Result 1A: Accuracy assessment and calibration of a few low-cost conductivity meters

Objective

The goal of this study was to evaluate the accuracy of commonly available water conductivity/total dissolved solids (TDS)/salinity testers. 5 devices were procured from different manufacturers on Amazon, costing a few dollars each. These devices would typically be used by hobbyists who manage acquariums or for household testing. They had a conductivity testing range of 0-9999 uS/cm, which is well suited for conventional water testing but conditionally suited for testing under saline environments. Fortunately, the Moira river (Goa, India) had a conductivity in the range of 2000-9000 uS/cm, so we could use samples collected for this study.

Methodology

 

 

• 5 samples were collected from the Moira river with the help of local fishermen.

• The samples were strategically sourced from clean and polluted stretches of the river, so that there is some variability in their conductivity.

• A clean, sterile and labeled glass container was selected for collecting the samples

• Once collected, a pic of the bottle was taken with a smartphone with location enabled and the sample number clearly seen.

• Each device was numbered and denoted as TDS 1, 2, 3, 4, and 5.

• Once the samples were collected, they were brought back to our testing site (Makers Asylum) and tested with all the 5 devices.

• The samples were then shipped to a local certified water testing lab, and the results were supplied to us within 1-2 days.

• The performance of each device was compared with the lab data and evaluated using mean relative error (MRE), coefficient of determination (R2) and Pearson's correlation.

 

Results

Although the correlations of all the 5 devices tested were >0.88, only two of the devices showed a correlation beyond 0.95.

The device with the lowest correlation i.e., TDS-3 exhibited a relative error of close to 60%, which is unacceptable. 3 of the devices showed an error under 20%. The corresponding R2 for the best two devices were 0.94 and 0.99. TDS-3 showed an R2=0.78, confirming that it was its performance was the poorest.

The performance of the best device i.e., TDS-4 is shown below.

 

 

 

Conclusions

Only 2 out of 5 devices were found to be reliable for testing the conductivity of saline rivers. When procuring such low-cost devices, there is a need to screen for their accuracy, especially when working with rivers with high levels of salinity. Poorly performing devices showed lower levels of correlation with lab data, low R2 and relatively poor linearity and high relative errors. The best performing devices showed excellent R2, correlation and relative errors <20%. There is a need to come up with a quick and inexpensive screening method that would not rely on access to an accredited lab.