Category: Applications

Conveyor sensing looks like one problem and is really two, with almost opposite requirements. Detecting the product on the belt is a precision job close to the material. Protecting the conveyor itself is a reliability job that has to work for years without attention. Specifying both the same way is the usual reason one of them keeps failing.

Job one: detecting the product

Counting, presence, gap detection, triggering a diverter or a reject. What matters here is the target.

If the product is metal, an inductive sensor is the cheapest and most robust answer. It ignores dust between itself and the target, which matters on an open conveyor.

If the product is not metal, which covers cartons, bottles, bags, plastics and timber, you need capacitive at short range or photoelectric at longer range. Photoelectric in through-beam mode is the reliable choice across a wide belt, because it does not depend on the product reflecting anything back. Diffuse mode on a dark or matt product is where most intermittent counting faults come from.

If the belt is wide, the distance decides the technology. Inductive stops being an option past about 30 mm. Our note on extended-range inductive sensing covers where that boundary sits and when to stop pushing it.

If the line is fast, check switching frequency against the actual product rate, and check that the controller input can keep up. Counts that are correct at low speed and drift low as the line speeds up are almost always the input scan rate rather than the sensor.

Job two: protecting the conveyor

This is the higher-value sensing, because a conveyor failure costs far more than a miscount.

Underspeed and belt slip. A drive turning while the belt is not is the classic conveyor failure. It wears the drive pulley, and in dusty plant the friction is a genuine fire risk. Detection is a magnetic or inductive sensor reading a toothed wheel or bolt heads on the tail pulley shaft, compared against the drive. Our guide to speed and underspeed monitoring covers target geometry and the controller requirements, which matter more than the sensor choice.

Belt drift and tracking. A belt running off to one side damages itself and the structure. Sensors either side of the belt edge detect it before it becomes expensive.

Blocked chute and transfer point detection. Where material backs up at a transfer, early detection prevents spillage and a cleanup. This is among the harshest positions on the whole conveyor: material flow, impact and dust all at once. Mount with standoff and protect mechanically.

Pull cord and guard status. Safety-relevant, and specified normally closed so a severed cable produces the stopped state.

What the environment does

Dust is the main enemy for optical sensing. Lenses foul, and in cement or grain handling they foul quickly. Where photoelectric is unavoidable, use through-beam, specify considerably more range than the gap requires so there is margin to lose as the lens dirties, and plan cleaning access. Inductive and magnetic sensing largely ignore dust, which is why they dominate on conveyor protection.

Vibration loosens mountings. A sensor that drifts out of position reads intermittently and gets blamed on electronics. Use locknuts both sides and check alignment as routine maintenance rather than after failure.

Impact happens. Material falls off belts. A guard bracket on any sensor in the drop zone costs very little.

Long cable runs. Conveyor galleries are long, and a sensor can be well over a hundred metres from the panel. Check the supply voltage actually reaching the sensor, use screened cable earthed at one end, and keep signal cable away from drive cabling. Where an analogue signal is involved over that distance, specify a current output rather than voltage.

Where this comes up

The protection requirements are broadly the same whether the conveyor is moving packages, ore, clinker, bagasse or grain. The product detection requirements change completely. Our industry guides cover the specifics for cement, sugar, steel and packaging.

Specifying

Tell us which of the two jobs you are solving, the product material and size, the belt width and speed, the mounting distance available, the environment, and what the output feeds. If this is underspeed protection rather than monitoring, say so, because the fail-safe requirement changes the specification.

On a machine tool the sensors are rarely what the machine is about, and they are reliably what stops it. The cutting process is controlled and repeatable. The environment around it, flood coolant, hot swarf, vibration and an operator with a hammer, is what sensors actually have to survive.

Where they go

Axis homing and limits. The reference position each axis returns to, and the overtravel limits either side. Repeatability matters here more than anywhere else on the machine, because the whole coordinate system is built on it. A homing sensor that drifts by half a millimetre puts every part out by half a millimetre.

Tool changer position. Carousel or turret indexing, arm position, tool present detection. Usually several sensors working as a sequence, where one misread stalls the whole change.

Chuck and clamp confirmation. Workpiece clamped, collet closed, tailstock engaged. These are interlocks, not conveniences, and they should be specified normally closed so that a broken cable prevents the spindle starting rather than allowing it.

Guard and door interlocks. Safety-relevant, and the same fail-safe logic applies.

Coolant and lubrication level. Usually capacitive sensing through the tank wall where the tank is non-metallic, which avoids a penetration and anything wetted.

Chip conveyor running. Rotation or underspeed confirmation, so a stalled conveyor is detected before chips back up into the machine.

What actually breaks them

Flood coolant. Directed at the cut, which means directed near the sensors. This is closer to a washdown condition than a splash condition, and IP67 rated for immersion is not the same as resisting a directed jet. Our note on what each IP rating actually tests is worth reading before specifying.

The usual failure is at the cable entry. Point the cable entry downward wherever possible. An upward-facing gland in a coolant environment collects standing fluid and fails eventually regardless of rating.

Coolant chemistry. Modern cutting fluids are chemically active, and they attack cable sheaths before they attack housings. If a sensor cable has gone hard and cracked, that is the coolant rather than the heat. Name the coolant when you enquire if you have had this before.

Swarf. Hot, sharp, ferrous and magnetic. It accumulates on an inductive sensing face and reads as a permanent target. Mount faces downward or vertically so swarf falls away, and keep them reachable for cleaning, because they will need it.

Mechanical damage. A sensor inside the working envelope will eventually be struck by a tool, a part or an operator. Where it cannot be moved out of the way, a simple guard bracket costs very little.

Specification habits that work on machine tools

• Shielded sensors, flush mounted in the machine casting wherever possible, so there is nothing to catch and nothing standing proud. See shielded versus unshielded.
• Normally closed for every interlock. Clamp confirmation, door, guard. A cut cable should stop the machine.
• Short-circuit protected outputs throughout. Cable damage inside a machine tool is routine.
• Design the gap at 70 to 80 percent of rated sensing distance, not at the limit. Temperature and supply variation both move the effective distance, and a homing sensor at the edge of its range is a repeatability problem waiting to happen.
• Connector versions where sensors are changed often, fixed cable in the right length where they are not.
• Standardise. A machine shop with four machines and fifteen sensor part numbers cannot hold meaningful spares. Rationalising to a handful means a breakdown is a ten minute job rather than a three day wait.

Target material matters more than people expect

Machine tool targets are frequently not mild steel. Cast iron, aluminium fixtures, stainless components and hardened tool steel all read differently, and the catalogue sensing distance is quoted against a standard mild steel target. Specifying an M12 with 4 mm sensing for an aluminium flag at 3.5 mm is a fault built into the design. The reduction factor table gives the corrections.

Where the target cannot be changed, the usual fix is a larger body with more sensing distance, which also buys clearance for swarf. The D30C range runs from M8 to M50.

Getting it specified

Send us the function, the target material and size, the available gap, whether the sensor sits in the coolant path, the output your control expects, and the mounting space. If you are replacing a sensor that keeps failing, tell us how it fails and how often, because the pattern usually identifies the cause faster than the specification does.

A resistance welding line is one of the few places where a proximity sensor can fail for two completely unrelated reasons at the same time, and where fixing one does nothing for the other. If you are replacing sensors on a weld cell more often than anywhere else in the plant, this is usually why.

Problem one: the magnetic field

Resistance welding passes very large currents for very short times. That current creates a strong magnetic field around the weld gun, the cables and the transformer, and an inductive proximity sensor works by sensing changes in an electromagnetic field.

The result is false switching. The sensor reports a target when the gun fires, or drops out when it should not. The characteristic symptom is a fault that only happens during welding and cannot be reproduced when the cell is jogged by hand, which is exactly the fault profile that gets blamed on the PLC for a year.

What actually helps:

• Distance from the weld cables. The field falls off quickly. Routing sensor cable away from weld cable, and moving the sensor itself where possible, does more than any specification change.
• Screened cable, earthed at one end only. Earthing both ends creates a loop that makes things worse.
• Never run sensor cable parallel to weld cable in the same tray or conduit. Cross at right angles if they must meet.
• Weld field immune construction, where the sensor electronics are designed to reject the interference rather than respond to it. This is the specification to ask for where the geometry cannot be changed.

Problem two: the spatter

Molten metal thrown from the weld lands on the sensor face, sticks, and builds up. Two things then happen. The sensing face accumulates ferrous material, which an inductive sensor reads as a permanent target. And the build-up physically damages the face.

What actually helps:

• A PTFE coating on the sensing face and barrel. Spatter does not adhere well to PTFE, so most of it falls away instead of sticking. This is the single most effective measure and it is the usual meaning of a spatter-resistant sensor.
• A ceramic or spatter-resistant face where the thermal exposure is severe.
• A mechanical deflector positioned so the sensor is out of the direct line of the spray. Cheap and frequently overlooked.
• Mounting the face downward or vertically wherever the geometry allows, so material falls off rather than collecting.

Note that these are two separate specifications. A sensor with excellent weld field immunity and a bare steel face will still fail on spatter, and a PTFE-coated sensor with ordinary electronics will still false trigger. On a weld cell you generally need both.

Problem three, which is really the cable

As in most hostile installations, the cable fails before the sensor does. On a weld cell it is exposed to spatter, to heat, and usually to movement if the sensor is on a moving fixture. A cable that is chafed, cooked or crushed produces exactly the same intermittent symptoms as weld field interference, so check it before changing the sensor specification.

Specify the cable length you actually need so the first joint is outside the cell, use a protective sleeve through the spatter zone, and support the cable so the gland carries no weight.

Where this applies

Automotive body shops and tier-one component lines are the obvious case, with sensors on weld fixtures, clamp position, part present detection and gun position. But the same two problems appear on general fabrication cells, on wire mesh and rebar welding, on seam welders in drum and cylinder manufacture, and anywhere a resistance welder sits close to automated handling.

Diagnosing which problem you have

Useful before buying anything:

• Fault only during welding, sensor face clean points to the magnetic field. Look at cable routing first.
• Sensor stuck on, visible build-up on the face is spatter. A PTFE face and a deflector will do more than better electronics.
• Fault that comes and goes with fixture movement is usually the cable.
• Sensor dead altogether is often the output destroyed by a short in a damaged cable, which is an argument for always specifying short-circuit protected versions.

Specifying one

Tell us the distance from the weld gun and from the weld cables, whether the sensor is in the direct spatter path, the target material and sensing gap, how the cable is routed, and whether the sensor moves with a fixture. A photograph of the cell with the sensor position visible answers most of it.

We manufacture proximity switches in Mumbai and build specials including coated faces and specified cable lengths, which on a weld line is usually the right route rather than adapting a catalogue part.

Most level enquiries that reach us do not actually need level measurement. They need to know when a tank is full, or when it is empty, so that a pump starts or stops or an alarm sounds. That is point level detection, and it is a much simpler and cheaper problem than continuous measurement.

Establishing which one you have is the first decision, because it changes everything that follows.

Point level or continuous?

Point level — a switch. Output changes at one defined height. Covers pump control, overflow prevention, dry-run protection, high and low alarms. This is the large majority of industrial requirements.

Continuous — a transmitter. Reports the level as a value across the range. Needed for inventory management, batching by volume, and closed-loop level control. Considerably more expensive.

Two or three point switches at the heights you care about frequently do the job that someone specified a continuous transmitter for, at a fraction of the cost and with far less to go wrong.

The wall material decides the technology

For point level in a tank, the single most useful question is what the tank is made of.

Non-metallic tank: capacitive, mounted externally

This is the one people are surprised by, and it is genuinely the best answer for a large class of applications.

A capacitive proximity switch mounted on the outside of a plastic, fibreglass or glass tank can detect the liquid inside through the wall. Set the sensitivity so the empty wall does not trigger it but wall plus liquid does, and you have a level switch with:

• No penetration of the tank
• No seal that can leak
• Nothing wetted, so nothing to clean and nothing to corrode
• No mechanical part inside to jam
• Free repositioning — move the bracket, change the switching height

For water tanks, chemical dosing tanks, plastic storage tanks and fibreglass vessels this is frequently the correct and cheapest solution. Our D32C range is M30 with 10 mm sensing, in NPN and PNP, normally open and normally closed.

The caveat, which matters. The wall consumes part of the sensing range, so thick walls may be beyond it. And if product clings to the inside of the wall, the sensor reads full permanently. Thin water-like liquids are ideal; heavy clinging syrups and slurries frequently are not. Assess this before ordering — see capacitive sensors for level detection.

Metal tank: the sensor has to be inside, or use a float

A metal wall blocks capacitive sensing entirely. Two routes remain.

A fitting through the wall, with the sensing element inside. Effective, but now you have a penetration and a seal, and the element is exposed to the process.

A magnetic float with an external sensor. A float carrying a magnet rides on the liquid inside a non-ferrous stilling tube or chamber, and a magnetic proximity switch outside detects it. Magnetic fields pass through stainless steel and aluminium, so this works where capacitive sensing cannot. Nothing electrical is inside the vessel.

For stainless steel process vessels and hygienic applications this is often the right approach.

What the liquid is doing matters

Four properties that change the answer:

Conductivity and dielectric constant. Water and aqueous solutions are detected easily by capacitive sensing. Oils and solvents have lower dielectric constants and are harder — possible, but the margin is smaller and setup is more critical.

Coating behaviour. Anything that clings to the wall or to a probe is the main enemy of level sensing generally. If the product coats, design around it.

Foam. Foam has a dielectric between air and liquid and can trigger a capacitive sensor. Where foaming is expected, either mount in a stilling arrangement or choose a technology that ignores it.

Turbulence. A sensor at a fill point sees splashing and chatters. Mount away from inlets, or add hysteresis in the control logic.

Powders and granulates

Capacitive sensing works for solids too, and hopper level, silo high and low level, and blocked chute detection are common applications. Two differences from liquids: solids have lower dielectric constants than water so detection is less strong, and solids do not self-level — material can bridge above a sensor and leave it reading empty when the hopper is not.

Mount where material actually flows, not in a corner where it can arch over.

What we supply, and what we do not

Being direct, because the search term that brings people here is broad. We manufacture capacitive and magnetic proximity switches in Mumbai and supply them for point level detection. That covers most industrial tank and hopper switching requirements in India.

We do not manufacture continuous level transmitters — radar, ultrasonic or guided wave. If your application genuinely needs continuous measurement across the full range, that is a different product category and we would rather tell you than sell you three point switches and call it level measurement.

What is worth checking first is whether you actually need continuous measurement. A great many specifications call for it out of habit when two point switches would control the pump perfectly well.

Specifying a level switch

• Tank material and wall thickness
• The liquid or solid, and whether it coats, foams or is abrasive
• Switching heights required, and how many
• Whether anything may be inside the tank, or whether it must be entirely external
• Temperature and pressure
• Cleaning regime, if any
• Output required and what it controls

A sketch of the tank with the switching heights marked answers most of this at once.

Speed monitoring covers three distinct jobs that get treated as one: counting pulses to measure rate, detecting that something has stopped or slowed when it should not have, and establishing which way a shaft is turning. The sensor requirements differ, and specifying for the wrong one is a common cause of intermittent faults.

Underspeed detection is protection, not monitoring

Worth separating clearly. A speed readout on an HMI is information. An underspeed trip on a bucket elevator or a mill roller is a protection function — it exists because a stalled or slipping drive under load damages itself, and in dusty plant can start a fire.

Protection functions deserve the fail-safe treatment. Ask what happens when the sensor cable is cut: if the answer is that the machine reads zero speed and trips, good. If it reads nothing and the logic interprets nothing as fine, the protection is not protection.

What limits the maximum countable speed

Three things, and only one of them is the sensor.

The sensor switching frequency. A rated figure in hertz. The sensor cannot produce more transitions per second than this, and running near the limit gives degraded, uneven pulses rather than a clean failure.

The target geometry. This is usually the real constraint and it is under your control. Pulses per second equals teeth multiplied by revolutions per second. A 60-tooth wheel at 3000 rpm demands 3000 Hz; a 6-tooth wheel at the same speed demands 300. If you are near the sensor limit, reducing the tooth count is easier than changing the sensor — provided the resolution still suits the application.

The controller input. An ordinary PLC digital input scanned every 10 ms cannot count 3000 Hz. It needs a high-speed counter input or a dedicated speed monitoring relay. This is the step most often missed, and it produces the characteristic symptom: counts correct at low speed, drifting low as speed rises.

Target design matters more than people expect

For an inductive or magnetic sensor counting a toothed wheel, the tooth and gap need to be comparable in size to the sensing face. Teeth that are too fine relative to the sensor produce a weak, mushy signal because the sensor never fully sees a gap.

Rules of thumb worth applying: make tooth width and gap width roughly equal, keep both at least as wide as the sensor face, and mount at a consistent air gap well within the sensing distance. A wheel that runs out radially gives a varying air gap and an output that is fine at one point of the revolution and marginal at another.

Direction sensing

A single sensor on a toothed wheel tells you the shaft is turning, not which way. Where direction matters — a reversible conveyor, a hoist, a mill that must not run backwards — you need two channels in quadrature, so the controller can compare which one leads.

The speed-and-direction variants in the D33C range provide this in a single sensor rather than requiring two units to be mechanically aligned. Given how difficult it is to hold quadrature alignment between two separately mounted sensors in an industrial environment, that is usually the better engineering answer.

The part numbers carry an SD suffix — D33C-18MD-NM-3S-SD is the M18 version, D33C-30MD-NM-3S-SD the M30.

Mind the output current

One specific and avoidable mistake. The D33C range is rated 30 mA, not the 200 to 300 mA of the general-purpose switching ranges. It is built to feed a controller input, not to drive a contactor coil or a relay directly.

Wiring one to a relay coil will either fail immediately or, worse, work marginally and fail later. If the speed signal needs to switch a load, take it through a controller or a speed monitoring relay. Our part number guide lists the output rating for each series.

Choosing the sensing technology

Inductive — for a ferrous toothed wheel, gear or keyway at moderate speed. Robust, cheap, tolerant of dust between sensor and target. The D30C range covers it.

Magnetic — senses through non-ferrous material, so the sensor can sit outside a stainless or aluminium cover with nothing exposed. This is why magnetic types dominate in wet, dusty and washdown positions.

Photoelectric — where the target is non-metallic or the standoff is large. Needs a clean environment, which rules it out of most speed monitoring positions in heavy industry.

Typical applications

• Bucket elevator underspeed — protection against belt slip, in dusty and fire-sensitive plant
• Conveyor belt slip and drift
• Mill roller underspeed in sugar and steel plants
• Fan and blower rotation confirmation, often interlocked with a process
• Reversible drive direction confirmation
• Production counting on packaging and assembly lines
• Shaft rotation confirmation on pumps and gearboxes

Specifying

Send us the maximum and minimum speed, the target — material, tooth count, tooth and gap dimensions, or whether it is a single keyway — the available air gap, whether direction is needed, the environment, and what the signal feeds. If the application is an underspeed trip rather than a readout, say so, because it changes the fail-safe requirement.

An inductive proximity switch sees metal and nothing else. That is not a limitation, it is the operating principle — it works by inducing eddy currents in a conductive target. Present it with a plastic bottle, a paper carton, a wooden pallet or a bin of powder and it will sit there indifferent.

Capacitive proximity switches solve that problem. They respond to a change in the dielectric in front of the sensing face, which means they detect essentially anything: plastic, liquid, powder, granulate, wood, glass, cardboard, and metal too.

How the sensing actually works

The sensing face forms one plate of a capacitor. The surroundings form the other. When a material enters the sensing field, the dielectric constant of the space in front of the face changes, the capacitance changes, and an internal oscillator circuit detects the shift and switches the output.

Two consequences follow, and they explain most of what is useful and most of what goes wrong.

Materials with a high dielectric constant are detected more easily. Water is detected extremely well. Metals behave as if they have an effectively infinite dielectric constant and are detected at the full rated distance. Dry plastics and powders have low dielectric constants and are detected at a reduced distance. This is why a sensor that detects a full bottle reliably may not detect an empty one.

The sensor cannot inherently distinguish target from contamination. A film of water, a dust build-up or a layer of product on the sensing face changes the dielectric just as a target does. This is the dominant failure mode in the field and it is a design problem rather than a device problem.

Sensing through a container wall

This is the application capacitive switches are genuinely unmatched at, and the one that brings most enquiries to us.

Mount the sensor on the outside of a non-metallic tank, hopper or pipe and set the sensitivity so that the wall alone does not trigger it but wall plus contents does. You then have a level switch with no penetration, no seal, no wetted part and nothing inside the vessel to clean. For food, pharmaceutical and chemical duty that is a substantial advantage.

What determines whether it will work:

• Wall thickness and material. The wall consumes part of the sensing range. A thin polypropylene wall costs little; a thick glass-reinforced one may consume most of it.
• The dielectric contrast between full and empty. Water against air is easy. Dry granulate against air is harder and needs care.
• Whether product clings to the inside of the wall. If it does, the sensor reads full permanently. This is the single most common reason a through-wall installation fails, and it must be assessed before ordering rather than after.

Our note on capacitive proximity sensors for level detection goes into the level application in more depth.

Where capacitive switches earn their place

• Packaging lines — detecting cartons, bottles, film and labels that inductive sensors cannot see.
• Level in non-metallic vessels — high and low level switching without breaching the tank.
• Powder and granulate handling — hoppers, silos, feed chutes and blocked-chute detection.
• Plastics processing — presence of mouldings, sprue detection, bin-full signalling.
• Timber and panel handling — board presence and edge detection.
• Glass handling — where optical sensing is defeated by transparency.

The Accent capacitive range

Our capacitive proximity switches are M30 bodies with 10 mm sensing range, available across the full output matrix. The ordering code is straightforward once you know the pattern:

• D32C-3010-NM-3S — NPN, normally open, 3-wire, 300 mA, short-circuit protected
• D32C-3010-NB-3S — NPN, normally closed, short-circuit protected
• D32C-3010-PM-3S — PNP, normally open, short-circuit protected
• D32C-3010-PB-3S — PNP, normally closed, short-circuit protected

The same four are available without short-circuit protection as the -3 suffix rather than -3S. In almost every case the protected version is the right choice — the price difference is small and a shorted output on an unprotected sensor destroys it.

Reading the code: N or P selects NPN or PNP, M or B selects normally open (make) or normally closed (break), and 3S indicates 3-wire with short-circuit protection. If you are unsure whether your input card wants NPN or PNP, our guide to PNP and NPN sensors covers it.

Two mistakes worth avoiding

Setting the sensitivity at commissioning and never revisiting it. A capacitive sensor adjusted on a clean, dry morning will behave differently in humid weather and differently again with six months of product film on the face. Set it with margin, and specify a mounting position where the face can be wiped.

Mounting it where splash or condensation lands on the face. Water has a very high dielectric constant. A sensor that gets splashed will trigger on the splash. Shield it, angle it downward, or move it.

Specifying one

Tell us the target material, whether you are sensing it directly or through a wall, the wall material and thickness if so, the required sensing distance, whether the target is stationary or moving and how fast, the output your controller needs, and the environment — washdown, dust, temperature. If the target is a powder or granulate, the bulk density is worth including.