Solenoid troubleshooting: an interactive fault-finding tree

Pick the symptom. The tree narrows to the most likely cause, the measurement that confirms it, and what to change. Every branch is drawn from the checks that decide real faults in the field, and every branch ends by pointing at the note that covers it in full.

What this is not. This tree does not calculate force, current or temperature, and it will not tell you how many newtons a design should produce. Published closed-form solenoid force calculations disagree with measured results by wide margins, because the magnetic circuit is non-linear, leakage changes with the gap, fringing has no closed-form solution, and the assembled gap is rarely the drawn gap. What follows is judgement about order and cause, not prediction.

Start

The complete decision tree, in text

This is the same logic the tool above walks, written out in full. It is here so the whole tree can be read, searched and copied without running anything.

1. Nothing happens at all — no click, no movement

What does the coil measure across its terminals, disconnected from the drive? — Meter on the ohms range, coil unplugged.

Open circuit → Open winding

Much lower than the rated value → Short to ground, or an inter-turn short

Close to the rated value

With the coil energised, what voltage do you measure AT the coil terminals? — At the coil, not at the supply. Probe the terminals themselves, not the connector upstream of them.

Zero, or a few tenths of a volt → The drive is not switching

Well below nominal — for example 9 V on a 12 V unit → Voltage at the coil is below the rated value

Nominal, and it still does nothing

Can you push the plunger through its full stroke by hand? — Power off. Note whether the mechanism is still coupled.

No, it is bound or very stiff → Mechanical bind or misalignment

Yes, it moves freely

Is this a latching (bistable, permanent-magnet) unit?

Yes → A latching unit driven with a single-polarity pulse

No, or I am not sure

Decouple the solenoid from the mechanism. Does it actuate on its own?

Yes — it works alone but not when coupled → The mechanism is binding, not the solenoid

No — still nothing even with no load → The real air gap is larger than the drawing

I have not measured it yet → Measure the coil resistance before anything else

2. It clicks, but the plunger does not move

Is the click a single event per energisation, or does it repeat?

One click each time power is applied

Can you push the plunger by hand from where it stalls to the fully seated position?

No — it binds part of the way through → Mechanical bind or misalignment

Yes, the rest of the stroke is free

How much of the stroke does it actually cover?

Barely moves at all → Current is far below what the coil should draw

Most of it, then stalls near the end → Force falls off faster than the load rises

Continuous buzzing or chattering → Chattering — the drive is dropping out

3. It moves, but there is not enough force

Where along the stroke is the force short?

At the start, pulling away from rest

With the unit fully energised and warm after 30 minutes or more, is the coil current below the cold value? — Copper resistance rises about 0.393 percent per kelvin, so a drop is expected — the question is how far.

Yes, noticeably below the cold value → Copper resistance rise — the correction every calculation skips

No, at or above the cold value

What does the supply do at the moment the solenoid energises?

The voltage sags or dips noticeably → The supply cannot hold voltage during pull-in

It holds steady

Is there anything between the pole faces — shim, plating, paint, gasket, or a burr?

Yes, something is in the gap → Something non-magnetic is sitting in the gap

No, faces are clean metal to metal → There is no margin — the unit was sized to the nominal case

I have not measured it → Copper resistance rise — the correction every calculation skips

Near the end, approaching the seated position → Force falls off faster than the load rises

Uniformly down everywhere

Measure the real air gap at the hold position with feeler gauges or a dial indicator.

Larger than the drawing says → The real air gap is larger than the drawing

Matches the drawing

4. It works when cold and fails once it is hot

Verdict: Thermal expansion is closing a clearance

5. It will not release after the power is removed

With the power off, does pulling the plunger away feel like a magnetic pull, or like friction?

A magnetic pull that lets go suddenly → Residual magnetism is holding the plunger

Friction, stick-slip, galling, or the load holding it → The release path — return force or friction

6. It buzzes, hums or rattles

Is the noise continuous while the unit is held energised, or only at the moment of switching?

Continuous while held

Is the coil driven from AC mains, or from DC?

AC → AC operation without an effective shading ring

DC → DC unit humming — PWM ripple, or a loose magnetic path

Only at switching on and off

What does it sound like?

A single sharp clack → Impact noise at the end stop

A rattle or a repeat chatter → Chattering — the drive is dropping out

7. The coil burned out, smells, or runs far too hot

How long did the coil survive?

Seconds to minutes

What drives it?

AC mains → AC coil energised with the stroke blocked

DC, or DC through a driver

During the long energised periods, is the plunger fully seated? — A blocked stroke, a jammed load, or a partly open position all count as not seated.

No — it holds part way or the load stalls it → DC coil held with the plunger unseated

Yes, it seats and stays seated → Overvoltage, or the wrong coil fitted

Weeks to months

Which of these is true?

It is energised longer, or more often, than the rated duty figure allows → The real duty exceeds the rated figure

The duty looks fine, but the mounting is plastic or enclosed → The heat has nowhere to go

The duty figure came from a catalogue and was never measured → The real duty exceeds the rated figure

8. It is slow to respond

Is it slow to pull in, or slow to release?

Slow to pull in → The L/R time constant is setting the response

Slow to release

Is there any suppression across the coil — a flyback diode, an RC snubber, a TVS?

Yes, a plain diode across the coil → The flyback diode is slowing the release

No suppression at all → The release path — return force or friction

9. It works on the bench but fails in the machine

Verdict: Works on the bench, fails in the machine

10. It is a valve, and it leaks

Where does it leak?

Past the seat while it should be closed

What is the fluid condition?

There is debris or scale, or the water is unusually soft or deionised → Debris, scale or aggressive water at the seat

The fluid is clean

Is it a direct-acting valve, or a pilot / diaphragm type?

Pilot-assisted or diaphragm → Pilot valve without enough differential pressure

Direct acting → A direct-acting valve needs sealing force, not just stroke force

Externally — body, threads, or out of the coil area → External leak — check the cable before the seal

11. It fails in a damp, wet or washdown environment

Does it also fail cold, or only after it has been running?

Only after it has been running, or after thermal cycling → Thermal breathing pulls water in

It fails even when cold → Corrosion from the environment, not from one bad part

12. Force is inconsistent between units in production

Verdict: Batch variation, or a design with no margin

Verdicts in full

Every verdict the tree can reach, with how to confirm it and what to change.

Open winding Coil

In a solenoid this is almost always a mechanical failure wearing an electrical costume. The break sits where the magnet wire leaves the bobbin and joins the lead-out, or at the first turn against the terminal pin. Vibration, a harness that pulls on the coil, or a soldered joint with no strain relief all produce it. An open from day one is a manufacturing escape; an open after months of service is nearly always that joint.

Confirm it. Measure across the disconnected terminals and compare with the datasheet. Then flex the lead-out gently while watching the meter — a reading that flickers confirms the joint rather than the winding.

What to change. Re-terminate the lead-out and add strain relief. If the break is inside the winding the coil is scrap; do not rewind one unit as a field repair, because the insulation damage behind it is still there. Then fix why the joint moved.

Short to ground, or an inter-turn short Coil

A resistance far below the rated value means turns are being bypassed. Two mechanisms look identical on a meter and need different responses: a short from the winding to the core or body, and a short between adjacent turns inside the coil. Inter-turn shorts usually start at a nick in the enamel, or where the winding was laid too tight against a sharp bobbin corner, and they grow with each thermal cycle until the current is high enough to cook the coil.

Confirm it. Measure terminal to body — a good unit reads open. For an inter-turn short, compare the measured resistance with the rated value cold, then again after a heat soak: a resistance that falls disproportionately with temperature points at shorted turns rather than a legitimate 0.393 percent per kelvin rise.

What to change. Replace the coil. Then look at what caused it: bobbin corner radius, winding tension, and whether the coil is being run past its insulation class limit, which the coil temperature note covers.

The drive is not switching Circuit

Zero volts at the coil with a healthy winding means the problem is upstream of the solenoid entirely. The usual suspects in order are the switching device, the connector, and the return path. A contactor that has welded open, a MOSFET that has failed short or open, a corroded connector pin, and a broken ground account for most of these. None of them are solenoid problems, and all of them look exactly like one.

Confirm it. Measure from one coil terminal to the supply positive, then from the other terminal to ground, while commanding the output on. That tells you which side of the coil is not being completed. Then measure at the driver output pin directly — if it is switching there, the loss is in the wiring or connector between them.

What to change. Repair the circuit, not the part. If a connector pin is corroded, find out why it is wet: a connector that admits water is a sealing problem that will return. Then check the load current against the driver rating, because an undersized driver that keeps failing is a specification error rather than a component fault.

Voltage at the coil is below the rated value Circuit

A solenoid can only produce the force its ampere-turns allow, and ampere-turns fall in direct proportion to voltage. Nine volts on a 12 V coil leaves roughly 75 percent of the rated current, and less than that in force, because holding force falls faster than current once the magnetic circuit is anywhere near saturation. This is the most common cause of a unit that works on the bench and fails in the machine, where cable runs are longer and connectors are older.

Confirm it. Measure at the coil terminals with the unit energised, not at the supply. Then measure the same two points on a bench supply set to nominal: if the bench works and the machine does not, the difference is entirely in the wiring.

What to change. Reduce the voltage drop rather than buying a bigger coil: larger conductor, shorter run, fewer connectors, and a supply that holds up under load. If the drop cannot be fixed, the honest choice is a coil wound for the voltage you actually have at the terminals.

Mechanical bind or misalignment Mechanical

The plunger is being asked to move through a clearance that is not there. Bore and plunger tolerance, concentricity between the bore and the mounting face, and the flatness of the surface the unit bolts against all contribute. A unit that binds only when bolted down is being distorted by the mounting; the same unit free in your hand is a different assembly. On a small bore, a running clearance below about 0.05 mm leaves very little room for any of this.

Confirm it. Push the plunger by hand with the unit clamped as installed, then with the mounting screws loosened. If loosening the screws frees it, the fault is flatness or distortion rather than the solenoid. Also stroke it hot if you can, because a clearance that is fine cold can close once the coil is at temperature.

What to change. Correct the mounting face flatness and the screw torque first, since that is usually the cheaper fix. If the bore itself is tight, the part is out of tolerance and the conversation with the supplier is about concentricity and surface finish, not about force.

A latching unit driven with a single-polarity pulse Drive

A latching solenoid holds by permanent magnet and changes state only when the coil current reverses. A single low-side switch driving it the same direction every time will set it once and then do nothing at all, which looks exactly like a dead coil. This is a design error rather than a fault, and it is common because latching coils are often specified late in a project by someone who chose them for the energy saving.

Confirm it. Measure the resistance — a healthy latching coil reads normally. Then apply a pulse of reversed polarity from a bench supply. If it changes state, the unit is fine and the drive is the problem.

What to change. Drive it from an H-bridge or a pair of changeover devices, and put the reservoir capacitor in. Then respect the upper energy limit as well as the lower one: a reverse pulse large enough to exceed the magnet coercivity weakens the magnet permanently, and that cannot be recovered in the field.

The mechanism is binding, not the solenoid System

The unit actuates freely on its own and does nothing once coupled, which means the load is not free to move. This is the point at which most field replacements are wasted. A solenoid is strong enough to move a clean mechanism and not strong enough to move one that has tightened up through thermal growth, dried lubricant, galling on a guide, or debris in a slot. The solenoid is doing its job; the evidence is that it works when nothing is attached.

Confirm it. Measure the force needed to move the load with the solenoid disconnected, using a spring scale or a load cell. Compare that with the force the solenoid produces at that position. If the load requirement is above the available force at any point in the stroke, the case is closed without touching the solenoid.

What to change. Fix the mechanism. Then write the measured load requirement down as a specification, because that number becomes the input for the next sizing exercise rather than a matter of opinion.

The real air gap is larger than the drawing Magnetics

Air dominates the magnetic circuit. Doubling the gap roughly halves the usable force, and the leak factor applied by hand underestimates the loss precisely because leakage grows as the gap opens. The gap in the machine is almost never the gap in the drawing: mounting face flatness, plating thickness, a paint layer, a gasket, and a burr rolled over a bore edge all add to it.

Confirm it. Measure the gap at the hold position with feeler gauges or a dial indicator, with the unit clamped as it will be in service. Then sum the contributors: face flatness, plating, gasket compression, boss height tolerance. The arithmetic usually shows where the gap went.

What to change. Close the gap before adding copper. Machining the stop face, removing non-magnetic coating from the working faces, and tightening the boss height tolerance are all cheaper than a larger coil, and they work at any temperature.

Something non-magnetic is sitting in the gap Magnetics

Nickel plating, paint, anodising, a gasket, and even a thick oxide layer all behave as additional air gap. On a small tubular unit, 20 micrometres of non-magnetic coating is 20 micrometres of gap, and on a gap designed at 0.15 mm that is over ten percent of the circuit. This is easy to miss because the part looks correct and measures correct.

Confirm it. Measure the gap both with and without the coating present, comparing a plated part against a bare sample. If the force returns when the coating is removed, you have the answer.

What to change. Keep non-magnetic finishes out of the working gap and plate only what needs plating. If corrosion protection is the reason for the coating, deal with it on the external surfaces and leave the pole faces bare where the environment allows.

Current is far below what the coil should draw Magnetics

Ampere-turns set the force, and current is what you can actually measure. A coil drawing a third of its rated current will not move anything, however good the magnetic circuit is. The causes are a supply that cannot deliver, a badly sized series element, a driver in a current-limited mode, or a coil wound to a different resistance than the label claims.

Confirm it. Measure current in the coil circuit while energised, and the voltage across the coil at the same instant. Divide to get the actual resistance and compare with the nameplate. A resistance far above rated is the wrong coil; a correct resistance with low voltage is a supply problem.

What to change. Fix the current source rather than the coil. If PWM is used for holding, check whether the pull-in interval really gets full supply — a driver that starts modulating immediately leaves the unit under-powered at the one moment it needs everything.

Force falls off faster than the load rises Magnetics

Force rises steeply as the gap closes, so a unit that stalls near the end of the stroke is not short of total force — it is short of force at that specific position. That usually means the load curve is rising too: a return spring compressing, a detent, or a mechanism that gets harder to move as it approaches its stop. Comparing peak force with required force hides this completely.

Confirm it. Plot both curves over the stroke: solenoid force at each position from the datasheet or a measurement, and load requirement at each position from a spring scale. The crossing point is the stall point. A single peak-force comparison will not show it.

What to change. Either use a design with a flatter force-stroke curve, which is where a tubular unit usually beats a flat-face one over a longer stroke, or reduce the load requirement near the end. Change the spring rate before changing the solenoid.

Copper resistance rise — the correction every calculation skips Thermal

Copper resistance rises about 0.393 percent per kelvin. A winding that stabilises 70 K above ambient is carrying about 78 percent of its cold current and producing roughly 61 percent of its cold force, because force falls with the square of current in the region that matters. This is not a fault and it is invisible on a cold bench test. It is the difference between a unit that passes incoming inspection and one that fails in the field three months later.

Confirm it. Measure coil current cold, run the unit to thermal equilibrium, then measure again. A current that has dropped by the percentage the temperature rise predicts confirms it. Any specification written from a cold reading has this error built in.

What to change. Write the specification in the hot state, and give the force figure at the temperature the product will actually reach. If hot-state force is short, the answer is more copper or better cooling rather than a different core material, since core material differences are usually within a few percent.

The supply cannot hold voltage during pull-in Circuit

At the start of the stroke the magnetic circuit has its lowest inductance, so the coil draws its highest current, and that happens at exactly the moment the force is most needed. A supply sized on steady-state holding current is therefore undersized for the only moment that matters. The symptom is a unit that pulls in weakly or not at all and then holds perfectly well, which sends people looking at the magnetics.

Confirm it. Scope the coil voltage from the instant the drive switches on, not after it settles. A voltage that dips and recovers has identified itself. Then compare the supply rating with the pull-in current rather than the holding current.

What to change. Add reservoir capacitance close to the drive, or raise the supply rating. If the drive is a battery, remember that a coin cell has enough energy for a short pulse and nowhere near enough current without a capacitor, and size that capacitor from the energy per operation rather than from the average current.

There is no margin — the unit was sized to the nominal case Specification

If the winding measures correct, the voltage is correct, the gap is as drawn, and the faces are clean, then the design has no allowance for the things that always happen: temperature rise, tolerance stack, a supply at the low end of its band, and a load at the high end of its range. A unit with no margin passes every bench check and fails in some fraction of field units, which is the definition of an intermittent complaint nobody can reproduce.

Confirm it. Build the worst case rather than measuring the nominal one: supply at its minimum, the relevant temperature extreme, and the maximum load in the tolerance band. Then compare with the minimum force the mechanism needs, not with the peak force available.

What to change. Decide the margin deliberately and write it down. The practical rule is that available force must exceed required force by a factor that covers the temperature derate, the tolerance stack and the load variation together, then verify it on a worst-case assembly.

Thermal expansion is closing a clearance Thermal

A unit that works cold and fails hot is almost never short of magnetic force at the moment it fails. Heat changes geometry: bore and plunger grow at slightly different rates depending on material, a plastic mounting boss grows far more than either, and a coil that expands pushes its own end faces outward. Clearances that are correct at 20 degrees are frequently closed at 90. This failure mode is invisible in any room-temperature test, which is why it survives so much incoming inspection.

Confirm it. Reproduce it hot rather than reasoning about it. Run the unit to thermal equilibrium, then measure the running clearance and the stroke at temperature. Failing that, measure force cold and again at equilibrium on the same fixture — a force that falls by more than the copper derate alone predicts points at geometry rather than magnetics.

What to change. Open the clearance and check the thermal expansion of the mounting rather than of the solenoid. If the unit is bolted to plastic, the plastic is usually the dominant term. Then add the hot-state check to the test plan so the next revision cannot regress.

Residual magnetism is holding the plunger Magnetics

Soft magnetic steel does not fully let go when the current stops. Remanence produces a small but real holding force that grows with the flux the material was last exposed to, so a unit held at high excitation sticks more than one that was pulsed. Free-cutting grades with high sulphur content make this worse, because their inclusions raise coercivity. The tell is that the plunger releases with a distinct snap once you overcome the initial pull, rather than easing away.

Confirm it. Measure the release force with a spring scale immediately after power-off, then again after demagnetising the plunger. A force that disappears after demagnetising was remanence. Materials with lower coercivity release more cleanly.

What to change. Give the return path more force than the remanence needs, which in most cases means a slightly stronger return spring rather than a different material. If the unit must release absolutely, choose a material with lower coercivity for the plunger and keep the working gap non-zero so the circuit is never fully closed.

The release path — return force or friction Mechanical

Release is a different design problem from actuation and usually gets far less attention. The plunger has to be moved back by something, and that something is either a spring or the load itself. Both can be marginal: a return spring has to overcome friction in the guides, any stiction from dried lubricant, and the residual force left in the magnetic circuit. Systems that rely on the load to return fail as soon as the load is absent, which happens on the bench and in the workshop.

Confirm it. With the power off and the unit mounted as installed, push the plunger to the energised position and let it go. Then repeat with the mechanism coupled and uncoupled. The comparison separates a spring problem from a load problem, and it takes under a minute.

What to change. Add or strengthen the return spring rather than polishing the bore as a first move. Then check lubrication and the guide surface for galling, because friction scatter between plungers commonly exceeds the magnetic scatter between them.

AC operation without an effective shading ring AC

On AC the force passes through zero twice per cycle, so an unshaded armature would be attracted and released 100 times a second at 50 Hz. A short-circuited ring set into part of the pole face delays the flux in that section by roughly a quarter cycle, so while one part of the face is at zero the other is still producing force. A missing, cracked or badly seated ring leaves the unit able to hum loudly and unable to hold cleanly. The fault looks electrical and is entirely mechanical.

Confirm it. Inspect the shading ring for a crack or a gap where it seats in the pole face. Then compare the hum against a known-good unit of the same type on the same supply. A ring that has lost its short-circuit path through corrosion behaves exactly like a missing one.

What to change. Replace the pole assembly or reseat the ring. This is not something to compensate for with more coil, and a larger coil on an unshaded AC unit will burn out rather than hold.

DC unit humming — PWM ripple, or a loose magnetic path Drive

A DC solenoid has no inherent reason to hum. If it does, something is modulating the current, or the magnetic path is moving. The first case is a PWM holding drive whose frequency sits in the audible band, which is a deliberate design choice that will be audible unless the frequency is chosen above it or the coil is held by a reservoir capacitor. The second is a laminated or bolted magnetic path being moved by the flux itself.

Confirm it. Look at the coil voltage on a scope rather than a meter. A visible ripple at a few kilohertz confirms the drive. If the waveform is clean, press on the stack and the frame while energised — a hum that changes when you press is a mechanical path problem.

What to change. Raise the PWM frequency out of the audible band, or hold with a capacitor instead of modulation. For a mechanical hum, tighten and where possible bond the stack; a bolted lamination stack that is not varnished will keep buzzing.

Chattering — the drive is dropping out Drive

Repeated clicking at a rate far slower than the supply frequency means the drive is losing and regaining command. Common causes are a supply that collapses under the inrush current and recovers, a controller with an undervoltage or overcurrent trip that resets itself, and a thermal cutout in series with the coil that cycles. Chattering damages a solenoid quickly, because each attempt draws the full inrush current while the plunger is still unseated.

Confirm it. Watch coil current and coil voltage together on a scope over several seconds. Chatter appears as a sawtooth: current rises, the supply dips, the command drops out, then the cycle repeats. The period points at the mechanism, since thermal cycling is slow and a supply collapse is fast.

What to change. Fix the drive margin. Add reservoir capacitance near the coil, check that the driver is rated for the inrush rather than the holding current, and remove any thermal cutout that is cycling by design.

Impact noise at the end stop Mechanical

A single sharp clack is the armature arriving and the energy of the moving mass being absorbed by the stop. That energy has to go somewhere: at a mass of about 8 grams arriving at 0.5 metres per second, each operation is roughly a millijoule, and a million operations is of the order of a kilojoule delivered into the same two faces. The consequence is not noise but geometry — the pole faces take edge damage, and once the gap has opened by 0.02 mm the force has already lost about twelve percent.

Confirm it. Measure the gap before and after an endurance run rather than listening to the noise. Gap growth is the real damage. Also check whether the two faces meet edge-first, which concentrates the impact on a small area instead of distributing it.

What to change. Decide deliberately what absorbs the impact: a resilient stop, a controlled deceleration, or explicit acceptance of face wear with a service interval. Do it as a stated design decision, because an undamped end stop is a brake made of the magnetic circuit.

AC coil energised with the stroke blocked AC

On DC the current is set by resistance, so a stalled plunger changes little. On AC the plunger position is part of the magnetic circuit: with the stroke open the inductance is low and the current is set largely by resistance, so pull-in current typically runs three to ten times holding current. Block the stroke and the coil never reaches the holding current at all — it sits at the inrush value and dissipates several times its rated power continuously until it fails. That takes seconds to minutes.

Confirm it. Measure coil current with the plunger free, then with the stroke mechanically blocked. The ratio between the two is the fault. A unit that measures normal while free and overheats when assembled has its stroke blocked by a jammed linkage, a binding damper, or an obstruction.

What to change. Restore the stroke. Do not fit a bigger coil, because the coil is not the problem and a larger one will fail the same way slightly later. Then add the blocked-stroke case to the specification, since it is what the field will produce eventually.

DC coil held with the plunger unseated Thermal

A DC solenoid has lower inductance with the gap open, so it draws more current unseated than seated. On a continuous-duty coil that is a modest increase and manageable. On a coil wound for a short pulse duty it is not: the wire was chosen for a brief high current, and holding even the seated current indefinitely will cook it. A mechanism that jams part way turns a correctly specified coil into a continuous overload, which is why the failure appears as a coil problem and the cause is mechanical.

Confirm it. Measure coil current seated and unseated, then compare both with the coil rating. Then check the actual duty: how long is the coil energised per cycle, and what fraction of that time does the plunger stay unseated?

What to change. Decide from the duty, not from the peak force. If long holding is required, use a holding current lower than the pull-in current, or move to a latching design where the state persists without power at all.

Overvoltage, or the wrong coil fitted Thermal

Power dissipated in a coil goes with the square of voltage divided by resistance, so a 24 V coil on a 36 V supply dissipates more than twice its rated power. Equally, a 12 V coil fitted in place of a 24 V one does the same thing with no supply fault at all. Both fail in seconds to minutes with a seated plunger and no mechanical problem. The winding resistance is the fastest way to catch it: if the resistance does not match the voltage on the label, the label and the coil disagree.

Confirm it. Measure resistance and compare with the rated value for that supply voltage. Then measure the actual supply under load, not the nominal figure printed on the supply case — an unregulated supply can sit well above its stated output at light load.

What to change. Fit the correct coil for the voltage, and if the supply tolerance is wide, specify for the top of the band rather than the nominal. Unregulated supplies and battery packs near full charge are the two cases where the nominal figure misleads most.

The real duty exceeds the rated figure Thermal

Duty cycle is the most commonly copied and least commonly verified number on a solenoid datasheet. It is set by the insulation class and by how quickly the coil can shed heat, and it changes with ambient and mounting. A figure quoted for an open frame unit in free air does not transfer to the same coil inside an enclosure at 50 degrees ambient. The coil then fails weeks later, long after everyone has stopped looking at the power supply.

Confirm it. Measure the coil temperature at equilibrium with the real duty, in the real enclosure, at the real ambient. Then compare with the insulation class limit minus a margin. Any duty figure that has never been measured on the actual assembly is an assumption, not a rating.

What to change. Either reduce the duty, improve the thermal path, or move to a class with a higher temperature limit. Reducing the duty usually means changing the mechanism so it holds its position without power, which is where latching designs earn their extra driver cost.

The heat has nowhere to go Thermal

A coil in free air sheds heat by radiation and convection from exposed surfaces. The same coil inside a closed shell, bolted to a plastic boss, loses both routes and is left with conduction along a narrow path into the mounting. At equal power the enclosed unit runs considerably hotter, and the resistance rise then takes force away in exactly the way the force calculation note describes. Enclosure design decides coil temperature at least as much as the winding does.

Confirm it. Measure coil temperature in free air, then in the real mounting, at the same power and the same ambient. The difference is the mounting, quantified. Repeat with the enclosure closed if there is one.

What to change. Give the heat a path: a metal mounting surface, a thermally conductive interface, or a larger contact area. Where that is not possible, derate the coil rather than pretending the free-air figure still applies.

The L/R time constant is setting the response Magnetics

How fast the current rises in a coil is set by inductance divided by resistance. The time constant depends on the core cross-section, core length, copper area and mean turn length, and rewinding with a different wire gauge leaves it unchanged, because turns and conductor area move together. So a slow pull-in cannot be fixed by changing the wire gauge. At three time constants the current has reached about 95 percent of its final value, which is a useful target to size against.

Confirm it. Measure the coil current rise on a scope and read the time to reach 63 percent of final value: that is one time constant. Compare the result with the response the mechanism actually needs. A mechanism that needs 20 milliseconds and a coil whose time constant is 40 will never get there on force alone.

What to change. Shorten the electrical time constant or overdrive it. Reducing turns reduces inductance but also reduces ampere-turns, so the honest routes are a higher pull-in voltage for the duration of the stroke, or a capacitor discharge that puts a large voltage across the coil briefly. Adding turns to get more force makes the response worse, and that trade-off belongs in the specification.

The flyback diode is slowing the release Drive

A plain diode across the coil protects the switching device by giving the stored energy a path. It also keeps current circulating through the coil until that energy is dissipated in the coil resistance, which delays the collapse of the field and therefore the release. Measured release times can be several times longer with a plain diode than with no suppression at all. This is a deliberate trade-off that is frequently made by accident.

Confirm it. Compare release time on a scope with the diode fitted, then with it removed or replaced by a Zener or a resistor in series. A release that shortens when the suppression is changed has identified the cause.

What to change. If release speed matters, replace the plain diode with a Zener or an RC snubber, which clamps the voltage while allowing the field to collapse faster. Then check the switch rating against the higher clamp voltage, because that is the trade being made.

Works on the bench, fails in the machine System

This pattern usually means the bench test was run at something other than the conditions the machine provides, and three measurements resolve it in almost every case: the voltage at the coil terminals in the installed harness, the real air gap with the unit clamped as installed, and the force needed to move the actual loaded mechanism. A bench supply holds nominal voltage, the bench fixture has a clean flat mounting face, and the load is often absent. All three of those differences favour the bench.

Confirm it. Measure at the coil terminals with the machine harness and the machine supply, not on the bench. Measure the gap with the unit torqued down in its real mounting. Measure the load with a spring scale. Those three numbers, taken in the machine, explain the discrepancy far more often than the magnetics do.

What to change. Move incoming inspection onto the mounting conditions the part will really see. A test that cannot fail the way the field fails it is not a test, it is a formality.

Debris, scale or aggressive water at the seat Fluid

A valve that starts leaking past the seat after a few months usually has something trapped at the seat rather than a worn seal. The counter-intuitive part is the water: softened and low-mineral water can be more aggressive to elastomers and metals than hard water, because the ions that would buffer or passivate have been removed. Scale on the seat, a spalled particle of elastomer, and a swollen diaphragm all produce a leak that looks like a seal failure.

Confirm it. Take the valve apart and look at the seat under magnification rather than replacing the seal by reflex. Then ask the customer what the water actually is: inlet temperature, hardness, and whether it has been treated. That question is the one that predicts the service interval.

What to change. Filter upstream of the valve, choose an elastomer for the actual fluid rather than for the generic case, and write the fluid condition into the specification so the service interval becomes a documented expectation rather than a complaint.

External leak — check the cable before the seal Sealing

The assumption is always that a seal has failed, and the most common path is frequently the cable. Water travels along the interstices between the strands of a multi-strand conductor by capillary action, entering at a damaged or undersized cable gland or a connector that admits water and arriving inside the coil. That path bypasses every O-ring in the assembly. Tellingly, it leaves the seal intact and the coil wet.

Confirm it. Open the unit and look for corrosion at the far end of the cable, inside the coil area, while the seal is still undamaged. Then check the gland for correct cable diameter sizing. A gland sized for a nominal diameter that the cable does not actually have will pass water and still look correctly installed.

What to change. Seal the cable entry by actual diameter rather than nominal size, and where the unit will be immersed or washed down, consider a potted termination. Then re-rate the product honestly against the real washdown regime, which the IP rating note covers.

Pilot valve without enough differential pressure Fluid

A pilot-assisted valve uses line pressure to hold the diaphragm closed and a small pilot orifice to relieve it. That means it needs a minimum pressure differential to operate at all, and it will leak or fail to seat below that point. The failure is therefore a system pressure problem presented as a valve problem. Zero differential and near-zero flow conditions are where this shows up, such as a closed-loop system at rest or a gravity feed with almost no head.

Confirm it. Measure the differential pressure across the valve at the moment it is supposed to close, and compare with the minimum differential on the datasheet. A valve that seals at 1 bar and weeps at 0.1 bar is behaving exactly as a pilot valve should.

What to change. Either guarantee the minimum differential in the system, or move to a direct-acting valve where the coil has to provide the full sealing force. That change costs force, which then has to be reflected in the coil specification.

A direct-acting valve needs sealing force, not just stroke force Magnetics

On a direct-acting valve the coil has to press the seal against the seat hard enough to hold the full line pressure, and that force has to be available at the seated position with the coil hot. The required force is pressure multiplied by the seat area, plus whatever the seal needs to deform into contact, plus margin. A design sized on the pull-in force at the start of the stroke can be entirely adequate there and inadequate at the seat, which is where the leak then appears.

Confirm it. Calculate the required sealing force from the line pressure and the seat diameter rather than measuring the valve. Then compare with the force the unit produces seated, at operating temperature. The two numbers are usually further apart than expected, because seat area is easy to underestimate.

What to change. Close the gap so more of the force survives to the seated position, and where possible reduce the seat diameter, since required force scales with its area. Both changes are cheaper than a larger coil.

Thermal breathing pulls water in Sealing

IP65 is a spray rating, not an immersion rating, and the reason it matters here is thermal cycling. A sealed cavity at 20 degrees heated to 90 expands its internal air by roughly a quarter of its volume at that temperature: a 5 cubic centimetre cavity needs to expel about 1.2 cubic centimetres or raise its internal pressure by roughly 0.24 bar. It usually does both, pushing past the seal on the way up and drawing moist air back on the way down. That air condenses, leaving of the order of a milligram of liquid water inside, which on a 20 micrometre gap is a film that will not drain away.

Confirm it. Run a powered thermal cycle and then open the unit, rather than spraying a cold one. Water found inside after a powered thermal cycle, with the seal intact, is a breathing problem. A standard spray test on a powered unit reproduces it; a spray test on a cold one does not.

What to change. Vent the cavity with a membrane that passes air and not water, or accept condensation and protect against it rather than trying to exclude it. The second option means conformal coating on the coil and a corrosion-resistant pole face, which is usually cheaper than a perfect seal.

Corrosion from the environment, not from one bad part Materials

Corrosion in a solenoid is a systems problem: humidity provides the electrolyte, dissimilar metals provide the couple, and inclusions in free-cutting steel provide the initiation sites. Hydrogen sulphide in the environment and sulphur inclusions in the steel are the same problem approached from two directions, which is why free-cutting grades machine beautifully and corrode sooner. A short salt spray test on a new unit will not reproduce the field failure, which means qualifying to that test gives false confidence.

Confirm it. Cycle the unit thermally in the humid state and inspect, rather than testing a fresh unit. Look specifically at the pole faces, the plunger surface and the lead-out joint, which is where initiation sites concentrate. Then check for condensation after a powered cycle.

What to change. Address the material and the cavity together: a low-sulphur grade where magnetics and corrosion both matter, a surface treatment that keeps non-magnetic coating out of the working gap, and either a membrane vent or conformal coating for the cavity.

Batch variation, or a design with no margin Production

A unit that is inconsistent in production is usually not a batch problem, it is a margin problem made visible. Batch variation in the magnetic material of a few percent, in the spring rate, in the gap after assembly and in the friction of the guides all exist on any line. If the design has enough margin, those variations stay invisible; if it has none, they become a yield problem that gets blamed on material. The diagnostic question is whether the failures cluster in a batch or scatter across them.

Confirm it. Measure the distribution rather than the average: force at the worst position of the stroke, over at least thirty units from several batches. Then measure friction separately, because friction scatter between plungers commonly exceeds the magnetic scatter between them and it is the parameter that decides repeatability.

What to change. Fix the margin first, because a design that is marginal will keep producing the same complaint with different excuses. Then tighten whichever of gap, spring rate or friction dominates the spread, in that order of likelihood.

How to use this without guessing

Two rules make the difference between using this tree and replacing parts until something changes. First, take the measurement before forming the theory: coil resistance, voltage at the coil terminals under load, the real gap with the unit clamped as installed, and the force needed to move the actual load. Second, when a branch does not fit, go back to the top rather than forcing an answer, because a fault with two causes will satisfy neither branch cleanly.

If you have taken those measurements and the tree still does not resolve it, the measurements themselves are the useful thing to send. What a useful fault report contains, and how to get in touch, is on the about page.