P 3.2 -smartgas- A new low cost infrared gas sensor for domestic and automotive applications Christian Stein, Gerhard Wiegleb, Gerhard Knaup 2, Wolfgang Schmidt 2, Fred Plotz 2 University of Applied Sciences Dortmund Institute for Micro Sensors Sonnenstrasse 96, D-4439 Dortmund 2 PerkinElmer optoelectronics GmbH Im Grohestück 2, D-65396 Walluf/Wiesbaden. Introduction Gas sensors for domestic applications are mainly used for the detection of natural gas (gas leakage) or air quality (e.g. carbon dioxide and carbon monoxide). Most of these sensors are based on solid state semiconductor gas sensors like tin dioxide [, 2]. Especially for applications in toxic and explosive gases, metal oxide based gas sensors are a preferable solution for low cost products [3, 4]. A major problem of these kind of gas sensors are long term stability (zero and span drift) and cross sensitivity to other gases (e.g. water vapour) in ambient air. For the detection of carbon dioxide (Air quality control) low cost solid electrolyte gas sensors are also available [7]. Warm up time, power consumption and cross sensitivity are the disadvantages of these types of gas sensors. For industrial applications the cost factor is not an issue. For a reliable and accurate measurement (Safety and Process Control) only the infrared technology is applied in this area. The new smartgas sensor (see Figure ) combines low cost requirements and high quality detection for the domestic and automotive market. The sensor is based on the infrared absorption in the 3.4 µm region for hydrocarbon detection (natural gas) and 4.3 µm for carbon dioxide (air quality) using a single beam principle. Electrically modulated infrared radiation from a simple tungsten filament is passing a small chamber and reflected backwards to the detector by a concave spherical mirror. We used a pyroelectrical Detector from Perkin Elmer (LHi 84) with an integrated FET preamplifier and a narrow band pass interference filter (3.4µm or 4.3µm, ~.µm) integrated in a TO-8 housing. Figure : Complete Sensor (smartgas PYM 5) including µc-electronics The gas sensor is capable to detect gas concentrations in the ppm range for CO 2 and < % of the LEL - Value. The temperature behaviour is compensated using an integrated semiconductor temperature sensor (LM 6). We used an 8-bit Microcontroller (ATmega 8) with an integrated -bit ADC for long term drift compensation. For domestic application the output (TTL Level) switched from zero to 5 Volt in case of a gas concentration higher than % of the LEL Value. 2. Sensor Design [5] In Figure 2 the optical design of the sensor is shown. The sensor consists of a miniaturized incandescent lamp with a tungsten filament. The lamp is powered with an AC voltage of Hz and 5 volt using an intelligent power operation. Compared with a DC or on-off operation the lifetime is much higher (see Figure 9). The radiation coming from the IR Source is reflected backwards by a concave mirror to the pyroelectrical IR detector with an integrated narrow band pass interference filter. The mirror is made in plastic and coated with a high reflective metal layer (e.g. Gold or protected Aluminum). The focal length f of the mirror is the half of the distance between the detector and the mirror. Low Explosion Level
IR Source IR Detector Mirror Figure 2: Optical design of the IR gas sensor using a mirror reflector. The distance between the IR Source and the mirror is app. mm so that we have 8 mm optical path length between IR Source and IR Detector. The gas flow inside the sensor is zero, which means the gas exchange is only due to the diffusion. Interference Filter Gas In Figure 3 the mechanical design of the gas sensor is shown. The optical components (IR Source, IR Detector) are located on the PCB board. On the same board the preamplifier and a temperature sensor is integrated, in order to avoid EMI because of the short distance. The preamp board is connected to a rectangular located µc-board (base board). This board is slide in to a slot in the sensor housing. The sensor housing is also made in plastic (ABS) using mould injection technique for production. The mirror has two mechanical fasteners for a tight connection between the sensor housing and the mirror itself. Furthermore the mirror has also a mechanical fastener in order to hold the µc board in position. On the top side of the sensor housing is a grid for the exchange of the sample gas (ambient air). In order to avoid condensation and deposition of dust a simple filter element (soft foamed plastic material) is located as a barrier for particles and other components. This filter element is pressed to the sensor housing using an additional plastic fastener. IR source IR detector Filter element Fastener Base PCB board Figure 3: smartgas Sensor Design [5] Sensor housing Mirror HC HC T [%] 8 6 Carbon Dioxide 2 2 3 4 5 6 7 8 9 Wavelength [µm] The absorption spectrum of natural gas in the Infrared region is shown in Figure 4. In general two different wavelengths are useful for the detection of Hydrocarbons (HC). At 3.4µm and 7.8µm the absorption strength is very similar. Since the glass bulb of the IR source has no transmission at 7.8µm, we used the 3.4µm absorption band of HC for our tests. Furthermore we can see the Carbon dioxide absorption band at 4.3µm due to the CO 2 concentration of app. 3 Vol.-% CO 2 in natural gas. Figure 4: Infrared Absorption Spectra of natural gas (L-Gas)
3. Test Results Gas sensors for safety applications in buildings have to meet the European Standard EN 594 [6]. All the requirements are related to the low explosion level (LEL) of natural gas (Methane) in ambient air. The LEL limit varies for different natural gases from different sources (e.g. Russia or North Sea), because natural gas is a mixture of more than 2 different components (Hydrocarbons, Nitrogen and Carbon Dioxide). For calibration purpose of the smartgas sensor different Methane mixtures ( % LEL) in air are used as a test gas. In Table the LEL concentration for the most important gases are shown. Gas LEL in Vol.-% HEL 2 in Vol.-% Gasoline.6 8. Propane.7.9 Ethylene 2.3 32.4 Methane 4.4 6.5 Table : LEL and HEL for different gases Due to the absorption of infrared radiation in the sample area we can notice a change of intensity (Sensor raw voltage). In Figure 5 the calibration curve of the smartgas sensor is shown. We applied gas mixtures in steps from to 5. Vol.- % Methane in Air. The calibration curve behaves like an exponential Output Voltage [mv] function due to the Lambert Bee 89 Law: 87 85 I(c) =I o exp. [ c L] () 83 8 79 77 y =,3333x 4-4,485x 3 + 22,528x 2-63,46x + 894,94 R 2 =,9999 75,5,5 2 2,5 3 3,5 4 4,5 5 Methane Concentration [Vol.- %] With I(c) = Intensity or raw voltage versus the gas concentration c. = coefficient of absorption L= optical path length I o = Intensity or raw voltage without gas (only air). Figure 5: Calibration curve of the smartgas sensor described by a polynomial of 4 th. degree The change of intensity due to 2% LEL (app. Vol.- % CH 4 ) is about 5% (see Figure 5). The temperature behaviour of the sensor is very important for practical applications since the ambient temperature is not constant. Regarding to the EN 594 the temperature range of operation is - C to C (T=5K). The change of Intensity due to this temperature variation is also 5%. That means the temperature influence is in the order of magnitude as the change for a gas concentration of 2% LEL. For temperature compensation of the raw voltage we applied the regression curve show in Figure 6 to calculate the compensated sensor voltage for further steps. The residual temperature error is negligible., S e n s o r V o l t a g e,9 9,9 8,9 7,9 6,9 5 c o m p e n s a te d S e n s o r V o lta g e S e n s o r S ig n a l ( R a w V o l ta g e ) v e r s u s A m b ie n t T e m p e r a tu r e y = 9 E - 6 x 2 -, 4 x +, 9 8 5 8 R 2 =,9 9 9,9 4 - - 5 5 5 2 2 5 3 3 5 4 A m b i e n t T e m p e r a t u r e [ C ] Figure 6: Temperature behaviour between - C to C 2 High Explosion Level
Another important requirement is the response time. In Figure 7 the experimental setup for testing the response time (t 9% ) is shown. We applied an adapter to the gas exchange area of the smartgas sensor (top side). The gas flow was < L/min. We used a PC based gas dilution system (Model: DIGAMAT, Woesthoff Germany) to set the gas concentration to 2% LEL. The rise time (change from air to test gas) was less than sec. and the fall time (change from test gas to ambient air) was app. sec. Digamat Gas input Adapter outut Methane Air smartgas Sensor 2 % LEL Laptop Figure 7: Experimental setup for determination of the time behaviour (response time) Rise time = sec. Fall time = sec.,5,995 Sensor Voltage (Raw Signal) %- LEL 35 3 Sensor Voltage n,99,985,98,975,97 Nitrogen Sensor Output (LEL Units) [appr. Vol.-% CH4] Nitrogen 25 2 5 5,965-5,96-2 6 8 2 6 Time [s] Figure 8: Test results for a gas step (air to test gas and vice versa) Stability test One important reason for the drift of the sensor signal is the degradation (aging) of the IR radiation source. Since the sensor must be capable to survive a continuous operation of more than years, it was necessary to get information about the long term behaviour of the sensor. The statistical life time of the used IR source is shown in Figure 9. We powered 45 lamps with DC, AC (on-off) and AC (intelligent) with different voltages (amplitudes). Using an intelligent power supply for the lamp it is possible to achieve a
percentage of survived lamps samples 2 8 6 2 Life time test @ different conditions 2 3 4 5 6 7 8 9 years of operation Figure 9: Life time test of the IR source using different power conditions DC on-off long life power life time of years. The drift behaviour during this time of operation is shown in Figure 9. The drift of the lamp intensity is less than % per year. That means the total lamp intensity drift is less than % within years. With a reference measurement (using an additional IR Detector and different Interference filter) it is possible to compensate this drift contribution. In our sensor design we applied an electronically compensation using the µc. Vs [volt],3,2,,,99 test @ 3.4µm,98,97,96 test @4.3µm,95 28..95 24.7.98 9.4. 4..4..6 Time Figure : Drift test in ambient air (laboratory environment) at different wavelength 4. Summary The described smartgas sensor is useful for a rugged operation in buildings for gas detection purpose. The sensor was designed for low cost applications and high volume production. The lifetime is about years for continuous operation. All the influences, due to temperature changes and lamp drift were compensated in the µc using an intelligent algorithm. References. Seiyama, T.: Chemical Sensor technology Vol.2 Elsevier Amsterdam 989 2. Göpel et.al.: Chemical and Biochemical Sensors Part I VCH Weinheim 99 3. Wiegleb, G. Heitbaum, J.: Solid State Gas Sensor based on Indium-Tin-Dioxide Experimentelle Technik der Physik 39(99)3, 227-237 4. Wiegleb, G. Heitbaum, J.: Semiconductor gas sensor for detecting NO and CO traces in ambient air of road traffic, Sensors and Actuators B, 7(994) 93-99 5. Wiegleb, G. Stein, C. Knaup, G. Plotz, F.: German Patent Application No. 24 28 23 4..24 6. EN 594 Klassifikation: VDE Teil 3-. Elektrische Geräte für die Detektion von brennbaren Gasen in Wohnhäusern Dezember 2 7. TGS 46 Figaro (Japan)