This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.
IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT
Abstract— A real-time turbidimeter based on time-correlated single photon counting (TCSPC) was developed to measure the low-level turbidity for drinking water. To improve measurement accuracy, we use a single-photon avalanche diode (SPAD) with high sensitivity to accurately detect the intensity of weak scattering light. A novel statistics principle-based TCSPC technique was applied in this system to reduce the fluctuation of measurement and improve the stability of turbidity measurement. Thanks to the SPAD with short response time and the digital output of single-photon detecting module, the real-time and steady measurement of low turbidity is finally implemented. Experimental tests for the turbidimeter’s performance were described and the results showed that 0.1 Nephelometric Turbidity Units (NTU) can be measured stably in the range of 0-400 NTU within 1 s.
On the basis of the theoretical analysis, a turbidity measurement model was proposed. It was found that a tradeoff between the high measurement resolution and wide linearity range should be considered adequately depending on the practical applications.
By adjusting the system parameters, we demonstrated that the linear range of measurement could be expanded in the regime of low turbidity, while maintaining high resolution of this system.
The proposed turbidimeter has advantages of high resolution, wide linear range, and short response time, which is sufficient for many applications, including the real-time online turbidity or particle concentration monitoring.
Index Terms— Particle scattering, single-photon avalanche diode (SPAD), time of flight, time-correlated single-photon counting (TCSPC), turbidity measurement.
T
I. I NTRODUCTION
URBIDITY is a comprehensive index of water quality.
It is also an important parameter in monitoring water pollution and water eutrophication. Studies reported by the
United States Environmental Protection showed that the spread of protozoa can be greatly reduced when the turbidity of water
Manuscript received July 24, 2014; revised September 23, 2014; accepted
September 29, 2014. This work was supported in part by the National Natural
Science Foundation of China under Grant 61201401 and Grant 61275165 and in part by the National Key Scientific and Technological Project under Grant
2011ZX05051004. The Associate Editor coordinating the review process was
Dr. George Xiao.
H. Wang is with the State Key Laboratory of Transducer Technology,
Institute of Intelligent Machine, Chinese Academy of Sciences, Hefei 230031,
China (e-mail: hqwang@iim.ac.cn).
Y. Yang and Z. Huang are with the State Key Laboratory of Transducer
Technology, Institute of Intelligent Machine, Chinese Academy of Sciences,
Hefei 230031, China, and also with the Department of Automation, University of Science and Technology of China, Hefei 230027, China.
H. Gui is with the Key Laboratory of Environmental Optics and Technology,
Anhui Institute of Optics and Fine Machanics, Chinese Academy of Sciences,
Hefei 230031, China.
Color versions of one or more of the figures in this paper are available online at http://ieeexplore.ieee.org.
Digital Object Identifier 10.1109/TIM.2014.2364703
is is less than 0.5 Nephelometric Turbidity Units (NTU). The protozoan removal rate is 99%, when the turbidity is lower than 0.3 NTU. When the turbidity is as low as 0.1 NTU, the removal rate can even reach 99.9% [1]. With the improvement of living standards, the demand for low turbidity of water has increased and many researchers pay attention to study the novel instrument for high-accuracy turbidity measurement
[2]–[15]. Postolache et al. [12], [13] designed an infrared
LED illumination turbidity sensor on the basis of photodiode detection whose measurement accuracy is
±
10 NTU in the range of 0–1000 NTU. Ker
¨ integrating sphere apparatus to measure the optical properties of dilute pulp suspensions. Juttula et al. [15] used digital camera images to determine the optical parameters of turbid media.
In recent years, optical fiber-based turbidity sensor become a popular technique in this field. García et al. [16] designed a low-cost four-beam turbidimeter by using optical fibers.
The usage of fiber optic can improve the signal-to-noise ratio (SNR) and has presented a flexible measurement, including the feasibility of remote and online measurement.
However, it will make the design require more sophisticated configuration. Moreover, the most important insufficiency for all these aforementioned techniques is that their measurement is based on normal detectors with low sensitivity, such as vacuum photodiodes, silicon photodiodes, cadmium sulfide photoconductors, and photomultiplier tubes [17], which is not suitable for the accurate measurement of low turbidity. When the measured turbidity is low, the intensities of scattering light will be very small and hard to be accurately and stably detected. In addition, all these detectors-based measurement methods suffer because the turbidity signal output from the front-end of detector is analog small signal and thus it needs multistage amplification and A/D conversion. As a result, several sources of noise and nonidealities are present and may be severe, which cause the low turbidity be difficult to measure accurately. At last, the time-consuming detection used to enhance the SNR to ensure enough accuracy makes the realtime measurement with high accuracy to be impossible. With
HACH 1720E turbidimeter as an illustration, its measurement accuracy is only
±
0.8 NTU in the range of 0–40 NTU and the detection time is as long as 1–5 min to get an average measured value, then another 30 s is needed to calculate and record the final stable value [18].
In this paper, we propose a real-time and highresolution turbidimeter based on time-correlated single-photon counting (TCSPC), which is a novel technique gradually used
0018-9456 © 2014 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.
See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
1

This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.
IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT 2 to measure the optical properties of scattering medium in the recent decade [19], [20]. The turbidity of water sample can be obtained by accurately measuring the intensity of scattering light. Compared with the conventional detectors based on turbidity measurement method, there are several advantages for our proposed method.
1) A high-sensitive single-photon avalanche diode (SPAD) operating under Geiger mode is used to accurately capture the intensity of weak scattering light.
By using the high avalanche gain inside device, the Geiger mode SPAD can directly generate large amplitude signal without multistage amplification, which has advantages of high sensitivity and short response time compared with the normal detectors [21]. Moreover, the lack of A/D conversions and multistage amplification can remove quantization errors and the usual nonidealities associated with these components [22], which can obviously improve the measurement accuracy of low turbidity.
In addition, the output of single-photon detecting module is digital signal, which makes our turbidimeter less sensitive to the noise and has a great capability of anti-interference.
2) As the TCSPC technique is based on the principle of statistics, the fluctuation of scattering intensity can be reduced in the process of measurement, which can finally improve the stability of turbidity measurement.
3) Owing to the high modulation frequency of light source and short response time of single-photon detector, the real-time measurement of turbidity can be achieved.
4) Because this turbidimeter has high detection sensitivity, the light source with low emitting power can be adopted in this system, which has advantages of low heat, stable output, long operating life, and no harm to human eye.
5) Simple structure and signal-processing circuit make this type of turbidimeter easy to be miniaturized.
Fig. 1.
Principle of TCSPC technique.
times, that is, the waveform of the optical pulse, builds up in the memory. As shown in Fig. 1, the horizontal coordinate of the histogram peak is time of flight, which stands for the location of measured points. The vertical coordinate of the histogram peak is the photon count, which is proportional to light intensity. In this paper, the intensity of weak scattering light is measured accurately by using TCSPC technique to determine the turbidity of water.
II. M EASUREMENT P RINCIPLE
For some high precision measurement, when the light intensity is so low that the probability of detecting one photon in one signal period is far less than one, it becomes difficult to be measured unless using TCSPC technique [23]. For TCSPC technique, it is not necessary to provide for the possibility of detecting several photons in one signal period. By making use of low-level, high-repetition-rate signals, it is sufficient for TCSPC to record the photons, measure their time in the signal period, and build up a histogram of the photon times.
Its principle is shown in Fig. 1.
The detector signal is a train of randomly distributed pulses corresponding to the detection of the individual photons. There are many signal periods without photons; other signal periods contain one photon pulse. Periods with more than one photon are very rare. When a photon is detected, the time of the corresponding detector pulse in the signal period is measured.
The events are collected in a memory by adding a ‘1’ in a memory location with an address proportional to the detection time. After many photons, the distribution of the detection
III. I NSTRUMENT D ESIGN
Generally speaking, there are three types of configurations for measuring turbidity through an optical system [24]:
1) measure the intensity of transmitted light; 2) measure the intensity of scattered light; and 3) measure and calculate the ratio of scattered light and transmitted light intensity. For the first type of configuration, the structure of instrument is relatively simple. However, the measurement error and a relatively large vibration will be introduced for turbidity measurement owing to the light absorption of organic material in the water. For the third type of configuration, it has high stability and is insensitive to water color, because the synchronous measurement of scattered light and transmitted light can reduce many interferences. However, both of the signal and instrument structures are complicated, because multiple detectors and processing circuits are required, which also make this type of turbidimeter difficult to be miniaturized. For the second type of configuration, the turbidity of water can be determined by the 90° scattered light intensity according to the scattering theory [25]. Currently, most of the turbidimeter adopts this type of configuration, which has relatively simple structure and a wider range of turbidity measurement. Furthermore, it shows high precision and sensitivity in the measurement of low turbidity. Therefore, we adopt this type of configuration to construct our turbidimeter in this paper.

This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.
WANG et al.: INSTRUMENT FOR REAL-TIME MEASUREMENT OF LOW TURBIDITY 3
Fig. 2.
Block diagram of TCSPC-based turbidimeter.
Fig. 3.
Picture of experimental setup.
According to the international water quality standard
ISO7027, the quantitative determination method can be realized when the measured scattering angle is 90°
±
2
.
5° [26].
This type of instrument has advantages of simple structure and excellent stability, and its measuring range is suitable for drinking water, mineral water, and so on. In our paper, the weak 90° scattering intensity is measured using TCSPC technique. To minimize the interference of background light and improve the detection accuracy, all of the measurement is implemented in a dark box. The block diagram of our proposed turbidimeter is shown in Fig. 2.
As shown in Fig. 2, a 650 nm laser diode (DL-5147-
242, Sanyo) with repetition frequency of f s
=
1 MHz is used as the light source in our measurement system. The full-width at half-maximum of each laser pulse is as narrow as 1 ns, and the average output power is 200
μ
W. An
Avalanche Photo Diode (APD) based single-photon detecting module (SPCM-AQRH-14, PerkinElmer) with low dark count
(no more than 30 counts) is used to measure the weak 90° scattering intensity when the laser beam is project into a round glass container full with turbidity solution. The typical width of the output pulses from this module is 15 ns. To reduce the noise of detector, a 650 nm narrowband pass filter with half-bandwidth of 10 nm has been used to filter the background light and an occluder with 1 mm aperture diameter has been adopted to limit the field of view (FOV).
A time digital converter (TDC) chip (TDC-GPX, ACAM) with time resolution as high as 82 ps is used to accurately measure the arrival time of the leading edge of echo pulses, according to which a FPGA chip (EP2C8Q208N, Alter) is used to synchronously record the echo photon counts in corresponding time slice. Then, a statistical histogram can be obtained by sorting the echo photons according to their arrival time. This measurement result is finally sent to PC for further display and processing by the USB interface. The picture of experimental setup is shown in Fig. 3.
It is worth mentioning that only a normal glass (not a special nonreflecting glass) container full with turbidity solution is our experiment. Therefore, some reflected light coexists with the useful scattering light and will seriously interfere with the measurement of low turbidity. Thanks to the small FOV of our turbidimeter limited by the occluder in 90° direction, most of the stray light reflected by the wall of glass container can be kept out of our optical system. Furthermore, our TCSPC-based turbidimeter with high time resolution can easily distinguish and discard those reflected photons by analyzing the distribution of photon arrival time. This advantage makes it to be a more effective and robust system in the practical applications, because there are also many possible reflected light generated by the reflection of container, tube, and object in environment.

4
This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.
IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT
Fig. 4.
Turbidity measurement results from 1 to 400 NTU. (a) Normalized photon counting value versus turbidity. (b) Typical statistical histogram of echo photons.
IV. I NSTRUMENT
A. Measurement Results
V ALIDATION
In our experiment, a standard 400 NTU formazin turbidity solution (particle size is 0.6
μ m) is diluted to obtain
27 different turbidity samples: from 1 to 10 NTU, the measurement interval is 1 NTU; from 10 to 100 NTU, the measurement interval is 10 NTU; from 120 to 400 NTU, the measurement interval is 40 NTU (note: the NTU is the abbreviation of nephelometric turbidity units and it has widely used as a standard unit when the turbidity is measured by a calibrated nephelometer. 1 NTU formazin turbidity solution can be prepared accurately by weighing and dissolving 5 g of hydrazine sulfate and 50 g of hexamethylenetetramine in 1 L of distilled water) [17]. Because the measurement system is difficult to ensure the same sampling times for different turbidity sample, the total photon counts of different turbidity samples have been normalized. The final result of measurement I is shown in Fig. 4. In the experiment, a series of statistical histogram of echo photons corresponding to each measured turbidity point is first obtained, according to which we can extract their envelope to obtain the peak of photon counting value N and the noise counting value N b
, as shown in Fig. 4(a). Fig. 4(b) shows a typical statistical histogram of echo photons corresponding to the measured turbidity is 100 NTU.
As shown in Fig. 4(a), the peak point A represents the total normalized photon counting value including the contribution of scattering light and noise light. Assuming that the noise light is uniform at the measured area in sample, the photon counting value of noise light at A point can be replaced by the average photon counting value at B point in the process of data processing. According to the experimental results shown in Fig. 4, some preliminary conclusions can be obtained.
1) With the increase of turbidity, both of the echo signal’s peak and SNR is increased.
2) In our experiment, we ensure the normalized value N of the total photon counts are equal for different turbidity measurement. Compare with the low turbidity samples, there are more photo counts in echo pulse for samples with high turbidity. Therefore, the sampling time t s will be decreased with the increase of turbidity, which causes the photon counts of noise N b t s in t s decrease constantly with the increase of turbidity, assuming that the photon count N b of noise in unit time is a constant at an unchangeable measurement environment.
As shown in Fig. 4(a), the photon counting value of noise in the low turbidity range (e.g., 0–20 NTU) is significantly higher than the one of high turbidity range (e.g., 100-400 NTU).
3) Because the system is based on TCSPC technique, it has high time resolution, which can be used to distinguish the scattered light from different measured areas. As shown in Fig. 4(a), for all the 27 sample points, the horizontal time coordinate of the histogram peak are the same, which means the measured areas are the same for all different samples. Therefore, the accuracy of turbidity measurement can be guaranteed.
4) A good linear relationship between the peak of photon count and the turbidity is shown in the regime of low turbidity ( < 20 NTU). Nevertheless, an obvious nonlinearity is presented in the regime of high turbidity, especially when the measured turbidity is higher than 20 NTU.
5) Because the modulation frequency f s of laser is as high as 1 MHz, one million echo signals will be acquired within 1 s. In our paper, 10 000 total photon counts are statistically detected (more photon count can further improves the measurement accuracy) to determine the turbidity of sample, thus 100 repeat measurements will be implemented in 1 s and their average value will be used to determine the final turbidity of sample.
Therefore, the real-time and stable online monitoring of turbidity with our proposed turbidimeter in flowing media is possible.
B. Turbidity Measurement Model Based on TCSPC
To explain the experimental results, a turbidity measurement model combined with the theory of scattering and single photon detection is proposed.

This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.
WANG et al.: INSTRUMENT FOR REAL-TIME MEASUREMENT OF LOW TURBIDITY 5
Fig. 5.
Experimental date fit for measurement I. (a) Experimental data with hyperbl fit for 1–400 NTU. (b) Experimental data with linear fit for 1–20 NTU.
The TCSPC technique considers the light is composed of many discrete photons. The relationship between the energy of a single photon
ε and the frequency of light
υ can be expressed as
ε = h υ = hc
λ
(1) where c is the speed of light in vacuum, h is the Planck’s constant, and λ is the wavelength of light.
Assuming that the energy of m photons can generate a photon counting pulse from detector, the optical power corresponding to N photon counting pulses can be calculated by
P
=
( m N ) · ε/ t
η
=
( m N ) · hc / t
ηλ
(2) where t is the sampling time corresponding to the photon counting value of detector is N , and
η is the efficiency of detector.
For the scattering light, when the photon counting value of detector is N s
, its power can be calculated by
P s
=
( m N s
) · hc
/ t s
ηλ
(3) where t s is the sampling time corresponding to the photon counting value is N s
.
According to the scattering theory [25], [27], the intensity of scattering light is proportional to the turbidity of sample
I
S
= K
S
T I
0
(4) where T is the turbidity of measured sample.
Moreover, the relationship between the light intensity and power for a certain wavelength of light can be determined by [28]
P
= k
·
F
= k
· ·
I (5) where k is the conversion factor between luminous flux and light power that is only related with the wavelength
λ
, is the solid angle, and I is the light intensity.
In addition, when the sample with T turbidity is illuminated by the incident light with I light can be expressed as
0 intensity, the power of scattering
P s
= k I s
= k K s
I
0
·
T
= k T (6) where I s k
= k K s is the 90° scattering intensity of light, and
I
0
.
Therefore, when the photon counting value of scattering light is N s time t s for the sample with T turbidity, the sampling can be obtained by combining (3) and (6) t s
=
( m N s
) · hc
ηλ k
·
1
T
= k
·
N s
T
.
(7)
The parameter k used in (7) is defined as: k
( m
· hc
/ηλ k
) = ( m
· hc
/ηλ k K s
I
0
)
.
=
Assuming that the photon counting value N b of noise in unit time is a constant, the total photon counting value N can be expressed as
N
=
N b t s
+
N s
=
N b k
·
N s
T
+
N s
.
(8)
The photon counting value of scattering light can be obtained by substituting (7) into (8), and be simplified as
N s
=
1
+
N
N b
T k
=
N
·
T mhcN b
ηλ k K s
I
0
+
T
.
(9)
If we define parameters a
=
N , b
= mhc N b
/ηλ k K s then (9) can be simplified as
I
0
,
N s
= a
·
T b
+
T
(10) which shows the relationship between the photon counting value N s of the scattering light and the turbidity T of sample.
As the measurement I shown in Fig. 4(a), the normalized photon counting value N s of the scattering light can be acquired, based on which we can obtain the relation between turbidity T and N s
. Fig. 5 shows the relation of turbidity T versus N s
, which is calculated from the experiment results shown in Fig. 4. According to (10) deduced from the turbidity theoretical model, a good fitting curve shown in Fig.
5(a) is obtained for measurement I, which demonstrates the correctness of our proposed model.
Fig. 5 shows as follows.
1) The functional forms after fitting is
N s
= [
P
1
T
/(
T
+
P
2
) ] = [
0
.
0605T
/(
T
+
72
.
594
) ]

This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.
IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT 6 which shows an excellent fitting result, because the
Adj. R-Square of the data to the fit is very close to 1.
2) It is necessary to note that the P
1 and P
2 obtained here are just fitting value, which are a weighted mean value related to the value of a and b for all the
27 different turbidity points. However, the value of b = mhc N b
/ηλ k K s
I
0 is a constant, which is only determined by measurement system itself, including the optical path parameters related with the system. Because the value of b is the same for 27 different turbidity measurement points, it equal to the obtained fitting value P
2
.
3) In the low turbidity regime (1–20 NTU), there is a good linear relationship between the turbidity of sample and the peak value of normalized photon counting (the correlation coefficient shown in Fig. 5 is greater than
99.7%). Nonlinearity will present obviously when the turbidity of sample is higher than 20 NTU.
C. Analysis of Linear Range and Resolution in Low Turbidity Regime
In practical application, a key requirement of turbidimeter is the linearity of its response signal over a large dynamic range, because a linear signal significantly reduces calibration effort (enabling one-point calibration) [29]. We have discussed the linearity of our proposed turbidimeter over a dynamic range of 0.1–400 NTU, which based on (10) derived from our proposed model.
For a measurement system, the expression of SNR has the form: SNR
= (
P s
/
P n
)
, where P s is the average power of the signal, and P n is the average power of the noise.
According to (2), the average power P expressed as n of noise can be
SNR
=
P n
=
( m N b
) · hc
ηλ
.
Combining (11) with (3) and (8), we can obtain
P
P s n
=
N
N s b
· t s
=
According to (10), the SNR of system can be simplified as
SNR =
N s
N
−
N s
=
N a · T b
+
T
− a
·
T b + T
The parameter b used in (13) is defined as b = mhc N b
ηλ k K s
I
0
.
N s
N − N s
=
T b
.
For (10), when T b, it can be simplified as
N s
= a b
T
.
.
(11)
(12)
(13)
(14)
(15)
According to (15), the precondition of the linear relationship between T and N s is that T should be far less than the parameter b (T b = mhc N b
/ηλ k K s
I
0
b). Because the value of is a constant, when the measured turbidity T is low, it will be satisfied with the condition of T b and a good linear relationship is shown in the low turbidity regime, as shown in Fig. 5(b). However, the nonlinearity will be gradually presented with the increasing of turbidity T, as shown in Fig. 5(a).
To expand the linear range, the value of b should be increased by changing the parameters of measurement system, especially the optical path parameters of system. In our system, the value of N b can be increased by increasing the light aperture diameter of occluder, and then the value of b can be increased according to (14). Therefore, a contrast test is investigated by increasing the light aperture diameters 1 mm of occlude in former optical system to 3 mm. Moreover, the repeatability of system, derived from comparing with a second measurement, also can be verified. The results of two measurements are shown in Fig. 6.
Fig. 6 shows as follows.
1) By fitting the raw data using (10), two new P
2 values
134.8 and 133.0 are obtained, which are about two times than the P
2 value obtained as shown in Fig. 5
(about 72.6). When the linear correlation coefficient of measurements II and III are the same to the measurement I (about 99.7%), the linear range is 1–40 NTU in the measurements II and III [as shown in Fig. 6(b) and Fig. 6(d)], which is twice expanded than that in the measurement I [as shown in Fig. 5(b)].
2) Because the fitting values of P
1 and P
2 in measurement II are almost the same to the measurement III, it shows excellent repeatability for the proposed measurement system. In addition, it also verified that the value of b is a constant for the system with the same optical parameters, because the value of b is equal to the obtained fitting value P
2
.
In practical application, a key requirement of turbidimeter is the linearity, which can be expanded by increasing the value of b. However, it means the SNR of system is too low according to (13), which will cause a high uncertainty in the photon counting and therefore a low measurement accuracy of turbidity [30]. If we cannot increase the emitting power of laser (the number of light pulses) and lengthen the sampling time of system to maintain the same SNR and measurement accuracy, a tradeoff between the high measurement resolution and wide linearity range must be considered adequately depending on the actual demand. Because this turbidimeter is designed to detect the low-level turbidity of drinking water, the expected resolution should be better than
1 NTU, which is the maximum allowable turbidity value of the World Health Organization drinking water quality standards [31].
To obtain the resolution and detection limit of our turbidimeter, a formula has been given out according
(12) and (13)
T
= b
·
SNR
= b
×
N s
N
−
N s
= b
×
N s
N b
· t s
.
(16)
Because the value of b is constant for a given system, the measured value of turbidity only relates with the scattered photon counting value N s and noise counting value N
In the ideal condition, the lowest values of N b t s and N s b t s are the
.

This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.
WANG et al.: INSTRUMENT FOR REAL-TIME MEASUREMENT OF LOW TURBIDITY 7
Fig. 6.
Experimental data fit for measurements II and III. (a) Experimental data with hyperbl fit for 1–400 NTU in measurement II. (b) Experimental data with linear fit for 1–40 NTU in measurement II. (c) Experimental data with hyperbl fit for 1–400 NTU in measurement III. (d) Experimental data with linear fit for 1–40 NTU in measurement III.
Fig. 7.
Experimental date of measurement IV. (a) Linear fitting result. (b) Uncertainty of 20 times measurements.
dark counts and shot-noise value of our single-photon detecting module, respectively. Therefore, according to (16), the theoretical detection limit T min of our turbidimeter is obtained
T min
= b
×
N
SN
N
DC
(17) where N
DC and N
SN are the dark counts and shotnoise value of single-photon detecting module, respectively. However, not only the shot-noise source and dark count, but also many other noise source will exist in the complicated measurement system, which will cause the practical measurement resolution of our system much higher than T min
.
To test the practical measurement resolution of our turbidimeter whose linear range has been expanded to 40 NTU, the measurement IV adopting the same light aperture diameter as the measurement II/III is carried out. In this paper,
1 NTU standard formazin turbidity solution (particle size is
0.6
μ m) is used to obtain ten different turbidity samples from
0.1–1 NTU by dilution method. In this paper, each turbidity point has been measured 20 times to obtain its average value and the final measurement results as shown in Fig. 7.

8
This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.
IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT
As shown in Fig. 7(b), the maximal uncertainty of multiple measurements in the range of 0.1–1 NTU is less than
0.07 NTU, which indicates the turbidity of 0.1 NTU can be measured accurately. Therefore, the measurement resolution of our turbidimeter is better than 0.1 NTU, which is lower enough than almost all of national standard for drinking water [31]–[33]. And the linearity, as shown in Fig. 7(a), is also excellent, which demonstrates the feasibility that a relatively wide linear range of measurement can be obtained by adjusting the system parameters, while maintaining high resolution of this turbidimeter.
V. C ONCLUSION
In this paper, a new TCSPC-based turbidimeter had been developed for accurate real-time measurement of the low turbidity of drinking water. The feasibility of this instrument was demonstrated by series of turbidity measurement experiments.
The results showed the turbidity could be stably measured as low as 0 .
1 NTU in the range of 0-400 NTU within 1 s.
With the theoretical analysis, a turbidity measurement model was proposed to explain the experimental results. It had been proved that the linear range of measurement could be expanded by adjusting the system parameter, while maintaining high resolution of this system. Accordingly, some suggestions about the design of this type of turbidimeter have been given out. For instance, a good compromise between the high measurement resolution and wide linearity range should be considered adequately depending on the actual demand. The proposed realtime turbidimeter in this paper is very robust and can be used to continuously monitor the turbidity or particle concentration in flowing media, including the turbidity of drinking water, sewage, and many chemical and pharmaceutical processes, such as crystallization, precipitation, and filtration, where the information of undissolved particles is of vital importance to stable control and optimization of these processes [29].
A CKNOWLEDGMENT
The authors would like to thank D. He (Key Laboratory of
Quantum Information, Chinese Academy of Sciences) at the
University of Science and Technology of China (USTC) for his valuable contributions and discussions.
R EFERENCES
[1] The Water Quality Standard of Urban Water Supply CJ/T206-
2005.
[Online].
Available: http://www.chinacitywater.org/rdzt/ sheshuibiaozhunjishu/shishipinglun/19713.shtml, accessed Jun. 1, 2014.
[2] S. M. J. Baban, “Detecting water quality parameters in the Norfolk
Broads, U.K., using Landsat imagery,” Int. J. Remote Sens., vol. 14, no. 7, pp. 1247–1267, 1993.
[3] A. G. Dekker and S. W. M. Peters, “The use of the Thematic Mapper for the analysis of eutrophic lakes: A case study in the Netherlands,”
Int. J. Remote Sens., vol. 14, no. 5, pp. 799–821, 1993.
[4] B. Forster, B. D. Xu, and X. W. Sha, “Satellite remote sensing of pollution and its distribution in near-coastal waters,” Asian-Pacific Remote
Sens. J., vol. 6, no. 1, pp. 1–10, 1993.
[5] S. Koponen, J. Pulliainen, K. Kallio, and M. Hallikainen, “Lake water quality classification with airborne hyperspectral spectrometer and simulated MERIS data,” Remote Sens. Environ., vol. 79, no. 1, pp. 51–59, 2002.
[6] H. S. Lim, M. Z. MatJafri, and K. Abdullah, “Algorithm for turbidity mapping using digital camera images from a low-altitude light aircraft,” in Proc. 2nd IEEE Int. Conf. Comput. Sci. Inf. Technol. (ICCSIT),
Aug. 2009, pp. 200–204.
[7] A. F. Omar and M. Z. MatJafri, “Consistency test on a newly develop water quality fiber sensor,” in Proc. 6th Regional IMT-GT (Indonesia-
Malaysia-Thailand Growth Triangle) UNINET Conf., Penang, Malaysia,
2008, pp. 388–392.
[8] C. G. Campbell, D. T. Laycak, W. Hoppes, N. T. Tran, and F. G. Shi,
“High concentration suspended sediment measurements using a continuous fiber optic in-stream transmissometer,” J. Hydrol., vol. 311, nos. 1–4, pp. 244–253, 2005.
[9] N. T. Tran, C. G. Campbell, and F. G. Shi, “Study of particle size effects on an optical fiber sensor response examined with Monte Carlo simulation,” Appl. Opt., vol. 45, no. 29, pp. 7557–7566, 2006.
[10] M. Borecki, “Intelligent fiber optic sensor for estimating the concentration of a mixture-design and working principle,” Sensors, vol. 7, no. 3, pp. 384–399, 2007.
[11] A. F. Omar and M. Z. MatJafri, “Development of optical fiber sensor for water quality measurement,” in Proc. Nat. Phys. Conf.-PERFIK Current
Issues Phys. Malaysia, 2008, pp. 398–402.
[12] O. Postolache, P. Girao, J. Pereira, and H. Ramos, “An IR turbidity sensor: Design and application [virtual instrument],” in Proc. 19th IEEE
Instrum. Meas. Technol. Conf. (IMTC), vol. 1. May 2002, pp. 535–539.
[13] O. A. Postolache, P. M. B. S. Girao, J. M. D. Pereira, and
H. M. G. Ramos, “Multibeam optical system and neural processing for turbidity measurement,” IEEE Sensors J., vol. 7, no. 5, pp. 677–684,
May 2007.
[14] V. T. J. Keränen, A. J. Mäkynen, M. Törmänen, and S. A. Prahl, “A scattering measurement system to determine the optical characteristics of industrial suspensions,” in Proc. IEEE Int. Instrum. Meas. Technol.
Conf., May 2009, pp. 570–573.
[15] H. J. Juttula, T. P. Kananen, and A. J. Makynen, “Instrument for measurement of optical parameters of turbid media by using diffuse reflectance of laser with oblique incidence angle,” IEEE Trans. Instrum.
Meas., vol. 63, no. 5, pp. 1301–1309, May 2014.
[16] A. Garcia, M. A. Perez, G. J. G. Ortega, and J. T. Dizy, “A new design of low-cost four-beam turbidimeter by using optical fibers,” IEEE Trans.
Instrum. Meas., vol. 56, no. 3, pp. 907–912, Jun. 2007.
[17] M. J. Sadar. (2013). Understanding Turbidity Science. [Online].
Available: http://www.hach.com/1720e-low-range-process-turbidimeterturbidity-sensor-only/product-downloads?id=7640457219&callback=bc
[18] Hach Company. (2013). 1720E Turbidimeter User Manual. [Online].
Available: http://www.hach.com/1720e-low-range-process-turbidimeterturbidity-sensor-only/product-downloads?id=7640457219&callback=bc
[19] D. Qin, H. Zhao, Y. Tanikawa, and F. Gao, “Experimental determination of optical properties in turbid medium by TCSPC technique,”
Proc. SPIE, vol. 6434, Feb. 2007, p.64342E.
[20] U. Gaetano, L. Maria, and L. I. Pietro, “Optical property measurements in scattering media by time-correlated single-photon counting system (TCSPCS),” Proc. SPIE, vol. 4160, Nov. 2000, pp.223–232.
[21] J. J. Degnan, “Photon-counting multikilohertz microlaser altimeters for airborne and spaceborne topographic measurements,” J. Geodyn., vol. 34, nos. 3–4, pp. 503–549, Oct. 2002.
[22] C. Niclass, C. Favi, T. Kluter, F. Monnier, and E. Charbon, “Singlephoton synchronous detection,” IEEE J. Solid-State Circuits, vol. 44, no. 7, pp. 1977–1989, Jul. 2009.
[23] W. Becker, “Overview of photon counting techniques,” in Advanced
Time-Correlated Single Photon Counting Techniques, 1st ed. Berlin,
Germany: Springer-Verlag, 2005, pp. 20–21.
[24] M. Sadar, “The basics of turbidity measurement technologies prepared for the methods and data comparability board QA/QC sensors group,”
Hach Company, Loveland, CO, USA, Tech. Rep., 2009.
[25] J. Wang, J. Yao, Y. Yu, J. Chen, and P. Wang, “Instantaneous twodimensional density measurements of gas flow by Rayleigh scattering,”
J. Optoelectron. Laser, vol. 12, no. 1, pp. 1005–0086, Jan. 2001.
[26] Water quality—Determination of Turbidity, ISO Standard 7027, 1997.
[27] B. Kong and S. Lu, “The dispersive intelligent liquid turbid degree apparatus’ manufacture,” J. Ningxia Univ., vol. 20, no. 4, pp. 332–334,
1999.
[28] Optical Power and Luminous Intensity.
[Online].
Available: http://wenku.baidu.com/view/daf6e8050740be1e650e9afa.html, accessed Jun. 1, 2014.
[29] A. Kramer and T. A. Paul, “Fiber-optic probes as sensors for diffuse backscattering,” in Proc. Adv. Photon. Renew. Energy, 2010, no. SThD2.
[30] A. F. B. Omar and M. Z. B. MatJafri, “Turbidimeter design and analysis:
A review on optical fiber sensors for the measurement of water turbidity,”
Sensors, vol. 9, no. 10, pp. 8311–8335, 2009.

This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.
WANG et al.: INSTRUMENT FOR REAL-TIME MEASUREMENT OF LOW TURBIDITY 9
[31] Earth Systems.
Clear Solutions Newsletter—Focus on Turbid-
ity.
[Online].
Available: http://www.earthsystems.com.au/wpcontent/uploads/2012/02/ClearSolutions_No1_2003.pdf, accessed Jun. 1, 2014.
[32] The Drinking Water Standard of Japan. [Online]. Available: http://www.water800.com/szbz/xbz/rbszjzxm.htm, accessed Jun. 1, 2014.
[33] EPA, Washington, DC, USA. Drinking Water Contaminants. [Online].
Available: http://water.epa.gov/drink/contaminants/index.cfm#3, accessed Jun. 1, 2014.
Zhe Huang was born in Anhui, China, in 1991.
He received the B.S. degree in information engineering from Anhui Normal University, Wuhu, China, in 2013. He is currently pursuing the master’s degree with the University of Science and Technology of
China, Hefei, China.
His current research interests include optical sensing, including particle monitoring, 3-D imaging, and single photon detection.
Huanqin Wang was born in Hunan, China, in
1982. He received the B.S. degree in applied physics and the Ph.D. degree in microelectronics and solidstate electronics from the University of Science and
Technology of China, Hefei, China, in 2004 and
2009, respectively.
He was in charge of several projects from the
National Natural Science Fund and the Strategic Priority Research Program of the Chinese Academy of
Sciences. He is currently an Associate Professor with the State Key Laboratory of Transducer Technology,
Institute of Intelligent Machine, Chinese Academy of Sciences, Hefei. He has authored over 30 papers and holds 27 patents. His current research interests include optical sensing, including particle monitoring, laser ranging finder,
3-D imaging, and single photon detection.
Yixin Yang was born in Anhui, China, in 1987.
He received the B.S. degree in measuring, control technology and instrumentations from China
Jiliang University, Hangzhou, China, in 2008.
He is currently pursuing the master’s degree with the University of Science and Technology of China,
Hefei, China.
His current research interests include optical sensing, including particle monitoring, laser ranging finder, and single photon detection.
Huaqiao Gui was born in Anhui, China, in 1979.
He received the B.S. degree from the Department of Physics, Anhui University, Hefei, China, in 2000, and the Ph.D. degree in optics from the University of
Science and Technology of China, Hefei, in 2006.
He was in charge of several projects from the
National Natural Science Fund, the National Special Fund for the Development of Major Research
Equipment and Instruments, and the Strategic Priority Research Program of the Chinese Academy of
Sciences. He is currently an Associate Professor with the Key Laboratory of Environment Optics and Technology, Anhui Institute of Optics and Fine Mechanics, Chinese Academy of Sciences, Hefei. He has authored over 30 papers. His current research interests include laser selfmixing.