Category: Industry Solutions

Packaging and pharmaceutical machinery asks sensors to do the two things they find hardest — detect transparent and low-contrast material, and do it fast — while surviving frequent format changeovers and, in pharma, a cleaning regime with documentation attached.

Here is how the specification usually needs to go.

Transparent material is the defining problem

Clear film, PET bottles, glass vials, blister foil backing and cellophane overwrap are all close to invisible to a diffuse photoelectric sensor. The light passes through, very little comes back, and detection is marginal in a way that varies with ambient light and film tension.

The reliable answer is through-beam. Emitter and receiver facing each other, the target detected by beam interruption rather than reflection. Even a transparent object attenuates the beam measurably, and with sensitivity set correctly this is stable where diffuse sensing is not.

Where a through-beam arrangement is not mechanically possible, retro-reflective is second best, but expect to spend commissioning time on sensitivity and expect it to need revisiting when the film supplier changes. The D31C range covers all three modes in M18 and M30.

Speed, and what it means for switching frequency

A packaging line running 300 packs a minute gives a sensor 200 milliseconds per pack, and the detection window inside that is much shorter. Two things have to be fast enough: the sensor, and the controller reading it.

Switching frequency is a specification parameter that gets ignored on slow applications and becomes decisive here. If the sensor is counting, specify against the maximum line rate with margin, not the nominal rate.

The failure mode is characteristic: counts that are correct at low speed and drift low as the line speeds up. If you are seeing that, the sensor or the input scan rate is the cause, not the mechanism.

Small targets and tight geometry

Packaging machinery is compact. There is rarely room for an M30 body, and the targets — a tablet, a cap, a label edge — are small. This pushes towards M8 and M12 inductive bodies for metal targets, and towards photoelectric with a well-defined beam for everything else.

Where the target is small relative to the sensing field, the sensor may respond weakly or inconsistently. Specify the target size when you enquire; it changes the recommendation more than people expect.

Changeover survives or it does not

A line that changes format several times a week puts mechanical stress on everything adjustable. Sensors get knocked, brackets get moved and not moved back, and cables get caught.

Three things help:

• Mount out of the changeover path wherever the geometry allows. The best protection is not being in the way.
• Use positive mechanical location — a dowel or a machined seat rather than a slot — so a sensor that is removed goes back in exactly the same place.
• Specify connector versions rather than fixed cable where sensors are routinely removed. An M12 connector survives repeated disconnection; a cable that is pulled through a machine frame every changeover does not.

Hygiene and cleaning in pharma

On pharmaceutical lines the cleaning regime is validated, which means it will not be adjusted to suit a sensor. Specify to survive it as it is.

• Stainless steel bodies, flush mounting, no crevices, no exposed threads collecting product, no horizontal ledges.
• Chemical compatibility with the actual cleaning agent — including the cable sheath, which is the component most often overlooked and quite often the first to degrade.
• Rating against the wash, not the process. Where high-pressure hot washdown is used, see IP69K washdown specification. The connector matters as much as the sensor: an IP69K sensor on an IP65 connector is an IP65 installation.
• Cable entries pointing down. Standing water on an upward-facing gland defeats any rating eventually.

Documentation

Pharmaceutical and export machinery frequently needs component documentation as part of the machine file — declarations of conformity, material certificates, and in some cases traceability to a production batch.

Establish what your quality system requires before ordering rather than during qualification. Components bought through unofficial channels are commonly untraceable, which is a difficult thing to discover late. Our note on buying direct versus through a channel covers how to avoid that.

Typical sensor map on a packaging line

• Product presence and gap detection — through-beam photoelectric, dark-on
• Counting — photoelectric with switching frequency specified against maximum line speed
• Metal cap, can and foil detection — D30C inductive in M8 or M12
• Carton, plastic and non-metal detection — capacitive, or photoelectric where the target moves fast
• Guard and interlock positions — normally closed, so a cut cable stops the machine
• Format part identification — inductive presence sensing to confirm the right tooling is fitted, which prevents an entire class of changeover error
• Servo and axis position feedback — position measurement rather than switching; see the Novotechnik linear range

Specifying

Send us the target — material, size, whether it is transparent or reflective — the line speed in units per minute, the available mounting space, the cleaning regime, and the output your controller expects. For transparent targets, a sample or a photograph is worth more than a description.

Cement plants present a combination that is unusual even among heavy industries: extremely abrasive dust everywhere, intense radiant heat in one area, and cable runs long enough that the electrical specification changes.

Each of those individually is manageable. Together they invalidate a few habits that work fine elsewhere.

Dust is the dominant problem

Cement dust is abrasive, it is fine enough to penetrate anything imperfectly sealed, and when it takes up moisture it becomes conductive and mildly cementitious. That last property is the one that catches people out — dust that has been damp and dried is not dust any more, it is a thin layer of set material, and it does not brush off.

Three consequences for specification:

Optical sensing is difficult. Lenses foul fast. Where photoelectric sensing is unavoidable, use through-beam mode, specify substantially more range than the gap requires so there is margin to lose, and plan for cleaning access. Diffuse mode in a cement plant is rarely a good decision.

Capacitive sensing needs care. Damp dust on the sensing face reads as a target. Where capacitive sensing is genuinely needed, mount so the face cannot collect material.

Inductive sensing is the workhorse, because dust between the sensor and a metal target affects it very little. This is why most of what actually survives in a cement plant is inductive.

Long cable runs change the electrical specification

Cement plants are physically large. A sensor on a conveyor gallery may be 150 or 200 metres from the nearest panel, and at that distance two things that are normally irrelevant become decisive.

Voltage drop. Over a long run at small conductor size, the supply reaching the sensor is measurably lower than the supply leaving the panel. Check the sensor still has adequate supply voltage at the far end, and size the conductor accordingly.

Noise pickup. A long cable running alongside VFD output cabling is an efficient antenna. Cement plants are full of large drives, and VFD carrier frequencies couple readily into signal cables.

The practical answers: use screened cable and earth the screen at one end only, physically separate signal and drive cabling, and where an analogue measurement is involved prefer a current output over a voltage output — a 4 to 20 mA loop is immune to the voltage drop and far more tolerant of induced noise. For long-distance position measurement, a digital or fieldbus interface removes the problem entirely; see CANopen position sensors.

Kiln area heat

Near the kiln, kiln hood and clinker cooler, radiant heat rather than ambient temperature sets the case temperature. The remedies are the same as in any hot process: increase distance, fit a shield, or provide cooling — in that order of preference, because each step adds something that can fail.

Measure the case temperature at the intended mounting position during normal operation before ordering. It takes ten minutes and prevents a recurring replacement.

Applications through the plant

Conveyors

Belt drift, belt slip, underspeed and pull-cord status account for a large share of the sensor population in a cement plant. Speed and underspeed monitoring is a protection function — a stalled belt under load is a fire risk and a mechanical one.

Magnetic speed sensors from the D33C range are appropriate here, and they tolerate dust better than optical alternatives because they sense through non-ferrous material.

Bucket elevators

Speed monitoring and belt alignment. An elevator that slips or misaligns damages itself quickly, and the sensors that detect it are among the most valuable in the plant.

Chutes and transfer points

Blockage detection. This is where plugged-chute events become spillage events, and where a sensor that works is worth a great deal. The environment is as bad as it gets — material flow, impact, dust — so mount with standoff and protect mechanically.

Gates, dampers and diverter positions

Straightforward D30C inductive switching, specified with generous sensing distance so the standoff can absorb dust build-up and mechanical slop. Normally closed for anything safety-related.

Silo and hopper level

Point level detection at high and low. Where the vessel is metallic, external capacitive sensing is not available and the solution needs to be inside — which changes the technology choice. Tell us the vessel construction when you enquire, because it determines the answer.

Damper and valve position feedback

Where the control system needs actual position rather than open or closed, this is position measurement rather than proximity switching. Rotary sensors from the Novotechnik range cover it, with touchless variants for the dustier positions.

The specification habits worth adopting

1. Bigger bodies, longer sensing distances, more standoff. M30 and above wherever space allows. The extra clearance is what tolerates dust build-up.
2. Short-circuit protected outputs — the -3S suffix — on everything. Cable damage is routine over these distances.
3. Screened cable, earthed one end, separated from drive cabling.
4. Current or digital outputs for anything analogue over distance.
5. Mount so faces shed material rather than collect it, and so they can be reached for cleaning.
6. Standardise. A plant this size with forty variants cannot stock spares properly.

If you want help with the last one, send us the plant sensor list and we will map it to the smallest set of equivalents that covers it.

Sugar is an unusual industry to specify sensors for, because the operating pattern dominates everything else. A mill runs continuously through the crushing season and then stops for months. During the season, unplanned downtime is expensive and cane is waiting. Outside it, there is time to do anything.

That shape should drive the specification. The question is not what is adequate — it is what will run to the end of the season without attention, because there is no window to fix it in February.

What the environment actually does

Fibre gets everywhere. Cane fibre and bagasse work into every crevice, pack around sensor faces, and bridge gaps that were designed to stay open. On a capacitive sensor, packed fibre reads as a permanent target. On an optical sensor it blocks the lens.

Juice is sticky, and it sets. Sugar-bearing juice dries to a hard deposit. A sensor face that is wiped clean weekly stays fine; one that is not accumulates a coating that changes its behaviour gradually rather than suddenly, which makes the fault hard to attribute.

Steam and humidity are constant. The boiling house runs wet and warm. Thermal cycling between a hot process area and cooler night-time ambient drives moisture into anything imperfectly sealed.

Mill drives vibrate heavily. Cane mills are high-torque, low-speed and rough. Mountings loosen and cables chafe.

Where sensors go, and what to specify

Cane carrier and feed

Belt or chain movement, level in the feed chute, and blockage detection. The requirement here is standoff — the sensor must be far enough from the material stream that fibre does not pack against it, which means specifying a larger body with a longer sensing distance than the geometry strictly needs. An M30 with 15 mm sensing mounted at 10 mm will outlast an M12 with 4 mm mounted at 3 mm, every season.

D30C inductive switches in M30 and above are the standard choice for metal targets on carriers and drives.

Mill roller speed and direction

Underspeed detection on mill rollers is a protection function, not a monitoring nicety — a slipping or stalled roller needs to be detected before something breaks. Magnetic speed sensors from the D33C range handle this, and the speed-and-direction variants are useful where reverse rotation matters.

Note the 30 mA output rating on that series: it feeds a controller input, not a contactor coil directly.

Juice, syrup and molasses level

This is where capacitive sensing earns its place. Juice has a high dielectric constant and is detected easily, and where the vessel wall is non-metallic the sensor can be mounted externally with nothing wetted and nothing to clean.

The caution is specific to this industry: syrup and molasses cling. If product coats the inside of the wall, an externally mounted capacitive sensor reads full permanently. Assess that before committing to the approach — on thin juice it usually works well, on heavy molasses often not.

Centrifugal position and interlocks

Lid position, discharge plough position and safety interlocks. Specify normally closed for anything that is a guard or a safety interlock, so a severed cable produces the stopped state rather than the running one.

Bagasse handling and boiler feed

Conveyor movement, chute blockage and level. Dusty, fibrous and in places warm. Photoelectric sensing is possible but needs through-beam mode and generous range margin, because lenses foul quickly in bagasse dust. Where a mechanical or inductive solution exists, it will usually need less attention.

Boiling house and pan floor

Warm, wet and steamy. Specify for washdown conditions rather than for the nominal ambient, and point every cable entry downward. Our note on IP67, IP68 and IP69K covers what each rating actually tests.

Specify for the season, not the day

Three practical rules that follow from the operating pattern:

Over-specify the sensing distance. The extra standoff is what buys tolerance to fibre packing and deposit build-up. This is the single highest-return decision in a sugar mill.

Standardise the part numbers. A mill with forty sensor variants cannot hold meaningful spares of any of them. A mill with eight can. During the season, being able to substitute from stores in ten minutes is worth more than an optimal specification for each position.

Specify short-circuit protected outputs. The -3S suffix. In an environment with this much vibration and cable chafe, output shorts happen, and an unprotected sensor is destroyed by the first one.

Use the off-season properly

The maintenance window between seasons is the time to do the things that prevent in-season failures: replace sensors that were marginal rather than failed, re-route cables that are chafing, fit guard brackets where something got hit, and rationalise the spares holding.

If you want to do that rationalisation, send us the current sensor list from the plant — part numbers and positions — and we will map it to the smallest set of equivalents that covers the mill. It is a more useful exercise than replacing like for like, and the off-season is when there is time to act on it.

A steel plant destroys sensors in a small number of very predictable ways. Once you know what they are, specification becomes a matter of designing around each one rather than hoping a high IP rating covers everything.

These are the five failure mechanisms we see, in roughly the order they cause problems.

1. Radiant heat, which is not the same as ambient temperature

The mistake is specifying against the plant ambient. A sensor mounted two metres from hot stock is not at ambient — it is receiving radiant heat continuously, and its case temperature can be far above the air around it.

Three responses, in order of preference:

• Move it. Distance is the cheapest heat shield there is, and a mounting position half a metre further back often solves the problem permanently.
• Shield it. A simple steel plate between the sensor and the radiating source blocks the line of sight. Effective and cheap.
• Cool it. Air purge or water cooling where the first two are not possible. Reliable, but now you have a service that can fail.

Measure the actual case temperature at the intended position during normal production before ordering. It is a ten minute job and it prevents a recurring replacement.

2. Scale

Mill scale is conductive, magnetic and abrasive, and it accumulates on everything. On an inductive sensor face a scale build-up can trigger the sensor permanently, which presents as a sensor that is stuck on and gets replaced unnecessarily.

Mount sensors facing downward or vertically where possible so scale falls away rather than collecting. Where the face must point up, plan for cleaning access, because it will need it.

3. Descaling water and cooling water

High-pressure descaling spray is closer to an IP69K condition than an IP67 one, and it comes with thermal shock: a hot sensor hit by cold water contracts rapidly and draws water past marginal seals. See our note on IP67, IP68 and IP69K for what each rating actually tests.

Specify for the spray, and point cable entries downward. An upward-facing cable gland in a wet area collects standing water and fails eventually regardless of its rating.

4. Vibration and mechanical damage

Mills vibrate, and cobbles happen. Two consequences worth designing for:

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

Things get hit. A sensor in an exposed position will eventually be struck by stock, by scale, or by a maintenance trolley. Where you cannot move it out of the way, a simple guard bracket costs very little and saves the sensor.

5. The cable, which is what usually actually fails

In our experience of mill installations, the sensor is rarely the first thing to go. The cable is — chafed on a sharp edge, crushed under something, cooked against a hot surface, or pulled out of the gland by thermal movement.

Specify the cable properly. Route it away from heat and away from sharp edges. Support it so the gland carries no weight. Leave a service loop. And where the run is long, consider specifying a made-to-length cable so the first joint is outside the hostile zone rather than inside it.

What to specify where

Position and limit sensing on cold or warm sections — D30C inductive switches, sized with generous sensing distance so the standoff can be increased for heat and scale clearance. M30 and above rather than M12 where space allows.

Hot stock detection and tracking — hot metal detectors, which sense radiated infrared and therefore need no proximity to the stock at all. Specification is a geometry problem; see what to specify before ordering an HMD.

Long-distance presence detection across a pass line — D31C photoelectric in through-beam mode, specified with substantially more range than the gap so there is margin as the lenses foul.

Roll and shaft speed, and direction — magnetic speed sensors from the D33C range, which detect through non-ferrous material and tolerate the environment better than optical encoders.

Screwdown, roll gap and hydraulic cylinder position — this is not proximity switching but absolute position measurement, and it wants a Novotechnik transducer. For high-cycle or wet positions the touchless rotary series are the appropriate choice.

Hazardous areas in the gas plant or by-products section — NAMUR switches with a suitable amplifier. See specifying intrinsically safe NAMUR loops.

Standardise the spares holding

A practical point that matters more than any individual specification. Mills accumulate sensor variety — a dozen manufacturers, forty part numbers, and a stores holding that cannot cover any of them properly.

Rationalising to a smaller number of variants across the plant reduces stockholding, makes substitution possible during a breakdown, and means the maintenance team knows the parts. If you want to do that exercise, send us the current sensor list and we will map it to the smallest set of equivalents that covers it.

Sensors on food, dairy and beverage lines do not usually fail because of the process. They fail because of the cleaning.

A clean-down cycle can combine water at 80 bar and 80 degrees Celsius with a caustic or acidic detergent, applied at close range by someone whose priority is a clean machine rather than an intact sensor. It is a more aggressive environment than the production process it is protecting, and it defeats hardware specified for the process alone.

IP67 and IP69K protect against different things

This is the distinction that matters, and it is routinely missed because the numbers look sequential.

IP67 certifies protection against temporary immersion — the sensor can sit under water at low pressure for a defined period. It says nothing about pressure, nothing about temperature, and nothing about a directed jet.

IP69K certifies protection against close-range, high-pressure, high-temperature water jets. The test applies water at approximately 80 bar and around 80 degrees Celsius from a short distance, at several angles, while the item rotates.

These are genuinely different failure mechanisms. Immersion applies gentle uniform pressure. A high-pressure hot jet applies concentrated force at a seam, and heat at the same time. Heat causes the air inside a sealed housing to expand and then, as the jet passes and the surface cools rapidly, to contract — creating a partial vacuum that actively draws water past any seal that is less than perfect.

That thermal cycling is why washdown failures often appear weeks after installation rather than immediately, and why a sensor can pass an IP67 test and still fail in a dairy.

Neither rating covers the chemistry

IP ratings test with water. Clean-down uses detergents, and typically alternates caustic and acid cycles. Chemical compatibility is a separate question from ingress protection and needs asking separately.

The materials that matter are the housing, the sensing face, the cable sheath and the seals. Stainless steel housings are standard for good reason; the cable sheath is the component most often overlooked and quite often the first to degrade. If your CIP regime uses a specific detergent, name it when you enquire.

What else to specify beyond the rating

• Hygienic mounting. Flush mounting with no crevices, no exposed threads collecting product, no horizontal ledges. In many plants this is an audit requirement, not a preference.
• Cable entry orientation. The most common ingress path is a cable entry pointing upward, collecting standing water. Point it down, or specify a connector version.
• The connector, if used. An IP69K sensor on an IP65 connector is an IP65 installation. The rating applies to the assembly, not the most impressive component in it.
• Temperature range including the clean cycle. The process may run at 4 degrees and the wash at 80. Specify against the wash.
• Cable length. A joint in the cable inside the wash zone is a failure point. Specify the full run to outside the zone.

What we supply for washdown duty

For position measurement in washdown environments, several Novotechnik series are rated to IP69 and are designed for exactly this kind of duty:

• RFX-6900 — absolute touchless redundant NOVOHALL heavy duty rotary sensor, rated to IP67 and IP69 variants
• RSX-7900 — absolute non-contacting redundant NOVOHALL heavy duty rotary sensor, also available to IP69
• IPX-7900 — absolute high-precision potentiometric rotary sensor for harsh operating conditions, available to IP69

These are touchless or sealed designs with no mechanical path into the measuring element, which is the structural reason they survive washdown rather than merely being rated for it.

For proximity switching duty on washdown lines, tell us the clean-down regime — pressure, temperature, detergent and frequency — and we will confirm what we can supply against it. We manufacture inductive, capacitive and photoelectric switches in Mumbai and can advise on the right specification for the wash conditions you actually run, rather than the ones the catalogue assumes.

A note on over-specification

IP69K is not free, and not every sensor on a food line needs it. A sensor inside a control panel, or one mounted above the wash zone and never directly jetted, does not. Map the wash zone honestly and specify IP69K where the jet actually reaches. Spending the budget on the sensors that get hit, rather than spreading it across all of them, usually produces a more reliable line.

Our overview of IP67, IP68 and IP69K ratings explains what each test involves in more detail.

A hot metal detector looks like a simple device. It watches for the infrared signature of hot stock and gives a signal when it sees one. In practice HMD installations fail more often than almost anything else on a rolling mill, and they fail for reasons that are decided at specification, not at installation.

This guide sets out what actually needs to be established before ordering, based on the enquiries that come to us and the ones that come back six months later.

What the device is doing

Hot stock radiates. A hot metal detector is an infrared receiver aimed at a defined point in the pass line, with the sensitivity and optics arranged so that it responds to the radiation from the stock and ignores everything else in the mill.

Everything difficult about HMD selection follows from that second clause. A rolling mill is full of hot objects: adjacent stands, furnace doors left open, glowing scale on the floor, steam backlit by something hot, a neighbouring line. The specification problem is not detecting the bar. It is not detecting anything else.

1. Field of view and mounting distance

These two are a single decision, because field of view at the target is a function of the optical angle and the distance.

Mount too close and you lose tolerance — a bar that wanders laterally leaves the field and the detector drops out mid-pass. Mount too far and the field widens until it takes in the stand behind, and the detector never drops out at all because it can always see something hot.

Establish the actual mounting distance before selecting the device, and establish the lateral wander the pass line really has, not the nominal centre line. On a roughing mill that difference can be substantial.

2. The smallest target you must detect

Specify by the smallest and coolest stock the line runs, not the typical product. A detector configured around a 100 mm slab at full temperature will miss the tail end of a thin section that has cooled on a delay. The awkward end of the product range sets the requirement.

3. Response time against line speed

On a fast finishing mill the stock may be in the field of view for a very short time, and the detector output has to be present long enough for the PLC scan to catch it. Two separate things need checking: that the detector responds fast enough, and that the control system samples fast enough to see the response.

An HMD that works perfectly on the roughing side and misses intermittently on the finishing side is almost always this problem rather than a faulty device.

4. Temperature threshold

Too sensitive and the detector picks up residual heat, scale and radiated background. Too insensitive and it misses cooled tail ends. The threshold should be set from the coolest stock that must be detected, with margin, and then checked against the hottest background that must be ignored. If those two overlap, the geometry needs changing — no threshold setting will resolve it.

5. The environment the device has to survive

A rolling mill is a hostile place for electronics. Three things matter:

• Ambient temperature at the mounting point. Not the mill ambient — the temperature where the device actually sits, which near a stand can be far higher.
• Cooling and air purge. Most mill installations need cooling for the housing and a purge to keep the optical window clear. An air purge is not an optional extra; a dirty window is the most common cause of a detector that gradually stops working.
• Water, scale and mechanical damage. Descaling spray, cobbles and maintenance traffic all reach places the designer assumed were protected.

Plan the purge air supply at the same time as the detector. Retrofitting it is expensive and frequently never happens.

6. Output and integration

Confirm what the control system expects — relay contact, transistor output, or an analogue signal — and the supply voltage available at the mounting point. On older mills the panel may have no DC supply where you need it, which changes the specification.

7. Alignment and maintenance access

The detector will need aligning during commissioning, and realigning after any mechanical work nearby. A device mounted where nobody can reach it safely with the line running will drift out of alignment and stay there. Specify the mounting bracket and access at the same time as the detector.

The three mistakes we see most often

Specifying from the catalogue instead of the geometry. The model number matters far less than field of view at the actual mounting distance.

No air purge. Works for three months, degrades slowly, gets blamed on the device.

Ignoring what else is hot. The single most common cause of false triggering, and the hardest to fix afterwards because it usually requires moving the mounting position.

Applications beyond rolling mills

Hot metal detectors are specified wherever hot stock has to be tracked without contact — continuous casting, forging lines, billet handling, furnace charging and discharge, shear control and cut-to-length. The specification logic is the same in each case: geometry first, timing second, environment third.

Getting a recommendation

Send us the mounting distance, the smallest and coolest stock to be detected, the line speed, what else is hot within the likely field of view, the ambient temperature at the mounting point, whether purge air is available, and the output your control system needs. A photograph or sketch of the intended mounting position is worth more than any of the individual numbers.

For background on how HMDs are used across a mill, see our overview of hot metal detectors in rolling mills.