Category: Sensor Selection

A paper machine runs for weeks between stops. Anything that causes an unplanned stop is expensive, and a sensor that fails in the wet end is both expensive and difficult to reach.

The sensing tasks themselves are mostly straightforward. The environment is what makes the specification non obvious.

The environment

Humidity and condensation. The wet end is saturated. Air that is humid at process temperature condenses inside anything cooler, including a sensor housing and a connector. Specify sealing properly and prefer a moulded fixed cable over a connector where you can, see connector or cable and IP ratings.

Paper dust. Fine, pervasive and mildly conductive when damp. It accumulates on every surface. For optical sensing this is the main problem, because a lens coated in dust loses signal. Favour through beam over diffuse wherever the geometry allows, because through beam has the most signal margin to lose.

Chemicals. The wet end involves bleaches, sizing agents and retention aids. Cable jackets and seals need to suit. State the chemistry in your enquiry.

Continuous running. You cannot stop to change a sensor. That pushes the decision toward connectorised parts in reachable positions, so a change takes minutes at the next brief stop, and toward holding spares for every fitted part, see the spares list guide.

The sensing tasks

Reel diameter and reel change. A sensor watching the reel surface or an arm that follows it, detecting when the reel has built to diameter or run down to core. An M30 inductive sensor on the follower arm is robust; the alternative is an analogue sensor or a position transducer on the arm, giving continuous diameter rather than a point.

Web break detection. Speed matters. A break has to be detected in the time it takes for the broken web to wrap a roll, which is a fraction of a second at machine speed. A through beam photoelectric across the web, or a speed sensor on an idler roll that stops turning when the web goes slack. The idler approach is often more reliable in dust, because it is an inductive sensor on a toothed wheel rather than an optical path through a dusty atmosphere.

Core and chuck position. Inductive sensing of the chuck engaged position, straightforward M18 work, specified normally closed so a failure reads as not engaged.

Splice detection. A metallised or foil splice tape detected inductively, or a visual splice detected photoelectrically. Tell us which the mill uses, because it decides the technology outright.

Roll rotation and underspeed. A D33C speed and direction sensor on each driven roll detects a roll that has stopped or slipped before it marks the web.

Guard and access interlocks. Note the caution in replacing limit switches: a guard interlock requiring positive opening action is not a proximity switch job.

Printing specifically

Printing presses share the dust and continuous running, and add two things.

Register and mark sensing is a contrast sensing task on a moving web at speed, which is a different class of optical sensor from a general purpose photoelectric switch. Tell us the mark colour, the substrate colour and the web speed when you enquire, because the answer depends entirely on contrast.

Solvent exposure in gravure and flexo. Housing material and cable jacket both need to suit the solvent in use, and this is worth stating explicitly rather than assuming a standard part copes.

Specification summary

• IP67 minimum through the machine, higher at the wet end
• Through beam over diffuse wherever the geometry allows
• Inductive over optical wherever a metal target can be arranged, because dust does not affect it
• Normally closed on anything that should stop the machine
• Cable supported within 100 mm of the sensor, protected where the web could wrap it
• Spares held for every fitted part number, labelled by machine position

Replacing what is installed

Most Indian mills run imported sensors specified by the machine builder, often decades ago and now slow or expensive to source. Send us the part numbers and we will cross reference, see cross referencing imported switches.

Where the original is genuinely special, we manufacture to specification in modest quantities, see custom and OEM manufacturing. For comparable heavy continuous process work, see our notes on cement plants and sugar mills.

Mechanical limit switches wear out. The actuator arm bends, the roller seizes, the contacts pit, and the switch either stops making or stops breaking. On a machine that cycles continuously that is a recurring maintenance item that a proximity switch removes entirely.

But the swap is not automatic, and doing it carelessly introduces faults that are worse than the one you solved.

Why the proximity switch usually wins

• No contact, no wear. Nothing touches the target, so there is no mechanical life to consume. The failure modes that dominate limit switch maintenance simply do not exist.
• No moving parts to seize. A roller lever in a dusty or wet plant seizes, and a seized roller skids and wears flat.
• Faster and repeatable. Operate point repeatability is far better than a lever and spring arrangement, which matters where the position gates a sequence.
• Properly sealable. A limit switch needs a moving shaft to pass through its housing, which is the point water gets in. A proximity switch is a sealed body with a cable, so IP67 and IP69K are achievable. See IP ratings.
• No impact damage. An overtravelling machine bends a limit switch lever. A proximity switch set back from the path of travel is untouched.

The three differences that bite

1. Contact rating. This is the big one. A mechanical limit switch has volt free contacts that will switch a contactor coil directly, often at mains voltage and several amps. A DC proximity switch has a solid state output rated a few hundred milliamps.

If the existing limit switch is driving a contactor, a motor or a solenoid directly, you cannot simply swap in a three wire DC proximity switch. You need either an interposing relay, or an AC two wire proximity switch rated for the load. Our AC two wire parts handle 200 mA, which covers most contactor coils but not a solenoid valve drawing more.

Check the actual load current before ordering. This is the single most common mistake in a limit to proximity conversion.

2. Fail safe behaviour. A mechanical NC contact opens when the lever is pressed, and also when the wire breaks or the switch is destroyed. That is inherently fail safe on a limit that stops motion.

A proximity switch has active electronics and needs a supply. Loss of supply gives no output, which on a normally open arrangement looks like the limit never being reached. Specify the normally closed variant on any limit that stops motion, and verify the behaviour on power loss rather than assuming it.

3. The target has to be ferrous. A limit switch is operated by anything solid. An inductive proximity switch needs a metal target, and a mild steel one for full sensing distance. If the existing switch is operated by a plastic cam or an aluminium casting, you need to add a steel flag, use a capacitive sensor, or correct the distance with the reduction factor table.

Where the mechanical switch is still right

• Safety interlocks requiring positive opening. A guard interlock to the relevant machinery safety standards needs a switch with direct opening action, where the contacts are forced apart mechanically. A proximity switch cannot provide that. Do not substitute one into a safety circuit on your own judgement.
• No supply available. A volt free contact in a circuit with no DC supply and no room in the panel. An AC two wire proximity switch may still work here, but a limit switch is simpler.
• Very high current switching. Where the switch directly carries a meaningful load and adding a relay is not acceptable.

Doing the conversion

1. Measure the load the existing switch carries. Decide between DC three wire plus interposing relay, or AC two wire direct.
2. Identify the target. Material and size. Add a mild steel flag if needed, at least as large as the sensor face.
3. Choose the body size from the gap you can hold. The sensing distance chart gives the options from M6.5 to M50. Choose generously, because the point of the exercise is to stop maintaining this position.
4. Position it out of the path of travel so the target sweeps past the face rather than into it.
5. Specify NO or NC on fail safe grounds, not on which is convenient to wire.
6. Protect the cable. Support it within 100 mm of the sensor, see connector or cable.

Worth doing first

If you are converting a plant progressively, start with the positions that are replaced most often, which are usually wet, dusty or high cycle. The maintenance log tells you where they are. Those positions pay back immediately, and they also build the case for the rest.

Send us the maintenance list and the loads and we will specify the replacements. For general selection, the supplier checklist and the troubleshooting guide are the practical starting points.

Once you need to know where something is rather than whether it has arrived, you are buying a transducer instead of a switch, and the cost step is significant. It is worth being sure you need it, and then worth choosing the right one.

If two end of stroke positions would do, fit two proximity switches and stop reading here.

Potentiometric linear transducer

A wiper runs along a resistive track. The output is a voltage proportional to position, usually conditioned to 0 to 10 V or 4 to 20 mA.

Strengths: genuinely absolute, so position is known the instant power is applied with no homing move. Simple to interface. Excellent resolution. The lowest cost per millimetre of the three for short and medium strokes.

Limits: there is mechanical contact, so there is a wear life expressed in cycles. On a slow moving axis that is effectively unlimited; on a fast cycling press it is a maintenance interval.

Specify it for: strokes up to around a metre, moderate cycle rates, machine axes, valve position, gate position, hydraulic cylinder feedback where cycle counts are moderate.

Our Novotechnik linear transducers cover this class, with the selection logic in linear transducer selection.

Magnetostrictive

A waveguide runs the length of the stroke and a permanent magnet travels along it on the moving part. A current pulse down the waveguide interacts with the magnet’s field and generates a return pulse; the time of flight gives position.

Strengths: no contact between the magnet and the waveguide, so no wear mechanism and essentially unlimited cycle life. Absolute. Long strokes, several metres, without losing accuracy. The rod version can be installed inside a hydraulic cylinder through a gun drilled piston rod, giving sealed feedback with nothing exposed.

Limits: the most expensive of the three. The resolution depends on the timing electronics. It needs a magnet properly positioned on the moving element, which is a mechanical design task.

Specify it for: long strokes, high cycle counts, servo hydraulic control, press position control, anything where a wear life is unacceptable or the transducer must be inside the cylinder.

Draw wire

A stainless wire unwinds from a spring loaded drum; the drum drives a rotary sensor. The output is proportional to the extracted wire length.

Strengths: very long strokes, several metres to tens of metres, from a compact housing. The mechanical installation is the simplest of the three because only the wire end attaches to the moving part, so it tolerates misalignment and non linear paths. Low cost per metre of stroke.

Limits: the wire is exposed and can be snagged, cut or fouled. Accuracy is lower than the other two because it depends on consistent wire tension and drum winding. Fast reversals can let the wire go slack and lose a turn.

Specify it for: long travel where accuracy requirements are moderate. Crane hook height, platform and lift position, telescopic boom extension, gate and barrier position, test rigs where a transducer has to be rigged temporarily.

Where the rotary element is the sensor, our rotary sensors and OEM kits and multi turn rotary sensors are the relevant parts.

Comparison

	Potentiometric	Magnetostrictive	Draw wire
Typical stroke	Up to about 1 m	Up to several m	Several m and beyond
Contact wear	Yes, cycle limited	None	Wire and drum
Accuracy	High	Highest	Moderate
Relative cost	Lowest	Highest	Low per metre
Install effort	Needs alignment	Needs magnet design	Simplest
In cylinder	No	Yes	No

Output and interface

Whichever you choose, settle the output early. Analogue 0 to 10 V or 4 to 20 mA is the common case and the easiest to interface; prefer 4 to 20 mA over any distance or in electrical noise, because current loops reject noise and a broken wire reads as zero current rather than as zero position. See analogue outputs explained.

Where the controller expects a bus, CANopen position sensors are available and remove the analogue conversion entirely.

And if you were considering an LVDT

LVDTs are excellent at very short strokes with very high resolution, typically in test and metrology rather than machine control. For a machine axis they are usually the wrong tool. See LVDT alternatives in India.

A capacitive proximity sensor responds to a change in dielectric near its face. It does not need a metal target, which means it detects plastic, glass, wood, grain, powder, granules and liquid. For dry bulk level in a hopper or silo it is usually the simplest answer available.

Sensing through the wall

The useful property is that a capacitive sensor can be mounted on the outside of a non metallic vessel and detect product on the inside. Nothing penetrates the vessel, nothing contacts the product, and there is nothing inside to clean or to break off into the batch.

Three conditions have to hold:

• The wall must be non metallic and reasonably thin. Plastic, glass or fibreglass. A metal vessel shields the field completely and the sensor sees only the wall.
• Wall thickness eats sensing distance. Our D32C capacitive sensor is M30 at 10 mm. If the wall is 5 mm thick you have roughly 5 mm left for the product, so this works on thin walled hoppers and not on thick ones.
• The product must have enough dielectric contrast with air. Most do. Low density, dry, fine powders are the marginal case.

Setting sensitivity

Capacitive sensors have a sensitivity adjustment and it is the whole job. Too low and the product is not detected through the wall. Too high and the sensor responds to the wall itself, to a hand near it, or to moisture on the outside.

The procedure that works:

1. With the vessel empty at the sensor height, turn sensitivity up until the sensor just switches on, then back it off until it switches off.
2. Back it off a little further, so there is margin against the empty condition.
3. Fill past the sensor and confirm it switches on with margin.
4. Empty it again and confirm it releases cleanly.

Do this with the vessel in its normal state, including any coating already on the inside wall, not on a clean new vessel. A sensor set against a clean wall drifts as the wall coats.

Build up, the recurring failure

Most capacitive level installations that fail do so because product has stuck to the inside of the wall at the sensor position. The sensor reads permanently on and the plant concludes the sensor has failed.

Defences, in order of effectiveness:

• Mount where material does not stick. On a sloping wall rather than a vertical one, and away from the fill stream where material is thrown against the wall.
• Set sensitivity with a realistic coating present, as above.
• Use the sensor as a point level, not a continuous indication. Capacitive gives you high level or low level, which is what a hopper control actually needs.
• Check it at cleaning intervals. Any level sensor in a sticky product is a maintenance item.

If the product is genuinely adhesive, wet, hygroscopic or changes moisture content between batches, capacitive is the wrong technology and you should be looking at a different measurement principle.

Moisture changes everything

Water has a very high dielectric constant compared with dry grain or plastic granules. A product whose moisture content varies between batches will present quite differently to the sensor, so a sensitivity set on dry material can read a damp batch at a different level, and condensation on the outside of the wall can trigger it with the vessel empty.

If moisture varies, set sensitivity on the driest material you run, which is the hardest condition to detect, and verify the empty state on the wettest day.

Typical applications

• Grain, pulses and seed in plastic hoppers on cleaning and grading lines
• Plastic granules at moulding machine hoppers, high level and low level
• Powders in pharmaceutical and food dosing hoppers, see packaging and pharmaceutical machinery
• Cement, fly ash and additives in small day bins, see cement plant sensors
• Sugar and bagasse handling, see sugar mill sensors

For liquids and larger tanks, see tank level sensing.

Ordering

D32C-3010-PM-3S is M30, 10 mm, PNP, normally open, three wire with short circuit protection. Substitute NM for NPN, PB or NB for normally closed. Choosing NC for a low level alarm means a broken wire raises the alarm rather than hiding it, which is usually what you want on a level interlock.

The ordering code guide covers each position, and capacitive sensing of plastics and non metals covers the non level applications.

Packaging lines are the densest photoelectric application in any plant. Bottle present, cap present, label present, carton gap, case count, jam detect, accumulation control. A line can carry dozens of sensors and the count at the end has to be right.

The clear container problem

Clear glass and PET are the hardest routine target. The light goes straight through, so a diffuse sensor sees almost nothing come back and a standard retro reflective sensor may see enough through the bottle to think the beam is unbroken.

Two reliable answers:

Through beam across the conveyor. Even a clear bottle refracts and attenuates enough to drop the received signal below threshold if the sensitivity is set correctly. Our D31C through beam parts cover 1 m to 15 m in the M18 body, so a conveyor width is comfortably inside range with signal to spare.

Polarised retro reflective. The corner cube reflector rotates the polarisation of the returning light and the receiver only accepts rotated light. Light passing through a clear bottle is not rotated, so the bottle is detected. This also solves the shiny foil and metallised film case.

A plain, non polarised retro reflective sensor on a clear bottle line will produce intermittent miscounts that look like a mechanical problem for weeks. If you are chasing a count that is slightly wrong, check the sensing mode first.

Counting accuracy

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

The beam is too wide for the gap. If containers are touching or nearly touching, the sensor must see the gap between them. A beam wider than the gap never goes clear and two bottles count as one. Through beam gives the tightest, best defined beam, which is why it is the counting default. Where the gap is very small, aperture the emitter down.

Bounce and jitter at the threshold. A signal sitting near the switching threshold chatters, producing several counts from one container. Set the sensitivity so there is real margin, and if the PLC allows it, apply a short debounce shorter than the minimum container pitch.

Switching frequency. On a high speed line, check the sensor switching frequency against the container rate. Photoelectric parts are generally fast enough, but at several thousand containers an hour it is worth confirming rather than assuming.

Where each sensor goes

• Infeed presence and accumulation. Retro reflective across the conveyor, mounted at container body height, not neck height, so a tilted container is still seen.
• Cap and closure check. Diffuse at short range looking down, or through beam across at cap height. Confirm against both the tallest and shortest container you run.
• Label present and label gap. A short range diffuse sensor looking at the label edge. Contrast between the label and the container decides whether this works, so test it on your actual product rather than a sample.
• Case and carton counting at the end of line. Through beam, with the beam positioned to see the gap between cases.
• Jam detection. A retro reflective sensor plus a PLC timer. If the beam stays broken longer than a normal container takes to pass, the line is jammed. Cheap, effective, and it catches a jam before it becomes a pileup.

Construction for a wet line

Bottling and food packaging are washed down, so the sensor has to survive the cleaning rather than the process.

Specify IP69K where there is high pressure hot water, IP67 where there is splashing. The difference between the two tests is explained in IP67, IP68 and IP69K, and there is more on construction in washdown sensors for food and beverage.

Two details that matter more on a wet line than a dry one. A water film on the lens scatters light and drops the signal, so mount the sensor so it drains and angle the face slightly downward where you can. And a connector is a joint that has to be cleaned and sealed, so on washdown lines consider a moulded fixed cable, see connector or cable.

Where photoelectric is not the answer

If you are sensing fill level inside an opaque container, or product level in a hopper, that is a capacitive job, because capacitive senses through a non metallic wall and photoelectric cannot see inside.

If you are sensing a metal part, a cap crimping tool or machine position, an inductive sensor is cheaper and more robust. Use photoelectric for product and inductive for machinery, and most lines sort themselves out.

Broader guidance on machine side sensing is in sensors for packaging and pharmaceutical machinery.

A photoelectric sensor detects an object by interrupting or reflecting a light beam. There are three ways to arrange that, and the choice sets the range, the reliability and how much installation work you are signing up for.

The useful rule: reliability runs in the opposite direction to convenience. Through beam is the most reliable and the most work. Diffuse is the least work and the least reliable.

Through beam

A separate emitter and receiver face each other. The object breaks the beam.

Range in our D31C range: 1 m, 3 m, 5 m, 10 m and 15 m in the M18 body.

Why it is the most reliable: the receiver sees the emitter directly, so the available optical signal is as strong as it gets. That surplus means dust on the lens, a misted window or a slightly misaligned bracket degrades the signal without stopping it working. On a dirty line, through beam carries on when the other two modes have stopped.

It also does not care what the object is made of. Black rubber, clear film, shiny metal, all break the beam the same way. The other two modes depend on how the object reflects.

What it costs: two devices, two brackets, two cable runs, one of them usually on the far side of the conveyor. Alignment matters, and on a long range it matters a lot.

Specify it for: dirty environments, long ranges, dark or transparent objects, and anything where a missed detection is expensive.

Retro reflective

Emitter and receiver in one body, with a reflector opposite. The object breaks the beam on its way back.

Range in our range: 1 m, 2 m, 3 m and 5 m in M18; 10 m, 15 m and 20 m in M30.

The compromise it strikes: one powered device, one cable run, and a passive reflector on the far side that needs no wiring. That removes most of the installation cost of through beam while keeping a defined beam path, which is why it is the default choice for conveyor edge detection and crane anti collision.

The trap: a shiny object can reflect enough light back to look like the reflector, and the sensor fails to detect it. On polished metal, foil wrapped product or glossy film, specify a polarised retro reflective sensor, which only accepts light with the polarisation rotated by the corner cube reflector. A shiny object does not rotate it.

Specify it for: general conveyor work, gate and door edges, crane anti collision, anywhere you want long range without a cable run on the far side.

Diffuse

Emitter and receiver in one body. The object itself reflects light back. No reflector, nothing on the far side.

Range in our range: 100 mm and 300 mm in M18; 500 mm, 1 m, 2.5 m and 3 m in M30.

Why people choose it: one device, one bracket, one cable, nothing opposite. On a machine where there is no far side to mount to, it is the only option.

Why it is the least reliable: the detection depends entirely on how much light the object sends back, which depends on its colour, finish and angle. A white carton at 500 mm and a black carton at 500 mm are different problems, and the black one may not be seen at all. A shiny object angled away reflects the light somewhere else entirely.

The quoted range assumes a standard white reference target. Against dark or matt material, expect a fraction of it.

Specify it for: short range, consistent product, where there is physically nothing to mount opposite. Avoid it where the product colour varies.

Choosing quickly

• Is the environment dirty? Through beam.
• Is the object transparent, black or variable in colour? Through beam.
• Is it over 5 m? Through beam, or retro reflective in M30.
• Is the object shiny? Through beam, or polarised retro reflective.
• Is there nothing to mount opposite? Diffuse, and test it on the darkest product you run.
• Everything else? Retro reflective.

Light on or dark on

Separate from the sensing mode, and frequently confused with it. Light on means the output is active when the receiver sees light, so the output drops when an object interrupts. Dark on is the reverse.

Our parts are ordered as one or the other, coded in the part number. D31C-18TB-10000-PL-2S is M18, through beam, 10 m, PNP, light on, three wire with short circuit protection. Change PL to PD for dark on, or to NL and ND for NPN.

Choose on failure behaviour. On a guard or a safety related interlock, arrange it so that a broken beam, a failed emitter or a cut cable all produce the same state as an object being present, which is the state that stops the machine.

Installation

Both bodies are available in M18 and M30 threaded barrels, so the mounting hardware matches your inductive sensors and a bracket tapped for one takes the other.

Three practical points. Mount the sensor so the lens faces away from falling dust. Avoid pointing a receiver at a window, a light fitting or direct sun, because ambient infrared degrades the margin. And on long range through beam, fit the brackets rigidly, because alignment that drifts with thermal movement will fail at one end of the day and work at the other.

We supply proximity switches to buyers in the UAE, Saudi Arabia, Qatar, Oman and Bahrain, mostly to oil and gas service companies, cement and aggregate plants, water treatment, and machine builders exporting onward.

The commercial attraction is straightforward: Indian manufacture at a price below European equivalents, with shorter lead times than importing from Europe or Japan. The technical specification needs two adjustments that buyers frequently miss.

1. Ambient temperature derating

This is the one that causes field failures. A sensor rated for a standard industrial ambient is being asked to work in a plant room in Dammam in July, inside an unventilated enclosure, in direct sun.

The ambient inside a south facing control box in the Gulf summer can sit well above the general plant figure that the specification assumed. Three consequences:

• Sensing distance drifts. Assured operating distance already accounts for temperature variation, but at the extreme end you are using all of that allowance. Set the gap at 70 percent of rated rather than 80.
• Cable jackets degrade. Standard PVC hardens and cracks under combined UV and heat. Specify PUR, and where cable is exposed outdoors, specify UV resistant and route it in conduit.
• Electronics life shortens. Every component has a temperature life relationship. A sensor that lasts a decade in a temperate plant will not.

Where the ambient is genuinely high, specify from the high temperature range rather than hoping a standard part copes. Tell us the actual measured ambient at the sensor, not the plant design figure.

2. Dust and sand ingress

Fine wind blown sand is more aggressive than industrial dust. It abrades the sensor face, works into cable glands and accumulates on the sensing face where it holds moisture during the night.

IP67 is the practical minimum for anything outdoors. Where there is also washing, IP69K. The IP ratings guide explains what each test actually covers, which is not what most people assume.

For sensors mounted facing upward outdoors, specify a connector with a proper seal and torque it, or use a moulded fixed cable, see connector or cable.

Certification and documentation

What is required depends on the end client, not the country.

• Oil and gas. Expect a hazardous area requirement. That means NAMUR sensors with intrinsically safe barriers, and certification evidence. See UL, CSA and Zone 0 NAMUR switches.
• Machine builders exporting onward to Europe. You need the CE documentation chain to hold. Our notes on sensors in exported machinery cover what the sensor supplier has to provide.
• General industrial. Declaration of conformity to IEC 60947-5-2, test data, country of origin certificate.

Tell us at the enquiry stage which of these the end client requires. Producing a certificate after shipment is slower and occasionally impossible.

Commercial mechanics

Exports from India are zero rated for GST, either under bond or with refund, so the price you are quoted is the price. You will need the commercial invoice, packing list, certificate of origin and the HS classification, which for proximity switches is set out in our HSN and GST guide. The Indian HSN and the international HS code share the first six digits, which is what your customs broker needs.

For repeat requirements, an annual rate contract with scheduled despatches is more efficient than quoting each order, and lets us plan production rather than treating every order as new.

What we would ask you

To quote properly we need more than a part number list, because an imported part number does not tell us the conditions it has to survive.

• Measured ambient at the sensor location, and whether it is enclosed or in sun
• Indoor, outdoor or washdown
• Area classification, if any
• Target material and the gap you can hold
• What the input card expects, PNP or NPN, or whether it is AC
• Which certificates the end client requires

If you are replacing an installed imported part, send the number and we will cross reference it, see cross referencing imported switches. We will also tell you where a cross reference is close but not identical, because in an export situation discovering that on site is expensive.

Hyderabad and the surrounding belt are dominated by pharmaceutical manufacturing, bulk drugs, process plant and a substantial engineering and defence sector. That mix changes which proximity switches matter, compared with an automotive region.

What the regional mix demands

Pharmaceutical and bulk drug. Washdown cleaning, solvent exposure and validation documentation. The sensor specification is driven by the cleaning regime more than by the sensing task. See IP69K washdown sensors and sensors in packaging and pharmaceutical machinery.

Process plant and bulk chemicals. Hazardous area classification. In a zone rated area a standard three wire DC sensor is not acceptable; you need a NAMUR sensor feeding an intrinsically safe barrier or a NAMUR control unit in the safe area. For certification, see UL, CSA and Zone 0.

Engineering and defence. Drawings, traceability and country of origin documentation, often with a preference for Indian manufacture on procurement grounds.

Washdown is the specification driver in pharma

The cleaning regime, not the machine, decides the sensor. Three questions:

• What pressure and temperature is the wash? IP67 handles splashing and temporary immersion. High pressure hot water washdown needs IP69K, which is a different test entirely. The distinction is in IP67, IP68 and IP69K.
• What chemicals? Caustic and solvent cleaning attacks cable jackets and seals. The sensor body material and the cable jacket both need to suit, and this is worth stating explicitly in the enquiry.
• Is there a crevice problem? In clean areas, a sensor with a connector creates a joint that has to be cleaned. A moulded fixed cable removes it, see connector or cable.

Non metallic detection

Pharma and process lines handle a lot of material an inductive sensor cannot see: tablets, powders, liquids, glass vials, plastic bottles and blister foil.

That is capacitive or photoelectric territory. Capacitive sensors detect level through a non metallic vessel wall, covered in powder and granule level sensing. Photoelectric handles counting and presence on a packaging line, see bottling, packaging and counting.

A common Hyderabad enquiry is a sensor that was specified as inductive and does not work because the target turned out to be aluminium foil or a plastic cap. If that is the symptom, the fix is a different technology, not a better inductive sensor.

Hazardous areas

If any part of the plant is zone rated, the sensor selection is constrained before anything else is considered. A NAMUR sensor is a two wire device with a defined current signature that a barrier or control unit interprets, and it is designed so the energy in the field wiring cannot ignite the atmosphere.

Specify the sensor and the control unit together. We supply both, see NAMUR suppliers in India and the control unit range. The D20C NAMUR parts cover M12 at 4 mm, M18 at 5 and 8 mm, and M30 at 10 and 15 mm.

Buying from Mumbai

We manufacture in Mumbai and supply across India. For Hyderabad that is a road or air freight question rather than a counter collection, so the sensible pattern is a planned order for consumption plus a holding of critical spares at your own stores. The spares list guide covers how to size that.

Telangana to Maharashtra is interstate, so IGST applies. The HSN code and GST treatment is set out separately.

Documentation

For pharma and defence buyers the paperwork is part of the deliverable, not an afterthought. We supply dimensional drawings, wiring diagrams, declarations of conformity to IEC 60947-5-2, country of origin and test documentation. For validated pharmaceutical equipment, tell us at the enquiry stage what documents the protocol requires so they come with the first delivery rather than being chased.

Pune and the Pimpri Chinchwad belt have one of the densest concentrations of automotive, auto component and machine tool manufacturing in India. All three run large numbers of proximity switches, and most plants in the region source them through local trading houses.

That works until you need something specific, a cross reference for a discontinued import, or a technical answer about why a sensor is failing.

What Pune plants typically run

The mix is predictable across the region:

• Automotive assembly and body shops. Heavy on weld immune sensors, because standard inductive parts near a spot welding gun fail from the magnetic field and the weld spatter. This is the single most common Pune enquiry we get.
• Machining lines and CNC. Tool position, door interlocks, chip conveyor monitoring. See proximity sensors on CNC machines.
• Auto component plants. Presses, transfer lines and assembly fixtures, usually M12 and M18 in quantity, often with non standard cable lengths.
• Machine tool builders. Buying for a build rather than for maintenance, so they need drawings, consistency across a production run and a stable part number.

Buying from Mumbai rather than locally

We are in Mumbai, roughly three hours away. For a Pune plant the practical questions are delivery and support.

Delivery. Mumbai to Pune is next day by road for stock items. For a planned order that is irrelevant. For a breakdown it matters, which is the argument for holding your own critical spares rather than relying on anyone’s delivery, local or otherwise. Our spares list guide sets out how to decide what to hold.

Support. This is where buying from a manufacturer differs from buying from a stockist. If a sensor keeps failing in a weld cell, we can look at the application, tell you whether the problem is the part or the installation, and change the specification if it is ours. A trading house can send you another one of the same.

The weld cell problem specifically

Because it dominates enquiries from the region, it is worth stating plainly. A standard inductive proximity sensor near resistance welding fails for three separate reasons:

• The magnetic field from the welding current saturates the sensor coil and causes false switching.
• Weld spatter accumulates on the face, and once there is ferrous material stuck to it, the sensor reads permanently on.
• The ambient temperature near the gun is higher than general plant temperature.

Weld immune and weld field immune parts, with a PTFE coated face that spatter does not adhere to, address all three. Fitting a standard sensor and replacing it monthly is more expensive within the first year. The detail is in weld immune proximity sensors.

Cross referencing what is already installed

Most Pune plants have a mix accumulated over decades, often imported parts specified by an OEM machine builder and now hard to source or priced for a captive market.

Send us the part numbers. We will tell you which map to our range, which are close with a difference you need to know about, and which we cannot match. See cross referencing imported switches and replacing Omron, Autonics and Pepperl Fuchs.

For machine builders, we also do non standard cable lengths and private label in modest quantities, see custom and OEM manufacturing.

Commercial details

Maharashtra to Maharashtra, so CGST and SGST rather than IGST. The HSN classification and GST rate for proximity switches is set out separately. We supply against purchase order with the usual documentation, and for repeat requirements an annual rate contract removes the quotation cycle from every order.

What to send us

A part list is ideal. If you do not have one, the application works: what is being sensed, what it is made of, the gap, the environment and what the input card expects. For a plant wide review, a photograph of the installed sensor and its surroundings usually tells us more than a specification does.

If you are sourcing proximity switches in India you will get quotations within a fairly narrow price band, and choosing on price alone is how plants end up with a drawer of failed sensors and no recourse.

Here is what we would ask, including the questions that are awkward for us to answer.

1. Do they manufacture, or do they import and relabel?

This is the first question and many suppliers are vague about it. Both models are legitimate, but they give you different things.

A manufacturer can change a specification, build a non standard cable length, adjust a sensing distance and tell you why a part failed. A relabeller can usually do none of these, and when the overseas source changes its design your supposedly identical replacement part behaves differently.

Ask directly: where is this made, and can I visit? A manufacturer will say yes.

2. Which standard, and can they show the test?

For proximity switches the standard is IEC 60947-5-2. Everyone claims it. Fewer can show you the test results for the batch you are buying.

Ask for the test report covering sensing distance, switching frequency, short circuit protection and ingress. If the answer is a generic certificate with no batch reference, you are being shown a marketing document rather than evidence.

If you are exporting the machine, you have a second set of requirements. See sensors in exported machinery and CE documentation, and for hazardous areas, UL, CSA and Zone 0 NAMUR.

3. What is the real lead time, and what is held in stock?

Distinguish three things: what is on the shelf today, what is built to order, and what is imported to order. The quoted lead time usually describes the best case of the first category.

Ask which of the parts on your list are stock items. For a plant spares holding this matters more than price, because the cost of an unplanned stop dwarfs the cost of the sensor. Our spares list guide covers how to decide what to hold yourself.

4. Can they cross reference what you already have?

Most plants are not starting fresh. They have installed sensors from several manufacturers and want to consolidate.

A supplier worth using will take your existing part numbers and tell you which of theirs matches, which is close but different in a way you need to know about, and which they cannot match. That last category is the honest answer and you should expect to hear it occasionally.

We publish our approach in cross referencing imported switches and replacing Omron, Autonics and Pepperl Fuchs parts.

5. Will they customise, and from what quantity?

Non standard cable lengths, specific connector orientations, particular sensing distances, custom housings, private labelling for an OEM. Ask what the minimum quantity is for each.

A manufacturer that will do a non standard cable length in tens rather than thousands is genuinely useful to a machine builder. See custom and OEM manufacturing.

6. What happens when a batch fails?

The question nobody asks at the quotation stage and everybody asks later.

Ask what the warranty actually covers, who pays for the replacement freight, and whether they will analyse a returned failed part and tell you what happened. That last one is the real differentiator, because a supplier who tells you the failure was caused by your installation is giving you information you can act on, and one who silently replaces it is not.

7. Is there engineering support before the order?

Will someone look at your application and tell you the part you have asked for is wrong? Our experience is that a meaningful share of enquiries specify a sensor that will not work, usually because of target material correction or flush mounting an unshielded part.

A supplier who sells you what you asked for is easier to deal with in the short term. A supplier who asks what the target is made of saves you a failure.

8. Documentation

Dimensional drawings in a usable format, wiring diagrams, declarations of conformity, and for export, country of origin and the HSN classification. If these have to be chased each time, factor that into the cost.

Our position

We build proximity switches in Mumbai and have done since 1985, to IEC 60947-5-2. We are not the cheapest quotation you will receive and we will tell you when a part we do not make is the better answer for your application.

If you want to test the questions above on us, send an enquiry and judge the response. That is a more useful comparison than a page of our own claims. There is also a printable supplier checklist and our notes on Indian against Chinese sourcing.

Every proximity sensor comes terminated one of two ways: a fixed cable moulded into the body, or a short pigtail ending in an M12 connector that mates with a separate cordset. People pick whichever the catalogue shows first. It is worth thirty seconds of thought, because this choice decides how the sensor fails and how long it takes to put right.

Note that an M12 connector is unrelated to an M12 barrel. A sensor with an M30 body can have an M12 connector, and frequently does.

Specify the connector when

• The sensor is removed for changeovers. On tooling that changes between products, a connector means unplugging rather than rewiring.
• Replacement time matters. A connectorised sensor is changed in two minutes by a shift fitter. A hard wired one means tracing the cable back to a junction box, and on a large machine that is an hour with a wiring diagram.
• The sensor is in a vibrating or flexing location. The cordset takes the abuse and is cheap to replace. Crane trolleys, moving arms, anything on a slide.
• The machine is exported or built in volume. Standard cordset lengths mean the wiring harness is repeatable and the build does not depend on cutting cable to fit.
• You want the diagnostic. Unplugging at the sensor and testing at the cordset end immediately separates a sensor fault from a cable fault, which is the first branch of our troubleshooting sequence.

Specify fixed cable when

• The sensor is fitted once and left. A position sensor inside a guarded enclosure that is opened at annual shutdown gains nothing from a connector.
• There is no room behind the sensor. An M12 connector plus the bend radius of the mating cordset adds perhaps 40 to 50 mm behind the barrel. In a tight pocket that space does not exist.
• The environment attacks connections. A properly moulded cable entry has no joint to fail. In continuous washdown, chemical mist or heavy dust, a fixed cable in the exact length needed removes the one component most likely to let water in.
• You want to eliminate a junction. Ordering the cable at the length that reaches the panel means no junction box, no terminal block, and two fewer places for a loose connection.

On that last point, we supply fixed cable at specified lengths rather than only standard ones. On a machine build it is usually worth asking, because it tidies the harness and removes connections.

Pin out

A three wire DC sensor on a four pin M12 uses:

• Pin 1 brown, positive supply
• Pin 3 blue, negative supply
• Pin 4 black, output
• Pin 2 white, second output on four wire parts, unused on three wire

This is standard across manufacturers, which is the main practical benefit of connectorising: a cordset from any supplier mates correctly. The colour convention matches the flying lead version, so a plant that mixes both does not need two wiring conventions. See two, three and four wire wiring.

Cordset selection

The cordset is a real component and specifying it carelessly undoes the benefit.

• Straight or right angle. A right angle connector saves the bend radius behind the sensor and is often the only thing that makes a connector fit.
• Jacket material. PVC for general use, PUR where there is oil or abrasion, and for continuous flexing on a drag chain specify a cable rated for it. Standard PVC in a drag chain fails within months.
• Shielded or unshielded. Where the cable runs alongside VFD output cabling, use shielded and earth the shield at the panel end only.
• Ingress rating. The connection is only as good as the mating. An IP67 sensor with a hand tightened connector is not IP67. Torque the coupling nut properly, and in washdown areas check that the cordset is rated to the same standard, see IP ratings explained.

The cost comparison people get wrong

A connectorised sensor plus a cordset costs more than a sensor with a fixed cable. That is the comparison usually made, and it is the wrong one.

The right comparison includes the cost of the second failure. One unplanned stop traced back to a chafed cable, on a line worth anything per hour, pays for connectorising every sensor on the machine. On a lightly loaded machine in a clean room the fixed cable genuinely is cheaper over the life. Decide per position rather than per plant.

Ordering

Both terminations are available across the D30C inductive, D31C photoelectric and capacitive ranges. Tell us which you want when you enquire, along with the cable length if you want it cut to size. The ordering code guide covers the rest of the specification.

On an EOT crane the sensor is doing safety related work: stopping long travel before the end stop, preventing two cranes colliding on a shared runway, limiting hoist height, detecting slack rope. When one fails the crane either stops producing or hits something.

The environment is the hard part. Continuous vibration, impact when the end stop is reached, temperature swings under a plant roof, and on a gantry, weather.

The sensing jobs

End of travel on long and cross travel. Usually a slowdown limit and a stop limit in series, sensing a steel flag or dog on the runway. This is the classic M30 application, because the gap between the trolley and the runway flag cannot be held tightly and 15 mm of sensing distance absorbs it.

Anti collision between cranes. Two cranes on one runway need to see each other before they meet. Inductive sensing is too short range for this; use photoelectric in retro reflective mode, which reaches 20 m in the M30 body, with the reflector on the opposing crane.

Hoist upper and lower limits. Rotary geared limit switches handle the counting, but a proximity switch on the drum or a flag on the rope guide gives an independent backup, which is worth having on a hoist that lifts over people.

Slack rope and rope drum rotation. A speed and rotation sensor on the drum detects a drum that has stopped turning while the motor is still commanded, which is what a snagged or slack rope looks like electrically.

Specifying for vibration

Vibration kills crane sensors in two ways and neither is the sensing element.

The mounting works loose. Locknuts on a threaded barrel will back off under continuous vibration unless they are properly torqued and locked. Use both nuts, torque them, and use a thread locking compound or a tab washer. Then check them at every scheduled inspection, because they will still move.

The cable fails at the gland. Continuous low amplitude vibration work hardens copper where the cable leaves the body. Support the cable within 100 mm of the sensor so the vibration is taken by the clamp rather than the gland, and use a connector version so a failed cordset can be swapped in minutes instead of rewiring back to the panel. See connector or cable.

Impact protection

A sensor sensing an end of travel dog will eventually be hit by it, usually because the dog has shifted or the trolley has overrun. Two defences:

• Mount the sensor so the dog passes across the face rather than towards it. A sensor that sees a flag sweeping past is never struck head on.
• Recess the face slightly behind a sacrificial non metallic plate and add the plate thickness to your gap calculation.

Choosing a larger body helps here too. The extra sensing distance of an M36 or M50 lets you set the sensor further back from the path of travel.

Weather and washdown

Indoor plant cranes are straightforward. Gantries in a yard, cranes in a steel plant near the water sprays, and anything in a coastal location need the ingress rating specified deliberately. IP67 covers temporary immersion, which is what driving rain amounts to. Where there is high pressure washing, IP69K. The difference is explained in IP67, IP68 and IP69K.

In a steel plant, add heat to the list. Near the caster or the rolling line, see high temperature sensors and our notes on steel plants and rolling mills.

Fail safe wiring

On a limit that stops motion, wire it normally closed. A broken cable, a disconnected plug or a failed sensor then reads the same as the limit being reached, and the crane stops. Wired normally open, the same failures read as the limit never being reached, and the crane runs into the end stop.

This is the single most important decision on the list and it costs nothing. Specify the NC variant at the order stage, not the NO.

Spares policy

A crane down for a sensor is a production line down. Hold spares for the parts fitted, not a generic selection, and label the spares with the crane and the position. Our plant spares list guide sets out how to build a list that is actually usable at two in the morning.

If the existing sensors are imported and hard to source, send us the part numbers. Most can be cross referenced to an equivalent we build in Mumbai and hold, see cross referencing imported switches.

Almost every automated machine needs to know whether a cylinder has extended or retracted. Interlocks depend on it, sequences depend on it, and a cylinder that reports a position it has not reached causes the kind of fault that damages tooling.

There are three ways to sense it, and they are not interchangeable.

1. Magnetic switch on the barrel

Most modern pneumatic cylinders have a permanent magnet set in the piston. A magnetic proximity switch clamped to the outside of the barrel detects the magnet through the wall as the piston passes.

Why it is usually right: no machining, no modification to the cylinder, adjustable anywhere along the stroke, and it senses the piston itself rather than something attached to it. You can also fit mid stroke positions, which the other methods cannot do easily.

What it requires: the cylinder must have a magnetic piston, and the barrel must be non magnetic. Aluminium and most stainless barrels are fine. A plain steel barrel will shield the magnet and this method will not work.

Where it fails: near a welding station or a large motor, stray fields can trigger it. Keep the switch and its cable away from welding returns, and route the cable separately from power cabling.

2. Inductive sensor on the rod end or the load

Fit a bracket and a steel flag to the rod clevis, the tooling or whatever the cylinder is moving, and sense the flag with a standard inductive proximity switch.

Why you would choose it: works on any cylinder regardless of piston magnet or barrel material, including hydraulic cylinders with thick steel walls. It also senses what you actually care about, which is the position of the load, not the position of the piston.

That distinction matters more than it sounds. If the tooling has worked loose on the rod, a barrel switch still reports the correct position while the tool is somewhere else. A sensor on the load catches it.

What it costs: a bracket and a flag, and the space for both. Use M12 or M18 depending on space, make the flag mild steel rather than stainless or aluminium, and make it at least as large as the sensor face.

3. Continuous position feedback

If you need to know where the cylinder is throughout the stroke rather than just at the ends, a switch will not do it. That is a linear position transducer job.

This applies to servo hydraulics, press position control, synchronising two cylinders, and anywhere the control loop needs the actual position rather than two discrete confirmations. Our Novotechnik linear transducers cover this, and there is a selection guide comparing draw wire, potentiometric and magnetostrictive approaches.

It is a different class of cost. Do not specify continuous feedback where two end of stroke switches would do.

Hydraulic cylinders specifically

Hydraulics change the calculation. The barrel is thick steel, so a magnetic switch through the wall is usually not an option. That leaves an inductive sensor on the rod end or a transducer.

Two practical points. First, hydraulic installations are dirtier and wetter, so specify the ingress protection properly, see IP67, IP68 and IP69K explained. Second, hydraulic cylinders generate real force and the bracket must be rigid, because a sensor that moves with the load tells you nothing.

Specifying it

• Output. PNP for most modern Indian panels, NPN on older machinery. Check the card, see PNP against NPN.
• Function. Normally open for a straightforward position confirmation. Normally closed where a broken wire should read as the unsafe state, which on an interlock is often what you want.
• Termination. On a moving cylinder specify a connector version or a cable rated for flexing, and support the cable close to the sensor. See connector or cable.
• Response time. On a fast indexing cylinder, check that the switching frequency of the sensor covers the cycle rate. Most inductive parts are comfortable here, but a cylinder cycling several times a second against a small flag is worth checking.

The fault that keeps coming back

A cylinder position sensor that works for a month and then goes intermittent is almost always mechanical. The bracket has loosened, the flag has bent, the cable has work hardened at the gland, or the gap was set at the limit of the sensing range and thermal drift has taken it outside.

Before replacing the sensor, measure the gap with the machine hot and check the bracket for movement by hand. Our troubleshooting guide works through the sequence.

Sensing distance scales with barrel diameter. That is the single rule behind most selection decisions, and once you have the table in front of you the choice usually makes itself.

These are the rated distances across our inductive range, measured against a standard mild steel target.

The table

Body	Thread	Shielded	Unshielded	Typical use
M6.5	M6.5	1 mm	2 mm	Grippers, end effectors
M8	M8 x 1	1 mm	2 mm	Tool changers, tight mechanisms
M12	M12 x 1	2 mm	4 mm	Indexing, small actuators
M18	M18 x 1	5 mm	8 mm	General plant workhorse
M30	M30 x 1.5	10 mm	15 mm	Heavy plant, conveyors, mills
M36	M36	On request	20 mm	Large standoff, AC two wire
M50	M50	On request	30 mm	Maximum standoff

Shielded or unshielded

The two columns are not just different distances, they are different parts with different mounting rules.

Shielded parts have a metal collar around the coil that focuses the field forward. They can be mounted flush in a metal bracket. You lose sensing distance and gain installation freedom.

Unshielded parts radiate sideways as well as forward. They sense further and must stand proud of the bracket with a clear zone around the face of roughly three barrel diameters. Fit one flush in steel and it will read permanently on. Our full comparison is in shielded against unshielded.

Three corrections before you use these numbers

The table gives rated distance. The gap you build to is smaller, for three reasons that stack.

1. Design margin. Rated distance is nominal. Effective distance allows for manufacturing tolerance, and assured operating distance allows for tolerance, temperature and supply variation together. Set the gap at 70 to 80 percent of rated and you are inside assured distance with room to spare.

2. Target material. Mild steel is the reference at 1.0. Everything else reads short:

• Stainless steel, roughly 0.7 to 0.85 depending on grade
• Brass and bronze, roughly 0.4 to 0.5
• Aluminium, roughly 0.35 to 0.45
• Copper, roughly 0.3 to 0.4

The reduction factor table has the detail. This is the most common reason a correctly specified sensor does not work.

3. Target size. The rated figure assumes a standard target at least as large as the sensor face. A smaller target reads short. If you are sensing a bolt head or a thin flag, allow for it or make the flag bigger, which is usually the cheaper fix.

A worked example

You need to sense a stainless steel cam at a gap of 6 mm, and the bracket can move about a millimetre.

Start with the gap plus the movement, so 7 mm worst case. Divide by the stainless factor of 0.75, giving roughly 9.3 mm of mild steel equivalent. Divide by 0.75 for design margin, giving roughly 12.5 mm of rated distance needed.

From the table, M30 unshielded at 15 mm covers it. M30 shielded at 10 mm does not. If the clear zone for an unshielded part is not available, you need to move the bracket closer or go to M36.

That arithmetic, done at the drawing stage, prevents most of the sensor failures we are asked about.

Other technologies in the same bodies

Inductive is not the only option in these barrels. Capacitive parts sense plastic, glass, grain and liquid, at 10 mm in M30. Photoelectric parts in M18 and M30 reach from 100 mm to 20 m depending on mode. If the inductive table does not give you the distance you need, the answer is often a different technology rather than a bigger barrel.

Below M12 you are into subminiature territory: M8 and M6.5 bodies sensing at 1 mm or 2 mm. They exist because sometimes there genuinely is no space, and they are unforgiving because there is no margin.

The numbers

Our D30C range covers both small bodies at two sensing distances each:

• M6.5: 1 mm and 2 mm
• M8: 1 mm and 2 mm

Both are available PNP or NPN, normally open, three wire, 200 mA. Note the 200 mA rating rather than the 300 mA of the larger bodies, which matters if you are driving anything other than a PLC input.

The tolerance problem

Design to 70 to 80 percent of rated and a 1 mm sensor gives you a working gap of 0.7 to 0.8 mm. That is less than the thickness of a business card. Everything now matters:

• A bracket that flexes under acceleration moves the gap by more than the gap.
• Thermal growth across a 100 mm arm is comparable to the whole sensing distance.
• A target in stainless rather than mild steel can halve what you have.
• Paint or plating on the target is a meaningful fraction of the gap.
• Swarf or a film of coolant on the face reads as a target.

So specify the 2 mm version wherever it will fit. The 1 mm parts are for cases where the mechanical envelope genuinely forbids anything else, and they want a rigid bracket, a mild steel target and a clean environment.

Where they earn their place

• Grippers and end effectors. Confirming jaw open and jaw closed, where the sensor travels with the arm and mass matters.
• Tool changers and turrets. Position confirmation inside a housing that was designed around the tooling, not around the sensor.
• Small pneumatic actuators. Though if the actuator has a magnet in the piston, a magnetic proximity switch on the barrel is usually the better answer. See sensing cylinder position.
• Textile, packaging and assembly machinery where the mechanism is dense and the sensor has to live inside it.

Mounting

The clearance rules scale down with the body, which sounds convenient and is not, because manufacturing tolerances do not scale down with them.

Use the shielded variant and mount it flush in the bracket. On a small body a flush mount in a well machined pocket is far more repeatable than a sensor standing proud on a locknut, because the pocket controls the position and the locknut does not. Where you can, design a shoulder the sensor seats against rather than relying on the nuts to hold a dimension.

Protect the cable at the back. On a moving arm the cable is the thing that fails, not the sensor, and on a subminiature body the cable exit is the weakest point. Specify a cable length that reaches the first fixed point without a joint, and support it within 50 mm of the sensor so the flexing happens in the cable and not at the gland.

When to go bigger instead

Before committing to M8, check honestly whether the bracket could be redesigned for M12 at 4 mm. The difference in reliability over a few years is large, and a bracket is cheaper to change at the drawing stage than a sensor is to replace quarterly once the machine is in production.

If the answer is genuinely no, then fit the 2 mm M8, use a mild steel target, make the bracket stiff, and accept that this sensor will need checking at every planned shutdown.

Ordering

D30C-0802-PM-2 is M8, 2 mm, PNP, normally open, three wire at 200 mA. D30C-0601-NM-2 is the M6.5 at 1 mm in NPN. The ordering code guide decodes every position, and the sensing distance chart shows how the small bodies sit against the rest of the range.

Most plants never need these. The ones that do need them badly, and usually find out after an M30 has been replaced three times because the real problem was that the gap could never be held.

M36 and M50 are large barrel inductive proximity switches. M36 senses at 20 mm, M50 at 30 mm. They exist for one reason: standoff.

When you actually need them

Four situations come up repeatedly:

• The target cannot be brought closer. A large rotating drum, a flywheel, a kiln shell, anything where the mechanical design fixes the gap and it is more than an M30 can reach.
• There is material in the gap. Scale, dust buildup, product carryover, a wear plate. The sensor has to see through the standoff plus whatever accumulates in it.
• Thermal growth moves the target. On a hot line the gap at operating temperature is not the gap you set cold. If the swing is several millimetres, you need a sensing distance that covers the whole range.
• The sensor must be protected. Setting it well back behind a guard or a non metallic wear plate costs you sensing distance, and a large barrel pays for it.

If none of those apply, fit M30 and save the money and the panel space.

What the size costs you

Be clear eyed about the trade. A large barrel needs:

• A bigger clear zone. Unshielded parts need a clear annular area around the face of roughly three barrel diameters. On M50 that is 150 mm of clear space, which on a crowded frame is a design constraint, not a detail.
• More spacing between sensors. At least twice the barrel diameter centre to centre, so 100 mm on M50.
• A heavier bracket. The sensor has mass and the mounting has to hold a gap within a millimetre or two. A bracket that was adequate for M12 will resonate under an M50.
• A larger hole. Retrofitting into an existing plate usually means a new plate.

AC two wire on large barrels

Our M36 and M50 parts are in the A20C AC two wire range, 200 mA, available normally open or normally closed. That matches where large barrel sensors are usually found: older heavy plant with relay logic and no 24 V DC supply at the sensing point.

Two wire AC switches in series with the load, so the usual cautions apply. There is a small leakage current when off and a voltage drop when on. Feeding a modern PLC input card directly from an AC two wire sensor is where people get into trouble, because the leakage can be enough to hold a high impedance input on. Switch a contactor or an interposing relay, and take the PLC signal from that.

If you do have 24 V DC at the point of use, say so when you enquire. A DC three wire part in a large barrel is a better match for a modern input card, and we can discuss what is available.

Mounting and commissioning

Set the gap at 70 to 80 percent of rated, so 14 to 16 mm on M36 and 21 to 24 mm on M50. Then verify it hot. On the applications where large barrels are specified, the cold gap and the running gap are routinely different by several millimetres, and the only way to know is to measure at temperature.

Correct for target material as usual. A large barrel does not exempt you from the reduction factor, it just gives you enough margin to absorb it.

Ordering

A20C-3620-M-2 is M36, 20 mm, normally open, AC two wire at 200 mA. A20C-5030-M-2 is the M50 at 30 mm. Substitute B for normally closed. The ordering code guide explains each position.

These are not fast moving stock items in the way M18 is, so talk to us about lead time before you design the bracket. If the application is marginal we will usually tell you to fit M30 instead, because a part we hold is better for you than a part we build.

M30 is what you fit when the gap cannot be held tightly. Rolling mills, crushers, conveyors carrying lump material, kiln areas, anywhere the target wanders or the bracket takes a knock. The extra barrel diameter buys sensing distance, and sensing distance is margin.

Thread and sensing distance

M30 proximity sensors use M30 x 1.5. Unlike M12 and M18, which use a one millimetre fine pitch, M30 is 1.5 mm, so do not assume the fine thread rule carries across sizes when you order locknuts.

Our D30C range covers M30 in two sensing distances:

• 10 mm, shielded. Flush mountable in a metal bracket.
• 15 mm, unshielded. Longer reach, needs a clear zone around the face.

Design to 70 to 80 percent of rated, so 7 to 8 mm on the 10 mm part and 10.5 to 12 mm on the 15 mm. On M30 that working margin is large enough to absorb a reduction factor and a bit of mechanical slop, which is exactly why it is specified on heavy plant.

Why the extra margin matters

On an M12 at 4 mm rated, a stainless target at a reduction factor of around 0.7 leaves you 2.8 mm rated and perhaps 2 mm of working gap. Build tolerance eats that. On an M30 at 15 mm rated, the same correction still leaves roughly 10 mm rated and 7 mm of working gap, which survives a bracket that shifts and a shaft with end float.

That is the real argument for M30 and it is worth making to whoever is trying to standardise the plant on one small size. See the reduction factor table for the multipliers.

Mounting clearance

• Shielded (10 mm): face flush with the bracket is fine. Keep unintended metal at least three times rated distance clear.
• Unshielded (15 mm): the face must stand proud. Allow a clear annular zone roughly three barrel diameters wide, so about 90 mm, which on a crowded frame is a real constraint.
• Side by side: at least twice the barrel diameter centre to centre, so 60 mm.
• Opposed: at least six times rated sensing distance.

The unshielded clearance is the one that bites. If you cannot give an M30 unshielded its clear zone, the surrounding steel will hold it permanently on and the fault will look like a failed sensor. Before replacing one that reads stuck, check what was built near it since.

Variants across the M30 body

M30 is the most heavily populated size in our catalogue, because it covers so much heavy plant. The same barrel is available as:

• D30C inductive, 10 mm and 15 mm, PNP or NPN, NO or NC, three wire with short circuit protection.
• D32C capacitive at 10 mm, for plastic, glass, grain, powder and liquid through a non metallic wall.
• D20C NAMUR at 10 mm and 15 mm, for intrinsically safe loops feeding a NAMUR control unit or barrier.
• D33C speed and direction at 30 mA, including the dedicated speed and direction variant for underspeed and rotation monitoring.
• D31C photoelectric in the same M30 barrel, diffuse to 3 m, retro reflective to 20 m.

That last point is useful on a standardised plant. If the bracket is tapped M30 x 1.5, you can change sensing technology without changing the mechanical design.

Heavy duty considerations

On mills and crushers the sensor rarely fails electrically. It fails because the face is abraded, the cable is crushed, or heat has cooked it. Three things to specify:

• Cable protection. Armoured conduit from the sensor to the first junction, not loose cable clipped to the frame.
• Face protection. Where there is impact risk, recess the sensor behind a non metallic wear plate and recalculate the gap to include the plate thickness.
• Temperature. Standard parts are rated to the usual industrial band. Near a furnace or kiln you want the high temperature range instead, and for hot metal detection the dedicated hot metal detector.

Ordering an M30

D30C-3010-PM-3S is M30, 10 mm, PNP, normally open, three wire with short circuit protection. D30C-3015-PM-3S is the 15 mm unshielded version. Substitute NM for NPN, PB or NB for normally closed. For AC two wire there is the A20C in the same body. The ordering code guide decodes the rest.

If you need more than 15 mm in an M30 body, ask about the extended range parts, or step up to the M36 and M50 barrels at 20 mm and 30 mm.

M12 is the size you specify when an M18 will not physically fit. Tool changers, small pneumatic actuators, indexing tables and anywhere a bracket has to sit inside a guard. It gives up sensing distance in exchange for the space it saves, and that trade is the whole decision.

Thread and dimensions

M12 proximity sensors use a fine thread, M12 x 1, not the coarse M12 x 1.75 of a general fastener. As with M18, the fine pitch exists so the locknuts give you fine adjustment of the sensing gap.

Order replacement locknuts as M12 x 1 explicitly. A hardware shop will hand you coarse by default and it will not run on the barrel.

Standard barrel lengths run around 50 to 60 mm for the cable version and longer for the connector version, because the M12 connector adds roughly 10 mm to the back end. If you are designing the bracket before the sensor arrives, allow for the connector and the mating cordset bend radius behind it, not just the barrel.

Sensing distance: 2 mm or 4 mm

Our D30C inductive range offers M12 in two variants:

• 2 mm, shielded. The face can be mounted flush in a metal bracket. Surrounding metal does not trigger it.
• 4 mm, unshielded. Double the distance, but it needs a clear zone around the face and cannot sit flush in metal.

Those are rated distances against a standard mild steel target. Design to around 70 to 80 percent of the rated figure, so 1.4 to 1.6 mm on the 2 mm part and 2.8 to 3.2 mm on the 4 mm. A sensor set at the limit of its range passes the commissioning test and goes intermittent the first hot week of May.

Then correct for your target material. Stainless steel, aluminium and brass all read short, and on an M12 the margin is already small enough that a reduction factor can take you below the gap you built. Our reduction factor table has the arithmetic.

Mounting clearance

This is where M12 catches people. The clearances scale with the body, so they are tighter than the M18 figures you may be used to:

• Shielded (2 mm): the face may be flush with surrounding metal. Keep the opposing metal, anything the sensor is not meant to see, at least three times the rated distance away.
• Unshielded (4 mm): the face must stand proud of the bracket, and you need a clear annular zone around it roughly three times the barrel diameter.
• Side by side: two M12 sensors facing the same way need a centre to centre spacing of at least twice the barrel diameter, or they will interfere with each other.
• Facing each other: at least six times the rated sensing distance.

If you cannot hold the unshielded clearance, use the shielded 2 mm part and move the bracket closer. That is almost always the better answer to a crowded installation.

Output and wiring

M12 is available as PNP or NPN, normally open or normally closed, three wire with short circuit protection. In India most new panels are PNP sourcing, because that is what Siemens, Allen Bradley and Mitsubishi input cards expect by default. NPN still turns up on older Japanese machinery and on retrofits where the existing card is sinking.

Check the card before you order. Our guide to PNP against NPN covers how to tell which you have, and two, three and four wire wiring covers the terminations.

There is also an AC two wire M12 in the A20C range, 2 mm and 4 mm, 200 mA, for direct switching of contactors and older relay logic where there is no DC supply in the panel.

Cable or M12 connector

Confusingly, an M12 sensor can have an M12 connector, and the two M12s are unrelated. One is the barrel thread, the other is the connector standard.

Specify the connector version where the sensor is removed for changeovers, because a fixed cable dragged through a machine frame fails at the gland eventually. Specify fixed cable, cut to the length you actually need, where the sensor is fitted once and left. We cover the trade in more detail in choosing the termination.

Ordering an M12

The code reads left to right. D30C-1204-PM-3S is an M12 body, 4 mm sensing, PNP, normally open, three wire with short circuit protection. Change 1204 to 1202 for the 2 mm shielded version, PM to NM for NPN, or PM to PB for PNP normally closed. The full ordering code guide decodes every position.

M12 is also available as a NAMUR version in the D20C range at 4 mm, for intrinsically safe loops, and as a D33C speed and direction sensor at 30 mA for underspeed monitoring.

When M12 is the wrong choice

If you have the room for M18, take it. Four millimetres of sensing distance is not much margin once you account for target material, thermal drift and a bracket that moves slightly under load. M12 is for when the space genuinely is not there, not for saving a little money on a sensor that then needs setting twice a year.

Going the other way, if you need more standoff than M12 gives, look at M30 at 10 or 15 mm, or the extended range parts.

Heat kills more proximity sensors in Indian plants than water does, and it almost always traces back to the same error: the sensor was specified against the plant ambient temperature rather than against the temperature it actually experiences.

A sensor mounted two metres from hot stock is not at ambient. It is receiving radiant heat continuously, and its case temperature can sit far above the air around it. Before you buy anything rated higher, it is worth understanding the three ways out, because they are cheaper in the order given.

First, measure what the sensor actually sees

Not the plant ambient. Not the design figure. Put a thermocouple or a thermal camera on the intended mounting position during normal production, in the hottest part of the shift, in the hottest month you can manage.

It takes ten minutes and it routinely produces a number nobody expected. It also tells you which of the three fixes below you need, and whether you need a higher-rated sensor at all.

Fix one: move it

Distance is the cheapest heat shield there is. Radiant heat falls off sharply, so a mounting position half a metre further back often solves the problem outright and costs a bracket.

This is worth exhausting before anything else. It needs no consumables, nothing to maintain, and nothing that can fail.

Fix two: shield it

A simple steel plate between the sensor and the radiating source blocks the line of sight. Radiant heat travels in straight lines, so interrupting the path is remarkably effective, and a shield costs very little.

Leave an air gap between the shield and the sensor so the shield itself does not conduct heat straight into it, and make sure the shield does not sit in the sensing field if the sensor is unshielded, or it will detect the plate.

Fix three: cool it

Air purge or water cooling, where the first two are impossible. Effective, and the standard approach around furnaces, kilns and rolling mills.

The caution is that you have now added a service that can fail. A cooling supply that stops will destroy the sensor quickly and quietly. If you go this route, plan the air supply at the same time as the sensor rather than retrofitting it, and consider whether loss of cooling should be alarmed.

What heat actually does to a sensor

Worth knowing, because the symptoms get misdiagnosed.

Sensing distance drifts. The oscillator circuit is temperature dependent, so the effective distance changes as the sensor heats. A sensor set at the edge of its range on a cool morning becomes intermittent by afternoon. This is why you design to the assured operating distance rather than the rated one, and why 70 to 80 percent of rated is the sensible mounting gap.

Seals age. Elastomer seals and cable sheaths harden and crack with sustained heat. The sensor then fails on ingress during the next washdown, months after the heat caused it, which makes the cause hard to see.

The cable fails before the sensor. In our experience of hot installations the cable is usually first, cooked against a hot surface or at the gland where it leaves the sensor. Route it away from the heat, support it so the gland carries no weight, and leave a service loop.

Where this comes up

• Steel plants and rolling mills, where radiant heat from stock is the dominant problem and hot metal detectors handle the stock itself
• Cement plants, around the kiln hood and clinker cooler
• Sugar mills, in the boiling house and around the boilers
• Forging, heat treatment and foundry, where the heat is intermittent but severe
• Plastics processing, around heated barrels and moulds, where the temperature is lower but constant
• Engine and vehicle test rigs

Getting a recommendation

Send us the measured temperature at the mounting point rather than the plant ambient, whether the heat is constant or cyclic, how close the sensor has to be to the target, whether purge air is available, and what the target is. A photograph of the position is worth more than a description.

We manufacture proximity switches in Mumbai and will tell you plainly whether the standard range covers your temperature, whether a special build is the answer, or whether the right fix is a bracket rather than a sensor at all.

Almost every proximity sensor sold is a switch. It tells you that something is present or it is not. An analogue proximity sensor does something different: it reports how far away the target is, continuously, as a voltage or a current.

That is a different product solving a different problem, and the most useful thing this page can do is help you work out which of the two you actually need, because the answer is often not the one people arrive asking for.

Switching or measuring: the question that decides everything

Ask what the control system does with the information.

If it acts on a threshold such as stop when the part arrives, or alarm if the gate is open, you need a switching sensor. Cheaper, simpler, more robust, and the whole D30C range is built for exactly this.

If it needs the actual distance to close a loop, control a position, measure thickness or detect gradual drift, you need continuous measurement.

A good number of analogue enquiries turn out to be threshold problems where someone wanted margin. If two or three switching sensors at the heights that matter would do the job, that is usually the better engineering and certainly the better price.

0 to 10V or 4 to 20mA

Where continuous measurement is genuinely needed, the output choice is mostly about cable.

0 to 10V is simple and feeds most PLC analogue input cards directly. Its weakness is cable length and electrical noise. Voltage drops along a long run and the reading drifts with it, and a signal cable running beside VFD output cabling picks up interference readily.

4 to 20mA is a current loop, and current does not change along the cable. It is immune to voltage drop, far more tolerant of noise, and it has a diagnostic property voltage does not: zero current is not a valid reading, so a broken wire is distinguishable from a genuine zero. On a voltage output, a severed cable reads as zero volts, which may be a perfectly plausible measurement.

For anything beyond a few metres, in any Indian plant with drives in it, specify 4 to 20mA. The only reason to prefer voltage is an input card that only accepts it.

The limitation nobody mentions

An analogue inductive sensor measures over a very short span, typically a few millimetres, and the output is not linear across it. It is also affected by target material exactly as a switching sensor is, so an aluminium target reads differently from a steel one at the same distance.

That makes analogue inductive sensing appropriate for a narrow set of jobs: measuring a small gap, detecting runout or vibration on a shaft, checking thickness, or monitoring wear. It is not a general distance measurement device, and buyers who arrive wanting one usually want something else.

What they usually actually need

If the measurement is over more than a few millimetres, the right product is a position transducer rather than a proximity sensor. This is the part of the range we supply as sole authorised distributor in India and SAARC for Novotechnik and Contelec.

• Linear travel, from short strokes to long ones, with the right mounting for the mechanism. See the linear transducer comparison.
• Rotary angle, contacting or touchless, up to 360 degrees. The RFC-4800 and RSC-2800 are the most specified.
• More than one revolution, where absolute position has to survive a power cycle. See the multi-turn range.
• Inside a hydraulic cylinder, where nothing can be exposed outside.

These measure over real distances, output 0 to 10V or 4 to 20mA as standard, and several offer CANopen or IO-Link if the machine is bus-based.

How to get the right answer

Tell us what the controller does with the number rather than what output you want. The range of movement, the resolution you genuinely need as a figure, what the target is, the environment, and whether this is a new design or a replacement. We will tell you whether it is a switching job, an analogue proximity job or a position transducer job, and we would rather say so than sell you the wrong category.