T90 is Not About Sensor Speed – It’s About Data Latency

Indonesia’s PermenLHK P.80/2019 demands that every SPARING system log wastewater data every 2 minutes. The rule is incredibly strict about the sampling frequency and the accuracy limits (pH ±0.1; COD, TSS, NH3-N, and flow ±10%). But it completely misses one critical detail: the regulation never sets a minimum standard for sensor response time.

The result? A slow sensor might just be starting to react to a chemical spike right when the data logger blindly grabs the reading. The number gets pushed to the portal successfully, but it has nothing to do with what is actually flowing out of the pipe at that second.

The metric that tells you how long a sensor takes to catch up is called T90. We are going to break down how to actually read a T90 spec, compare the real-world numbers across SPARING parameters, and show you how to use T90 as an early warning system for maintenance.

What Are T90, T63, and T95?

Open any pH, COD, or NH3-N sensor datasheet. You will find a spec labeled T90 or response time. The IEC 60746-1:2003 standard defines T90 as the exact time a sensor takes to reach 90% of its final reading after the water chemistry suddenly changes. The industry originally wrote this rule for electrochemical pH meters, but everyone uses it for process sensors now.

Sometimes a manufacturer will print T63 instead. That is the time it takes to hit 63.2% of the final reading. To convert it, just remember: T90 is roughly 2.3 times the T63. If a brochure claims T63 = 7 seconds, the sensor actually takes about 16 seconds to reach T90.

Companies like Mettler Toledo and Endress+Hauser sometimes use T95 (95% of the final value), which is roughly 3 times the T63. But here is the catch with all these numbers: they are measured in a laboratory. The temperature is perfect, the flow rate is steady, and the pressure is controlled. When you bolt that same sensor into a raw effluent channel, it will respond slower because temperatures swing and the lens gets covered in biofilm.

Whenever you look at a T90 spec, look for the fine print about how they tested it. What was the water temperature? What calibration fluid did they use? How fast was the flow? Change the test conditions, and you change the number.

Real T90 Numbers Across SPARING Parameters

Let’s look at the actual T90 ranges from major manufacturers for the mandatory SPARING parameters (pH, COD, TSS, NH3-N, and flow):

Parameter	Sensor Technology	Typical T90	Notes
pH	Glass electrode	5–15 seconds	In-Situ <15 s, YSI <5 s, Hach <10 s
COD	UV254 (single wavelength)	1–10 seconds	MZD <2 s, ABB AWT420 <10 s
COD	UV-Vis spectrum (multi-wavelength)	up to 30 seconds	E+H CAS80E: spectrum scan + computation
TSS	Optical (scattered light)	1–30 seconds	Adjustable via averaging setting
NH3-N	ISE (ion-selective electrode) process	60–180 seconds	Hach AN-ISE <120 s, E+H CAS40D <120 s, YSI AmmoLyt <180 s

COD UV254 sensors like the RK500-25 are incredibly fast. They react in seconds. Glass pH electrodes do the same. But look at the bottom row. An NH3-N ISE sensor takes anywhere from one to three full minutes to produce a valid number. YSI is completely transparent about this—their AmmoLyt Plus spec sheet openly lists a T90 of under 180 seconds.

Ignore colorimetric (reagent-based) COD analyzers for this. Those machines take 5 to 15 minutes to run a single chemical reaction cycle. That is an intermittent cycle time, not a continuous T90. Because they are so slow, nobody uses them for fast-paced SPARING deployments anyway.

The 2-Minute SPARING Gap Nobody Talks About

This brings us to the technical loophole in PermenLHK P.80/2019. It forces you to pull data every 2 minutes, but it completely ignores how long the sensor takes to wake up and read the water.

Imagine you install an NH3-N ISE sensor with a T90 of 180 seconds. That sensor needs 3 full minutes to recognize an ammonia spike in the wastewater. But your GEOVOS 1000 data logger is programmed to pull a reading every 2 minutes to satisfy the regulators. At the 2-minute mark, the logger grabs the data. The ISE sensor is still climbing toward the real number. The bad data hits the compliance portal, the regulators see a green light, and you just logged a number that completely missed reality.

Look at how Europe handles this. The EU JRC Air Quality Sensor Protocol demands that a sensor’s T90 must be less than a quarter of the reporting interval. If we applied that logic to SPARING’s 120-second interval, a sensor would need a T90 of 30 seconds or less.

Under that rule, pH (5–15 seconds) and COD UV254 (1–10 seconds) pass easily. Optical TSS (1–30 seconds) usually survives. But process NH3-N ISE sensors (60–180 seconds) fail immediately. Some of them are slower than the logger’s 2-minute polling rate.

I am not saying you should avoid NH3-N ISE sensors. I am saying you need to know exactly how much your data is lagging. You have to factor this delay into how you judge water quality and how you run your field calibration checks.

T90 Gets Worse Over Time. Here is How to Catch It.

The T90 on the brochure is for a brand-new sensor in a clean lab. The second you drop it into a wastewater channel, fouling starts ruining that number.

Biofilm builds a physical wall across the sensor face. Ammonia ions have to push their way through that slime before they ever touch the electrode. Rinn, Sacher & Hedrich (2025) proved that the T90 of an ammonia sensor skyrockets during its first week underwater, before finally leveling off after a month. Samuelsson (2018) saw the exact same thing with optical sensors: biofilm blocks the light path, forcing the sensor to take longer to read anything.

Do not wait for the sensor to throw an error code. Watch the T90:

• The 2-minute DO limit: YSI specifically warns that if the T90 on a membrane DO sensor crosses the 2-minute mark, the membrane is dead and needs replacing immediately.
• Mettler Toledo’s 50% rule: If your T90 is now 50% slower than the day you bought it, you need to schedule preventive maintenance right now.
• Track the Baseline: Write down the T90 the day you commission the sensor. Check it again every time you run a routine calibration.

T90 is the heartbeat of your instrument. If your NH3-N sensor originally hit its mark in 120 seconds but now drags out to 180 seconds, your SPARING data is reporting history, not the present.

Ask These Questions Before You Buy

Before you sign a PO for any of the 12 mandatory SPARING industries, put the vendor on the spot with these three questions:

1. What is the exact T90, and what standard did you use to test it? Make sure they tested to IEC 60746, not some internal lab trick.
2. What were the test conditions? Temperature, concentration, and flow speed change everything. Ask to see the actual lab certificate, not just the marketing PDF.
3. Do you provide a procedure for checking T90 in the field? A vendor who knows what they are doing will give you a baseline and a monthly check procedure so you aren’t guessing when the sensor starts to fail.

Fortuna Argatech builds SPARING systems with pH, COD, TSS, NH3-N, and flow sensors that meet PermenLHK P.80/2019. We use the GEOVOS 1000 data logger to lock in the 2-minute reporting interval. Our RK500-25 COD sensor uses UV254 technology, which reacts in seconds. If you need to talk about how sensor selection and T90 affect your measurement uncertainty, reach out to the Argatech technical team.