Testing our new low-cost turbidimeter 3


Well we’re two days late, but the data on our new turbidimeter setup were worth the wait. WASH For All members Alex Krolick and Chris Kelley tested the current build (single-beam IR LED with TSL230R sensor) against a handheld MicroTPI turbidimeter freshly calibrated at the factory. The two driving questions were: (1) does this perform to expectation for the intended use range (0-100 NTU) and how turbid can a water sample be before the setup stops accurately reporting. To explore these questions eight readings were taken with the experimental setup and the MicroTPI at each of 60 turbidity intervals from 0-1,000 NTU (there went Saturday afternoon). The chart below compares the average NTU measurements from the MicroTPI turbidimeter with the average arbitrary units (AU) readings of our prototype.

Turbidimeter prototype challenge test data


A very promising correlation between a commercial handheld turbidimeter and our prototype.

The curve shows two highly linear regions with an inflection point at roughly 300 NTU. Needless to say we are pleased, and pleasantly surprised that the prototype faithfully measured up to 1,000 NTU. This setup achieved the following  accuracy measures:

± 0.5 NTU in the range of 0-20 NTU

± 2.0 NTU in the range of 20-200 NTU

± 2% in the range of 200-1000 NTU

These achievements surpass most of our design goals for this summer’s public release, and fall just short of goal for the range 20-50 NTU. We are busy at work refining this prototype and will have more to report soon, including photos of the sleek new 3D-printed casing!

Here is the attachments of this Post

Leave a comment

Your email address will not be published. Required fields are marked *

3 thoughts on “Testing our new low-cost turbidimeter

  • Anthony

    In your goals you stated you wanted to take automated readings every 15 minutes. How do you intend to accomplish that given the design? Are you considering some sort of a pump to bring water from a source into the tester? Or will the tester be submersible?

    Also, have you found that ambient light plays a role in your numbers, meaning have you tried to take readings in a bright sunlit room, and in the dark?

    • iamchriskelley Post author

      Anthony, thanks for comment and please forgive the the long-delayed reply. The inline version of our turbidimeter will have an immersible sensor, just as you suggested. The handheld model will also meet those battery life requirements, though we don’t expect many people would take measurements every 15 minutes by hand.

      As for ambient light, it’s a huge concern in obtaining accurate measurements. It’s important to take a sensor reading with the device’s light source turned off (the so-called “dark count”) to differentiate background light versus scattered light from the sample analysis. Even with this information though, minimizing background light is crucial. Our handheld turbidimeter accomplishes this with a few millimeters of coated ABS plastic, and we are experimenting with a few different approaches for the inline model.

  • Chris R

    HI There,
    I stumbled upon your website and think you idea is great especially for those in need of clean water.
    I have been in the water industry for years and I actually have a lot of experience with turbidity.

    I have looked at your design and I have noticed you used a white housing please use a matt black housing as white will cause reflections.

    Zero your sample with RO water or distilled water not ion exchange water.

    Please calibrate your own gear with Formazine anything under 20NTU is unstable and will deteriorate very quickly. so I advise a 20 and 100ntu calibration. Make sure it meets Nata stds or Nist up your way.

    Please invert Formazine back and forth to create a homogenous solution as it settles quite quickly

    Your algorithm is only taking one calibration point after 5 secs. You must take at least 10 for each measurement and take the avg of the result for both points as fast as you can.

    The more you take samples the better the reading.

    Your nanometers needs more refinement as im sure they use 860nm or something like this you will need a filter to tune the nm of near infra red. look up hach they will have the nm’s.

    Remember this is a ratiometric measurement so you must find the correlation between the 2 detectors you will find the 90 for Raleigh scattering is the key. As I have never really played with the algorithms im sure with known stds you should be able to correlate the light adc counts.

    Also dont focus on the high end of resolution lets face it dirty water is easy to detect focus on the low end of the scale so you can see when your filters are failing, and require backwashing.
    Good dual media I have regularly seen at .03 NTU.
    I believe you will get to 0.1 NTU which can easily be good potable water as the std used to be .5 once upon a time.

    Inside most turbs is a half sphere focusing lens which concentrates the bean through the vial. Use glass quartz even better.

    Krohne has a good cal stds kit for your findings as it has a known .02NTU std very good to know your on the mark a 10ntu and 100ntu also included. Shelf life 2 years.

    Contact me if you would like one I can help you here.

    Please note also just because you have done a calibration does not mean you did a good calibration always check with a known std.

    For online you can use the same vial as a flow through cell, you will need to attach ultrasonic vibration to the vial to stop build up and make sure no air air is not your friend will cause error in the readings. you will need a small head with back pressure and a de- bubbling chamber.

    Hope this helps, you have my email call anytime.

    Chris R