Aspects of TGS 813 Gas Sensor´s Use

Ioan Silea, Dorina Petrica

Department of Automation and Applied Informatics, Faculty of Automation and
Computers, "Politehnica" University Timişoara, Romania, Fax. + 04 0256 4032,
Tel. + 04 0256 403347
e-mail: [email protected], [email protected]

Abstract: Aspects of air polution and health protection are now prioritary problems for
researchers and establishing the dangerous gas concetration in the air allows the initiation
of actions destined to avoid both types of problems. Due to the mentioned reasons, the
sensors, equipments and methods of establishing the grade of dangerous substances
mixture in the environment are of great interest in the scientific world. In this paper are
presented aspects regarding the use of TGS183 sensor for the detection of the concetration
level of denagerous gases in the air, namely approximation of the characteristics,
dependent on the temperature and relative humidity in the measure point. The obtained
results facilitate the use of these sensors in embedded systems dedicated to the measure of
gas concentration in the air, realised with the help of microcontrollers.

Keywords: sensor, gases, characteristics, nonliniarity, measure

1 Introduction
In many places is necessary to detect burned gases (carbon monoxide) or gases
(methane, butan, hydrogen), which together with air form a mixture, that can
explode. In enterprises where such gases are used, the monitorisation of their
concentration is very important in order to avoid catastrophes.
Not only in production departments is a gases´ monitorisation important, but also
in all-day life; we reach situations in which it would be good to know, for
example, how much the carbon monoxide concentration in a room is, because not
rarely, it´s exceeding has had tragical consequences. The concentration´s increase
over 880 ppm, namely 0.08% is dangerous for humans. Taking into consideration
the danger presented, there were concerns, and equipments for gas detection
already exist. [1]
In the technical world there are sensor producers [2], who detect either only one
type of gas or more types. In this paper we will refer to the posibilities of using a
sesnor that can detect methane, butane, hydrogen and carbon monoxide [3],
connected to an equipment based on a microcontroller.
2 Sensor’s Characteristics
According to catalogue data [3], the sensor used (TGS 813) to detect the gas
concentration has the nonlinear sensitivity feature (se Fig. 1) (representation is in
logarithmic scale) and is dependent on the temperature and humidity of the
environment (see Fig. 2) where it is placed. It is noted: R0 = the sensor´s resistance
in 1000 ppm methane, R= the sensor´s resistance to different concetrations of
various gases.

Figure 1
TGS#813 sensor’s sensitivity feature
 R/R0

Figure 2
TGS#813 dependence on relative temperature and humidity

3 The Sensor´s Use

In Fig. 3 a basic gas concentration measure circuit is presented with sensor
TGS#813.

Figure 3
Measure circuit with TGS sensor
The fluctuation of the TGS sensor´s resistance is indirectly measured like the
tension fall appeared on the reading resistance RL. In working regime, the current,
that passes through the sensor, which is serial with the RL resistance, is constant,
and if a gas like methane,butane, propane etc. comes in contact with the sensor´s
surface, it´s endurance reduces in correlation with the present gas concentration.
The tension fall on the RL resistance is changing, so that together with the tension
on the sensor the VC value results, namely the continous supply tension (or
altenative tension). As it follows, this sensor become very approchable for users
because it has the exit in tension and it is very easy to mesure this parameter. To
work with the features of the provided sensor, in this case, Figaro firm, we have to
process this exit tension (VRL) and to achieve the sensor´s endurance RS with the
ecuation (1):

RS =
V C xR L
− RL (1)
V RL

The configuration of the sensor´s connected circuits has to ensure the following
conditions:
• VC can be 5,6,12 or 24 V
• VH heating tension has to be 5 V±0.2 V
• The power supply on the sensor maximum 15 mW
If we determine the sensor´s sensitivity feature in standard testing conditions, this
characteristic has to coincide with the sensitivity feature, given to us by the
producing Figaro firm, because this characteristic is determined with a relative
representation of the sensor´s resistance.
Standard testing conditions, indicated by the producer, are:
• Atmospheric conditions: temperature 20 ºC±2 ºC and relative humidity
65%±5%
• VC: 10 V±0.1 V, VH: 5 V±0.05 V, RL: 4 K±1%
• Time for the sensor´s supply maintenance 7 days or more
• Testing gas: methane
Sensor´s parameters:
• Heating resistance: 30 Ω±3 Ω
• Sensor´s resistance: 5 k ~15 KΩ methane 1000 ppm
• Resistance´s report: R S inmethane 3000 ppm = 0.6±0.05
R S inmethane 1000 ppm

Once the value R0 is calculated, the sensor´s endurance at different concentrations

of various gases will be next determined.
The sensor´s resistance for a given concentration is calculated as it follows. We
admit the the snesor´s endurance at 1000 ppm methane is 10 KΩ and we want to
determine it´s endurance at 4000 ppm hydrogen. From Fig. 1 arrises for R/R0 the
value 0.3, so it is only left for us to multiply 0.3 with 10 KΩ and we obtain that
the sensor´s endurance value is 3 KΩ. Because we will supply the sensor´s heating
circuit with a stabilised tension source, it is not relevant to discuss the influence of
the supply (heating) tension´s variation on the sensor´s resistance. It is imposed to
disuss about the influence of other parameters like temperature and relative
humidity.
The determination of a gas´ concentration, if we want to eliminate the influence of
temperature and relative humidity, is as it follows: with relation (1) we calculate
RS, then form Fig. 2, for temperature and relative humidity, we read the value
R/R0 in the given conditions and we obtain a value x. We multiply value x with R0
and we obtain a value y. We divide RS with y and we get R/R0, uninfluenced by
temperature and relative humidity, with whose help, from Fig. 2, we will
determine the appropiate concentration. The sensor´s features also depend on the
working time period, like in Fig.4.

Figure 4
Dependenţa rezistenţei senzorului TGS#813 de timpul de funcţionare

When the snesor starts functioning, after a long not working period, a time interval
has to pass until the sensor´s resistance reaches a stable regime. This time interval
depends on the atmospheric conditions of the place the sensor was kept. It is
important not to calibrate any of the circuits that contain this sensor (TGS#813),
before the resistance stabilises, because we can obtain wrong results. A sensor´s
lifetime is 8-10 years.[3].
The feature illustrated in Fig. 2 has to be approximated in order to make a very
exact calculation of the temperature´s and relative humidity´s influence on the
value of the ensor´s endurance.
 (cx,cy)

(xa,ya) (xm,ym) (xb,yb)

h
Figure 5 Approximation through arcs of circle

We try to approximate these curves on the feature through circle arcs. We detail
on base of Fig. 5. We need the coordinates of the ending points (T, R/R0) at –10
ºC, noted (xa,, ya) and at +40 ºC, noted (xb,, yb) and we also need another
parameter, the arc´s height, noted (h). At first, the coodinates (xm, ym) are
established, through a simple calculation, with the formulas:
xa + xb ya + yb
xm = , ym = (2)
2 2

We determine the line’s equation through the points (xa, ya), (xb, yb), which has the
form:
y = a * x + b (3)

values a and b are obtained with the relations:

y a − yb
a= ; b = y b − xb * a (4)
x a − xb

We calculate the equation of the orthogonal line on the previous determined line
that passes through the point (xm, ym), which if of form (3) with parameters a1,
b1, which can be determined with the following relations:
1
a1 = − , b1 = y m − x m * a 1 (5)
a

We calculate the distance between (xa,, ya), (xb, yb) noted (dist):

dist = (x a − xb )2 + (y a − yb )2 (6)

Having dist and h we can calculate the circle’s radius, noted Rc, like this:
2
⎛ dist ⎞
⎟ + h
2
⎜
Rc = ⎝
2 ⎠ (7)
2 * h
Having this informations we’ll determine [4] the coordinates of the circle’s centre
under the condition that it’s radius is Rc and the centre located on the orthogonal
line on the determined one by the points (xa,, ya), (xb, yb), in the coordinates point
(xm,ym):
(x − c x )2 + (y − c y )
2
= R c
2
(8)
y = a 1 * x + b1

If we explicit cy from the line’s equation: c y = a 1 * c x + b 1 (9) and replace it in

The circle’s equation, we obtain the second degree equation:
2 2 2 2
(1 + a1 ) * cx + 2 * (a1 * b1 − y * a1 − x) * cx + y2 + x2 − 2 * y * b1 + b1 − Rc = 0 (10)

With the condition that the equation (10) has to contain the point (xa, ya), the
equation becomes:
2 2 2 2
(1 + a1 ) * c x + 2 * (a1 * b1 − ya * a1 − xa ) * cx + ya + xa − 2 * ya * b1 + b1 − Rc = 0
2 2
(11)

We determine from this equation cx with the relation:

2 2 2 2 2
−(a1 *b1 − ya *a1 − xa ) + (a1 *b1 − ya *a1 − xa )2 −(1+a1 )*(ya + xa −2* ya *b1 +b1 −Rc ) (12)
cx = 2
(1+a1 )

As a last step we calculate cy with the relation (9).

Having the circle’s equation, radius and center’s coordinates we can easily
calculate for any temperature and relative humidity, with which we have made the
determinations, the sensor’s resistance value compared to normal conditions.
Misfortunately, this calculation is for a certain value of the relative humidity, but
in the main this is how the circle’s parameters are determined, having the
coordinates of the two points and the arc’s height.
We still haven’t solved the problem because we have to make the calculation for
the relative humidity. This is made tabling. Each curve’s parameters for a certain
relative humidity and then these parameters will be approximated in accordance to
the relative humidity with a third degree equation. But let’s see these parameters,
Table 1.
RH H ya yb
0 0.01 1.83 1.6
20 0.08 1.73 1.07
40 0.12 1.68 0.92
65 0.15 1.6 0.86
100 0.18 1.57 0.82

Table 1
Parameter’s values at different values of relative humidity
In Figs. 6 and 7 we have the approximations of these parameters with third degree
equations, together with the original curve. The approximation was made in
EXCEL , this is why when we work with approximation equation instead of x, we
have to use an adjusted value, namely the interval [0;20] ->[1;2] ; (20;40] ->(2;3];
(40;65]->(3;4]; (65;100] ->(4;5].
With the help of these approximation equations, we can calculate for a value of
relative humidity the parameters h, ya and yb. Knowing that xa=-10 °C and
xb=+40°C, we can always establish the circle’s parameters with the relations
(1)÷(10), which has as the arc, the respective curve and we can make the
calculation.
The calculation procedure is the following:
• For the relative humidity value we determine ya,yb,m
• We establish with these values the circle’s parameters
• For the temperature value we determine R/R0”
• We calculate R with the relation (1) and we divide with R0 obtaining
R/R0

R / R0 =
R / R0 (13) With this value we can go to the sensitivity feature and
R / R0"
calculate the concentration for the desired gas, taking into consideration that it is a
logarithmic feature.
 2

1.8

1.6
y = 0.0079x2 - 0.1121x + 1.932 ya
yb
1.4
Poly. (ya)
Poly. (yb)
1.2

0.8 y = -0.03x3 + 0.3464x2 - 1.3436x + 2.624

0 20 40 65 100

Fig. 6. Coordinates ya,yb depending on the relative humidity

0.20

0.18
y = 0.0025x3 - 0.0304x2 + 0.1471x - 0.114
0.16

0.14
0.12
h
0.10
Poly. (h)
0.08
0.06

0.04

0.02

0.00
0 20 40 65 100

Fig. 7. The height of the circle´s arc h, dependeing on the relative humidity
If we implement the characteristic relative resistance – temperature – relative
humidity with a two dimensional search table, then we will have Table 2.
T/RH
0 10 20 30 40 50 60 70 80 90 100

-10 1.83 1.781 1.74 1.7011 1.6668 1.6422 1.6201 1.6031 1.590 1.5791 1.569

-5 1.803 1.7143 1.644 1.5899 1.5461 1.5169 1.4912 1.471 1.455 1.4392 1.4234

0 1.7784 1.651 1.556 1.4869 1.4352 1.4025 1.3744 1.352 1.333 1.3141 1.2935

5 1.7543 1.5917 1.4747 1.3922 1.3341 1.2991 1.2696 1.246 1.225 1.2038 1.1792

10 1.7306 1.536 1.399 1.3058 1.2428 1.2066 1.177 1.1531 1.132 1.1083 1.0805

15 1.7073 1.4848 1.33 1.2277 1.1614 1.125 1.0963 1.0732 1.052 1.0275 0.9974

20 1.6844 1.437 1.268 1.1579 1.0897 1.0544 1.0278 1.0064 0.986 0.9616 0.93

25 1.662 1.394 1.212 1.0963 1.0277 0.9946 0.9713 0.9527 0.934 0.9103 0.8781

30 1.64 1.354 1.1623 1.043 0.9756 0.9458 0.9268 0.912 0.896 0.8739 0.8418

35 1.62 1.318 1.119 0.998 0.9333 0.908 0.8945 0.8844 0.872 0.8522 0.8211

40 1.6 1.28 1.0824 0.961 0.9008 0.881 0.8741 0.8698 0.861 0.8452 0.816

Table 2
Search table for the characteristic relative resistance – temperature – relative humidity

As a columns index we have the samples of relative humidity from 10% to 10%,
as rows index the temperature samples from 5 °C to 5 °C, and as value in the table
we’ll have relative resistance. There are problems concerning the edge only in
temperature case and because of this fact we express the following rules:
• if the temperature is lower than -10°C, then it is considered as if it would
be in the interval [-10 °C, -5 °C]
• if the temperature is higher than +40°C, then it is considered as if it
would be in the interval [+35 °C, +40°C]
 Figure 8. The measurement results

As an implementing algorithm, at first we determine in which intervals are

temperature and relative humidity located, considering previous reasons and we
choose the four points, which mark these intervals. We’ll have four triples for
example for RH=75%, T=12 °C:
(10,70,Ta[indt,indrh]), (10,80,Ta[indt,indrh+1]),
(20,70,Ta[indt+1,indrh]), (20,80,Ta[indt+1,indrh+1]),
where indt=T<-10?0:T>=40?9:floor((T+10)/5); indrh=floor(RH/10); Ta – search
table. The relative resistance value will be calculated with relation (14).
(T − indt * 5 − 10 )
R / R0 = * (Ta [ indt + 1, indrh ] − Ta [ indt , indrh ]) + (14)
5
( RH − indrh * 10 )
+ * (Ta [ indt , indrh + 1] − Ta [ indt , indrh ]) + Ta [ indt , indrh ]
10

Applying the facts presented it can proceed to the software implementation of the
gas ´ concentration sensor´s features.
Conclusions
The obtained results, applied to a measuring system carried out with such a sensor
are illustrated in Fig. 8. Methane concentration, temperature and humidity at the
measure place are presented.
In the paper, the temperature’s and humidity’s influence is emphasized – when a
precise measure of a gas´ concentration is desired – and there are elaborated the
relations through which the parameters are taken into consideration, but also how
they can easily be implemented in embedded systems [5]. It facilitates the use of
these sensors in dedicated air gas measuring systems, realised with the help of
microcontrollers. It appears the possibility to monitories more measuring points
with dedicated systems, using the internet.
Bibliography
[1] www.allgasdetectors.blagspot.com/atom.xml
[2] Paul E. Keller, Lars J. Kangas, Lars H. Liden, Sherif Hashen, Richard T.
Kouzes: Electronic Noses and Their Application, in Proceedings of IEEE
Northcon/Technical Applications Conference (TAC’95), Portland, Oregon,
USA, October 12, 1995, pp. 159-165
[3] http://conrad.de/index/183474-da-01-en-Gassensor_TGS_813.pdf
[4] Pavel Naslau, Romeo Negrea, Liviu Cadariu, Bogdan Caruntu, Dan
Popescu, Monica Balmez, Constantin Dumitrascu: Matematici asistate de
calculator, Ed. Politehnica 2005, Timisoara, ISBN 973-625-234-5
[5] Ayala J. Kenneth: The 80251 microcontroller, in Prentice-Hall 2000, Ohyo,
ISBN 0-13-907551-8, USA