On a brownfield PLC upgrade, a capacitive proximity sensor is normally chosen over an inductive sensor when the target is non-metallic — plastic pellets, wood, cardboard, glass, paper, oil, or the level through a sight-glass or tank wall — and a typical product family from suppliers such as Lanbao lists six functional variants: standard, extended distance, frequency-enhanced, corrosion-resistant, pipeline liquid-level, and high-temperature material-level [S2].
The same site groups the sensors by housing — M12, M18, M30, plus rectangular and flat-block styles — so the thread-out cutouts on existing brackets and conveyor side-frames do not have to be re-machined [S2].
Match the variant to the physical job before anything else
Capacitive variants differ more in usable sensing distance and in environmental tolerance than in electronics. The standard function class is the swap-in for simple object detection at short range (typically a few millimetres on a metallic target, less on plastic). Extended-distance units roughly double that nominal Sn and are the right pick when the target is small, when the mounting geometry is fixed by legacy brackets, or when the target sits a few mm further than the original inductive switch could see [S2].
Frequency-enhanced capacitive sensors are the upgrade pick for high-speed conveyors and small-object detection — they run a higher oscillator frequency so the response time is short enough for label, splice, and tab detection at line speeds above about 60 m/min. Corrosion-resistant variants, almost always stainless or PTFE-faced, replace legacy inductive switches on washdown bottling and CIP skid retrofits; the housing chemistry, not the sensing principle, is the spec driver [S2].
For brownfield tank and hopper work, two families dominate. Pipeline liquid-level capacitive sensors mount through a gland into the pipework and detect water, coolant, or light oils through a stainless or plastic sight-glass wall — the same gland boss is normally present from a previous float or conductivity probe, so the upgrade is mechanical reuse, not new pipework. High-temperature material-level units extend that duty up to siloed powder, hot asphalt, and molten-resin service where the housing temperature rating is the first thing to check against the existing purge air or heat-sink arrangement [S2].
Wiring and PLC I/O compatibility — where brownfield upgrades actually fail
The electrical half of the upgrade is binary: the new capacitive sensor must produce the same output type the existing discrete input card is wired for. PNP (sourcing) and NPN (sinking) are the two DC three- and four-wire flavours seen on retrofits; a 24 V DC PNP sensor will not register on an NPN input and vice versa, and mis-wiring is the most common reason a retrofit "doesn't work" even when the sensor itself is correctly ranged. Open-collector two-wire units are still seen on older European lines and need a leakage-current check against the input card's OFF-state threshold before sign-off [S2].
Capacitive units are used for the non-safety auxiliary signals feeding the standard PLC (level good, label seen, splice detected), not the E-stop or guard-door path. The brownfield decision then is to keep the legacy safety chain untouched and add the new capacitive signals to the spare points on the standard I/O rack [S2].
For IO-Link retrofits on brownfield plants, capacitive sensors with an IO-Link COM3 interface can sit on the same 3-wire M12 drop cable as the original discrete sensor and feed diagnostics back to the PLC without rewiring the trunk. The catch is the field gateway: an existing brownfield PLC that does not support IO-Link on the input card needs a 4- or 8-port IO-Link master upstream, and the cable length is constrained to the standard 20 m COM3 limit, which usually forces the cabinet relocation otherwise avoided in a brownfield scope [S2].
Looking at the six functional classes against four decision criteria — sensing distance, housing material, response time, and the typical brownfield replacement — gives a usable spec map. Standard function: short Sn, nickel-plated brass or plastic, normal response, drop-in for M12/M18 inductive swaps. Extended distance: roughly 2× Sn, same housing options, drop-in when the old inductive was a long-range LR08/LR12 unit. Frequency-enhanced: same housing options, response time in the sub-millisecond class, the only choice for high-speed label and splice detection on packaging retrofits. Corrosion-resistant: stainless body, lower Sn penalty, mandatory on washdown and CIP. Pipeline liquid-level: pipe-thread gland, the swap for a legacy float or conductivity probe in the same boss. High-temperature material-level: stainless body, derated Sn at high temperature, the only choice for hot bulk-solid service [S2].
For OEM-style proof-of-concept work on a brownfield prototype, the open-source CapacitiveSensor library (artefact-group) on GitHub remains a useful reference for the charge-time / discharge-time measurement principle that commercial capacitive proximity ICs are built on, even though it targets an Arduino-class MCU rather than an industrial PLC front-end [S1].
Selection criteria for the controls engineer
Five criteria reliably sort the candidates. Second, sensing distance: pick a unit whose nominal Sn is at least 1.5× the worst-case installation gap, so that build-up on the active face does not push the target out of range before the next planned cleaning [S2].
Third, housing and ingress: on a brownfield washdown site, IP67 is the floor, IP69K is the right ask for bottling and CIP skids. Fourth, electrical interface: PNP/NPN, two-wire, NAMUR, or IO-Link must match both the existing cable and the input card; mis-spec here is the most common rework cause.
Who should NOT pick the mainstream capacitive option
Capacitive sensors are the wrong pick when the target is metallic and the required sensing distance is long — an inductive sensor is cheaper, more repeatable, and immune to humidity and material build-up on the active face. They are also the wrong pick where the wall material is thick conductive metal: a capacitive sensor sees the wall, not the target, and the only fix is a flush-diaphragm or guided-wave approach, which moves the spec out of the standard product family [S2].
For non-contact distance or position feedback on a retrofitted servo axis, a capacitive point sensor is also a poor substitute for a displacement sensor; the latter is specified for analogue output over a defined range, not for a binary threshold at a single point. On brownfield work where the upgrade is a packaging or bottling line, the drag chain cable selection criteria for packaging line retrofit and the industrial buzzer selection criteria for packaging line retrofit are the typical downstream constraints on the same upgrade window.
Failure modes that show up after a brownfield swap
Build-up of product or condensate on the active face shifts the effective Sn and can hold the output in the ON state until the next washdown; the mitigation is a sensitivity pot trimmed to the dry, clean state so that the soiled state still releases. Humidity drift on outdoor or unconditioned sites moves the trip point by a few percent per day, which is invisible on a single threshold but adds up on a slowly-filling tank — the fix is a unit with temperature compensation, available in the higher-tier extended-distance and high-temperature families. EMI from a VFD on the same cable tray adds high-frequency noise that can hold a capacitive oscillator in saturation; the fix is the same as for any other sensor — segregated routing and, where the existing tray cannot be re-run, a screened cable with the screen bonded at the cabinet gland only [S2].
Final shortlist logic for the controls engineer
Use a three-question shortlist. Q1 — is the target non-metallic AND the duty binary? If no, pick an inductive sensor for metallic binary work, or a displacement sensor for analogue position feedback. Q2 — is the duty washdown, hot, or corrosive? If yes, restrict to the corrosion-resistant, pipeline liquid-level, or high-temperature variants and check the IP/IP69K rating against the existing cabinet [S2].
Q3 — does the existing input card accept the sensor output? PNP for a PNP card, NPN for an NPN card, NAMUR for a NAMUR card, IO-Link for an IO-Link master. With those three answers, the controls engineer can drop one of the six capacitive families into the spare points on the existing rack without re-engineering the cabinet, the cable, or the safety chain — which is the whole point of a brownfield upgrade [S2].