-negligible power consumption and includes numerous components that can introduce inaccuracies, such as the presence of a comparator that introduces nonlinearity.

[0030] Figure 1a and Figure 1b Simplified views of the components of a temperature sensor (1) with and without cascaded current mirrors are shown respectively. The temperature sensor (1) includes a current source (2) configured to emit a current proportional to temperature. The temperature-based current represents the absolute temperature of the electronic circuit whose temperature is being measured, and preferably represents a proportional absolute temperature, and the electronic circuit integrates the temperature sensor (1). Implementing such a current will be within the capabilities of those skilled in the art, and several solutions are known in the art.

[0031] The current sensor includes a first transistor (4) and a second transistor (6), the first transistor (4) typically being a field-effect transistor (preferably a PMOS as in this example) and the second transistor (6) typically being a field-effect transistor (preferably a PMOS as in this example). If a transistor other than a PMOS transistor is used, it is advantageous to modify the transistor connections described below accordingly, especially when using cascaded circuitry to significantly reduce the impact of the VDD power supply on the IPTAT output current.

[0032] The source terminals of the first transistor (4) and the second transistor (6) are coupled together and connected to the power supply voltage VDD. The drain terminal of the first transistor (4) is coupled to the current source (2). In, for example... Figure 1bIn the case of a cascaded circuit, the gate terminal of the first transistor (4) is connected to its drain or the drain of another transistor (5), wherein the other transistor (5) is then inserted between the first transistor (4) and the current source (2). The gate terminals of the first transistor (4) and the second transistor (6) are coupled together such that the IPTAT current flowing through the first transistor (4) is replicated in the current flowing through the second transistor (6), and thus this current is also called IPTAT. The drain terminal of the second transistor (6) is connected to a power node (8) having an excitation voltage VDD_RO and having an IPTAT current forming the excitation signal converted through it, or the drain terminal of the second transistor (6) is connected to a power node such as Figure 1b In the case of a cascaded circuit, it is connected to the source of another transistor (7), which is arranged between the second transistor (6) and the power supply node (8). The gate terminal of transistor (5) is connected to the gate terminal of transistor (7) and can be connected to another component (not shown).

[0033] Figure 2 Two graphs are shown to illustrate the effect of cascading in the current mirror on the temperature linearity of the sensor according to different values ​​of the power supply voltage VDD.

[0034] More specifically, graph A shows the percentage error of the non-cascaded temperature sensor based on the temperature (in °C) for two different supply voltages VDD: 1.1V (curve 20) and 1.3C (curve 21).

[0035] The bottom graph B shows the percentage error of the cascaded temperature sensor based on the temperature (in °C) for two different supply voltages VDD: 1.1V (curve 22) and 1.3V (curve 23).

[0036] As can be seen, the cascading allows the unit to be almost unaffected by changes in the supply voltage VDD. In fact, it is advantageous that once the sensor is calibrated, its conversion function does not change with its supply voltage. Therefore, the presence of cascading is preferred.

[0037] return Figure 1a and Figure 1b The temperature sensor (1) includes a ring oscillator (10) configured to be supplied with an excitation signal, and the ring oscillator (10) is configured to generate a square pulse signal Crx having a square pulse frequency depending on the excitation signal. The ring oscillator (10) includes a power node (8) having an excitation signal (e.g., IPTAT current) reaching it.

[0038] like Figure 1aAs shown, the ring oscillator (10) includes an odd number of inverters (12) connected in series or cascaded to form a ring. For example, these inverters are NOT gates, NOR gates, or NAND gates. For example, depending on the transistor forming the inverter, the inverter (12) can be a simple PMOS gate, NMOS gate, or CMOS gate, or it can be an inverter differential amplifier assembly.

[0039] The inverters (12) are biased or powered using the IPTAT current and therefore the excitation signal. Each inverter (12) outputs the logical NOT of its input signal. However, when the input signal is switched, the output of the inverter (12) does not switch immediately. There is a switching delay, and this switching delay depends on the strength of the bias current of the inverter (12) and, therefore, on the strength of the temperature-dependent IPTAT current, which serves as the excitation signal, in addition to other parameters (especially the width-to-length ratio W / L of the inverter).

[0040] The output signal of an odd number of cascaded inverters (12) is the logical negation of the input signal. However, the loop connection (i.e., closed loop) introduces feedback to the output signal, which generates a periodic oscillation phenomenon that causes a square pulse signal (voltage) to flow through the ring oscillator (10), characterized by the square pulse frequency. Furthermore, the ring oscillator (10) is sometimes referred to as a square pulse generator.

[0041] The frequency of the square pulse depends on the bias current of the inverter (12), and therefore on the intensity of the temperature-dependent IPTAT current, which serves as the excitation signal. Thus, the square pulse frequency depends on the temperature via the excitation signal. Typically, this dependence is positive at high bias frequencies; that is, the square pulse frequency increases with increasing temperature.

[0042] The square pulse frequency also depends on the number of inverters (12) in the ring oscillator (10): the more inverters (12) there are, the more significant delay they introduce, and thus the lower the square pulse frequency.

[0043] To simplify, Figure 1a The example shows three inverters (12), but the preferred number of inverters (12) is greater than three, and for example in the range of five to thirteen. The oscillator's power supply voltage VDD_RO can also be modulated to modify the square pulse frequency; a higher voltage reduces the delay and thus increases the frequency.

[0044] like Figure 1bAs shown, the ring oscillator (10) may include logic gates (18) arranged in the ring of the ring oscillator (10), replacing the inverters (12) and thus connected in series with other inverters (12) in the cascade. The sum of the number of inverters (12) and the number of logic gates (18) thus remains odd. For example, the logic gates (18) are OR gates or AND gates controlled by control signals (not shown), or NOR gates or NAND gates.

[0045] The logic gate (18) can be used as a switch for the ring oscillator (10) enabled by a control signal to start or initiate oscillation, thereby starting or initiating the generation of a square pulse signal.

[0046] Typically, the logic gate (18) is used to start the ring oscillator (10) after the establishment of the IPTAT current and to stop the ring oscillator (10) before the interruption of the IPTAT current. Therefore, it is possible to ensure that the square pulse frequency measurement effectively represents the temperature via the established IPTAT current, unaffected by transient effects. Since the logic gate (18) only enables the use of the ring oscillator (10) when the temperature expectation is known, which is typically intermittent rather than continuous, this configuration also minimizes power consumption.

[0047] It should be noted that the cascaded components are independent of the presence or absence of logic gate (18), and Figure 1a and Figure 1b The difference is presented here only to illustrate the possibility. Logic gate (18) can be replaced... Figure 1a The inverter (12), and Figure 1b The ring oscillator (10) can have an odd number of inverters (12) without logic gates (18).

[0048] The square pulse signal Crx can be sampled from the output (14) located at any point in the ring oscillator (10). The output (14) can be configured with components that do not interfere with the oscillation in the ring oscillator (10), such as a buffer amplifier (voltage follower or level shifter (not shown)).

[0049] The square pulse signal Crx is a periodic AC signal consisting of an all-or-nothing pattern formed by the repetition of a pattern, meaning a high voltage level followed by a low voltage level. The alternation between high and low levels is performed according to the square pulse period, which is determined by the square pulse frequency. Depending on the configuration, this square pulse frequency depends on the temperature.

[0050] The square pulse signal Crx from output (14) is provided to counter (16) to calculate the count value derived from the level transitions in the square pulse signal. Counter (16) can be a flip-flop-based register circuit or another type of register circuit. For example, counter (16) can count each transition, count only the rising edge or the falling edge, or only its multiples. The changes in the square pulse signal Crx considered by counter (16) are referred to as events below.

[0051] The counter (16) receives a clock signal Clk, which is temperature-independent, and then includes a series of transitions between high and low levels to define the clock frequency. The clock signal Clk is typically provided by a clock generator; usually, a piezoelectric quartz-based oscillator can be integrated into the temperature sensor (1), or more advantageously, it belongs to the electronic circuitry containing the temperature sensor (1), and the clock signal Clk is further used by other components of the electronic circuitry.

[0052] A counter (16) is configured to count the number of events in a square pulse signal Crx between two events of a clock signal Clk. Typically, the counter (16) increments for each event of the square pulse signal Crx and resets for each event of the clock signal Clk. Therefore, the resulting count value directly represents the ratio of the square pulse frequency to the clock frequency. Since the clock frequency is known and assumed to be a constant independent of temperature, the count value reveals the square pulse frequency, thus enabling a square pulse frequency measurement. The occurrence of an event in the clock signal Clk results in the reading of the count value before the counter (16) is reset. Therefore, a square pulse frequency measurement is possible.

[0053] Typically, the square pulse frequency is higher than the clock frequency, and more specifically, high enough within its expected range of variation to extend over the entire range. A sufficient number of square pulse cycles occur within a single clock cycle, for the number of transitions during the clock cycle, to derive fine temperature variations.

[0054] Compared to the square pulse frequency, the lower the clock frequency, the more accurate the temperature measurement. For example, the square pulse frequency can be at least 1000 times higher than the clock frequency, preferably at least 10000 times higher.

[0055] Conversely, the square pulse frequency can be lower than the clock frequency. In this case, the counter (16) is configured to count the number of events of the clock signal Clk between two events in the square pulse signal Crx. The resulting count value still directly represents the ratio of the square pulse frequency to the clock frequency, but the corresponding actions of the clock signal Clk and the square pulse signal Crx are opposite to those described above.

[0056] As mentioned above, linking the square pulse frequency to temperature allows square pulse frequency measurements to be used to determine temperature. This relationship is established using relational data.

[0057] The relational data can be stored in the register of any of the actual temperature sensors in the actual temperature sensor (1) or in a computer memory (e.g., flash memory type), or in the electronic circuitry that receives the temperature sensor (1) to allow the measurement (e.g., count value) of the square pulse frequency to be converted into a temperature value.

[0058] For simplicity, the relational data in first order represents the affine relationship between the square pulse frequency measurement and the temperature, and the relational data may include a scaling factor A, which will be applied to the square pulse frequency measurement to obtain the corresponding temperature.

[0059] The relational data typically also includes at least one constant used to obtain the temperature. For example, the temperature can be obtained directly by multiplying a square pulse frequency measurement and adding the constant. Multiple constants, such as reference frequency measurements (e.g., reference count values) and corresponding reference temperatures, can be stored.

[0060] To obtain the temperature, the interval between the square pulse frequency measurement and the reference frequency measurement is multiplied by a scaling factor A to obtain the interval between the measured temperature and the reference temperature, which makes it possible to deduce the measured temperature.

[0061] Typically, the excitation signal changes positively with temperature, the square pulse frequency increases with temperature, and correspondingly, the proportionality coefficient A is positive.

[0062] The relationships established from the relational data depend on the specific characteristics of each temperature sensor (1) and vary between the two transistors based on their dimensions and due to unavoidable variations in manufacturing methods. Therefore, relational data must be established individually for each temperature sensor (1) or at least for each manufacturing batch (e.g., each “wafer”). Thus, it is desirable to implement a calibration method to determine these relational data.

[0063] The fact that affine relationships are established based on data is inherently determined by the very limited availability of data to establish a relationship between each square pulse frequency and its corresponding temperature. Typically, only two measurement points (square pulse frequency measurements at known temperatures) are available for calibration. These two calibration measurements are usually performed during quality control at the temperatures at which the electronic circuitry is submitted to determine its correct operation at points representing its operating range. These two measurements determine the proportionality coefficient A and the associated constants.

[0064] Figure 3This is a simplified diagram illustrating the steps of a calibration method according to a possible embodiment of the present disclosure. The calibration method can be performed in advance by calibrating the current source (2) (step S0), which will be described in further detail below.

[0065] A temperature sensor (1) is placed at a first temperature T1. The method includes acquiring (step S1) a first square pulse frequency measurement FT1 at the first temperature T1 from a square pulse signal. The first square pulse frequency measurement FT1 and the first temperature T1 form a first measurement point. The first square pulse frequency measurement FT1 and the first temperature T1 are stored.

[0066] Temperature sensor (1) is placed at a second temperature T2. The calibration method includes acquiring (step S2) a second square pulse frequency measurement FT2 at the second temperature T2 from a square pulse signal. The second square pulse frequency measurement FT2 and the second temperature T2 form a second measurement point, and can also be stored or directly processed in the determination of relational data.

[0067] The first temperature T1 and the second temperature T2 are different; therefore, one temperature is lower than the other. However, the first temperature T1 is not necessarily specified as being lower than the second temperature T2. The designation of the first and second temperatures distinguishes the measurement points, and the order in which the two measurement points are acquired is not predetermined.

[0068] The first temperature T1 and the second temperature T2 are preferably sufficiently far apart to represent the range of operating temperatures of the electronic circuit. Preferably, the second temperature T2 is at least 50°C higher than the first temperature T1, and more preferably at least 70°C higher. For example, the first temperature T1 can be in the range of 10°C to 40°C, and the second temperature can be in the range of 100°C to 150°C. Of course, these are merely indications, and the temperatures depend on the type of electronic circuit, the application of the electronic circuit, and especially on the temperature at which the test is performed.

[0069] As described above, before acquiring the first square pulse frequency measurement FT1 and the second square pulse frequency measurement FT2, it is possible to calibrate the current source (2), and more specifically, to calibrate the resistance value R involved in determining the ratio of the temperature-dependent IPTAT current to the temperature-based voltage VPTAT (IPTAT = VPTAT / R) (step S0). This calibration of the resistance value R is performed at a given known temperature, which is either a first temperature T1 or a second temperature T2.

[0070] The determined resistance value R is stored and can be used as part of a relational dataset establishing a relationship between each square pulse frequency and temperature. However, without a comparator, a comparator calibration is no longer necessary.

[0071] The two measurement points of the ring oscillator (10) define the measurement scaling factor of the measurement affine relationship between the first measurement point and the second measurement point: A = (T2-T1) / (FT2-FT1).

[0072] The determination of the relational data includes determining the scale factor A' to be used relative to the measurement scale factor A (step S3). Then, the scale factor A' thus determined is stored and used in the relational data (step S4). This determination of the scale factor A' can be indirect, as it primarily implies the determination of the measurement scale factor A, or direct, without implying the determination of the measurement scale factor A. This will be combined with... Figure 6 and Figure 7 These steps (S3, S4) will be described in further detail.

[0073] The frequency-temperature relationship of temperature sensors is usually in a more complex form than a simple affine relationship, which is only an approximation. Furthermore, when the temperature deviates from the temperature used to measure the point, the temperatures of the two measurement points are usually not correctly determined by their ability to describe the frequency-temperature relationship.

[0074] The accuracy of the measurement proportionality coefficient A depends on many characteristics of the temperature sensor (1) and may exhibit non-negligible variations, thus necessitating calibration. In particular, the structure of the transistors forming the inverters of each ring oscillator significantly affects the measurement proportionality coefficient A. Besides knowing the exact value of the measurement proportionality coefficient A, it may be desirable to ensure that this value is close to a target value. Therefore, despite the potential for variation, it is desirable to select the value of the measurement proportionality coefficient A to a certain extent.

[0075] Figure 4 This is a simplified view of a temperature sensor (1') according to a possible embodiment of the present disclosure, wherein multiple oscillators are arranged in a ring. As an example, the configuration shown is... Figure 1a The configuration, but Figure 1b The configuration can be preferably used or another configuration. The difference from the previous configuration is that the temperature sensor (1') includes a plurality of ring oscillators 10a, 10b, ..., and 10n, which are configured to be supplied with the same IPTAT current to form an excitation signal, and each ring oscillator generates a square pulse signal Crxa, Crxb, ..., and Crxn using a corresponding square pulse frequency.

[0076] The ring oscillators 10a, 10b, ..., and 10n differ from each other in the channel width-to-length ratio (or W / L) of the transistors forming the inverters of each ring oscillator. For example, the first ring oscillator 10a may have an inverter with a W / L ratio of 15 μm to 0.4 μm, the second oscillator 10b may have an inverter with a W / L ratio of 8 μm to 0.4 μm, and another oscillator may have an inverter with a W / L ratio of 10 μm to 0.25 μm, and so on. Therefore, each ring oscillator 10a, 10b, ..., and 10n has a W / L ratio specific to itself and different from the others. The width, length, or both of the two ring oscillators 10a, 10b, ..., and 10n may differ.

[0077] Preferably, the number n of the ring oscillators 10a, 10b, ..., and 10n' in the same temperature sensor (1') is less than 10, preferably less than 5, to avoid consuming too much surface area. Typically, there are two to four ring oscillators 10a, 10b, ..., and 10n. Preferably, all ring oscillators 10a, 10b, ..., and 10n are used during calibration, but not all may be used. Different W / L ratios of the inverters of these ring oscillators 10a, 10b, ..., and 10n are selected to span a range of values ​​for the measurement scaling factor A, which may cover the target scaling factor A. target The corresponding value is the target scaling factor A. target It is a target value previously defined in the design (through simulation) or by characterizing a certain number of parts.

[0078] As previously described, the generation of the IPTAT current is performed via a current mirror, which means that the current source (2) is connected to the drain of the first transistor (4). The gate terminal of the first transistor (4) is connected to its drain terminal and to each of the gate terminals of a plurality of second transistors 6a, 6b, ..., and 6n, each second transistor being arranged on a branch through which the IPTAT current flows. Each branch forms a power supply node 8a, 8b, ..., and 8n with the drain of the second transistors 6a, 6b, ..., and 6n connected thereto, and has excitation voltages VDD_RO1, VDD_RO2, ..., and VDD_ROn specific to each ring oscillator (10a), 10b, ..., and 10n applied thereto. The excitation voltages employ different values ​​due to variations in behavior caused by differences in the W / L ratios of the ring oscillators 10a, 10b, ..., and 10n.

[0079] Figure 5 This shows that when there are multiple such Figure 4 The diagram shows a simplified representation of the steps of a calibration method according to a possible embodiment of this disclosure when a ring oscillator is shown.

[0080] The calibration method then includes the same steps S1 and S2 as before, except that each ring oscillator 10a, 10b, ..., 10n is measured at temperatures T1 and T2. In step S1, as many square pulse frequency measurements FaT1, FbT1, ..., FnT1 are obtained at the first temperature T1 as for the ring oscillators 10a, 10b, ..., 10n. Similarly, in step S2, as many square pulse frequency measurements FaT2, FbT2, ..., FnT2 are obtained at the second temperature T2 as for the ring oscillators 10a, 10b, ..., 10n. Therefore, there are as many first measurement points as for the ring oscillators 10a, 10b, ..., 10n, and as many second measurement points as for the ring oscillators 10a, 10b, ..., 10n.

[0081] The calibration method then includes step S2a, where measurement scaling factors Aa, Ab, ..., An are determined for each of the ring oscillators 10a, 10b, ..., 10n from the measurement points of the ring oscillators 10a, 10b, ..., 10n. Therefore, there are as many measurement scaling factors Aa, Ab, ..., An as there are ring oscillators 10a, 10b, ..., 10n. Two measurement points for each ring oscillator 10a, 10b, ..., 10n define the measurement scaling factor Ai, which is an affine relationship between the measurements of the first measurement point and the second measurement point of the ring oscillators 10a, 10b, ..., 10n: Ai = (T2 - T1) / (FiT2 - FiT1), and i = a, b, ..., n.

[0082] Figure 6 The diagram illustrates the relationship between the square pulse frequency measurement (horizontal axis in Hertz) and temperature (vertical axis in °C) for a single ring oscillator (Figure A above). For multiple ring oscillators (Figure B below), in Figure A above, curve 30 shows a simplified example of the real relationship between the temperature of the ring oscillator (10) and the square pulse frequency. It is further characterized by the first measurement point (T1, FT1) and the second measurement point (T2, FT2) of the ring oscillator, where straight line 31 is defined as a solid line passing through these two measurement points, and has a measurement scaling factor A for the ring oscillator (10) as the slope.

[0083] In Figure (B) below, curves 30a, 30b, ..., and 30n show a simplified example of the real relationship between the temperatures (vertical axis in °C) of the multiple ring oscillators 10a, 10b, ..., and 10n of the temperature sensor (1') and the square pulse frequencies (horizontal axis in Hertz). The same first temperature T1 has multiple corresponding first square pulse frequency measurements FaT1, FbT1, ..., and FnT1, and the second temperature T2 has multiple corresponding second square pulse frequency measurements FaT2, FbT2, ..., and FnT2, which allows as many straight lines 31a, 31b, ..., and 31n as possible to be shown on the solid line passing through the pair of measurement points, with a measurement scaling factor Ai for each ring oscillator 10a, 10b, ..., and 10n as the slope.

[0084] During the selection step S2b, a measurement scale factor Ai is then selected from the thus determined measurement scale factor Ai. The measurement scale factor Ai is selected as the one closest to the previously defined target scale factor A. target .

[0085] exist Figure 6 In the example of Figure (B) below, line 33 (dashed line) has the target scale factor A. target As the slope, the line closest to line 33 is line 31b, which has the measurement scale factor Ab of the second ring oscillator 10b defined by the measurement points (T1, FbT1) and (T2, FbT2) as its slope. This measurement scale factor Ab is therefore selected as the new measurement scale factor A.

[0086] More systematically, Figure 7 The simulation results of multiple real relationships linking square pulse frequency measurements and temperature are presented in the form of curves, based on the expected changes in the characteristics affecting this relationship.

[0087] exist Figure 7 In Figure (A) above, the curves show the temperature errors (in °C) (vertical axis) (resulting from the approximation of the affine relation and the measurement scale factor A for seven temperatures ranging from -40°C to 130°C. The two measurement points correspond to temperatures of 30°C and 110°C, so logically, the errors at these temperatures are zero. It can be observed that while the approximation exhibits a small error for high temperatures (greater than or equal to 30°C), the error increases significantly for low temperatures (below 20°C), particularly for temperatures close to or below zero.

[0088] Therefore, based on the first and second measurement points, the calibration method does not use the measurement scaling factor A, but instead determines (step S3) the scaling factor A' to be used, which is derived from the measurement scaling factor A, but is different from the scaling factor A.

[0089] Similar to the measurement scaling factor A, the scaling factor A' used defines the affine relationship between the two square pulse frequency measurements and temperature. However, the scaling factor A' used is modified relative to the measurement scaling factor A that links the measurements at the first and second measurement points. More specifically, the scaling factor A' used is the theoretical scaling factor A defined from two measurement points at two temperatures T1 and T2, and from previous characterization in the design (through simulation) or by a determined number of parts. max It is derived. The proportionality coefficient A' used defines a more accurate affine relationship between the square pulse measurement frequency and temperature than the measurement proportionality coefficient A, that is, it has a smaller significant error above the usable range.

[0090] The method may include adjusting the measurement scaling factor A to reduce temperature error for temperatures below a first temperature T1. When the square pulse frequency increases with temperature, the scaling factor A' used is smaller than the measurement scaling factor A, but has the same sign. Therefore, the measurement scaling factor A is modified by subtracting an amount smaller than the measurement scaling factor A from the measurement scaling factor A, which varies according to the measurement scaling factor A.

[0091] For example, the scaling factor A' used can correspond to: A' = AK(A max -A), K is a positive factor and A max These are the previously defined theoretical proportionality coefficients. The positive constant factor K and the theoretical proportionality coefficient A are chosen. max This ensures that the scaling factor A' used has the same sign as the measurement scaling factor A. Preferably, the positive constant factor K is strictly in the range of zero to two, and still preferably, the positive constant factor K is strictly in the range of zero to one.

[0092] In a direct manner, by replacing A in the formula above with (T2-T1) / (FT2-FT1), the proportionality coefficient A' to be used can be determined directly from the measurement point without the intermediate determination of the proportionality coefficient A by measuring it.

[0093] exist Figure 6 In the figure above (A), line 32 (dashed line) corresponds to the affine relationship between square pulse frequency measurement and temperature, which is defined by the measurement scaling factor A' used to modify the affine relationship between the measurement of the first and second measurement points shown by line 31 (solid line).

[0094] The proportionality coefficient A' thus determined is used to determine (step S4) the relational data. This relational data, representing the affine relationship between the square pulse frequency measurement and temperature, is stored in memory and used during subsequent temperature measurements. Typically, when the temperature T is subsequently derived from the frequency measurement FM: T = A'.FM + B, the proportionality coefficient A' used can be used as the relational data, and B is a constant that also forms part of the relational data, for example corresponding to B = (T1.FT2 - T2.FT1) / (FT2 - FT1). The relational data may also contain neither the proportionality coefficient A' nor the constant B, but rather a lookup table between the square pulse frequency measurement and temperature, constructed using the proportionality coefficient A' and the constant B.

[0095] A lookup table stores multiple square pulse frequency-temperature measurement pairs, which are determined by the scaling factor A' used and the two measurement points (e.g., via a constant B). For each measured frequency, there may be a sufficient number of pairs to be directly assigned the associated temperature. Alternatively, interpolation may be performed between two square pulse measurement-temperature pairs around a specific square pulse frequency measurement that was just taken.

[0096] After calibration, during the operation of the temperature sensor (1), relational data representing the affine relationship between the square pulse frequency measurement and temperature is stored and used. This affine relationship is defined by the scaling factor A' used, rather than by the measurement scaling factor A' that links the measurements of the first and second measurement points. This calibration reduces the error of the temperature sensor (1) in response to low temperatures (i.e., temperatures below the first measurement point).

[0097] Figure 7 Figure (B) below shows, in the form of curves, simulation results of multiple real relationships between the square pulse frequency measurement and temperature, based on the expected changes in the characteristics affecting this relationship. Figure 7 The curve in Figure (B) below shows the temperature error (in °C) approximated by the scaling factor A' used for seven temperatures in the range from -40 °C to 130 °C (vertical axis). It can be observed that the maximum error is less than... Figure 7 The error in (A). More specifically, the error in the upper figure (A) extends over the range from +1.4℃ to -0.7℃, while the error in the lower figure (B) extends over the range from +0.4℃ to -0.6℃, which is more than twice the range obtained using the uncorrected measurement scale factor A.

[0098] After calibration, the relevant data is stored in the temperature sensor (1) or in the electronic circuitry of the integrated temperature sensor (1). When the temperature is desired, a current source (2) is started by establishing an IPTAT current, which then turns on the ring oscillator (10) via logic gate (18). The output square pulse signal (14) is delivered to a counter (16), which determines the square pulse frequency measurement in the form of a count value. This square pulse frequency measurement is then converted into temperature using relational data (e.g., a calculator). The temperature thus measured can be used. The ring oscillator (10) is then stopped by logic gate (18).

[0099] This disclosure is not limited to the embodiments described in the accompanying drawings. Modifications are still possible, particularly from the perspective of the composition of various technical features or by substitution with technical equivalents, without departing from the scope of this disclosure.

[0100] Although the description has been detailed, it should be understood that various changes, substitutions, and alterations can be made without departing from the spirit and scope of this disclosure as defined by the appended claims. In different figures, the same elements are denoted by the same reference numerals. Furthermore, the scope of this disclosure is not intended to be limited to the specific embodiments described herein, as it will be readily understood from this disclosure by those skilled in the art that existing or future processes, machines, manufactures, compositions of matter, apparatuses, methods, or steps can perform substantially the same functions or achieve substantially the same results as the corresponding embodiments described herein. Therefore, the appended claims are intended to include such processes, machines, manufactures, compositions of matter, apparatuses, methods, or steps within their scope.

[0101] Therefore, the specification and drawings are simply regarded as a description of this disclosure as defined by the appended claims, and are intended to cover any and all modifications, variations, combinations or equivalents falling within the scope of this disclosure.

Claims

1. A method for calibrating a temperature sensor, the method comprising: An excitation signal proportional to temperature-based current is transmitted via a current source. An inverter of a ring oscillator generates square pulse signals based on the excitation signal, each square pulse signal having a corresponding square pulse frequency, and the inverter is powered by the excitation signal; as well as Temperature is determined based on relational data, including: Based on the temperature sensor being at a first temperature, a first measurement of the frequency of the square pulse is obtained from the square pulse signal; this first measurement and the first temperature form a first measurement point. Based on the temperature sensor being at a second temperature higher than the first temperature, a second measurement of the frequency of the square pulse is obtained from the square pulse signal. This second measurement and the second temperature form a second measurement point. The relationship data is determined from the first measurement point and the second measurement point. The relationship data represents an affine relationship between square pulse frequency measurement and temperature. The affine relationship is defined by a scaling factor used, which is modified relative to the measurement scaling factor linking the measurements of the first measurement point and the second measurement point.

2. The method of claim 1, wherein the typical square pulse frequency increases with temperature, the measurement scaling factor is positive, and the scaling factor used is less than the measurement scaling factor.

3. The method according to claim 1, wherein the measurement scaling factor and the scaling factor used have the same sign.

4. The method of claim 1, further comprising determining the used scaling factor by subtracting an amount that varies according to the measurement scaling factor from the measurement scaling factor.

5. The method of claim 1, wherein the relationship data is determined without using any measurement points other than the first measurement point and the second measurement point.

6. The method of claim 1, wherein the second temperature is at least 50°C higher than the first temperature.

7. The method of claim 1, wherein the relational data includes the scaling factor used, or the relational data includes a square pulse frequency table comprising temperature pairs constructed from the scaling factor used.

8. The method of claim 1, wherein the temperature sensor comprises the current source and the ring oscillator, the ring oscillator comprising an inverter, and the inverters are arranged in series.

9. The method of claim 8, wherein the temperature sensor further comprises a plurality of ring oscillators, each ring oscillator having an inverter arranged in series, the ring oscillators being configured to be provided with the excitation signal and configured to generate a square pulse signal having a square pulse frequency dependent on the excitation signal, the ring oscillators differing from each other in the channel width-to-length ratio of the transistor forming the inverter of each ring oscillator, obtaining a first square pulse frequency measurement at the first temperature for each ring oscillator, and obtaining a second square pulse frequency measurement at the second temperature for each ring oscillator, each ring oscillator having a first measurement point and a second measurement point, the method further comprising: For each of the ring oscillators, a measurement scaling factor for each ring oscillator is determined based on an affine relationship that couples the first measurement point and the second measurement point of each ring oscillator. Prior to the determination of the relational data, a measurement scale coefficient is selected based on the proximity of the measurement scale coefficient to a previously defined target scale coefficient.

10. A temperature sensor, comprising: A current source is configured to transmit an excitation signal proportional to a temperature-based current. A ring oscillator, including inverters arranged in series, is configured to be supplied with the excitation signal and to generate a square pulse signal having a square pulse frequency depending on the excitation signal; as well as The memory is configured to store relational data representing the affine relationship between square pulse frequency measurement and temperature, and the temperature sensor is configured to: Based on the temperature sensor being at a first temperature, a first measurement of the frequency of the square pulse is obtained from the square pulse signal; this first measurement and the first temperature form a first measurement point. Based on the temperature sensor being at a second temperature higher than the first temperature, a second measurement of the frequency of the square pulse signal is obtained from the square pulse signal. This second measurement and the second temperature form a second measurement point. Temperature is determined from the first and second measurement points based on relational data representing the affine relationship between square pulse frequency measurement and temperature. The affine relationship is defined by a scaling factor used, which is modified relative to the measurement scaling factor linking the measurements of the first and second measurement points.

11. The temperature sensor of claim 10, further comprising a plurality of ring oscillators, each ring oscillator having an inverter arranged in series, the ring oscillators being configured to be provided with the excitation signal and configured to generate a square pulse signal having a square pulse frequency dependent on the excitation signal, the ring oscillators differing from each other in the channel width-to-length ratio of the transistor forming the inverter of each ring oscillator, a first square pulse frequency measurement at the first temperature being obtained for each ring oscillator, and a second square pulse frequency measurement at the second temperature being obtained for each ring oscillator, each ring oscillator having a first measurement point and a second measurement point, the temperature sensor further configured to: For each ring oscillator, a measurement scaling factor for each ring oscillator is determined based on an affine relationship that couples the first measurement point and the second measurement point of each ring oscillator. Before determining the relationship data, the measurement scale coefficient is selected based on its proximity to the previously defined target scale coefficient.

12. The temperature sensor of claim 10, wherein the typical square pulse frequency increases with temperature, the measurement proportionality coefficient is positive, and the proportionality coefficient used is less than the measurement proportionality coefficient.

13. The temperature sensor of claim 10, wherein the measurement proportionality coefficient and the used proportionality coefficient have the same sign.

14. The temperature sensor of claim 10, wherein the temperature sensor is further configured to determine the used proportionality coefficient by subtracting an amount that varies according to the measurement proportionality coefficient from the measurement proportionality coefficient.

15. The temperature sensor of claim 10, wherein the second temperature is at least 50°C higher than the first temperature.

16. The temperature sensor of claim 10, wherein the relationship data includes the scaling factor used, or the relationship data includes a square pulse frequency table comprising temperature pairs constructed from the scaling factor used.

17. A temperature sensor, comprising: A current source is configured to transmit an excitation signal proportional to a temperature-based current. Multiple ring oscillators, each ring oscillator having an inverter arranged in series, and each ring oscillator being configured to be provided with the excitation signal, and each ring oscillator generating a square pulse signal having a square pulse frequency depending on the excitation signal; as well as The memory is configured to store relational data representing the affine relationship between square pulse frequency measurement and temperature, and the temperature sensor is configured to: At a first temperature, a first measurement point is determined based on the first temperature and a first measurement from a first square pulse signal; At a second temperature, a second measurement point is determined based on the second temperature and a second measurement from a second square pulse signal, wherein the second temperature is higher than the first temperature, and Temperature is determined based on relational data from the first and second measurement points, the relational data representing an affine relationship defined by a scaling factor that is modified relative to the measurement scaling factor linking the measurements of the first and second measurement points.

18. The temperature sensor of claim 17, wherein the second temperature is at least 50°C higher than the first temperature.

19. The temperature sensor of claim 17, wherein the frequency of each square pulse increases with temperature, and each measurement scaling factor is positive.

20. The temperature sensor of claim 17, wherein the first temperature is in the temperature range of 10°C to 40°C, and the second temperature is in the temperature range of 100°C to 150°C.

Images