WO 2012/166585 PCT/US2012/039539 1 Re-Calibration of AB NDIR Gas Sensors Field of the Invention The present invention is in the field of measuring instruments, and 5 specifically relates to re-calibrating non-dispersive infrared (NDIR) gas sensors whose outputs have drifted over time and no longer correctly reflect their measurement accuracy. Background of the Invention 10 Output stability or drift over time leading to measurement inaccuracies has long been a major deficiency for gas sensors irrespective of what technology or methodology is used for their conception or realization. Output software correction may alleviate the problem somewhat but it is in many instances inaccurate and not even always applicable. Software correction has proven to be 15 somewhat successful so far only to NDIR C02 gas sensors used in Demand Control Ventilation application to save energy in the HVAC&R industry. It has long been the objective of many researchers in this field to overcome this problem fundamentally and for good SUMMARY OF THE INVENTION 20 An apparatus and method of using a dual-beam non-dispersive infrared (NDIR) gas sensor calculates a gas concentration ("P") of a sample gas in a sample chamber through the use of a calibration curve P = F(x). F is a polynomial function of x = G/Go where G is the ratio of the signal channel output ("Vs") of a dual beam NDIR gas sensor divided by the reference channel output ("VR") or G = 25 VsNR and Go being the value of G when there is no sample gas present in the sample chamber. The dual-beam NDIR gas sensor is of an Absorption Biased ("AB") designed type that uses an identical spectral narrow band pass filter for wavelength selection for both the Signal channel and the Reference channel, An absorption bias is applied to the Signal channel by making its path length longer 30 than that of the Reference channel, The dual-beam NDIR gas sensor has no WO 2012/166585 PCT/US2012/039539 2 moving parts for effecting the interposition of spectral filters, an absorbing cell or a non-absorbing cell to create both the Signal channel and the Reference channel. A re-calibration method is described in which the output P of an AB designed NDIR gas sensor is compared to a second gas concentration of the 5 sample gas P, determined by a Calibration Master, itself an AB designed NDIR gas sensor, in the close environ of the sensor to be re-calibrated. If the difference between P and Pc exceeds a preselected threshold, the calibration curve of the sensor can be adjusted by adjusting its Go value based upon a reversed calibration curve algorithm. The reversed calibration curve algorithm first 10 expresses P = F(x) reversely as x = F 1 (P) and calculates xc = F'(Po) for the correct value of Pc as determined by the Calibration Master. The algorithm then determines a new adjusted Go or GON, such that GOaN = G/xc = Go F- 1 (P)/xa where G = GoF'(P), the non-normalized ratio of VsNR for gas concentration P as measured by the sensor to be recalibrated. 15 The recalibration method can use a master NDIR gas sensor which itself is an AB designed NDIR gas sensor which obtains an air sample from a close environ air space proximate the sample chamber of the NDIR gas sensor being recalibrated through the use of an air sampler. An AB designed gas sensor can also be self-recalibrated by using a stored 20 standard gamma ratio and a measured standard gamma ratio and a self calibration algorithm to correct the calibration curve for a difference between the stored standard gamma ratio and the measured standard gamma ratio when their difference exceeds a preselected threshold. The stored standard gamma ratio and the measured standard gamma ratio are obtained at different points of time, 25 the standard gamma ratio being the ratio of signal to reference outputs from a standard signal detector located in the Signal channel and a standard reference detector located in the Reference channel. The standard signal and reference detectors are equipped with an identical narrow band pass filter with the same Center Wavelength ("CWL") and Full Width Half Maximum (FWHM) neutral to the 30 absorption of the gas of interest. Exactly like Go, the standard gamma is independent of the amount of gas of interest present in the sample chamber of the sensor and its value changes only when changes in the sensor components are WO 2012/166585 PCT/US2012/039539 3 detected. The standard gamma can therefore be used proportionally to correct for the changes in the value of Go, thereby readjusting the calibration curve and rendering the sensor to be self-calibrating over time. Accordingly, it is a primary object of the present invention to provide 5 improved NDIR gas sensors that are easily recalibrated to prevent output drift over time. This and further objects and advantages will be apparent to those skilled in the art in connection with the drawings and the detailed description of the invention set forth below. 10 BRIEF DESCRIPTION OF THE DRAWINGS Figure 1 shows the optical component layout for the Absorption Biased methodology for NDIR gas sensors. Figure 2 shows respectively the output curves for the Reference and Signal channel detectors as a function of CO 2 in the sensor sample chamber. 15 Figure 3 shows the ratio of the output of the Signal channel detector over the Reference channel detector output at sensor block temperature Br as a function of CO 2 in the sensor sample chamber Figure 4 shows the normalized ratio of the output of the Signal channel detector over the output of the Reference channel detector at sensor block 20 temperature BT as a function of CO 2 in the sensor sample chamber. Figure 5 depicts the sensor calibration curve expressed the CO 2 concentration in the sample chamber for the Absorption Biased (AB) NDIR gas sensing methodology as a third order polynomial of the normalized ratio of signal output/reference output. 25 Figure 6 depicts the sensor reverse calibration curve expressed the normalized ratio of signal output/reference output for the Absorption Biased (AB) NDIR gas sensing methodology as a third order polynomial of the CO 2 concentration in the sample chamber. Figure 7 portrays a typical scenario wherein an Absorption Biased (AB) 30 designed NDIR gas sensor is being recalibrated by a Calibration Master using the Effortless Recalibration (ERC) technique without the use of an air sampler.
WO 2012/166585 PCT/US2012/039539 4 Figure 8 portrays a typical scenario wherein an Absorption Biased (AB) designed NDIR gas sensor is being recalibrated by a Calibration Master using the Effortless Recalibration (ERC) technique with the use of an air sampler. Figure 9 depicts the component layout and construct of a specially 5 designed air sampler guaranteeing at all times the accuracy of using the ERC technique to recalibrate an AB designed NDIR gas sensor with a Calibration Master. Figure 10 shows the details of an air-tight telescopic tube which is part of the specially designed air sampler, 10 Figure 11 depicts the optical component layout for a self-commissioning Absorption Biased NDIR gas sensor. DETAILED DESCRIPTION OF THE INVENTION The present invention only applies to NDIR gas sensors and not to other 15 technology types of gas sensors. The present invention builds upon the inventor's earlier disclosure of an Absorption Biased (AB) methodology for NDIR gas sensors set forth in U.S. Patent No. 8,143,581, the disclosure of which is specifically incorporated herein by reference. This AB methodology can be reviewed briefly as follows. First of all, this methodology is based upon a 20 conventional Double Beam Configuration Design for NDIR gas sensors. Two channels or beams are set up, one labeled Signal and the other Reference. Both channels share a common infrared source but have different detectors, each of which is equipped with the same or identical narrow band-pass filter used to spectrally define and detect the target gas of interest. Both detectors for the two 25 channels share the same thermal platform with each other and also with the sample chamber and the common infrared source mount for the sensor. An absorption bias is deliberately established between the Signal and Reference channels by having the sample chamber path length longer for the Signal channel than that for the Reference channel. By so doing, the detector output of the 30 Reference channel is always greater than that of the Signal channel when there is target gas present in the sample chamber. This is due to the fact that there is WO 2012/166585 PCT/US2012/039539 5 more absorption taken place in the Signal channel because of its longer sample chamber path length. By applying this absorption bias between the Signal and Reference channels, one is able to calibrate the sensor even when both channel detectors have the same and identical narrow band-pass filters. 5 Figure 1 shows the optical component layout for the Absorption Biased methodology for NDIR gas sensors. As shown in Figure 1, both the signal channel detector 1 and the reference channel detector 2 are entrapped with 100% dry nitrogen 3 and have the same narrow band-pass spectral filter 4 which is used to detect the gas of interest in the sample chamber 5. As an example, the filter 10 designed to be used for the detection of C02 gas will have a center wavelength (CWL) = 4.
2 6p and a full width half maximum (FWHM) = 0.1 4 p. Notice that both detectors 1 and 2 are thermally connected to the entire sensor body 6 through their respective waveguides 7 and 8 and consequently they always share the same thermal platform with each other. In other words, the entire sensor body 6, 15 which is in essence a composite of aluminum parts comprising the infrared source mount 9, sample chamber 5 and the waveguides 7 and 8, respectively, for the signal and reference channels, provides an excellent common thermal platform for detectors 1 and 2. As shown in Figure 1, the sample chamber path length LR, 10, associated 20 with the reference channel is approximately one-half of the sample chamber path length Ls, 11, associated with the signal channel. A common infrared source 12 is used to illuminate both the signal and the reference channels. The output of detector 1 for the signal channel is always less than that of the detector 2 for the reference channel irrespective whether or not there is any amount of the gas of 25 interest in the sample chamber 5. The respective detector outputs can be determined by using the well-known Beer-Lambert Absorption Law for the particular gas of interest, the designed characteristics for the narrow band-pass filter 4 and the physical dimensions of LR 10 and Ls 11. Following the conventional NDIR Double Beam design, it is always the ratio 30 value of the Signal channel detector output over the Reference channel detector output that is used to process the different gas concentrations present in the sample chamber. The Absorption Biased (AB) methodology for NDIR gas WO 2012/166585 PCT/US2012/039539 6 sensors recognized the significance of this zero target gas ratio called "Gamma0 (Go)" that is unrelated to the Physics of this gas measurement technique because there is no gas absorption taken place. By normalizing the ratio of the outputs for the Signal and Reference channels with Go and plotting this normalized ratio value 5 as a function of the target gas concentration in the sample chamber to obtain the calibration curve, one is in essence separating the invariant Physics treatment of the NDIR gas sensing principle from the other inevitably changing components treatment of the sensor over time. In other words, any changes in the calibration curve for an AB designed NDIR gas sensor will only be reflected in the changing 10 value of Go over time. It will not be reflected in the Physics measurement principle of such an NDIR gas sensor, which is supposed to always remain invariant If the output of the infrared source for any NDIR gas sensor is changing spectrally over time due to whatever reason, and it is delivered to the Signal and Reference channel detectors, and these detectors have different spectral narrow band-pass 15 filters, this changing spectral output of the source will destroy the invariance of the absorption Physics treatment for the sensor. This is because the ratio of the two channels at the very beginning establishes spectrally the absorption Physics for the gas measurement based upon the spectral output of the source. Such is actually the case for non-AB designed Double Beam NDIR gas sensors since the 20 Signal and the Reference channel detectors, unlike the AB-designed gas sensors, each has its own and different spectral narrow band-pass filters instead of identical ones. Figure 2 shows the graph 13 depicting the output VR(BT) of the reference channel detector 2 as a function of C02 concentrations in the sample chamber 5. 25 Graph 14 of Figure 2 shows the output Vs(BT) of the signal channel detector 1 as a function of C02 concentrations in the same sample chamber. Note that both outputs of the detectors are individually a function of the sensor block temperature BT, which is linked to ambient temperature T wherein the sensor is located. Since the signal channel path length is longer than that for the reference channel, Vs(Br) 30 changes more than VR(BT) for any amount of C02 in the sample chamber 5. An NDIR C02 gas sensor implementing the Absorption Biased methodology processes the values for the ratio G (BT) = VS(BT) I VR(BT) as a function of C02 WO 2012/166585 PCT/US2012/039539 7 concentrations in the sample chamber 5. Such a functional relationship between G(BT) and the C02 concentrations in sample chamber 5 is the de facto calibration curve for the sensor as depicted by graph 15 in Figure 3 for a particular sensor block temperature BT. Note that the value of G(BT) depends on sensor block 5 temperature BT and BT must therefore be kept unchanged during calibration for the sensor when concentrations of C02 are made to vary in sample chamber 5 in order to obtain corresponding G(BT) values. It is most important to note that the value of G(BT), other than being dependent upon the value of C02 concentration in the sample chamber of the 10 sensor and its block temperature BT, is invariant over time since both the signal and reference channels of the sensor have similar detectors with identical spectral filters and share the same thermal platform at BT. As a matter of fact, at any BT, the value of G(BT) is governed only by the NDIR gas absorption Physics for a particular gas of interest and is therefore invariant over time. However, while this 15 is indeed true in theory, it is not quite exact in reality. This is because the components of the sensor will not be time invariant and their performance characteristics can and will inevitably change over time. For example, a sagging filament for the aging light bulb resulting in an output radiation pattern change or the responsivity of the signal channel detector changes differently over time from 20 that of the reference channel detector, these changes are not related to any spectral changes of the source that are immune to causing any adverse effects to the calibration curve for the sensor implementing the Absorption Biased methodology. But when any of these component characteristics changes, they will affect the value of G(BT) and the calibration curve for the sensor will change 25 resulting in output drifts for the sensor over time. The Absorption Biased methodology recognizes two distinct domains that constitute the sensor's realistic calibration curve. The first is the invariant NDIR gas absorption Physics domain discussed before and the second is the variant sensor component characteristics domain discussed below. As shown before, the 30 invariant NDIR gas absorption Physics domain is represented by a functional relationship between G(BT) = Vs(BT) / VR(BT) and the concentrations of the gas of interest (e g. C02) in the sensor's sample chamber. The variant sensor WO 2012/166585 PCT/US2012/039539 8 component characteristics domain is represented by value of G(BT) when there is no gas of interest present in the sensor's sample chamber or GQ(BT) = Vs(BT) / VR(BT) ... 0 concentration of gas of interest in sensor sample chamber 5 Note that in this case the role of any NDIR gas absorption Physics for the gas of interest is eliminated since no gas is involved leaving Go(Br) strictly dependent only upon the sensor component characteristics. By normalizing G(BT) with Go(BT) to form x(BT) = G(Br) / Go(BT) and plotting the gas concentration (e.g. in ppm) as a function of x(BT), one combines 10 the two domains together to formulate the realistic calibration curve for the sensor as P (ppm) = PX [x(BT)] = PX [G(Br)/Go(BT)] (1) By plugging in the value of x(BT) into the function PX, one can get CO 2 concentration in ppm. Graph 16 of Figure 4 shows the de facto calibration curve 15 for the sensor linking the value of X(BT) to the gas concentration (in this case C02) in the sample chamber. Note the value of x(BT) starts off with unity when there is zero concentration of the gas (CO 2 ) in the sample chamber. The function PX[x(BT)] can be expressed as a polynomial of x(Br) to the nth order (e.g. n = 3 or the third order as depicted by graph 17 in Figure 5). Conversely, the same plotted 20 data can also be used to generate the inverse de facto calibration curve for the sensor or XP[P(ppm)] linking C02 gas concentration in the sample chamber P(ppm) to the value of x(Br). By plugging in the value of P(ppm) into the function XP, one can get the value of x(BT) or x(BT) = XP [P(ppm)] (2) 25 XP[P(ppm)] can also be expressed as a third order polynomial of P(ppm) as depicted in graph 18 of Figure 6, As stated earlier, at a particular BT of the sensor, the value of G(BT) is invariant as far as the gas absorption Physics is concerned. But since GO(BT) is also dependent upon BT, the calibration curve as shown in Equation (1) above for the sensor combining both the invariant Physics 30 domain and the variant sensor components domain is valid only if G(BT) and Go(BT) are measured at the same temperature of BT. As a matter of fact, G(BT) can be determined at any temperature BT as long as Go(BT) is also determined at WO 2012/166585 PCT/US2012/039539 9 the same temperature for determining x(Br). Because of this fact, we must determine Go(BT) as a function of BT or Go(BT) = Q (B) (3) where the function Q(Br) expresses the behavior of GO(BT) as a function of BT. 5 Now for the sensor to make a gas measurement, one first notes the sensor block temperature BT. One then measures G(BT) which is the ratio of the signal channel detector output over the reference channel detector output at Br. Using Equation (3) above to determine the value of Go(Br) at BT one then obtains the value of x(BT) = G(Br)/Go(Br). By plugging in the value of x(Br) into the 10 polynomial PX of Equation (1) above, one obtains the gas concentration P(ppm) in the sample chamber. Conversely, one can also plug a known P(ppm) of gas value into the polynomial of Equation 2 above to obtain the corresponding value for x(Br) at temperature Br. The formulation of the calibration curve in the NDIR Absorption Biased gas 15 sensing methodology by separating it into two distinct domains, one being invariant and the other variant, leads to a very significant advantage when the sensor needs to be re-commissioned or recalibrated. In this case one needs only to refresh the variant domain without having to deal with the invariant domain. Therefore in the calibration curve expressed earlier in Equation (1) as 20 P(ppm) PX [x(Br)] = PX [G(BT)/Go(BT) (1) only Go(BT) needs to be refreshed. Furthermore, one only needs 0 pprn gas or 100% dry nitrogen for the recalibration because the determination Go(BT) requires that there is zero concentration of gas in the sample chamber. But even the need for carrying a standard certified gas like 100% dry nitrogen in order to perform a 25 re-commissioning or recalibration task can still be very labor intensive and cumbersome, It would be extremely advantageous if no standard certified gas is needed at all for this purpose. This is achieved by the present invention's Re calibration methodology for Absorption Biased designed NDIR gas sensors. In this innovative technique, the gas concentration in the immediate 30 neighborhood or surrounding of the sensor to be re-commissioned or recalibrated will first be accurately determined by a "Calibration Master". Needless to say, this so-called "Calibration Master" is a gas sensor that must live up to its name as WO 2012/166585 PCT/US2012/039539 10 being able to measure accurately the gas concentration in the vicinity of the sensor to be re-commissioned or recalibrated. (The Calibration Master can be another gas sensor whose accuracy has been checked or re-calibrated prior to the time it is being used by its operator to make rounds checking multiple gas 5 sensors.) This information is then sent wirelessly via WiFi or via infrared under direct visual contact from the "Calibration Master" to the sensor in question. Using that information and a special algorithm within (described below), the sensor will know how to re-commission or recalibrate itself according to this information for the accurate gas concentration level of its environ that it receives from the 10 Calibration Master. In the present invention's Re-calibration methodology for Absorption Biased designed NDIR gas sensors, the calibration curve of an AB designed NDIR gas sensor is transformed into a curve that expresses the amount of the target gas present in the sample chamber, P(ppm), as an nth order polynomial of the 15 normalized ratio, x, of the Signal channel detector output over the Reference channel detector output. For a third order polynomial, which is plenty accurate for most applications, this calibration curve transformation can be quantitatively expressed in terms of P(ppm), x and Go as follows: P (ppm) = AO + Ajx + A 2 x 2 + A 3 x 3 (4) 20 Go = VSONRO (zero target gas in sample chamber) (5) x = (Vs/VR)/Go (6) where Vs and VR are respectively the Signal and Reference channel detector outputs when there is target gas in the sample chamber, Note that in this transformation of the calibration curve for the sensor, P (ppm) and Go of 25 Equations (4) and (5) above represent respectively the invariant Physics principle portion and the inevitably variant components portion of the methodology. But since the parameter x is a function of Go [see Equation (6)], when there is a change in the value for Go over time that is not corrected, x will be affected and the calibration curve for the sensor will change accordingly leading to sensor 30 output drifts. However, if for whatever reason the change in Go over time is known, the value of x can be corrected back to its proper value, and the original calibration curve for the sensor as represented by Equation (4) will still be valid.
WO 2012/166585 PCT/US2012/039539 11 Under this circumstance, no output drifts should be detected from the sensor and it will stay accurate over time. In order to achieve a simple, easy and inexpensive re-calibration methodology for AB designed NDIR gas sensors, the expression of P(ppm) as a 5 third order polynomial of x [see Equation (4) above] is reversed into one where x is expressed as a third order polynomial of P (ppm) without changing the value of Go as shown below x = B 0 + B, x P + B 2 x P 2 + B 3 xP 3 ; Go unchanged (7) All AB designed NDIR gas sensors manufactured with this re-calibration 10 methodology will carry both polynomials, namely Equation (4) and Equation (7) along with the Go value obtained during initial calibration in their Central Processing Unit (CPU) memory. Assume now that an NDIR gas sensor, e.g. C0 2 , is calibrated with a calibration curve characterized by a third order polynomial with coefficients (AO, 15 A,, A 2 , A 3 ) and Gamma0 = Go as shown in Equations (4) and (5). As time goes by we recognize that the sensor no longer accurately detects CO 2 and we wish to restore this sensor to its original accuracy or calibration curve. Since we do not want to use any gas standards such as 100% Nitrogen or a certified CO 2 concentration (e.g, 1,000 ppm) admixed with Nitrogen to achieve this, we must 20 however prepare an acceptable gas standard for this sensor in order that it can be recalibrated. An acceptable gas standard for this purpose could just be the concentration of the gas of interest (e.g. CO 2 ) that surrounds the sensor to be recalibrated. In order to do this, we need an accurate AB designed NDIR gas sensor acting as a Calibration Master to determine the concentration of the gas of 25 interest surrounding the sensor to be recalibrated. Furthermore, the Calibration Master must be sensing the same air sample in the air space surrounding the to be-recalibrated sensor. Since the air sample is never stationary but is quite dynamic with or without any air current in the vicinity of the relevant sensor, it is also very important that the to-be-recalibrated sensor and the Calibration Master 30 be sensing the same air sample and also during the same time period. The objective here is to make sure that both the sensor to be recalibrated and the WO 2012/166585 PCT/US2012/039539 12 Calibration Master sense or detect the same gas concentration value within the same place and within the same time period. The Calibration Master first sends a command to the relevant sensor to measure the concentration of the gas of interest in the immediate space 5 surrounding it for a certain time period, e.g. 120 seconds. At the same time the Calibration Master also commences to measure the same in the same air space and for the same time period itself via the use of an air sampler. At the end of the specified time period, the Calibration Master requests from the relevant sensor the measured concentration value for the gas of interest, Upon comparing the 10 received value with the one measured by itself and if the gas concentration values between the two are found to be within the expected accuracy specification (e.g. +/- 50 ppm), nothing else will be carried out by the Calibration Master indicating that the relevant sensor is accurate. However, if the compared values lie outside of the expected accuracy specification, the concentration value of the gas of 15 interest as measured by the Calibration Master will be sent to the relevant sensor and it will attempt to recalibrate itself automatically as outlined below. Using the correct gas concentration value Pc received from the Calibration Master, the relevant sensor first attempts to calculate the corresponding xc value using the stored reverse calibration curve ([Equation (7)], namely (B 1 , B 2 , B 3 , B 4 ). 20 In other words, xc = F 1 (Pe). For the value of P it measured during the same time period as the Calibration Master, which is incorrect, the corresponding normalized ratio x = F-(P) = G/Go is also obtained. Since G represents the absorption Physics portion of the calibration curve and is therefore invariant over time and clearly x # xe, the recalibration algorithm consists simply of determining a new 25 adjusted Go or GON, such that G/GON = xc. Expressed alternately GON can be rewritten as GON = G/x, = GoF(P)/xc = GoF 1
(P)/F
1 (Pe). Since the reversed calibration curve x = F(P) and Go are stored in the sensor to be recalibrated, once it receives the correct P, and uses its own measured P and Go values, the adjusted GON can readily be determined. 30 By carefully reviewing the above described procedures for the successful design of Absorption Biased (AB) NDIR gas sensors and the formulation of a convenient re-calibration technique for AB designed NDIR gas sensors without the WO 2012/166585 PCT/US2012/039539 13 need of standard gases, one might recognize that the key concept that makes them possible is the acknowledgement that the calibration curve for these sensors can be separated into two portions, one portion is based upon the NDIR gas measurement Physics which is invariant over time and the other portion is based 5 upon the inevitably variant components of the sensors that will change over time, Furthermore, if the sensor is not making any target gas measurement, i.e. when there is no target gas present in the sample chamber, the ratio of the Signal channel detector output (Vso) over the Reference channel detector output (VRO), which is designated as Go = Vso/VRO, belongs uniquely only to the variant 10 components portion of the calibration curve and will change as the component characteristics of the sensor inevitably change over time, for example from aging. By normalizing the ratio of the Signal channel detector output (Vs) over the Reference channel detector output (VR) by Go, designated as x = (VS/VR)/Go, one can combine the two portions of the calibration curve together to obtain the 15 complete calibration curve for the sensor. Recognizing the fact that it is only the Go for the sensor that can change over time, the re-calibration methodology for AB designed NDIR gas sensors is a procedure that works by updating the Go of the sensor to be re-calibrated. Now that we understand the theoretical Physics principle behind the validity 20 of what we now address as the "Effortless Re-Calibration" (ERC) technique specifically applicable only to Absorption Biased (AB) designed NDIR gas sensors, we will go into the procedural details and special equipment useful in order to carry out such a recalibration routine accurately all the time which is also an object of the current invention. 25 Figure 7 portrays a typical scenario wherein an Absorption Biased (AB) designed NDIR gas sensor 19, e.g. a C02 sensor, is to be recalibrated with a Calibration Master 20 held by an operator 21 using the Effortless Re-Calibration (ERC) technique described earlier. Operator 21 is standing just a few feet in front of sensor 19 which is hung roughly in the center and close to the top of a wall 22 30 which might typically be 20 ft. wide and 10 ft. tall. According to the teaching of the ERC technique, operator 21 using the Calibration Master measures the concentration of the gas of interest (e.g. C02) in the immediate environ of the WO 2012/166585 PCT/US2012/039539 14 sensor to be checked and/or recalibrated. Through the commands of the Calibration Master along with its own action, the concentration of the gas of interest surrounding the relevant sensor is determined by both it and the Calibration Master wirelessly within the same air space 23 and also within the 5 same time period. If operator 21 finds out that the gas concentration level as obtained from sensor 19 and that from Calibration Master 20 do not agree to within a predetermined accuracy specification, the operator 21 determines that sensor 19 needs recalibration. Operator 21 then uses the Calibration Master 20 to send its 10 measured gas concentration value to sensor 19 so that the latter can recalibrate itself according to the ERC procedure described earlier. Although the ERC maneuver to recalibrate sensor 19 just described is technically correct, it might not be very accurate at all. The reason is that the concentration of the gas of interest in the common environ of air space 23 (see 15 Figure 7) as measured by sensor 19 and the Calibration Master 20 might not always be the same. The basic assumption that the concentration level of the gas surrounding the sensor 19 is the same as that surrounding the Calibration Master 20 can only be true if there is no air flow of any kind in the air space 23 shared by the two sensors during the recalibration maneuver. Furthermore, since this air 20 space 23 is closer to the operator 21 who exhales quite a bit of C02 gas into air space 23 while working, the concentration of the gas of interest in the shared air space 23 may be non-uniform with higher gas concentration level leaning towards the operator 21 holding the Calibration Master 20. This gas concentration non uniformity plus the fact that the still air condition in air space 23 during the 25 recalibration routine cannot always be guaranteed in real life situations lead to the inevitable conclusion that performing the ERC this way might not always be accurate. Potential shortcomings of the above situation can be remedied by providing a specially designed air sampler 24 built into Calibration Master 20 as illustrated in 30 Figure 8 in order that both sensor 19 which is to be recalibrated and Calibration Master 20 can now measure the concentration level of the gas of interest in the same close environ air space 25 immediately close to sensor 19. Consequently if WO 2012/166585 PCT/US2012/039539 15 sensor 19 correctly reads the gas concentration value in close environ air space 25 within a certain time period, it should be substantially the same value as that measured by Calibration Master 20 held by operator 21 during the same time period with the aid of the special air sampler 24, With the use of this air sampler 5 24, the potential error that the sampled air surrounding sensor 19 is not the same as that surrounding the Calibration Master 20 is eliminated, or at least reduced to the point that it will not interfere with the recalibration procedure, Figure 9 portrays the details of the components layout for an especially preferred embodiment of a specially designed air sampler 24 encompassing the 10 Calibration Master 20 in the same package. The specially designed air sampler 24 comprises a small air pump 26 whose inlet 27 is connected to one end 28 of an air-tight telescopic sampling tube 29 (see Figures 9 and 10). Outlet 30 of air pump 26 is connected to inlet 31 of an air-tight confined space 32 wherein an AB designed NDIR gas sensor 33 is located. Outlet 34 of confined space 32 leads to 15 free space 35 outside of specially designed air sampler 24. Air pump 26 is powered by a battery pack 36 and controlled by an ON/OFF switch 37 all located inside the air sampler unit 24. Also confined inside air sampler unit 24 is Calibration Master 20 whose printed circuit board (PCB) (not shown in Figure 9) interfaces with AB designed NDIR gas sensor 33 on one side and a LCD display 20 38 and a keypad 39 on the other. Whereas LCD display 38 shows operator 21 of Calibration Master 20 what is going on at any one time, keypad 39 allows operator 21 to issue functional commands to Calibration Master 20 in order for it to carry out the ERC routine. An air-tight telescopic sampling tube 29 (see Figure 10), when not in use, is 25 lodged by two clamps 40 and 41 located on the right-hand-side of air sampler unit 24. Unlike an ordinary telescopic tubing where the joints of its sections may not normally be designed to be air-tight, the telescopic tubing 29 (see Figure 10) of air sampler 24 is, in an especially preferred embodiment, specifically designed to have substantially air-tight sections so that air does not get into air sampler unit 24 30 during sampling except through inlet 42 of telescopic sampling tube 29 (see Figures 9 and 10). The air-tight telescopic sampling tube 29 might be 6 ft. long when fully extended with 9 sections and an outside diameter of 0.5. When all its WO 2012/166585 PCT/US2012/039539 16 sections are drawn back, its length is around 8". In theory this air-tight sampling tube 29 can be of any length and any diameter as long as it is convenient to use for air sampling under all circumstances. So far the ERC procedure has been described in terms of how it can be 5 accomplished in the field. It should be noted that the ERC procedure can be accomplished very quickly, without the need for using standard gasses, which greatly reduces the cost of the procedure In practice, it is important to realize that the ERC procedure allows a technician to check calibration of large numbers of sensors in short periods of time, a limiting factor being the time necessary to move 10 between sensors and a short amount of time needed for an ERC procedure. In an especially preferred embodiment of the ERC procedure, each gas sensor has a unique identification number. A Calibration Master can address a particular gas sensor via its unique ID number and can request instantaneous data from it in order to ascertain whether the gas sensor is accurate. 15 To increase the efficiency of the ERC procedure, in an especially preferred embodiment, software is included in Calibration Master 20 (e.g., in processor memory or other memory media) to facilitate the ERC process and also allow Calibration Master 20 to interact with a computer (e.g., by use of the Internet, a LAN, a WAN or hardware device) where information from Calibration Master 20 20 can be collected and utilized with one or more computer program modules to track compliance with scheduled calibration checks. Thus, for example, each time an ERC procedure is performed, Calibration Master 20 can create and store a data file containing desired information such as the unique identifier of the gas sensor being checked, the gas concentration detected by the gas sensor, the date and 25 time of the procedure, whether the gas sensor was recalibrated and any other desired information. If desired, automatic reports documenting the ERC procedure, and its results, can be generated, stored or sent to one or more additional locations electronically, such as through, for example, an Internet connection. Because the information used to generate such results is stored 30 electronically, human error is minimized and, if desired, the system can be configured with sufficient safeguards so as to prevent doctoring of calibration WO 2012/166585 PCT/US2012/039539 17 results, thus guaranteeing better information regarding long term stability results of gas sensors subjected to the ERC procedure. It is also worth pointing out that a Calibration Master can be configured so that it can be used to test multiple gas sensors used to sense different types of 5 gasses or a single gas sensor that can detect multiple gasses. For example, a single gas sensor might be configured so that it can detect both CO 2 and water vapor, and a single Calibration Master can be designed to calibrate the sensor for both gasses. Accordingly, the present invention has now advanced a novel Re 10 calibration methodology applicable only to AB designed NDIR gas sensors and apparatus that can be used to perform such methodology. The final portion of the present invention will now address how a specially designed AB NDIR gas sensor can be made to recalibrate itself without the need for using a Calibration Master as described earlier to carry out a re-calibration procedure. 15 Using the optical component layout for an Absorption Biased NDIR gas sensor as depicted in Figure 1, the first step is to install a "Standard" Signal channel detector 43 and a "Standard" Reference detector 44 both equipped with the same and identical band-pass filter 45 neutral to the detection of the target gas respectively next to the Signal channel detector 5 and the Reference channel 20 detector 6 as shown in Figure 11. As disclosed earlier, both Signal channel detector 5 and Reference channel detector 6 are equipped with the same narrow band-pass filter 8 which is used to detect the gas of interest in the sample chamber 9 (see Figures 1 and 11). Detectors 5, 6, 43 and 44 are all of the same kind but each has its own spectral filter. Detectors 5 and 6 have the same 25 spectral filter for the detection of the target gas whereas detectors 43 and 44 have the same filter that is neutral to the detection of the target gas, i.e. passing no radiation that would be absorbed by it. As a matter of fact, detectors 5 and 6 in the component layout configuration for an AB designed NDIR gas sensor as shown in Figure 1 are single channel detectors. When detectors 5 and 43 and 30 also detectors 6 and 44 are installed next to each other together as pairs, they can be, respectively, two dual-channel detectors 46 and 47 (see Figure 11). The values for the CWL and FWHM for filter 8 depend upon which target gas the WO 2012/166585 PCT/US2012/039539 18 sensor is designed to detect. The CWL for neutral band-pass filters 45 (see Figure 11) can be at 220p, 3 .91p or500p with a FWHM of -0.1p. None of the common gases encountered by the general public everyday including those in the atmosphere have absorption bands at these wavelengths within the specified 5 spectral pass-band of -0,1p. A new sensor parameter called "Standard GAMMA" which is the ratio of the output of the "Standard" Signal channel detector 43 over the output of the "Standard" Reference channel detector 44 (see Figure 11) is now defined and created. First of all, the value of "Standard GAMMA" is independent of the 10 presence of the target gas in the sample chamber since the spectral filters that the "Standard" detectors carry are neutral to the detection of the target gas. In other words, the radiation passed by these filters will not be absorbed by the target gas in the sample chamber of the sensor. The "Standard GAMMA" is therefore unrelated to the measurement Physics of the AB designed NDIR gas sensor but 15 serves to monitor the performance characteristics of all the sensor components over time. Should there be any change at all in the performance characteristics of the sensor components over time, e.g. due to aging, the value of "Standard GAMMA" will change accordingly. The value of the regular Go of the AB designed NDIR gas sensor will also change when the performance characteristics of the 20 sensor components change over time and hence affect the calibration curve of the sensor. But the only way to compensate for the change of the Go value in order to restore the measurement accuracy of the sensor is to update it from time to time. This can be done by flowing 100% dry N 2 through the sample chamber of the sensor and re-determine the correct Go value or to execute the re-calibration 25 methodology disclosed earlier above. The present invention advances a third way to update the value of Go when there are changes in the performance characteristics of the sensor components over time by taking advantage of the definition and creation of the concept for "Standard GAMMA". As it turns out, since both values of the regular Go and "Standard GAMMA" 30 are affected only by the changes in the performance characteristics of the sensor components over time and are both independent of the measurement Physics of the AB designed NDIR gas sensor, they actually are directly proportional to each WO 2012/166585 PCT/US2012/039539 19 other. Because of this fact, any change taking place in the regular Go can be corrected by knowing the change in the value of "Standard GAMMA" over the same period of time. As a matter of fact, by measuring the value of "Standard GAMMA" and storing it along with the initial calibration curve, namely (A,, A 2 , A 3 , 5 A 4 ) (see Equation (4) above] and the regular Go, the "Standard GAMMA" can be used to update the regular Go when the performance characteristics of the sensor components change over time. It can update proportionally the value of the regular Go with the change it detects in itself in order to preserve the measurement accuracy of the sensor going forward in time. In other words, such 10 a sensor has now become self-commissioning, namely knowing how to correct any performance characteristics changes in the sensor components over time thereby restoring the measurement accuracy of the sensor since its initial calibration. A sensor according to the present invention is ideally suited for use with the 15 HVAC&R industry, especially when numerous such sensors are networked together in a single structure, such as a building. The accuracy gained by continued self-commissioning allows networked sensors to now fulfill a long-felt need for stable sensors. In addition, multiple sensors can be combined within a single sensor unit, by adding one or more additional pairs of detectors, one of 20 which is in the signal channel, the other of which is in the reference channel, such additional pairs of gas detectors meeting the requirements of an AB designed NDIR gas sensor-namely, that this new pair of detectors is equipped with the same or identical narrow band-pass filter used to spectrally define and detect a different target gas of interest. In other words, just as Figure 11 illustrates two 25 pairs of detectors, as compared to Figure 1, such a sensor would now have three detectors in each of the signal and reference channels, two of which function to detect two different target gasses, and one of which serves as the Standard in accordance with the teachings of this invention. Note that a single pair of Standard detectors can be used to calibrate multiple pairs of different target gas 30 detectors. Thus, a single sensor can be used to detect two or more gasses, such as 002 and water vapor, and the information obtained from the Standard can be WO 2012/166585 PCT/US2012/039539 20 used to self-commission the multiple gas detectors contained in the same single sensor. In summary, the present invention discloses a powerful new NDIR gas sensor that is self-commissioning, that can detect one or more target gasses, 5 which can be networked for inclusion in sophisticated networking applications that have gone unused to date for want of suitable sensors. The self-commissioning sensors disclosed herein ensure that such sensors will represent a major advance in the field of NDIR gas sensors. But, as important as self-commissioning is, it is still possible that sensors 10 according to the present invention may ever so slowly drift over time, albeit in an amount of time much longer than presently encountered within the industry. The reason for this is the lack of a perfect source. The present invention ensures that changes in the intensity or spectral content of the source will be corrected by self commissioning. Yet, if there is physical change in the source that affects its 15 radiation pattern, which might theoretically occur if, for example, there is sagging of a filament in an incandescent light bulb or possible bubbling on a MEMS source, there is a possibility of a very slight drift over a long period of time that cannot be corrected by self-commissioning. Luckily, however, this theoretical problem can be overcome by also using the re-calibration methodology disclosed 20 earlier in this application. So, in conclusion, when a sensor according to the present invention is also equipped to take advantage of re-calibration methodology that uses a Calibration Master NDIR gas sensor to calculate a master gas concentration which is used to recalibrate the sensor, or multiple master gas concentrations if the sensor is being 25 used to detect multiple gas concentrations, a drift-free sensor is truly obtained which, if it ever does drift, can easily be recalibrated. And, even if the sensor never does drift, its users will know it can quickly be checked and recalibrated if need be. This then represents about as perfect an NDIR sensor as there ever has been, one that can only be improved with respect to drift by use of a perfect 30 source. The invention has been described herein with reference to certain earlier disclosures by the author presented for illustration and explanation only should not WO 2012/166585 PCT/US2012/039539 21 limit the scope of the invention. Additional modifications and examples thereof will be obvious to those skilled in the art having the benefit of this detailed description. Further modifications are also possible in alternative embodiments without departing from the inventive concept. 5 Accordingly, it will be apparent to those skilled in the art that still further changes and modifications in the actual concepts described herein can readily be made without departing from the spirit and scope of the disclosed inventions as defined by the following claims.