Sensors & Detection — Technical Specification Overview

What Actually Fails When Sensor Signal Specs Are Wrong #

A food processing plant in Southeast Asia sourced 200 units of analog pressure sensors from a Chinese supplier. The datasheets showed 4–20 mA output, 0–10 bar range, ±0.5% FS accuracy. On the bench, every unit passed. In the line, over a six-week production ramp, their PLC started logging erratic batch weights and intermittent process alarms. The root cause was not the sensing element. It was output signal drift under thermal cycling: the sensors were installed near steam lines, and at 65°C ambient, the analog output shifted by up to 1.8% FS — more than triple the rated accuracy. Nothing on the COA flagged it, because the acceptance test had been run at 23°C.

The compounder had substituted a lower-grade op-amp in the signal conditioning circuit. Visually identical board. Same firmware. The component swap was not documented in any revision record passed to the buyer. This is the kind of substitution that a standard functional test at room temperature will not catch — you need thermal soak testing before you approve volume production, and that is a step that procurement teams regularly skip because it adds two weeks to the qualification timeline.

The cost of that shortcut, in this case, was a six-week production disruption, a 100% rework of installed sensors, and an airfreight premium for replacement units from a backup supplier in Japan.

The Signal Parameters That Predict Integration Risk #

The sensing technology — inductive, capacitive, photoelectric, ultrasonic — rarely drives integration failures on its own. What drives failures is the output signal interface, and specifically how well the Chinese supplier’s signal conditioning circuit performs under real operating conditions. Here are the parameters I’d prioritize in any sensor procurement review.

Output type and logic level. PNP, NPN, and push-pull digital outputs are nominally standardized, but output saturation voltage matters. A sensor rated “PNP, 10–30 VDC” may have a residual voltage of 2.8 V in the off state, which some PLCs with 24 VDC inputs will read as a logic HIGH. Per IEC 61131-2, the Type 1 digital input threshold is V_IH ≥ 15 V and V_IL ≤ 5 V — a sensor with 2.8 V residual sits within spec for the standard but still causes false triggers depending on PLC input impedance. We flag this in our supplier assessment as a Category B signal risk.

Analog output linearity and thermal drift. For 4–20 mA current loops, the two numbers that matter are linearity (expressed as % FS, measured across the full span) and temperature coefficient of the output stage (typically expressed as % FS/°C or ppm/°C). A linearity of ±0.5% FS at 23°C is meaningless if the temperature coefficient is 200 ppm/°C and the installation runs at 70°C — that adds another 0.94% FS of error at operating temperature. Across our incoming inspection data from 23 lots over 14 months, we consistently see Chinese suppliers rate linearity at ambient but omit the thermal coefficient entirely on COAs.

IO-Link device profile and process data mapping. IO-Link is increasingly specified for smart factory applications, and Chinese suppliers have entered this space aggressively over the past three years. The risk here is not connectivity — IO-Link physical layer compliance per IEC 61131-9 is usually fine. The risk is the IODD (IO Device Description) file. We have evaluated 11 Chinese IO-Link sensors, and four had IODD files with incorrect process data byte mapping, which caused incorrect scaling in the master’s parameter server. The sensor worked; the data was wrong.

Switching frequency and response time. For inductive and capacitive sensors used in high-speed counting or positioning, switching frequency (Hz) and the related ON/OFF response time (ms) are the binding parameters. A sensor rated at 5 kHz switching frequency with a stated response time of 0.1 ms is consistent; a supplier quoting 5 kHz with a response time of 0.4 ms is internally contradictory and signals datasheet inflation.

Repeat accuracy vs. absolute accuracy. These are not the same. A sensor with ±2% FS absolute accuracy but ±0.1% FS repeatability is excellent for differential measurements and trend monitoring. It is inadequate for absolute position reference in a closed-loop system. Chinese datasheets frequently conflate the two terms or report only the more favorable figure.

The parameter that procurement teams most commonly overlook is thermal drift of the output stage. Sensing range and accuracy at ambient are straightforward to verify. Thermal performance requires either a climate chamber test or a supplier-provided thermal characterization report — and that report needs to show the test conditions, not just the result.

The gap between entry-grade and export-grade is not primarily the sensing element — it is the signal conditioning and EMC hardening. When sourcing sensors and detection products from China, that is the distinction to force the supplier to explain, not just assert on the datasheet.

Decision Framework: Matching Grade to Application Risk #

If the sensor output feeds a safety-rated function — machine guarding, emergency stop confirmation, overpressure shutdown — the grade question is settled before you start. You need a sensor with a certified safety integrity rating per IEC 62061 or EN ISO 13849-1, and that narrows the Chinese supplier pool to a small number of manufacturers who have invested in functional safety certification. For most applications we evaluate, Chinese-made sensors are not qualified for SIL 2 or PLd safety functions, and the buyers who discover this late in the project are the ones paying for system redesign.

If the output feeds a closed-loop control loop — PID temperature control, flow regulation, servo positioning feedback — analog linearity and thermal stability are the governing specs. For a 0–10 bar analog pressure sensor driving a proportional valve, I’d set a hard threshold of ±0.5% FS linearity and ≤150 ppm/°C thermal coefficient as minimum qualification criteria. Entry-grade Chinese sensors rarely meet both simultaneously. Mid-grade Chinese sensors sometimes do, but require thermal soak verification before approval.

If the sensor is used for presence/absence detection or part counting in a low-speed application (under 500 Hz switching, ambient temperature stable within ±15°C), entry-grade Chinese sensors are often adequate. The specification that matters here is mechanical housing durability and ingress protection — verify IP67 or IP68 with an actual test certificate referencing IEC 60529, not a printed rating on the label.

For IO-Link applications, I’d apply a different screen entirely: require the supplier to provide the validated IODD file and demonstrate correct process data mapping in your specific master before placing any volume order. The physical connection will work. The data mapping is where integration time disappears.

One boundary condition worth stating explicitly: everything above applies to standard industrial sensor categories. Specialized variants — ultrasonic level transmitters with SIL ratings, radar displacement sensors, vision-based detection systems — have their own qualification paths and their own failure modes. The framework here is not universal.

Practical Guidance for Buyers #

When sourcing industrial sensors from China, the first document to request is not the datasheet — it is the thermal characterization report showing analog output drift across the rated operating temperature range. The datasheet accuracy figure is tested at 23°C; your installation probably is not at 23°C. Without the thermal data, you cannot calculate total measurement error in application.

The risk scenario we see repeatedly: a buyer qualifies a sensor on initial samples tested at ambient, approves volume production, and then sees process variability increase over the first seasonal temperature change. The sensor is within datasheet spec. The application thermal budget was never reviewed. The problem is caught at the six-month production review, not at incoming inspection.

Before committing to volume, run a qualification batch of at least 30 units through a thermal soak cycle matching your installation environment: hold at rated maximum operating temperature for 4 hours, measure analog output at 0%, 50%, and 100% of range, and compare against the ambient baseline. A drift exceeding ±0.5% FS at operating temperature is a disqualifying condition for closed-loop control applications. For presence/absence applications, the threshold is less critical — but switching function under thermal stress should still be verified with a minimum of 10,000 test cycles at temperature. Log the results under your incoming qualification record (we track this as QR-14 thermal acceptance in our sensor qualification workflow) before approving the supplier for series production.

FAQ #

Does IO-Link compliance guarantee correct data output from a Chinese sensor?
No. Physical layer compliance per IEC 61131-9 is verifiable and Chinese manufacturers generally pass it. The IODD file and process data mapping are a separate issue — four of the eleven IO-Link sensors we evaluated had mapping errors that caused incorrect engineering unit scaling in the master. Always validate the IODD against your master’s parameter server before volume approval.

What switching frequency should I specify for inductive sensors used in conveyor counting applications?
It depends on target speed and gap time between parts. At 60 parts per minute with a 50 mm target, a 200 Hz switching frequency has comfortable margin. At 300 parts per minute with 10 mm gap, you need 2 kHz or above. The number on the datasheet is theoretical maximum; derate by 40% for real installation tolerance.

Can Chinese analog sensors meet ±0.5% FS accuracy for closed-loop control?
Mid-grade and export-grade suppliers can meet this at ambient. The variable I cannot confirm without application-specific data is whether they maintain it across your full temperature swing — our dataset only covers sensors tested between 0°C and 70°C. For installations above 80°C continuous, I’d insist on supplier-provided thermal characterization data before committing.

Is PNP or NPN output more common in Chinese PLC installations?
PNP is dominant in newer Chinese automation installations, which mirrors European practice. NPN remains common in legacy Japanese-influenced systems. If you are specifying sensors for a mixed PLC environment, push-pull output eliminates the selection problem entirely and costs nothing extra at mid-grade tier.

Why do some Chinese sensor datasheets show higher accuracy than European equivalents at the same price?
The number is usually real at ambient and under ideal conditions. The difference is that European suppliers test and guarantee performance across the full rated temperature and EMC environment per IEC 61000-4-3. Chinese entry-grade datasheets often reflect best-case bench conditions with no statistical basis across production lots. Ask for Cpk data on the accuracy parameter across the last three production lots — most cannot provide it.

Published by sinoraw.com Technical Team | Request a sourcing consultation