Turbidity
Turbidity — Turbidity is the reduction of a light beam's transmission through a water sample caused by suspended and dissolved matter, read as scattered light at a fixed angle against a reference suspension.
Nephelometric measurement
Nephelometric measurement — A nephelometric turbidity measurement detects light scattered at 90 degrees to the incident beam, so the signal comes from particles that remain suspended in the optical path at the moment of measurement.
| Nephelometric, 90 degree scatter | 0.02 - 40 NTU |
|---|---|
| Ratio, scattered over transmitted | 1 - 400 NTU |
| Transmitted light, attenuation | 40 - 4000 NTU |
| Wavelength, ISO 7027 practice | 860 nm |
| Wavelength, classical nephelometric practice | 650 nm |
| Process band | Arrangement | Mechanism behind the choice |
|---|---|---|
| Filtered water, sub-NTU | Nephelometric, 860 nm | Scattered light still stands above the stray-light floor of the instrument at low solids |
| Raw water, tens of NTU | Ratio, dual path | The ratio of the two channels cancels most of the error caused by a film on the window |
| Clarifier blanket, hundreds of NTU | Transmitted light | Scatter saturates once particles shadow each other, attenuation still grows with solids |
| Coloured effluent | 860 nm rather than 650 nm | Dissolved colour absorbs at the red end and adds apparent turbidity there |
- Illustrative span drift between cleanings, unfiltered sample
- 6.5% of span
- Source: Illustrative sample dataset, not a measured value.
Can one turbidity sensor cover both 0.05 NTU and 800 NTU?
Not with a single optical path. The low end is set by stray light and by scatter from the window itself, the high end by multiple scattering once particles begin to shadow each other, so an instrument quoting both ends is switching range or path, and the region around the switch is where its specification is weakest.
Why does filtered-water turbidity read high before the sensor is due for calibration?
Because the optical surface, not the electronics, usually moves first. A film of biofilm or iron oxide scatters light into the 90 degree detector and the reading rises; lifting the sensor out of the flow and reading a certified standard in a beaker separates a fouled window from a drifted calibration in under five minutes.
Every numeric value on this page is an illustrative sample value. Replace each of them with your own measured data before publishing, and treat none of them as a specification.
The four decisions, in the order they should be made
Turbidity instruments are selected wrongly in four distinguishable ways, and all four are visible at the desk before anything is bought. The first is a measuring band that does not contain the real process range. The second is an optical geometry chosen for the wrong solids concentration. The third is a cleaning and calibration interval shorter than the fouling rate of the actual sample. The fourth is sample plumbing that removes the particles the sensor is supposed to see. Take them in that order, because the first answer changes the technology and the last three only change the specification of a sensor you have already chosen.
Start from a range of grab samples taken across the process, including startup and upset conditions, not from the range printed on the previous purchase order. A band chosen from routine data will not contain the turbidity spike that arrives after a rain event or a coagulant dose error, and the sensor will clip exactly when the measurement is worth something.
Where nephelometric scatter stops and attenuation starts
A nephelometric detector sits at 90 degrees to the beam and counts light deflected toward it by particles in suspension. Its signal grows with solids only while each particle is illuminated and seen independently. Once the sample becomes dense enough that particles shadow one another and light bounces more than once, the scatter channel compresses and eventually turns over, so a dirtier sample can read flatter or even lower. That is the physics behind the upper end of the scatter band.
Attenuation measures the opposite quantity: the fraction of the direct beam that survives the sample path. It follows an exponential decay in solids, which means it has plenty of signal where scatter has none, and almost no sensitivity at the low end, where the lost light is smaller than the noise of the detector. A ratio instrument measures scatter and transmission at once and reports their quotient; a film on the window attacks both channels in the same direction, so the quotient moves far less than either channel alone. That is why ratio designs are the practical choice where the sample is dirty enough to foul the glass but not so dirty that attenuation is the only usable signal.
What really sets the zero, and how to check it in five minutes
The floor of a low-range measurement is set by stray light inside the housing and by scatter from the optical window, not by the amplifier. This is why a broad-range instrument cannot verify a sub-NTU filtered water: at the bottom of its band it is reporting its own optics. The same mechanism explains upward drift between calibrations. Iron oxide, manganese oxide and biofilm all scatter into the detector, and all of them grow on a wetted window in the direction of a rising reading.
Air works the other way in one specific respect. Entrained bubbles are strong scatterers, so they push a nephelometric channel up, and they also refract the direct beam out of the transmitted channel, so both channels rise and a ratio instrument only partly cancels the error. If readings rise during a pump-start period and settle later, look at degassing before you look at the sensor.
The field test costs nothing: read the process, then lift the sensor into a beaker of a certified standard at the same temperature and read again. A reading that returns to the standard means the optical path was fouled; a reading that stays high by the same offset means the calibration or the electronics moved.
Coloured samples: the 650 nm and 860 nm trade
Two documented practices differ mainly in wavelength. The classical nephelometric approach uses a red source near 650 nm; ISO 7027 practice uses near-infrared light near 860 nm. Dissolved colour absorbs light at the red end of that pair, so a humic-coloured waste water reads noticeably higher at 650 nm than its true scatter. Moving to 860 nm removes most of that absorption, and the penalty is that near-infrared scatter responds more strongly to particle size and refractive index. Switching wavelength after a coagulant change therefore shifts the correlation between the instrument and the laboratory result, and the correlation has to be rebuilt with your own samples rather than assumed.
Installation faults that produce a stable but wrong number
A sample cell with a dead zone settles solids out of the beam, which makes the sensor read below the true value with excellent repeatability - the worst kind of error, because nothing looks wrong. Keep the flow through the cell in the range the manufacturer specifies, mount the cell so nothing ponds when the flow stops, and never place the sensor where a valve discharges turbulence straight into the beam. On a filtered-water line, the sample line itself is usually the instrument: a long horizontal run with low velocity grows a biofilm that sloughs into the cell, and filter-to-waste leakage during the first minutes after backwash arrives at the sensor as a turbidity peak that has nothing to do with the filter bed. Check the plumbing before you recalibrate.