Arrangement for linearizing a non-linear sensor
Field of the invention
The invention relates to an arrangement for linearizing a non-linear sensor. More particularly, an arrangement for converting the output of a sensor, whose output follows a non- linear response curve, to an output of linearized readings. Even more particular, the invention relates to an arrangement for linearizing an NTC thermistor over a very wide temperature range, i.e. -400C - 2000C.
Background of the invention
To display the value of a physical parameter, such as temperature, pressure, etc., it is necessary to employ a transducer that responds to that parameter by changing an electrically-detectable characteristic. Such characteristics include temperature-sensitive resistance variations and thermocouple voltages. Most sensors display significant non- linearities in their outputs, i.e. sensor output voltages or currents deviate appreciably from a linear relationship with the input phenomenon being monitored. It is well known that thermistors and thermocouples exhibit nonlinear output increments in response to linear temperature increments.
A system for linearizing a non-linear sensor output is known from US5274577. A processor is described for monitoring a sensor whose output signal follows a non linear curve in response to a linear change in a monitored phenomenon. The processor converts the output signal to a linear output value proportional to the phenomenon monitored. The non-linear sensor signal is linearized after digitizing.
Furthermore, a variety of classes of analog linearizing devices exist for correcting non-linearity over input ranges. Analog linearizing devices provide adequate accuracy for a limited number of sensor types over restricted input ranges. The known analog linearizing devices store the required correction information in the values of resistors which set break points between various segments of the response curve and the slope/offset peculiar to each segment of the response curve. Analog linearizers typically have eight or fewer segments, and are capable of linearizing most temperature sensing elements, within a few degrees, over commonly-experienced input ranges.
US 6099163 discloses a circuit for correcting a non- linear output of an electronic temperature sensor.
"Thermistor-Based linear temperature-to-voltage converter" my Jain L. C, Measurement, Institute of measurement and control. London, GB, vol. 7, no. 3, 1 July 1989, page 132/133, and US-A-5116136 discloses an arrangement for linearizing a non-linear sensor by using a logarithmical amplifier. The logarithmical amplifier performs an amplification wherein the output voltage is K times the natural log of the input voltage.
WO2006/135977 discloses a method for fitting the output of a sensor to a predetermined relationship. A microprocessor reads the voltage across a thermistor via an AD converter. A number of bias resistances is under control of the microprocessor connected in series with the thermistor. The effective bias resistance is changed from moment to moment. The microprocessor reads one at a time the voltage across the thermistor associated with one of the bias resistances. The microprocessor forms a weighted sum of the read voltages. When using any of the bias resistances, the voltage across the thermistor supplied to the AD converter has a characteristic that follows a curve in response to a linear change in temperature in the disclosed temperature range, wherein the curve is non-linear.
CO2 Air Conditioning (AJC) systems can support both "cooling" and "heat pump" functionality. Heat pump functionality allows the AJC system to heat instead of only cooling the interior of for example a car. In "cooling" mode, the radiator cools the CO2 in the AJC system. The AJC system will be most efficient when the temperature of the CO2 in "cooling" mode is approximately 1500C. In "heat pump" mode, the radiator has to heat-up the CO2 in the AJC system. The lower the radiator temperature the more efficient the system will be. However, the radiator will only work properly when the temperature of the CO2 at the inlet of the radiator is above 00C. As soon as the temperature is below 00C, ice will be formed on the radiator, resulting in degradation of the heating functionality of the radiator. Therefore, in CO2 A/C systems, a temperature sensor is needed, which can accurately measure temperature around both 00C and 1500C. Furthermore, it is desired that in both temperature areas the temperature sensitivity of the output of the sensor is linear and in the same range.
Summary of the invention
It is an object of the present invention to provide an improved arrangement for linearizing non-linear sensor outputs.
It is another object of the invention to provide a high accuracy, wide-range linearization arrangement for non-linear sensor outputs that allows to have one temperature sensor to measure accurately low and high temperatures allowing a CO2 AJC system to include both cooling and heat pump functionality.
It is still another object of the invention to provide a linearizing arrangement that provides a digital output signal having a response curve which slope is substantially constant and without discontinuities across the whole range of the phenomenon.
An arrangement is described for linearizing a non-linear sensor whose output signal follows a non-linear curve in response to a linear change in a monitored phenomenon in a predefined range. The non-linear curve can be approximated by an exponential or logarithmic function. The arrangement includes an analog conversion unit to convert the signal from the non- linear sensor to obtain a converted signal, wherein the conversion has a substantially logarithmical or exponential transfer characteristic and the converted signal has a characteristic that follows a linearized curve in response to the linear change in the phenomenon in the predefined range, wherein the linearized curve can be approximated by a linear function. Furthermore, the conversion unit comprises a pn-junction and the conversion circuit is arranged to measure the current through the non- linear sensor and to exchange said current into a current through the pn-junction. The current through the pn-junction has a linear relationship with the current through the non-linear sensor and the converted signal is the voltage across the pn-junction. In prior art analog linearization circuits cited above, a log amplifier is used to linearize the non- linear voltage signal representative of the temperature. An log amplifier comprises at least an operational amplifier, resistor and a transistor. A transistor comprises at least two pn-junctions. According to the invention at minimum one pn-junction is needed in the conversion unit to perform the linearization of the non- linear sensor signal. The current/voltage characteristic of a pn- junction is used to convert the current through the non-linear sensor which follows an exponential curve into a voltage across the pn-junction. This conversion is a logarithmic function. In this way the exponential transfer curve of the current through
the non-linear sensor is transformed into a voltage signal having a transfer curve which is more or less linear.
According to an embodiment, the arrangement further comprises an AD- converter having a sensor signal input to receive the converted signal and to digitize the converted signal to obtain a digital signal value, a first reference signal generator arranged to generate is first reference signal wherein the first reference signal corresponds to the voltage across a pn-junction. The pn-junction matches the pn- junction of the conversion unit. A first predefined current is fed through said pn- junction. The arrangement further comprises a second reference signal generator arranged to generate is second reference signal wherein the second reference signal corresponds to the voltage across a pn-junction. The pn-junction of the second reference signal generator matches the pn-junction of the conversion unit. A second predefined current is fed through said pn-junction. The AD-converter comprises a first reference signal input coupled to the first reference signal generator and a second reference signal input coupled to the second reference signal generator. The AD- converter is arranged to generate digital signal values corresponding to the voltage difference between the sensor signal input V
pn ntc and the first reference signal input Vpni and the voltage difference between the second reference signal input V
pn2 and the first reference signal input V
pnl. It is commonly known that the voltage across a junction are dependent on the saturation current and the temperature of the junction. Furthermore, it is commonly known that the saturation current is temperature dependent. Normally, the ambient temperature can fluctuate. These fluctuations will result in fluctuations of the junction temperature making the converted signal dependent of the ambient temperature. These features provide two signals. One depending on the phenomenon measured by the sensor and the junction temperature of the pn-junction of the first reference signal generator and the other signal only depending on the junction temperature of the pn-junctions both the first and second reference signal generator. In both signals the influence of the saturation current is reduced significantly, making the signals less sensitive to temperature fluctuations within the conversion unit. In a further embodiment, the pn-junction of the converter and pn-junctions of the first and second reference signal generator are on the same die. These features are employed to reduce the temperature sensitivity of the saturation
current of the pn-junctions of the conversion unit in the digital signal values. Junctions on the same die will have substantially the same temperature. As a result of this, the dependency of the saturation current and its corresponding temperature dependency is substantially eliminated. According to an embodiment, the AD-converter is further arranged to generate digital signal values corresponding to the voltage difference between the first reference signal input V
pn2 and the sensor signal input V
pn ntc and the voltage difference between the first reference signal input V
pni and the second reference signal input V
pn2. These features enable us to eliminate drift in the differential AD-converter. According to an embodiment, the arrangement further comprises a processing unit arranged for processing the digital signal values to obtain a value for the phenomenon. In an embodiment, the processing unit is arranged to perform the following equation: r
ati
o
wherein ratio = a value indicative of the phenomenon,
Vpn me - Vpni = a digital signal value corresponding to the difference between the sensor signal input Vpn ntc and the first reference signal input Vpni ,and
VPn2 - Vpni = a digital signal value corresponding to the difference between the second reference signal input Vpn2 and the first reference signal input Vpni . These features are employed to eliminate the temperature dependency of the pn-junctions and to provide a digital value that is insensitive for temperature variation of the pn-junctions.
According to an embodiment, the processing unit is arranged to calculate a value for a digital output signal representing the sensed phenomenon by applying to the value of ratio to a polynomial equation in the digital domain to further linearize the curvature of the digital output signal in response to a linear change in the phenomenon in said predefined range. These features help to reduce further the non-linearity in the digital signal due to a mismatch between the exponential curve of the non- linear sensor and the logarithmic I-V function of the pn-junction. According to an embodiment, the non-linear sensor comprises a sensing element made of material which conductivity rises with increasing value for the phenomenon.
In a further embodiment, the non-linear sensor is an NTC thermistor. The present invention provides a solution to linearize the response of a non- linear enabling us to measure accurately temperature in a wide range i.e. -40 - 2000C with at least 3 to 5 bits per degree Celsius with inexpensive electronic circuitry, which could be made by a standardized process with standard components.
Brief description of the drawings
The present invention will be discussed in more detail below, using exemplary embodiments, with reference to the attached drawings, in which Fig. 1 shows a circuit diagram of an arrangement for linearizing a non- linear sensor according to the present invention;
Fig. 2 shows the non- linear characteristics of an NTC and a linearized signal Vpn_NTC; and
Fig. 3 shows the relation between the linearized signal Vpn_NTC and the temperature of the PN junction.
Detailed description of exemplary embodiments
Figure 1 shows a circuit diagram of an arrangement for linearizing a non- linear sensor RNTC- A non- linear sensor according to the present invention is a sensor that provides a signal that follows a non-linear curve in response to linear changes in a phenomenon. The non-linear sensor could be any transducer that responds to a phenomenon, such as temperature, pressure, air flow, etc, by changing an electrically- detectable characteristic. Such characteristics include thermocouple voltages and temperature-sensitive resistance variations. Non-linear sensors with a negative phenomenon coefficient comprise a sensing element made of a material which conductivity rises with increasing value for the phenomenon. Non-linear resistive sensors can be used in temperature sensors, gas sensors and humidity sensors. In the following description the non-linear sensor is a resistor RNTC with negative temperature coefficient (NTC). In the embodiment shown in figure 1, the voltage across the RNTC is kept stable by means of an operational amplifier OPA. A reference voltage Vref is supplied to a non- inverting input of the operational amplifier OPA. The output signal of the
operational amplifier will control a FET 112 in such a way, that the voltage at the inverting input of the operational amplifier will be similar to the reference voltage Vref at the non-inverting input of the operational amplifier OPA. In this way, the voltage across the NTC will kept constant. The current through RNTC is feed through a diode DNTC- Figure 2 shows the relation of the current through RNTC and the temperature and the voltage across diode DNTC when the current through RNTC is fed through the diode DNTC- The X-axis indicates the temperature, the left Y-axis indicates the current values and the right Y- axis indicates the voltage across the diode DNTC- It can be seen that the current is not sensitive at low temperature and very sensitive at high temperature.
The voltage across the diode can be approximated by means of the following pn- junction equation:
wherein: Vp
n = voltage across the pn-junction, k = Boltzmann's constant q = the elementary charge
T = the absolute temperature of the pn-junction in kelvins I = pn-junction current and Is = the saturation current.
The operational amplifier OPA, FET 112 and diode DNTC form an analog conversion unit 101. The conversion unit 101 is arranged to measure the current through the non linear sensor, by means of an input formed by the connection to the NTC and to exchange said current received at said input into a current through the pn- junction of the diode DNTC- In the embodiment shown in figure 1, the current through the pn-junction is similar to the current through the non linear sensor RNTC- By means of a current-mirror, not shown and well known in the art, the current through the sensor can easily be multiplied by a predefined factor to obtain another current having a linear relationship with the current through the sensor. In this way, the current to be fed through the diode DNTC can be adapted into suitable current values.
The diode DNTC will convert in the analog signal domain the temperature characteristic of the current through RNTC into a converted signal which corresponds to the voltage Vpn ntc across the pn-junction of the diode. By means of equation [1] a logarithmic conversion is applied on the current through the NTC. Figure 2 shows the result of the logarithmic conversion by means of equation [1] of the current through the non-linear sensor into the voltage across the pn-junction. The temperature characteristic of the voltage across DNTC is linearized with respect to the non-linear characteristic of the current though RNTC-
The voltage across DNTC could be converted by means of an AD converter to obtain a digital signal value representative of the temperature. This voltage can easily be digitized by a simple 11 or 12 bit AD converter while still meeting the required resolution of 3 - 5 bits per degree Celsius along the whole temperature range. The digital samples could be processed further to improve the linearization. By means of the conversion unit 101, the NTC current is amplified logarithmically opposite to its own nonlinear curve in the analog domain. This results in an analog signal that is close to linear and that can be digitized by a linear AD converter with much higher resolution per degree Celsius than the original non-linear signal having a linear relationship with the NTC current. The digitized signal obtained from the voltage Vpn ntc could be used to measure temperature with sufficient accuracy in a large temperature range if the temperature of the pn-junction could be kept constant.
However, the logarithmic conversion by means of equation [1] comprises two temperature dependent parameters i.e. the absolute temperature of the pn-junction in Kelvin T and the saturation current Is. The values of both parameters depends on the temperature of the die on which the diode is present. In operational conditions, the temperature of the die has a relationship with the environmental temperature. In a car, the environmental temperature of the sensor could be in a range of -20 - 300C when starting the engine and could rise up to a range of 100 - 1600C when the engine is warmed up. Figure 3 shows the relation of the Vpn ntc curve shown in figure 2 with respect to the temperature of the die. The X-axis indicates the temperature, the Y-axis indicates the voltage across the diode DNTC- Shown are the curves that can be obtained when the temperature of the die Tdie, i.e. pn-junction of diode, is kept at a defined
temperature. The amplitude of the curve shifts over the whole temperature range in dependence of the die temperature Tdie.
To eliminate the temperature dependency due to the saturation current Is, a first reference signal generator 105 is provided. In figure 1, the first reference signal generator 105 comprises a serial connection of a current source providing a predefined current IR^ and a diode Dl . The predefined current lRefi provided by the current source is independent from the temperature variations and provides at each temperature in a predefined range of for example -40 - 2000C substantially the same current. The predefined current Irefi is fed through the diode Dl. The anode of the diode Dl is coupled to an input of a differential AD converter
104 to measure the voltage potential at the anode. In figure 1, the anode of diode DNTC is coupled to the other input of the differential AD converter 104. Furthermore, the cathode of both diodes DNTC and Dl are coupled to ground. The differential AD converter 104 measures the voltage difference between the voltage across diode DNTC being the sensor signal input Vpn ntc and the voltage across diodeDl being a first reference signal input Vpni .
The diode Dl has a pn-junction which matches the pn-junction of diode DNTC such that the saturation currents of both pn-junctions have the same temperature dependency. By subtracting the Vpn voltages across the matched diodes DNTC and Dl, the temperature dependency due to the saturation current is removed. This can be proved by the following equation:
Vpn ntc = voltage across pn-junction Dntc,
Vpni = voltage across pn-junction Dl,
Intc = current through pn-junction Dntc, and
Irefi = current through pn-junction Dl .
From equation [2] can be seen that the temperature dependency due to the saturation current Is is removed and the temperature dependent parameter T still exists. By measuring the temperature of the die accurately, the variations of the voltage across the diodes due to parameter T could be compensated. According to the present invention, temperature dependent parameter T could be eliminated by providing a second reference signal generator 103. In figure 1, the second reference signal generator 103 comprises a serial connection of a current source providing a second predefined current IReβ and a diode D2. The second predefined current IReβ provided by the current source should be independent from temperature variations and provides at each temperature in a predefined range of for example -40 - 2000C substantially the same current. The predefined current Iree is fed through the diode D2.
The anode of the diode D2 is coupled to an input of differential AD converter 104 to measure the voltage potential at the anode. Furthermore, the cathode of diode D2 is coupled to ground. The other input of differential AD converter 104 is coupled to the anode of diode Dl. Differential AD converter 104 measures the voltage difference between the voltage across diode D2 being the second reference signal input Vpn2 and the voltage across diode Dl being a first reference signal input Vpnl.
The diode D2 has a pn-junction which matches the pn-junctions of diode DNTC and Dl such that the saturation currents of the pn-junctions have the same temperature dependency. The pn-junction of the converter 101, i.e. pn-junction of diode DNTC, and the pn-junction of the first and second reference signal generator 105, 103 are preferably on the same die. In that case, the temperature of the pn-junctions will be substantially similar.
By subtracting the V
pn voltages across the matched diodes D2 and Dl, the temperature dependency due to the saturation current is eliminated from the difference signal as proved by equation [3].
wherein:
VPn2 = voltage across pn-junction D2, Vpni = voltage across pn-junction Dl,
IreG = current through pn-junction D2, and
Irefi = current through pn-junction Dl .
The die temperature Tdie can easily be determined from the voltage difference (V
Pn2- V
pni) as shown in equation [4]: Λ
wherein:
VPn2 = voltage across pn-junction D2,
Vpni = voltage across pn-junction Dl,
IreG = current through pn-junction D2, and Irefi = current through pn-junction D 1.
Consequently the die temperature can be determined by dividing the measured voltage difference with a constant.
Both voltage differences (Vpn ntc- Vpni) and (Vpn2- Vpni) are digitized by an AD converter 104 and supplied to a processing unit 110 for further processing. The AD converter 104 could be a 16 bit AD-converter with 14 effective bits. A 16 bit AD- converter 104 is sufficient to digitize the voltage difference along the whole temperature and to provide enough resolution with respect to temperature. The processing unit 110 is arranged to process the digital signal values supplied from the
AD-converter 104 and to calculate a value for a digital output signal representing the temperature measure by the non- linear sensor R
NTC- The processing unit 110 performs the following equation: [5]
wherein ratio = a value indicative of the temperature Vp
n ntc - Vpni = a digital signal value corresponding to the difference between the between the sensor signal input V
pn ntc and the first reference signal input V
pn2 ,and V
Pn2 - Vpni = a digital signal value corresponding to the difference between the second reference signal input V
pn2 and the first reference signal input V
pni .
By applying the equations [2] and [3] in equation [5] we arrive at equation [6]:
It can be seen from equation [6] that the value of ratio depends only on the temperature dependent variable current INTC through the sensor and further the fixed currents Irefi and Iree, which are temperature independent. The value of ratio is a dimensionless number which only catches the NTC current, wherein the die temperature dependent parameters are completely removed. The value of ratio is indicative of the measured temperature and follows a curve similar to the curve of Vpn NTC shown in figure 2. However, contrary to the curve of Vpn NTC the curve of ratio is independent of the temperature of the pn-junctions of the circuitry. The curve of ratio, as shown in figure 2, has a non-linearity of 10 - 15%, which could be sufficient for some applications.
The curve of ratio could further be linearized by applying a polynomial equation to the calculated value of ratio. By applying a polynomial equation, a digital output signal can be calculated which sensitivity for a linear temperature change along the whole range of the measuring range is substantially constant. By means of a fourth order polynomial temp = ax4+ bx3+ cx2+ dx!+ e, wherein temp is a value indicative of the temperature, a, b, c, d and e are predetermined constant values and x corresponds to the value of ratio, a value for temp can be calculated with an accuracy smaller than 1°C over the whole temperature range from -400C up to 2000C. Preferably, the calculated value for temp is in a digital format which corresponds to the measured temperature. If so, the arrangement for linearizing a non-linear sensor according to the invention can easily connected to a display unit to show the actual temperature or other phenomenon without further processing the outputted data.
The voltages across the diodes DNτc, Dl and D2 could be digitized by three different AD converters. In figure 1 only one differential AD converter 104 is provided to digitize the voltage differences (Vpn _nte- Vpnl) and (Vpn2- Vpnl). To enable this, a selector 106 is provided which switches under control of a control signal generated by a control signal generation unit 108 between the voltage across diode DNTC and diode D2. Furthermore, a chopper 102 is provided at the input of the AD converter 104. The chopper interchanges under control of a control signal generated by the control signal
generation unit the input signal supplied to the AD converter 104. In this way a digital value for (Vpn _ntc- Vpnl) and (Vpnl - Vpn ntc) is made available at the output of the AD converter 104. A bias drift in the digital signal values at the output of the AD converter can be eliminated in the processing unit 110 by determining the average value of (Vpn ntc- Vpni) and -(Vpnl - Vpn ntc) in processing unit 110. The same method will be applied for determining the digital values for (Vpn2- Vpnl). The control signal generation unit 108 is arranged to generated appropriate control signals to the selector 106, chopper 102, AD converter 104 and processing unit 110, to synchronize the respective components. It should be noted that equation [5] can be executed with sufficient accuracy when the analog values of the voltage differences (Vpn ntc- Vpni) and (Vpn2- Vpnl) can be sampled with sufficient accuracy and the digitized value for (Vpn2- Vpnl) corresponds to a value which allows the processing unit 110 to perform an accurate division. It has been found that the current through a standard thermistor RNTC is in a range from 50OnA up to 3.2mA when IV is applied across the thermistor. The first predefined current Irefi is preferably in the lower part of said range and the second predefined current Iree is preferably in the upper part of said range. In an embodiment the first reference current Irefi is 50μA and the second reference current Iree is 0,5mA. Furthermore, the pn-junctions present in the conversion unit 101 and the first and second reference signal generators 105, 103 could be made by standard CMOS Mixed Mode processes.
The arrangement according to the invention can be applied in sensors for measuring one phenomenon. The arrangement could also be applied in sensors for measuring two or more phenomena. An example of such an sensor is a combined temperature pressure sensor.
Furthermore, in the embodiment of the invention described above, the exponential function of the NTC over the temperature range -40 - 2000C is linearized by the logarithmic conversion provided by the pn-junction and which is described by equation [I]. A person skilled in art would know how to adapt the conversion unit of arrangement to convert a signal having a characteristic that follows a curve in response to a linear change in a phenomenon in a predefined range, wherein the curve can be approximated by a logarithmic function in to a conversion unit that converts
exponentially the logarithmic signal from the non-linear sensor to obtain a linearized converted signal that can be digitized by a linear AD-converter with sufficient resolution.
Several embodiments of the invention have been described above by way of exemplary embodiments. Various modifications and variations for the elements described with respect of these embodiments may be made by the skilled person without departing from the scope of the present invention, which is defined by the appended claims.