BATTERY POWERED COST EFFECTIVE TDS LOGGER INTENDED

FOR WATER TESTING

I. A. IVAN1,2, V. STIHI1, M. IVAN1, C. STIHI1, M.RAKOTONDRABE2, A. JELEA3

1
Valahia University of Targoviste, B-dul Unirii 18-20, 130082, Targoviste, Romania
E-mail: ivan@valahia.ro, cstihi@yahoo.com
2
Femto-ST Institute, AS2M Dpt., 24 Rue Alain Savary, 25000 Besancon, France,
E-mail: alex.ivan@femto-st.fr
3
SC Catalyx Engineering, 021388 Bucharest, Romania, E-mail: jelixadrian@gmail.com
Received October 21, 2009

The paper presents a cost-effective device designed for measuring and monitoring the TDS
(total dissolved solids) level of drinkable, surface (lakes, rivers) and/or industrial waters. Providing a
first reading of potential water pollutions, the device is dedicated to the sectors of environment and
consumer protection. The device was implemented and a series of continuous measurements is
depicted, discovering some abnormalities in the quality of Targoviste city water utility.
Key words: water quality, conductivity-meter, TDS, microcontroller.

1. INTRODUCTION

In recent years, the need for water quality sensors spread from industry to
environment and domestic applications. One of the most direct methods is the TDS
(Total Dissolved Solids) measurement.
TDS devices are, in reality, conductivity meters. Pure water shows
intrinsically a very reduced electrical conductivity (0,00548 mS/m). When
impurities are mixed and dissolved into water, ions separate and conductivity
increase. Upon application of a voltage between two or more electrodes made of
non-reactive materials, positively charged ions (e.g. Na+, Ca++, Mg++, H+ etc.) and
negatively charged ions (e.g. Cl–, SO4– –, HCO3– etc.) will constitute an electrical
current. The conductivity of water may afterwards be converted into a measure of
the total quantity of ions or dissolved solids. The method is not selective, meaning
that the nature of ions cannot be determined but only the total quantity, from where
the term TDS (Total Disolved Solids). TDS meters also give a relatively accurate
reading of some colloidal compounds such as silver [1].
Our device is a modified version of a TDS meter initially designed for the
control of a brackish water desalinating system utilizing the novel capacitive
deionization technology (CDT). This second version of the meter module is

Rom. Journ. Phys., Vol. 56, Nos. 3–4, P. 540–549, Bucharest, 2011
2 TDS logger intended for water testing 541

hermetically sealed and is capable to sample, convert and record in-situ

conductivity, temperature and TDS values over extended periods of time (up to 21
days of logging and with 60 minutes sample interval).
Instead of using DC current, we propose to use an electrical field that
oscillates very rapidly, where there is no overall movement of ions toward the
electrodes. In fact, the ions move one way for a tiny fraction of a second, and then
back the other direction for the second half of the cycle. Overall, the solution and
electrodes stay unchanged and the conductivity is accurately measured. Based on
8-bit Microchip PIC18F architecture [4], the devices use a cycle of alternated
charging-discharging small capacitors. Threshold voltage detection and derived
period of time is made using onboard CMOS trigger and 16-bit counter. The
electronic schematic is thus simplified, with no need of A/D converters or voltage
references.
Given the fact that the conductivity of ions in water depends nonlinearly
upon temperature, the conversion of sampled conductance (mS/cm) in TDS units
(ppm) is compensated by aid of internal EEPROM stored look-up tables. The
designed TDS device has been calibrated with standard “442” solution, and
readings were compared with a commercial Hach CO150 TDS meter. First
measurements were carried-out in the city of Targoviste water utility system.

2. EXPERIMENTAL METHODS

2.1. WATER CONDUCTIVITY MEASUREMENT PRINCIPLE

At a first view, simplistically speaking, water and ions appear as an electric

resistance and we might tempt to measure its conductivity in direct current, which
is a wrong method. By applying a constant electric field between the two
electrodes, ions will migrate and deposit on the electrodes, finally electrically
isolating the positive electrode. If the applied voltage is larger than 1.3 volts then
the electrolysis phenomena appear between the two electrodes and the conductivity
measurement will be completely erroneous.
The measurement of conductivity of liquids may be done in several ways [3],
provided that no continuous level of voltage is applied, nor impulse voltage. The
frequency of the alternating signal should be larger than 3 kHz, as above this
frequency water does not decompose nor plate the electrodes, and as the applied
voltage amplitude is no longer limited to 1.3V.
The proposed method is simple and ready to be implemented in logic circuits
– no analog circuits being required. A microcontroller PIC18F1220 [4] was chosen
for its fast setting of pins state (input/output) and for its symmetry of output ports –
they sink and source the same current which is internally limited. The performance
of reduced 8-bit architecture is convenient for this type small and low-cost
application.
542 I.A. Ivan et al. 3

The measurement schematic is shown in the figure 1. Rx is the unknown

water resistance and C1 is a fixed small capacitor (3,3nF). On the left, C+, C- and
EC are some pins of the microcontroller [4] (noted uC).
The measurement algorithm of the Rx value is based of the following steps:
1. Pin EC is set as input (high impedance state) then C+ is put on high logic (3V)
and pin C- is driven in low state, charging the capacitor very close to the Vcc
supply voltage (5V or 3V battery).
2. Pin C+ is set as input and EC pin is set as output on low state. Pin C- continues
to stay on low level. Practically C- and EC are “linked to ground” and in this
configuration the charged capacitor and water resistance are in parallel and
referenced to the ground. The capacitor will discharge following the known
exponential decay law. Time constant is proportional to the factor Rx⋅C1.
3. Microcontroller inputs are Schmitt triggers, switching low when the voltage
reaches less or equal a 1/3 of supply voltage value. Time of capacitor discharge
between Vcc and one third of Vcc is measured and recorded as T1. In the case
that T1 exceeds a maximum preset value, T1 will be considered equal to the
preset value. This value is determined from the minimum frequency condition of
3 kHz that prevents ions to plate the electrodes. It will only be in the case of
ultra-pure water or electrodes outside water, and corresponding TDS value will
afterwards be reported as zero.
4. To prepare the next measurement step, pin EC is set to input and pins C- and C+
are set both on high output states. Thus, the capacitor will be completely
discharged.
5. Pin C+ is brought to the low state, charging the capacitor to an opposite voltage
than that of step (1).
6. Pin C+ direction is set to input (high impedance) and pin EC is set as output
(high logic level). In this way, Rx and C1 are put in parallel and referenced to
Vcc. The discharge time T2 in which the C+ voltage rises to 2/3 of Vcc is
measured.
7. Pin EC is set as high impedance, pins C- and C+ are set to low level output.
Temperature value is read from the sensor. Next cycle can start from step (1).
A number of 16 cycles will be performed and finally the all-corresponding
T1 and T2 time values will be averaged and converted by means of look-up tables
stored in EEPROM, as in this case it is conversely proportional to Rx value.

Fig. 1 – Basic measurement schematic.

4 TDS logger intended for water testing 543

Two voltage waveforms across Rx (electrodes) are presented in figure 2 for

different water resistivities. The curve in magenta is for a higher water conductivity
value than that of the grayed curve.

Fig. 2 – Electrodes signal waveforms for two different water conductivities.

The geometry of electrodes is important and has to be respected; otherwise

the cell constant k will change and the device will need to be recalibrated. In our
case the electrodes were small stainless steel rods of 1 mm in diameter, of 5 mm in
length and placed in parallel at a distance of 3,5 mm between their symmetry axes
(figure 3).

Fig. 3 – Photographs of the reference CO150 conductivity meter and of the designed device.
Close-up of the sensing probe.
544 I.A. Ivan et al. 5

Recalibration is also recommended if tests are performed on different water

sources with quite different dissolved compounds. In figure 4 a calibration
characteristic is presented. It is known that different conductivities might be
reported for different solutions of the same TDS.

Fig. 4 – TDS measured by device versus conductivity measured by the referencing device
(in a mixture of KCl+NaCl solutions at 20 0C).

2.2. THE BATTERY-POWERED DEVICE: DESIGN AND CALIBRATION

The device is powered from a 3V lithium coin cell battery. Most of the time
the device is on the stand-by mode, requiring a very small power supply. The data
interface is a simple serial asynchronous (TTL levels) for reading the logged data
or writing the calibration table into an external EEPROM or in the internal flash
memory. The microcontroller is configured with two oscillators: the first is a
32kHz low power oscillator serving for regularly waking-up the microcontroller
from stand-by (60 seconds intervals) and keeping the date and hour, while the
second oscillator uses a high speed 10 MHz resonator for the clock source and to
count the small time intervals required during acquisition.
Two versions were designed (figures 5-8), the first one is temperature-
uncompensated while the second one uses a TC77 temperature sensor featuring
internal A/D conversion and SPI interface. In figures 9 and 10 are presented the
characteristics conductivity versus temperature for the different concentration of
the KCl and NaCl standard solutions used in this work. The conversion algorithm
implemented on the second device was modified to compensate the readings. As
seen, there is a large difference between figures 11 (without reading compensation)
and 12 (with reading compensation) taken at different temperature values.
6 TDS logger intended for water testing 545

Fig. 5 – Schematic of the first uncompensated model. Fig. 6 – Printed circuit board.

Fig. 7 – Schematic of the second temperature compensated model. Fig. 8 – Printed circuit board.

Fig. 9 – Conductivity to temperature Fig. 10 – Same graph re-scaled.

characteristics for different concentrations.
546 I.A. Ivan et al. 7

Fig. 11 – Non-compensated model TDS Fig. 12 – Compensated model TDS

to temperature characteristic. to temperature characteristic
(various concentrations).

3. EXPERIMENTAL RESULTS AND DISCUSSIONS

Water plays an intrinsic role in living environment. Water of suitable quality

and quantity is essential to all life forms. Water is never in a pure status, it always
contains minerals and microorganisms [8,9]. Human activities can contaminate
water with various substances. But water should be clean in order to ensure a good
health. Actually, some minerals present in water can have benefit consequences.
Surface water is an important component of fresh water systems and surface water
monitoring [10-12] is essential in attaining a comprehensive understanding of the
physical, chemical, and biological characteristics of aquatic systems.
Total Dissolved Solids (TDS) are solids in water that can pass through a
filter. Therefore, TDS can be used to measure the amount of material dissolved in
water. This material can include carbonate (HCO3–), chloride (Cl–), sulphate (SO42–),
calcium (Ca++), magnesium (Mg++), sodium (Na+), potassium (K+), iron (Fe++,
Fe+++), nitrate (NO3–) and other ions. A certain level of these ions in water is
necessary for aquatic life. However, if TDS concentrations are out of admissible
limits, the growth of many aquatic lives can be limited, and death may occur [1]. A
high concentration of TDS may also reduce water clarity, contributes to the
decrease of photosynthesis, may combine toxic compounds and heavy metals and
leads to an increase in water temperature.
Table below presents the normal variation of conductivity for several types of
water, according to [2].
8 TDS logger intended for water testing 547

Table 1
Conductivity range of common aqueous solutions [2]

Experimental measurements were conducted on the water utility system of

Targoviste city. Usually, logged data showed minor daily fluctuations within the
normal range (figure 13). During the afternoon when the domestic water flow
increases, a slight decrease of the TDS level is recorded. We noticed that on the 5th
of July (day 8 on recordings) at 22 hrs the pumping switched off for a current
repair, which lasted until the next day in the morning at 5 AM. As consequence, the
recorded level of TDS (figure 14) showed a significant increase, and it took about
30 hours to retrieve the initial level.

Fig. 13 – Daily normal TDS fluctuations.

548 I.A. Ivan et al. 9

Fig. 14 – Abnormal fluctuation due to the water utility shutdown between 23:00 and 5:00.

5. CONCLUSIONS

The aim of this paper was the design and test of a cost-effective TDS logger
intended for water quality monitoring. Such battery-powered devices are intended
for large scale monitoring, e.g. along rivers or in urban water systems. It is based
on a 8-bit microcontroller architecture and a small EEPROM memory. An
additional temperature sensor allows the reading compensation. Data exchange is
performed over a simple SPI serial synchronous interface. Up to 512 readings are
possible with the provided memory capacity. The autonomy at a 1 hour sampling
rate is of 21 days and it can be easily improved by replacing the memory chip.
First off-line TDS measurements and logs were performed on one end-point
of the Targoviste city water utility. The results were within the limits except for a
single day when the utility was shut-down for a night for maintenance. In the
morning there were noticed abnormally high TDS values. It should be noticed that
according to the Romanian standard [6] the maximum allowed level of 2000 mg/l
was not attained; however we can notice that the recommended value of 1 mS/cm
shown in Table 1 or 670 mg/l was exceeded for a period of 7 hours after the utility
resumed the pumping. We appreciate that a utility shutdown of a longer period,
let’s say of one day, would raise even further the TDS level, making the water
undrinkable.
Finally, the developed low cost device for continuous monitoring can provide
a quick test of the water quality, preventing unwanted health problems. As further
improvements, these devices could be connected in wired or wireless distributed
networks for reporting instant alarms.
10 TDS logger intended for water testing 549

REFERENCES

1. Craun GF, McGabe LJ., Problems associated with metals in drinking water, Journal of the
American Water Works Association, (1975).
2. ****, CO150 Conductivity Meter Model 50150 Instruction Manual, Hach Company, (1998).
3. Singh T, Kalra YP., Specific conductance method for in situ estimation of total dissolved solids,
Journal of the American Water Works Association, (1975).
4. ****, PIC 18Fxxxx Microcontroller Data Sheet, Microchip Technology Comp. (2005).
5. Wagner, R.E. (ed), Guide to Environmental Analytical Methods, 4th edition, Genium Publishing
Corporation, Schenectady, NY, (1998).
6. ****, National Standard STAS 4706 For Surface Waters, Quality Categories and Conditions
(1988).
7. ****, International Organization for Standardization ISO 7888: Water quality - determination of
electrical conductivity, Geneva, (1985).
8. Oprean L., Chicea D., Gaspar E., Lengyel E., Results of Physical and Chemical Parameters
Monitoring of the “Râul Mare” River, Rom. Journ. Phys., Vol. 53, Nos. 7–8, P. 947–953 (2008).
9. Anton M.C., Baltazar Rojas M.M., Aluculesei A., Marguþa R, Dorohoi, D, Study Regarding the
Water Pollution in Romania and Spain, Rom. Journ. Phys., Vol. 53, Nos. 1–2, 157–163 (2008).
10. C. Stihi, I. V. Popescu, A. Bancuta, V. Stihi, Gh. Vlaicu, Inductively Coupled Plasma (ICP) and
Total Dissolved Solids (TDS) Measurements of Surface Waters from Ialomiţa River, Romanian
Journal of Physics, Vol. 50, Nos. 9–10, 977–981 (2005).
11. Gabriel Dima, Ion V. Popescu, Claudia Stihi, Calin Oros, Sergiu Dinu, Laur manea, Gheorghe
Vlaicu, Fe, Mn and Zn concentration determination from Ialomita river by atomic absorption
spectroscopy, Romanian Journal of Physics, Vol 51, Nos 5–6, 633–638 (2006).
12. Gonca Coskun H., Tanik A., Alganci U., Cigizoglu H.K., Determination of Environmental
Quality of a Drinking Water Reservoir by Remote Sensing, GIS and Regression Analysis,
Water Air Soil Pollut 194:275–285 (2008).