Patent Application
20100169050
This invention relates to systems that their response to a stimulus approaches equilibrium exponentially and devices that measure their responses. Specifically, the invention relates to temperature sensors and thermal mass flow meters.
A temperature sensor does not provide an immediate output corresponding to a change in its input. The response of such a sensor to a sudden change or a step input gradually approaches a final equilibrium value. This slow response is due to thermal inertia and varies exponentially relative to time.
A slow sensor may cause instability and reduced accuracy in a control system. By the time the sensor signals the control element to respond to a change in a parameter, the parameter may have already changed to a new value. In other words, the control system lags the change in the parameter. Larger lag is equivalent to less accurate control or instability.
In temperature control systems the accuracy of maintaining a specific temperature depends on the response time of the temperature sensor.
Thermal mass flow meters and controllers use heat transfer and change in temperature of a sensor system as the measuring means for the flow. U.S. Pat. Nos. 4,464,932, 4,984,460 and 5,461,913 are examples of these meters and controllers. Therefore, the measured flow has the same exponential profile as the response of temperature sensor. In a similar way, the accuracy and responsiveness of a flow controller using thermal mass flow meter depends on the response time of the temperature sensor.
A sensor with faster response allows better assessment and control of parameters and processes.
The objects of the invention include:
Essentially eliminating the delay in measurement of a parameter such as temperature of a system by a sensor due exponential response of the sensor. In other words, instantaneously providing the actual value of the parameter or the final value of the sensor response.
Instantaneously providing the final value of a system parameter such as temperature varying exponentially relative to time.
Improving the accuracy of temperature measurement and control systems beyond the existing limits.
Improving the accuracy of thermal mass flow meters and controllers beyond the existing limits.
The invention provides a system that essentially eliminates the delay in measurement of temperature by a sensor due to exponential response of the sensor. The system finds the rate of change of the sensor signal, multiplies it by the time constant of the sensor signal and adds the result to the instantaneous value of the sensor signal to predict the final value of the sensor signal or the actual value of the temperature. When the sensor is permanently attached to an object, the final temperature of the object is predicted in a similar manner using the combined time constant value of the sensor and the object. The system can be software, analog hardware or digital hardware. The system also allows more accurate temperature control without undershoot or overshoot by providing a control signal proportional to the difference between a desired value and the predicted final value of the sensor signal. The system similarly eliminates the time delay in measurement and control of flow in thermal flow meters.
The theory behind the invention is the mathematical relation developed by the inventor for an exponential function. According to this mathematical relation, in an exponential function, the asymptotic value or the final value of the function can be predicted by adding the instantaneous value of the function to the product of the derivative of the function and its time constant.
Based on the above, according to the invention there is a device that receives a signal, which is varying exponentially with time. The device differentiates the input, multiplies the differentiated input by the time constant of the signal and adds the result to the value of the input signal and delivers the result as the output. As an example, the signal can be the voltage from a temperature sensor system.
Suppose v is an exponential function of time t with a final value of v∞ and time constant T.
v=v
∞(1−e−t/τ)
dv/dt=(v∞/τ)e−t/τ
v+τdv/dt=v
∞(1−e−t/τ)+τ(v∞/τ)e−t/τ
v+τdv/dt=v
∞
This relation is shown in
This is a general relation and holds true for all conditions. When the function v is decreasing, the value of τdv/dt is negative which reduces the instantaneous value of v to provide the lower final value v∞ of v.
As a more general case, when the initial value of v is not zero,
v=(v∞−vi)(1−e−t/τ)+vi
dv/dt=((v∞−vi)/τ)e−t/τ
v+τdv/dt=(v∞−vi)(1−e−t/τ)+vi+τ((v∞−vi)/τ)e−t/τ
v+τdv/dt=v
∞
The general method of the invention is shown in a block diagram in
As time passes by the slope of the function v approaches zero, and therefore, the product τdv/dt approaches zero. Thus, any inaccuracy is diminished upon time and the predicted value will be same as the actual final value.
The method described above can be performed by an analog circuit, a digital circuit or computer software.
The product CR1can be chosen to represent τ, the time constant of the sensor. Thus, the multiplication step of the method is performed concurrently at the differentiation stage.
Differentiating amplifier 16 functions in inverting mode and has negative output. For addition stage of the method, input signal v is also inverted to −v by the inverting amplifier 19. Signal v is applied to the inverting terminal of amplifier 19 through resistor 20. Feedback resistor 21 connects output of amplifier 19 to its inverting input. Resistors 20 and 21 are equal. The non-inverting terminal of amplifier 19 is grounded.
The inverted outputs of amplifiers 16 and 19 are added together by the inverting amplifier 22 to obtain the sum of v and CR1dv/dt as the final value of sensor signal or v∞.
v+CR
1
dv/dt=V
∞
Outputs of amplifiers 16 and 19 are applied to the inverting terminal of amplifier 22 through resistors 23 and 24 respectively. Feedback resistor 25 connects the output of amplifier 22 to its inverting terminal. Resistors 23, 24 and 25 are equal. The non-inverting terminal of amplifier 22 is grounded.
Resistor R1 is preferably variable to allow for adjustment or calibration of the circuit to match the time constant of the sensor. All other resistors can have the same value R2.
As time passes by, the sensor signal slope approaches zero, and therefore, the product CR1dv/dt approaches zero. Thus, any probable error is diminished upon time and the predicted value will be same as direct sensor value.
Calibration of the system is best performed by exposing the sensor to a known condition such as a piece with a fixed temperature. Resistor 18 is adjusted such that the output of the system represents the known value. Similarly, in a digital or software system, a value is entered as time constant τ such that the output of the system represents the known value.
For automatic calibration, the system is provided with additional hardware or software to record sensor signal and its derivative at a specific time. When the slope of the sensor signal approaches zero, the system records the sensor signal as the final value v∞. The time constant is then calculated by the following formula and stored in the system.
v+τdv/dt=v
∞
When the sensor is permanently attached to an object, the final temperature of the object is predicted in a similar manner using the combined time constant value of the sensor and the object.
The value of τ depends on thermal inertia and coupling. However, in many cases it remains essentially invariant for the application of the invention. Examples include a sensor probe touching a solid surface for measuring the temperature of a solid body, a sensor immersed in fluid inside a container and a sensor attached permanently to a body of material with fixed weight and boundary conditions.
The possibility of predicting the final value allows improving control accuracy and stability by one or two orders of magnitudes. In a conventional control system, the set value is compared with the instantaneous value of the parameter and the control signal is proportional to their difference. In a control system using the principle of the invention, the set value is compared to the predicted final value. Thus, the system is controlled by the essentially exact value of control signal. There would be no overshoot or undershoot as in a conventional control system.
The block diagram of a control system based on the principle of the invention is shown in
v
c
=v
set
−v
∞
To reach the desired value vset faster, a proportional system may be used until the instantaneous value reaches a given percentage of the desired value and then the predicted final value is used for precise and smooth approach to the desired value.
The method described above can be performed by an analog circuit, a digital circuit or computer software.
An analog circuit for the control system of
v
c
=v
set
−v∞
at the output 32 of amplifier 23.
The differencing amplifier circuit built around amplifier 23 is a typical example. Different variations are possible. For example an instrumentation amplifier may be used for differencing without the need for resistors 27, 29, 30, 31.
The principle and different embodiments of the invention can be similarly used in thermal mass flow meters and controllers and any system that measures or controls an exponentially varying parameter.
