Molten salt level measurement is one of the few level applications where the wetted-part metallurgy matters more than the sensor principle. At 565 °C in a concentrated solar power (CSP) cold tank — that is the cold side — solar nitrate salt is corrosive to ordinary stainless steel, hostile to PTFE-bearing radar antennas, and hot enough to melt a polymer-bonded capacitive probe. Pick the wrong probe alloy and the level signal vanishes within weeks; pick a fluoride-salt probe for a nitrate-salt service and the corrosion mechanism is reversed and equally lethal. This guide names the three salt families an engineer actually meets, the failure modes that matter at temperature, and the three transmitter technologies that survive — guided-wave radar with HT probe, air-cooled non-contact pulse radar, and remote-seal differential pressure with capillary fill.

Contents

• Three Salt Families Engineers Actually Encounter
• What Fails at 565 °C: Probe, Seal, Process Connection
• Guided-Wave Radar with HT Probe — First Choice for Nitrate
• Air-Cooled Pulse Radar — Backup for Splashing Service
• DP with Capillary Seal — When Radar Will Not Survive
• Standards, Certifications, and Acceptance Tests
• Featured Molten-Salt Level Transmitters
• FAQ

Three Salt Families Engineers Actually Encounter

Three molten-salt families dominate industrial level service today. Solar salt — a 60 wt% NaNO3 / 40 wt% KNO3 eutectic — is the workhorse of the CSP tower and trough plants commissioned since 2010, operating between 290 °C in the cold tank and 565 °C in the hot tank. Chloride salts (typically MgCl2-KCl-NaCl ternary at 500–800 °C) are the next-generation CSP and Generation IV reactor candidate, more thermodynamically efficient but also more aggressive on Ni-Cr alloys. Fluoride salts — FLiBe (LiF-BeF2) at 460–700 °C — are exotic but real in molten-salt reactor (MSR) test loops.

Salt family	Composition	Operating range	Density	Dielectric constant εr	Compatible probe alloy
Solar salt (nitrate)	60% NaNO3 / 40% KNO3	290–565 °C	1830 kg/m³	~22 (estimated)	Inconel 600, Alloy 800H, 347H SS
Hitec / HitecXL	NaNO3-KNO3-NaNO2 ternary	140–500 °C	~1900 kg/m³	~25	347H SS, 321H SS
Chloride MgCl2-KCl	MgCl2-KCl-NaCl ternary	430–800 °C	1660 kg/m³	~6	Hastelloy C-276, Alloy 617, Inconel 625
FLiBe (fluoride)	LiF-BeF2 (66/34 mol%)	460–700 °C	1940 kg/m³	~9	Hastelloy N (UNS N10003) only

The dielectric constant matters because it sets guided-wave radar sensitivity. Nitrate salts at εr ≈ 22 give a strong reflection — comparable to water (εr = 80 at 20 °C, falling at high temperature). Chloride and fluoride salts at εr 6–9 give weaker reflections and demand a high-end transmitter with end-of-probe tracking enabled, otherwise the signal is lost in the noise. For a primer on how dielectric constant governs radar response, our dielectric-constant influence on level measurement guide covers the underlying physics. A related extreme-temperature deployment is the 80 GHz radar on ironworks blast furnace service, where the same flange-cooling architecture applies.

What Fails at 565 °C: Probe, Seal, Process Connection

Three components break at solar-salt cold-tank temperature, in this order: the polymer process seal, the probe centring spider, and the antenna PTFE window. The transmitter electronics, mounted on the cold side of the process connection, never see the salt and are usually fine.

1. Process-connection seal. Standard FKM (Viton) and even FFKM (Kalrez) elastomer seals fail above 327 °C. A molten-salt process connection must use metal-to-metal sealing — graphite-foil with a stainless-steel jacket is the field standard, with a Belleville-stack live load to maintain seating force across thermal cycles.
2. Probe centring spider. GWR coaxial probes use a centring disc to hold the inner conductor concentric in the outer tube. PTFE discs creep and lose alignment above 250 °C; the inner conductor sags, and the radar reflection scrambles. Specify a ceramic (alumina) centring disc, or switch to a single-rod probe with no centring required.
3. Antenna window (non-contact radar). Pulse-radar antennas with PTFE-clad horn windows are common at lower temperature; PTFE softens and salt vapor accelerates the failure. At 565 °C use a metallic-only horn (no PTFE) with a flange-mount air purge or a sapphire window if vapor traffic is moderate.
4. Process connection thermal-bridge length. The transmitter electronics are rated to about 80 °C ambient. The standoff between the salt-tank flange and the transmitter head must be long enough that conduction down the wetted-part wall does not roast the head. For 565 °C service, 250–400 mm cooling neck is typical; verify the manufacturer’s rating curve.

Guided-Wave Radar with HT Probe — First Choice for Nitrate

Guided-wave radar with a high-temperature probe is the default level technology for solar nitrate salt. The reasons are practical: nitrate’s εr ≈ 22 gives a clean top-of-liquid reflection; the alloy choice (Inconel 600 or Alloy 800H) is widely available; and a single GWR transmitter delivers ±5 mm continuous level over a 0.3–25 m measuring range with no recalibration through the daily charge–discharge cycle.

• Probe geometry. Single-rod for tanks with smooth walls and no internal obstructions; coaxial for narrow standpipes where surrounding signals would interfere. Single-rod is more forgiving on solar-salt service because the inter-tube gap of a coaxial design tends to crystallise during cool-down events.
• Probe alloy. Inconel 600 for nitrate up to 565 °C is the field standard. For Hitec ternary salts at 500 °C, 347H stainless suffices and is cheaper. Avoid 304/316 grades — sensitised in service and cracking has been documented on field returns.
• Probe-end tracking. Critical at start-up when the tank is empty and the only echo is the probe-end reflection. Disable probe-end tracking only after commissioning confirms the empty-tank fingerprint is captured. Our guided-wave radar calibration procedure covers the four-step empty/full/DK/threshold sequence.
• Cooling neck. 300 mm minimum for 565 °C service. The longer-neck variant (HT-extended) handles continuous service at the upper temperature limit; the standard neck handles 400 °C continuous and 565 °C with 50% duty.

Air-Cooled Pulse Radar — Backup for Splashing Service

Non-contact pulse radar (typically 26 GHz or 80 GHz) earns its place when GWR is mechanically impractical — for example, on a hot-tank dump line where the salt level surges during a discharge event and a probe would be subject to flow-induced vibration. The air-cooled metallic horn antenna eliminates the PTFE window failure mode; a continuous instrument-air purge keeps salt vapor and crystallised particulates off the antenna.

Antenna design	Max temperature	Beam angle	Best for	Risk
Metallic horn, air-purged	600 °C	10–15°	Open-tank dump/recovery	Loss of purge = blocked horn
Metallic horn, sapphire window	800 °C	8–12°	Closed tanks, low vapor traffic	Sapphire fracture under thermal shock
Air-cooled drop antenna	500 °C	4–6°	Narrow standpipes, side mount	Side-wall echoes, requires false-echo mapping
80 GHz parabolic, air-cooled	450 °C	3–5°	Long-range silos/towers	Cost, parabola fouling

The frequency × aperture trade-off matters here too — a 26 GHz horn gives a 12° beam at the 6-inch standard aperture; an 80 GHz parabolic compresses to 4°. For a deep, narrow molten-salt charging silo, the narrow 80 GHz beam clears the inner agitator without false echoes; for the wide hot tank with no internals, 26 GHz is cheaper and equally accurate. The trade-offs are similar to those discussed in our radar antenna types selection guide.

Differential Pressure with Capillary Seal — When Radar Will Not Survive

Some molten-salt installations rule out radar entirely — vacuum-blanketed receiver tanks where probe penetrations cannot be tolerated; closed pressurised loops where vapor turbulence destroys non-contact echoes; chloride salts at 700+ °C where GWR probe alloys fall outside reliable service. The fallback is differential pressure with remote diaphragm seals and a capillary fill suitable for the temperature.

• Diaphragm material. Hastelloy C-276 for chloride salts; Inconel 600 for nitrate; tantalum for the most corrosive fluoride-salt service if the budget allows.
• Capillary fill fluid. Standard silicone oil (DC-704) is rated to 315 °C and fails on prolonged 400+ °C service. Use Syltherm 800 (Dow) or KN-86 (Solutia) for 400 °C continuous; for 500 °C+ continuous, NaK alloy or Galinstan liquid-metal fill is the only option, with a sealed bellows isolator.
• Capillary length. Long capillaries respond slowly and amplify ambient temperature error. Keep capillaries under 6 m where possible; insulate and trace-heat them in cold-climate installations to prevent fill-fluid solidification at the cold side.
• Span calibration. Recalibrate at process temperature, not in the workshop, because the fill-fluid density changes with temperature and the tank pressure at the diaphragm shifts accordingly. Document the calibration temperature on the tag.

Standards, Certifications, and Acceptance Tests

Two standards are routinely cited in molten-salt level specifications. ASME BPVC Section II Part D defines allowable stress for the wetted-part alloys at temperature — the engineer must verify Inconel 600 at 565 °C is rated for the design pressure of the tank flange; the standard tables show stress allowables falling sharply above 540 °C. NACE MR0175 / ISO 15156 governs sour-service material selection where the salt loop has trace H2S or sulfide contamination; this matters in next-generation chloride loops where corrosion products can include sulfide species.

Acceptance tests for a new molten-salt level transmitter typically include: (1) pre-installation calibration on a cold dummy tank with the actual probe, (2) a hot-side commissioning test recording empty / quarter / half / three-quarter / full readings across one full thermal cycle, (3) probe inspection after 1000 hours service for crystallised salt deposit and alloy attack at the centring discs.

Featured Molten-Salt Level Transmitters

FAQ

What is the maximum temperature for guided-wave radar on molten salt?

Continuous service to 565 °C is achievable with an Inconel 600 single-rod probe, ceramic centring discs, and a 300 mm cooling neck. Above 600 °C the probe alloy creep rate accelerates and frequent recalibration is required. For chloride-salt service above 700 °C, switch to non-contact pulse radar with a metallic horn or to a remote-seal DP transmitter.

Can capacitive level sensors handle molten salt?

No, not reliably above 250 °C. Polymer-bonded probe insulators (typical PEEK or PTFE) creep and lose dielectric integrity at solar-salt temperatures. Some ceramic-insulated capacitive probes are rated to 400 °C, but their accuracy drifts as crystallised salt deposits change the effective dielectric path.

Why is solar salt usually 60% NaNO3 / 40% KNO3 specifically?

That eutectic composition gives the lowest melting point (around 220 °C) of the binary nitrate system, maximising the operating temperature window between freeze and decomposition (~600 °C). It is also low-cost and abundant. Hitec ternary salt (NaNO3-KNO3-NaNO2) extends the lower limit to 140 °C but trades against thermal stability above 500 °C.

What dielectric constant does molten nitrate salt have?

Solar salt at 290–565 °C shows a dielectric constant in the range εr ≈ 18–25, close to that of water at room temperature. This is high enough that guided-wave radar gives a clean top-of-liquid reflection without needing the high-end transmitter modes. Chloride salts at εr ≈ 6 require probe-end tracking and a more sensitive transmitter.

How do I prevent salt freezing in the probe assembly?

Three measures combine. First, electric trace-heating on the cooling neck and any horizontal pipework, set to at least 50 °C above the salt freezing point. Second, insulation on all wetted-part standoff sections, with the trace heater under the insulation. Third, a documented start-up procedure that pre-heats the probe assembly before the salt is melted into the tank — bringing the salt up around a cold probe is the most common cause of cracked probes and lost commissioning weeks.

Is bubbler level measurement still used on molten salt?

Rarely, and only as a backup. The bubbler injects an inert purge gas (nitrogen or argon) at the bottom of the tank and measures the back-pressure. The hot purge gas accelerates salt freezing in the dip-tube tip, and the pressure transmitter at the cold side requires a long capillary that is itself failure-prone. Modern installations use GWR as primary and a DP capillary system as backup; bubblers persist only on legacy installations and on metallurgical melt furnaces with much higher temperature than CSP.

What is the typical accuracy specification for a CSP molten-salt level transmitter?

Project specifications normally call for ±10 mm or 0.1% of measured level, whichever is greater, on a 12 m tall hot tank. Modern HT-GWR transmitters meet ±5 mm; pulse radar meets ±10 mm; DP with long capillaries meets ±15 mm at best because of the capillary thermal-error budget. For inventory accountability the spec tightens to ±0.05% of full scale, typically only met by GWR plus an independent verification reading via load-cell weighing.

For a salt-specific selection — chemistry, temperature range, tank geometry, and standards alignment — our application engineers will scope the probe alloy, antenna or capillary configuration, and acceptance-test plan within one working day.

Request a Quote

Choosing a radar antenna is choosing a beam. Beam angle, near-field length, focusing gain, build-up tolerance, and chemical compatibility all flow from the antenna geometry — not from the radar electronics. This guide compares the five antenna types used on industrial radar level meters (horn, planar/lens, parabolic, rod, and PTFE-clad), then maps each to the tank, dielectric, and process condition it actually fits. See blast furnace radar antenna selection for the lens-antenna + purge-air configuration used on 1200 °C ironworks service.

Contents

• The One-Sentence Antenna Decision
• Five Antenna Types: Beam, Gain, Cost — Side by Side
• Horn Antenna: the General-Purpose Default
• Planar / Lens Antenna for 80–120 GHz Narrow-Beam Tanks
• Parabolic Antenna for Long-Range Open-Air Service
• Rod Antenna for Small Vessels and Branch-Off Nozzles
• PTFE-Clad Antenna for Acid, Caustic, and Sticky Service
• Frequency × Antenna: Why 80 GHz Changed the Decision
• Featured Radar Level Meters by Antenna Type
• FAQ

The One-Sentence Antenna Decision

The antenna controls the beam angle, the beam angle controls how much of the tank wall you have to live with in the radar return, and the wall returns are the single biggest cause of false echoes — so the antenna decision boils down to: get the beam narrow enough that no internal obstruction sits inside the cone. Everything else (build-up tolerance, chemical resistance, cost) is a constraint that limits which narrow-beam option is available.

Once that one rule is in place, the rest of this guide is a lookup: given the tank diameter, the obstruction map, the medium, and the temperature/pressure envelope, which antenna keeps the beam clean.

Five Antenna Types: Beam, Gain, Cost — Side by Side

The matrix below is the field-engineer’s cheat sheet. Beam angle assumes the most common operating frequency for that antenna; the figures shift if you pair an antenna with a non-standard frequency.

Antenna	Typical frequency	Beam angle (–3 dB)	Max range	Gain	Build-up tolerance	Process T / P	Best for
Horn (DN50–DN200)	26 GHz	8°–18°	30 m	20–28 dBi	Medium — flush bell collects light dust	250 °C / 40 bar	Most liquids and slurries; default choice
Planar / lens	80–120 GHz	3°–4°	30 m (liquid), 80 m (solid)	27–36 dBi	High — flat smooth face	150 °C / 16 bar	Narrow nozzles, internal obstructions, small tanks
Parabolic	5.8 / 10 GHz	4°–6°	70 m	30–35 dBi	Low — open dish collects build-up	80 °C / atm	Open-air silos, water reservoirs, mining stockpiles
Rod (PTFE/PP)	6 GHz	30°–40°	10 m	10–12 dBi	High — non-stick rod	120 °C / 6 bar	Small vessels, branch nozzles, acidic liquids
PTFE-clad horn	26 GHz	8°–14°	20 m	18–24 dBi	High — chemical-resistant lining	200 °C / 25 bar	HCl, H₂SO₄, NaOH, latex, syrups

Two numbers in the matrix carry most of the design weight: the beam angle and the build-up tolerance. The next five sections explain when each antenna deserves the spec line.

Horn Antenna: the General-Purpose Default

The horn antenna is the sensible default on a 26 GHz radar level meter. The conical bell focuses the beam to 8°–18°, and the steel construction handles process temperatures up to 250 °C and 40 bar without an exotic seal.

Geometry rules of thumb for a horn:

• Bell diameter sets the beam width. A DN50 horn at 26 GHz gives ≈18° half-power beam; DN100 gives ≈10°; DN200 gives ≈6°.
• Nozzle clearance. The nozzle bore must be at least the bell diameter, and the bell must extend 10 mm past the inside of the tank roof. A horn that sits inside the nozzle generates a doughnut of false echoes from the nozzle wall.
• Vertical alignment. Tilt of the horn axis must be <1° from vertical. A 5° tilt on a 25 m tank shifts the beam centroid by 2.2 m, which can drag the cone onto an agitator or ladder.

Horns lose to planar/lens antennas when the tank has crowded internals or the nozzle is undersized for an adequate bell — that’s where the next antenna type wins.

Planar / Lens Antenna for 80–120 GHz Narrow-Beam Tanks

An 80 GHz or 120 GHz planar/lens antenna delivers a 3°–4° beam from a face only 50–100 mm across. That combination — narrow beam, small footprint — solves three failure modes that defeat horn radars.

• Tanks under 1.5 m diameter. A horn’s 10° cone clips the wall in any vessel narrower than 1.5 m at 5 m measurement distance; the planar’s 3° cone clears that down to 0.4 m diameter.
• Tanks with internal obstructions. Agitator shafts, heating coils, anti-vortex baffles. A 4° beam threads through gaps that a 12° beam cannot.
• Build-up service. The flat lens surface stays cleaner than a horn bell; condensate runs off and dust does not collect in a bell-shaped pocket.

The trade-offs are temperature and pressure: PTFE/PEEK lens windows top out around 150 °C and 16 bar. Above that, you go back to a horn. For solids tanks and wide silos, a high-frequency lens has changed the economics — see our notes on silo radar level measurement.

Parabolic Antenna for Long-Range Open-Air Service

The parabolic dish is the long-range specialist. A 600 mm dish at 10 GHz produces a 4° beam with 35 dBi gain — enough echo budget to measure 70 m of stockpile at the edge of a quarry crusher.

Parabolic earns its spec line in three jobs:

• Open-air bulk solids. Cement, coal, aggregate stockpiles, fly ash. The high gain handles the diffuse return from a sloped solid surface.
• Water reservoirs and reservoirs. 20 m+ depth with no upper structure to anchor a horn.
• Mining bin level (top-of-pile). Where the dust load and the conical pile angle would defeat any other antenna.

The downside is the dish itself: it collects rain, dust, and bird debris, and it cannot survive a closed pressurized vessel. A parabolic outdoors needs a feed-horn weather cap and a quarterly visual inspection — not a fit-and-forget install.

Rod Antenna for Small Vessels and Branch-Off Nozzles

Rod antennas are the workhorse for the cheap end of the radar level meter market — small vessels, hot-water tanks, condensate receivers, branch-off nozzles where a 1.5 m horn cannot fit.

The geometry trade is clear: the radiating element is a rod of PTFE or PP, 100–250 mm long, that floods the nozzle with a 30°–40° wide beam. This makes rod antennas tolerant of a misaligned mount, but vulnerable to false echoes from anything inside the cone.

• Use a rod antenna when the vessel is <1 m diameter, the nozzle is DN25–DN50, and the medium is a low-corrosion liquid (water, oil, glycol).
• Avoid a rod antenna when the tank has internals, when measurement range exceeds 10 m, or when accuracy needs to beat ±10 mm — the wide beam smears the echo.

The rod’s PTFE construction also gives chemical resistance for free, which carries over to the next category.

PTFE-Clad Antenna for Acid, Caustic, and Sticky Service

PTFE-clad horns and rods are the answer when the medium attacks bare 316L. Hydrochloric acid, sulfuric acid, sodium hydroxide, ferric chloride, latex, syrups, polymer slurries — all eat 316 within months. PTFE shrugs them off.

Two configurations cover most acid/caustic tanks:

• PTFE-clad horn (extended antenna). The horn bell is overlaid with 2–4 mm of PTFE that extends past the flange face. Antenna gain drops by 2–4 dBi vs bare steel, but corrosion vanishes. Used on 26 GHz radars in chemical tanks up to 200 °C / 25 bar.
• PTFE rod (drip-tip). All-plastic radiating element with a teardrop shape that resists wetting. Self-cleaning under condensate and droplet flow. Used on storage tanks for HCl, NaOH, and corrosive condensate streams.

PTFE-clad antennas pair well with the chemical tank scenarios discussed in our sulfuric acid storage tank level radar guide, where the same antenna geometry argues against straight-stainless probes.

Frequency × Antenna: Why 80 GHz Changed the Decision

Antenna and frequency are coupled by the same physics — beam angle is roughly proportional to λ/D, where λ is the wavelength and D is the antenna aperture. Doubling the frequency halves the wavelength, so the same aperture gets you a beam half as wide. This is why 80 GHz radars rewrote the antenna selection table:

Aperture	Beam at 6 GHz	Beam at 26 GHz	Beam at 80 GHz
50 mm	~70°	~18°	~6°
100 mm	~35°	~10°	~3°
200 mm	~18°	~5°	~1.5°

The practical consequence: a 100 mm planar 80 GHz antenna delivers a tighter beam than a 200 mm 26 GHz horn — at a quarter of the nozzle footprint. That is why narrow-beam, small-footprint planar antennas have become the default for new tanks under 5 m diameter and for any retrofit where the existing nozzle is undersized.

For a deeper look at how operating frequency choices interact with antenna selection, see our radar level transmitter frequency range reference.

Featured Radar Level Meters by Antenna Type

One sample sensor for each of the three antennas that cover the most common application classes — horn (default liquids), parabolic (long-range solids), and planar 120 GHz (narrow-beam, small-footprint).

FAQ

What is the most common radar level meter antenna?

The 26 GHz horn antenna is still the installed-base default for liquid storage tanks, with beam angles from 8° to 18° depending on bell diameter. The 80–120 GHz planar/lens antenna is the fastest-growing category for new builds, especially small vessels and retrofits with undersized nozzles.

How do I choose between a horn and a planar antenna?

Pick a horn when the process is hot (above 150 °C) or pressurized (above 16 bar) and the nozzle is generous (≥DN80). Pick a planar/lens when the tank is small (<1.5 m diameter), the internals are crowded, the nozzle is undersized (≤DN50), or the medium tends to coat the antenna.

Why does a narrower beam matter?

Anything inside the beam cone returns a radar echo. A 12° beam at 10 m range covers a 2.1 m diameter footprint on the surface — easily wide enough to clip an agitator shaft or a heating coil. A 3° beam at the same range covers 0.5 m, which threads through most internals and removes the false-echo source at the geometry level.

Can a parabolic antenna be used in a closed tank?

No. Parabolic dishes are an open-air technology — they cannot be sealed inside a flange and they collect product on the dish surface in any closed vessel. For closed-tank long range, use an 80 GHz planar antenna with a high-gain lens, not a parabolic.

Does a PTFE-clad antenna lose accuracy?

The PTFE layer absorbs about 2–4 dB of antenna gain compared with bare steel, which costs 3–6 m of maximum range. Accuracy at typical operating distances (under 15 m) is unchanged. The chemical-resistance benefit easily outweighs the gain loss for any acid, caustic, or sticky-medium service.

What antenna should I use on a small condensate tank?

For a tank under 1 m diameter, use either a PTFE rod antenna at 6 GHz (cheap, condensate-tolerant, range under 10 m) or a planar 80 GHz with a 3° beam (more expensive but more accurate, immune to nozzle reflections). Avoid a horn: the bell will not clear the nozzle and false echoes will dominate.

If you have an existing radar-level installation that loses lock or shows a wandering reading, the antenna and the nozzle geometry are usually the first place to look — send the tank drawing and the radar model and our team will diagnose within one business day.

Request a Quote

Related: stilling well pipe sizing.

Cooling tower water-level control is four separate jobs in one basin: refill the evaporated water, dump the concentrated water (blowdown), guard the pump from running dry, and stop the basin from overflowing. One sensor cannot do all four reliably — the right pick depends on which job, the basin geometry, and how aggressive the water chemistry is. This guide maps the four control loops to the right sensor technology, then gives the install and wiring rules that keep cooling tower level sensors out of the maintenance log.

Contents

• The Four Water-Level Jobs in a Cooling Tower Basin
• Sensor Technology by Job: a Selection Matrix
• Float Switches for Makeup and Anti-Overflow
• DP / Hydrostatic Transmitters for Continuous Basin Level
• Ultrasonic Sensors When Scale and Spray Permit
• Capacitive and Conductivity Probes for Low-Low Alarm
• Water Chemistry: Why Cooling Tower Sensors Foul Faster
• Install Position, Baffles, and Wiring Rules
• Featured Cooling Tower Level Sensors
• FAQ

The Four Water-Level Jobs in a Cooling Tower Basin

A cooling tower basin level moves continuously, and four separate control jobs act on it. Treating them as one loop is the most common reason a sensor specification ends up over- or under-built.

1. Makeup water control. Replaces water lost to evaporation, drift, and blowdown. A typical 100 ton tower evaporates roughly 1.6 USgpm at full load (≈3% of recirculation per 10 °F approach). The makeup loop only needs an on/off command between two setpoints (refill start and refill stop), so a two-stage float or a level switch with hysteresis is enough.
2. Blowdown control. Dumps a fraction of the basin water to keep the cycles of concentration (CoC) below the corrosion limit — usually 4–6 cycles depending on the makeup hardness. Blowdown is triggered by conductivity, not by level, but the blowdown valve must be inhibited if level drops below the low setpoint.
3. Low-low (pump protection) cutout. Prevents the circulation pump from cavitating if the makeup line fails. This is a safety interlock and must be a separate sensor from the makeup level — sharing one sensor for control and trip is a single point of failure that ASHRAE 188 explicitly flags for Legionella risk.
4. Anti-overflow / high-high alarm. Stops makeup if a stuck float feeds water past the overflow weir. The overflow loss costs treatment chemicals, not just water.

Four jobs, three setpoints (low-low, refill on/off, high-high), and at least two independent sensors. The selection matrix below maps each job to the technologies that deliver the right combination of accuracy, cost, and resistance to scale.

Sensor Technology by Job: a Selection Matrix

Match the sensor to the loop, not the loop to whatever sensor is on the shelf. The following table is the short version we hand to plant engineers when they ask which technology to specify per job.

Job	Output type	Best technology	Why	Avoid
Makeup on/off	2 discrete contacts	Side-mounted reed float, multi-stage cable float	Cheapest reliable hysteresis; no analog signal needed	Continuous DP — over-spec for on/off
Low-low pump trip	1 normally-closed contact	Independent float switch or vibrating fork	Must be galvanic-independent of makeup loop	Sharing the makeup float — single point of failure
High-high overflow	1 normally-open contact	Vibrating fork, rod-end conductivity	Tolerates spray and foam better than reed	Top-mounted ultrasonic — spray washout
Continuous basin level (BMS)	4–20 mA	Submersible piezoresistive, side-mount DP, 80 GHz radar (open basin only)	Hydrostatic types ignore foam; radar tolerates fouling	Capacitive — coating drift in 6 weeks
Cold-weather sump (heated)	Discrete or 4–20 mA	Magnetostrictive in stilling well	Ice on the float surface does not affect reading	External ultrasonic on a frozen sidewall

Float Switches for Makeup and Anti-Overflow

For the on/off loops a float switch is still the right answer in 2026. The water-side electronics are sealed in the float body, the contact is magnetically actuated, and there are no electronics in the wet path that scale can corrode.

Two configurations cover almost every cooling tower:

• Side-mounted, single contact. Used as the high-high overflow guard. Mount on a 3/4″ or 1″ NPT half-coupling tapped through the basin sidewall, 25–50 mm above the overflow weir. Hysteresis is fixed by float geometry, typically 6–10 mm — fine for an alarm.
• Top-mounted, multi-stage cable float. Used as the makeup controller. Two switches on the same cable: one trips when basin drops below normal operating level (refill on), the second trips when basin recovers (refill off). The dead band is the cable distance between switches, set anywhere from 50 mm to 250 mm depending on tower size — bigger band means fewer makeup-valve cycles per hour.

Both must be installed in a stilling tube if the basin sees splash from the spray distribution headers. A 50 mm PVC tube with 6 mm holes drilled below low-low level is enough — see our stilling well design rules for the geometry.

DP / Hydrostatic Transmitters for Continuous Basin Level

For a continuous 4–20 mA reading to the BMS, a hydrostatic differential pressure (DP) transmitter is the workhorse on cooling towers. It does not care about foam, drift, or surface waves, and the sensing diaphragm sits below the worst of the scale-forming surface layer.

Two sub-types matter for cooling tower service:

• Submersible piezoresistive transmitter (cable-suspended). Range 0–1 m to 0–10 m water column, accuracy ±0.25% FS, 316L body with hydrophobic vent in the cable. Drop it into a stilling tube to keep the cable away from the recirculation suction. Cleanable in 5 minutes by lifting the cable.
• Side-mount diaphragm-seal DP transmitter. Flush flange or extended diaphragm seal mounted on the basin sidewall at the lowest point. Reads 0–500 mbar (≈0–5 m H₂O). Use this when the cooling tower is buried (no access to drop a cable from above) or when the makeup-water treatment chemistry attacks 316L over a 5-year horizon and a remote-seal Hastelloy diaphragm is justified.

Both deliver continuous level data for the building automation system without exposing electronics to the spray zone above the basin.

Ultrasonic Sensors When Scale and Spray Permit

Ultrasonic level sensors are tempting because they are non-contact, but cooling tower service has three failure modes that quietly degrade them.

• Spray and condensation on the transducer face. Even a thin water film on the piezo crystal attenuates the return echo by 6–10 dB. After 6 months on a humid tower, the loss adds up to lost lock and a frozen reading.
• Mineral scale ring on the basin wall. If the transducer is mounted flush with a wall return, the false-echo from the scale ring shows up as a steady reading at the scale height — often within 50 mm of true level, which is the worst possible failure: undetectable until calibration day.
• Wind-driven foam. Outdoor towers in summer build a foam layer that ultrasonics see as a soft reflection. The reading walks downward as foam builds.

Ultrasonic still works in two cooling tower roles: indoor closed-loop chillers with low spray carry-over, and open basin sumps where a stilling well shields the sensor from spray. Mount 300 mm minimum above maximum level, set the dead band to cover the spray-carry zone, and schedule a monthly wipe-down of the transducer face.

Capacitive and Conductivity Probes for Low-Low Alarm

Capacitive level switches and conductivity rod sensors handle the safety-interlock job — point detection of low-low for pump trip, or of high-high if a float is unreliable. They are simple, have no moving parts, and the rod can be cut to length on site.

Cooling tower water has 1000–4000 µS/cm conductivity, so both technologies see a strong wet/dry signal. The trade-off:

• Conductivity rod (2- or 3-electrode). Cheapest sensor that exists for water level. Drift comes from scale bridging the electrodes — typically 6–18 months between cleanings, depending on hardness. Use the 3-electrode version with a reference rod when the makeup is variable hardness.
• RF capacitive switch with guard driver. Tolerates a thin coating because the guard electrode rejects current through the build-up layer. Cleaning interval doubles vs straight conductivity, but the sensor cost is 4–6× higher. Use only on the low-low cutout, not the operating loop.

Avoid straight (non-RF) capacitance for cooling tower water. Untreated capacitive probes drift by 30–60 mm of indicated level after 8 weeks of carbonate scale build-up, which is enough to trip a pump that is actually running fine.

Water Chemistry: Why Cooling Tower Sensors Foul Faster

Cooling tower water is the harshest level-sensing environment in HVAC. Because the system concentrates dissolved solids by 4–6× as water evaporates, every minor problem in the makeup is amplified.

Issue	Typical level	Effect on sensor	Mitigation
Calcium carbonate (CaCO₃) scale	200–600 ppm as CaCO₃	Coats wetted parts; jams floats; biases capacitive	Acid wash quarterly; scale-resistant probe geometry
Biological growth (algae, slime)	Visible film < 4 weeks if biocide off	Coats floats and ultrasonic faces	Continuous biocide dose; weekly visual inspection
Suspended solids (dust, pollen, scale fragments)	10–80 ppm	Plugs stilling tube vent holes	2 mm side-vent slots, no bottom plug
Dissolved iron (corrosion)	0.3–2 ppm	Stains 304 SS; pits if combined with chloride	Specify 316L minimum; Hastelloy in seacoast service
Chloride (sea air, brine carryover)	50–500 ppm	Stress-corrosion of austenitic SS at >200 ppm	Use 316L with cathodic protection or PTFE-clad probes

The practical takeaway: any sensor going into a cooling tower needs a defined cleaning interval. Build a quarterly wipe-down into the operations checklist before the install date, not after the first false trip.

Install Position, Baffles, and Wiring Rules

Sensor placement on a cooling tower is constrained by spray, cold-weather freezing, and the geometry of the basin. The rules below come from field installations and apply to most square or rectangular cooling tower basins.

• Distance from spray distribution. Mount sensors at least 600 mm horizontally from the spray pan edge. Closer than that, the falling-water drag pulls floats off the wall and creates negative-pressure zones that bias DP transmitters.
• Distance from the recirculation suction. Keep sensors 1.5× suction-pipe-diameter away from the suction. Otherwise the local drawdown reads as a phantom low-low, especially during pump start.
• Stilling tube for cable-mounted sensors. 50–100 mm Schedule 40 PVC, slotted on the lower 300 mm with 6 mm vent holes, top open. The tube damps surface waves to ±2 mm and shields submersibles from cross-flow.
• Freeze-protection for outdoor towers. Heat-trace the sensor cable above 0 °C from the basin to the conduit elbow. The basin water itself stays warm during operation but freezes within 60 minutes of pump shutdown in –10 °C ambient.
• Wiring class. Discrete float and capacitive switches: 24 V DC, normally closed for fail-safe (open contact = trip). Continuous transmitters: 2-wire 4–20 mA loop powered, twisted shielded pair, shield grounded at the BMS panel only.
• Lightning and surge. Outdoor cooling towers attract surges. Add a Type 2 surge protector on the 4–20 mA cable at the BMS-panel entry. A blown sensor with a clear sky overnight is almost always a surge from a distant storm.

For piping and instrumentation drawings, the cooling tower level loop should show two physically independent sensors driving the makeup valve and the pump-trip relay, and a third sensor or float to drive the high-high overflow alarm — see also our notes on water tank level sensor selection for the wider sensor family.

Featured Cooling Tower Level Sensors

The three sensors below cover the makeup, continuous, and non-contact roles for a typical 50–500 ton cooling tower.

FAQ

What is the most common cooling tower level sensor?

The cable-suspended multi-stage float switch is still the most common, because the makeup-water loop only needs on/off control between two setpoints. Hydrostatic DP transmitters dominate the continuous-reading role for sites with a BMS. The two technologies coexist on most modern towers.

Can one sensor do both makeup control and pump trip?

No, and ASHRAE 188 explicitly cautions against it. The makeup loop and the low-low pump-trip loop must be galvanic-independent so that a single fouled probe cannot disable both refill and protection at once. Use two physically separate sensors at different elevations.

Why does my ultrasonic level reading drift downward over time?

Three causes account for nearly all cooling tower ultrasonic drift: a water film on the transducer face from spray, a mineral-scale false echo at a fixed wall height, or a foam layer that returns a softer echo than open water. Wipe the face, then re-baseline. If drift returns within 30 days, switch to a hydrostatic DP transmitter or move the ultrasonic into a stilling tube.

What basin level should I use as the makeup setpoint?

Standard practice is to keep the operating level 50–100 mm below the overflow weir, with the refill-start setpoint 50–150 mm below the operating level and the refill-stop setpoint 25–50 mm below the operating level. Larger dead band reduces makeup-valve cycling, which extends solenoid life.

How often should cooling tower level sensors be cleaned?

Quarterly is the minimum for typical municipal-water towers (200–400 ppm hardness). Float switches and DP diaphragms can stretch to 6 months if biocide is well-controlled. Conductivity rod electrodes need monthly inspection in hard-water service. Build the interval into the cooling tower preventive-maintenance schedule from day one.

Is a guided-wave radar suitable for a cooling tower?

Guided-wave radar is overkill for an open atmospheric basin and the probe accumulates scale where the spray hits it. Reserve guided-wave radar for closed-loop chiller tanks or condensate receivers. For open cooling towers the cost-effective continuous solution is hydrostatic DP or a sealed submersible piezoresistive transmitter.

Need help specifying the right cooling tower level sensor for your basin geometry, makeup chemistry, and BMS interface? Send the basin sketch, makeup conductivity, and required outputs and our team will reply with a wired drawing within one business day.

Request a Quote

Guided wave radar (GWR) calibration is a four-step procedure: tank-empty zero, full-distance reference, dielectric (DK) entry, and threshold tuning. Done in that order, a probe-tip locked GWR transmitter delivers ±3 mm accuracy on liquids down to ε_r = 1.4 and ±0.5% of span on solids. Skip a step and you end up chasing the same false-echo for a week. This page walks through each step with the exact menu paths used on most GWR brands, the dielectric values that decide which threshold strategy to pick, and a fault-symptom-cause-fix table for the calls we get most often.

Contents

• Pre-Calibration Checks Before You Touch the HMI
• Step 1: Empty Calibration and Probe-End Reference
• Step 2: Full-Span Calibration With a Wet Reference
• Step 3: Dielectric (DK) Entry and the Low-DK Trap
• Step 4: Threshold Tuning and False-Echo Suppression
• GWR Fault-Symptom-Cause-Fix Table
• Two Field Cases: Asphalt Tank and Sulfuric Acid Storage
• Featured Guided Wave Radar Products
• FAQ

Pre-Calibration Checks Before You Touch the HMI

Three mechanical checks come before any menu navigation. Skipping them turns a 30-minute calibration into a 30-hour debugging exercise.

1. Probe length entered correctly. Open the head, measure the probe from the process flange face to the probe tip with a tape, and confirm the value matches the parameter “Probe length” in the transmitter. A 50 mm error here pulls the entire level reading by 50 mm even before any electronic tuning.
2. Reference distance to the flange. Most brands store an “Upper blocking distance” or “Reference offset” between the antenna face and the process connection. The factory default is right for the supplied flange, but if the user has installed an extension nipple, the offset changes. Measure and update.
3. Probe geometry verified. Coaxial probes need the inner rod centred and not touching the outer tube anywhere along the length; twin-rod probes need a 14–18 mm rod separation; single-rod probes need at least 300 mm clearance from the tank wall. A bent probe or a probe touching a heating coil will pin the level reading at the contact point.

Once those three items are confirmed, write down the live process: liquid, vapor temperature, vapor pressure, and the level on a sight glass or tape-down. The calibration steps below depend on knowing those four numbers.

Step 1: Empty Calibration and Probe-End Reference

Empty calibration teaches the transmitter where the probe tip is. With the tank drained below the probe end, navigate to Setup → Empty calibration → Apply. The transmitter sweeps the probe, picks up the end-of-probe (EOP) reflection, and stores its time-of-flight as the zero reference.

Two situations require a different approach. If the probe cannot be uncovered (deep tank, no drain), use the “Tank linearisation” or “Manual empty” command instead, which lets you enter the empty distance manually from the flange-face. And if the EOP is missing because the probe sits in a vapor with high ε_r (saturated steam, ammonia at 50 bar), run the calibration after a vent-and-purge cycle that brings the vapor ε_r below 1.05.

A successful empty calibration logs an “EOP signal strength” between 60 and 95 mV typically. If the value is below 30 mV the probe is dirty, mechanically deformed, or the EOP reflection is being absorbed by build-up; clean the probe and retry.

Step 2: Full-Span Calibration With a Wet Reference

Full calibration teaches the transmitter how the level reflection scales with the immersed probe length. Fill the tank to a known level, ideally between 70% and 90% of working span, take a tape-down reading, and enter that value at Setup → Full calibration → Enter measured value.

The transmitter computes the gain factor between the level reflection’s time-of-flight and the manual reference. After commit, the displayed level should match the tape-down within ±3 mm on coaxial probes and ±5 mm on twin-rod / single-rod probes. If it does not, the cause is almost always one of: a wrong DK value (next step), a stilling-well coupling that delays the wave, or a probe tip touching the bottom and pulling EOP into the level signal.

Step 3: Dielectric (DK) Entry and the Low-DK Trap

The dielectric constant of the product determines how much of the radar pulse reflects from the surface. High-DK products (water at 80, acids at 30+) reflect 30–60% of the pulse and the calibration is forgiving. Low-DK products (LPG at 2.0, hexane at 1.9, light oils around 2.2) reflect only 1–3%, and the threshold settings have to compensate.

Product	εr	Pulse reflection	Threshold strategy
Water (clean)	80	~60%	Default; no probe-end trick needed
30% sulphuric acid	32	~45%	Default
Methanol	33	~45%	Default
Diesel / fuel oil	2.1	~3%	Probe-end tracking ON; reduce noise threshold
LPG / propane	1.6–1.9	~1.5%	Coaxial probe required; probe-end tracking ON
Hexane / light naphtha	1.9	~2%	Coaxial probe; lower noise threshold; verify EOP signal >120 mV

For low-DK products, enable “Probe-end tracking” or “End-of-probe-mode” in the menu. The transmitter then computes the level from the apparent time delay of the EOP reflection — when liquid covers part of the probe, the dielectric slows the pulse over that length, and the EOP arrives later than its dry-tank reference. This trick works down to ε_r = 1.4 even when the surface reflection itself is too weak to lock.

Enter the actual product DK to two significant figures. A wrong DK by 0.5 produces about 5% level error in probe-end tracking mode and about 1% error in surface tracking mode.

Step 4: Threshold Tuning and False-Echo Suppression

Once the level reading agrees with the tape-down, the calibration is mathematically right. The remaining failure mode is false-echo capture — the transmitter locking onto a tank fitting, agitator blade, or weld seam instead of the product surface.

Open the echo curve display from the HMI or HART handheld. The curve shows signal amplitude vs distance from the antenna. The expected pattern: a strong reference pulse near zero, the level reflection at the surface position, and the EOP reflection at the probe tip. False reflections show up as bumps between the reference and the level peak. Two corrective steps handle them:

1. Run “Tank mapping” or “False-echo suppression.” The transmitter samples the empty-tank echo curve and stores it. During normal operation, the stored curve is subtracted from the live curve so that the fixed bumps disappear and only the moving level peak remains.
2. Set a “Search range” or “Window.” If the product surface is always between 0.5 m and 4 m from the flange, restrict the search range to those bounds. The transmitter then ignores echoes outside that range, including any internal weld at 4.5 m or any EOP reflection at 5 m.

One more knob is worth knowing: “Threshold curve” or “Detection threshold.” This is the minimum amplitude the transmitter accepts as a real echo. The default is set conservatively for high-DK service. On a low-DK product, lower the threshold by 30–50% in the EOP region so the weaker level peak is not rejected. On a foamy product, raise it by 20% near the antenna so foam is treated as noise rather than level.

GWR Fault-Symptom-Cause-Fix Table

Symptom	Most likely cause	Fix
Level pinned at probe end (100% or 0%)	EOP reflection captured as level peak; product DK below threshold	Enable probe-end tracking; lower detection threshold near EOP
Reading jumps 100–500 mm during agitator motion	Level peak loss in turbulence; transmitter walks to a false-echo on the agitator shaft	Run tank mapping with agitator running; restrict search range
Level reads 5–15% high after a fast fill	Foam layer above true liquid; foam dielectric is detected as level	Add a stilling well or a coaxial probe; enable foam-mode (raises threshold near antenna)
Reading drifts 10–30 mm/day with no process change	Build-up at the probe top growing over time; pulse reflects from the deposit instead of the antenna face	Schedule a probe wipe-down or upgrade to a probe with hydrophobic coating
Loss of signal alarm during steam blow-down	High-DK vapor (saturated steam at 8 bar has εr ~ 1.10) absorbs the pulse	Switch to coaxial probe; pause the level controller during blow-down
Reading correct at low level, wrong above 70% span	Antenna or upper-rod build-up shifting the reference reflection	Recalibrate empty after a manual cleaning cycle

Two Field Cases: Asphalt Tank and Sulfuric Acid Storage

Two cases from recent commissioning visits illustrate the procedure.

Asphalt tank, 12 m, 180 °C, single-rod probe. The site complained of a 1 m offset between the GWR reading and the manual hatch dip. Pre-calibration check: probe length entered as 12.0 m, actual 12.05 m. Probe-end tracking was off; product DK was set to default 5.0 (asphalt at 180 °C is closer to 2.6). After updating the probe length, entering DK = 2.6, enabling probe-end tracking, and running tank mapping with the heating coils visible in the curve, the reading converged to within 12 mm of the dip in 5 minutes.

30% sulphuric acid storage, 6 m, ambient, coaxial probe. The site reported a slow upward drift — 4 mm per day — over six weeks. The acid was being slowly diluted by condensate dripping into the tank, so the actual DK had drifted from 32 to 38. The fix was to set up a quarterly DK update procedure rather than re-tuning the threshold. After the first quarterly update, drift dropped to under 1 mm per day. For dedicated radar guidance on sulphuric acid storage, see our sulfuric acid tank level radar guide.

Featured Guided Wave Radar Products

For wider radar context, see the storage-tank guided wave radar overview and the radar antenna selection guide. Build-up rejection comparisons against capacitance probes are covered in our RF admittance level sensor explainer.

FAQ

Do I need to drain the tank to calibrate a GWR?

Only if you want a true empty calibration. Most brands accept a manual empty calibration where you enter the actual probe-tip distance from the flange. Combined with a wet full-cal at 70–90% level, the manual approach reaches the same accuracy as a tank-empty cal on liquids with ε_r > 3.

What is the lowest dielectric constant a GWR can measure?

With probe-end tracking enabled and a coaxial probe, GWR works reliably down to ε_r = 1.4 (LPG, butane, light hydrocarbons). Below 1.4 the reflection is too weak even in tracking mode and a different technology, normally float or magnetostrictive, is preferred.

Can a GWR be calibrated through the HART handheld instead of the local HMI?

Yes, all major brands expose empty cal, full cal, DK entry, and false-echo suppression as HART commands. The local HMI is convenient because it shows the live echo curve in graphical form; the HART handheld shows the same data as numerical samples and works on any GWR with a 4–20 mA loop.

How often should a GWR be recalibrated?

Annually for most clean liquid services. Quarterly for products with a documented DK drift (sulphuric acid concentration changes, sugar syrup density changes). After any cleaning, probe replacement, or tank internal modification, redo at minimum the empty cal and the false-echo suppression.

Why does my GWR reading drop to zero during a steam blow-down?

Saturated steam at high pressure has an ε_r around 1.10, which absorbs the pulse before it reaches the surface. The pulse cannot return, the transmitter sees no echo, and it falls to its loss-of-echo state. Use a coaxial probe (which traps the pulse inside the rod-tube annulus and survives high-DK vapor) and pause the level controller during blow-down.

Is GWR the same as guided microwave or TDR level?

Yes — “GWR,” “guided microwave,” “guided wave radar,” and “TDR (time domain reflectometry) level” all describe the same technology. The pulse travels along a probe instead of through free space. Vendor brand names differ but the calibration steps in this guide apply to all of them.

Can I install a GWR vertically through the side of a tank?

Top-mounted vertical install is standard. Side-mount is only practical for chamber installation (a vented bypass chamber outside the tank), not for direct insertion into the main vessel. Side-mounted probes pick up sloshing waves and weld-seam reflections that destroy the calibration.

Request a Quote

A capacitive fuel level sensor measures diesel, petrol, kerosene, or jet fuel by treating the tank as a parallel-plate capacitor and reading the change in capacitance as fuel rises and falls. It is non-mechanical, has no moving parts inside the probe, and resolves level to about ±1 mm in clean fuel — which is why it has displaced reed-switch float chains in most over-the-road tankers and standby gensets built since 2015.

This guide explains how the capacitance reading is converted to a fuel volume, what the probe actually looks like inside, where capacitive technology beats float / ultrasonic / magnetostrictive alternatives, and where it does not. Worked numbers use Pa, mm, and pF so you can repeat the math against your own datasheet.

Contents

• How a Capacitive Fuel Level Sensor Works
• Inside the Probe: Coaxial Electrode Design
• Dielectric Constant and Why Diesel ≠ Petrol
• Capacitive vs Float, Ultrasonic, and Magnetostrictive
• Where Capacitive Wins — and Where It Doesn’t
• Calibration and Field Diagnostics
• Where We See It Installed
• Featured Capacitive Level Products
• FAQ

How a Capacitive Fuel Level Sensor Works

A capacitive fuel level sensor uses a vertical metal rod (or rod-and-tube) inserted into the tank as one capacitor plate, and the tank wall — or a concentric outer tube — as the second plate. Fuel acts as the dielectric between them. As fuel level rises, the dielectric in the gap changes from air (relative permittivity ε_r≈1.0) to fuel (ε_r≈2.0–2.2 for diesel, 1.9–2.1 for petrol). Capacitance rises in direct proportion to the wetted length.

For a coaxial probe the relationship is C = 2π·ε₀·ε_r·L / ln(b/a), where a and b are the inner and outer electrode radii, L is the wetted length in metres, and ε₀ is the vacuum permittivity (8.854×10⁻¹² F/m). A 600 mm probe with a=2 mm, b=8 mm reads roughly 60 pF empty and 120 pF full of diesel — a 60 pF span the transmitter electronics resolve to 12-bit (≈0.015 pF), which is why ±1 mm resolution over 600 mm is realistic.

The transmitter applies a 1–10 kHz AC excitation across the electrodes and measures the resulting current. AC is mandatory: a DC reading would be swamped by the leakage resistance of fuel additives and water contamination, and would also polarise the electrodes over time. Modern transmitters use synchronous demodulation to reject 50/60 Hz pickup from nearby alternators and switching gear.

Inside the Probe: Coaxial Electrode Design

Two electrode geometries dominate fuel applications. Coaxial probes use an inner rod inside an outer perforated tube — the outer tube fixes the gap, so the reading is independent of tank wall material and the probe can be installed in plastic, fibreglass, or rubber-lined steel tanks. Rod-to-wall probes use the metal tank itself as the second plate; they are cheaper and shorter but only work in conductive metal tanks and are sensitive to off-axis mounting.

The inner rod is usually 316L stainless steel for diesel and biodiesel-blend service. PTFE or PFA insulation on the outer surface of the rod is non-negotiable — it stops the probe from short-circuiting if conductive water settles to the bottom of the tank. The PTFE jacket is also what gives capacitive probes their tolerance for ethanol-blend gasoline; bare metal probes corrode in E10/E15 fuel within months.

Probe lengths up to 4 m are standard. Above 2 m we recommend a centring spider every 600 mm to keep the rod parallel to the outer tube — vibration during transport will otherwise let the rod swing 1–2 mm off-axis, which moves the apparent capacitance enough to produce a 30–50 mm level error.

Dielectric Constant and Why Diesel ≠ Petrol

Different fuels have different dielectric constants, so the same probe gives a different full-scale capacitance for each. A transmitter calibrated for diesel and then filled with petrol will read about 7 % low. The values below are typical at 20 °C — they drift roughly −0.001 ε_r/°C with rising temperature.

Fuel	Dielectric constant εr	Notes that affect calibration
Diesel (No.2 / EN 590)	2.0 – 2.2	Reference fuel for most factory calibrations
Biodiesel B100	3.1 – 3.2	Reads ~50 % high if not recalibrated
Petrol / gasoline (RON 95)	1.9 – 2.1	Reads ~5–7 % low on a diesel calibration
E10 (10 % ethanol)	4.5 – 5.5	Recalibration mandatory; ethanol is polar
Jet A-1 / kerosene	1.7 – 2.0	Lowest end of the range
Hydraulic oil	2.1 – 2.8	Varies with additive package

If the tank serves multiple fuels, look for a transmitter with a wet-cal procedure — fill the tank to a known level, push a button, and the firmware re-derives ε_r from the measured capacitance. Two-point wet calibration (empty and full) handles ε_r drift caused by additives, water emulsion, and seasonal temperature swings without manual datasheet lookup.

Capacitive vs Float, Ultrasonic, and Magnetostrictive

For diesel and petrol tanks under 4 m, four technologies dominate. The decision usually comes down to accuracy required, tank cleanliness, and whether the tank is on a vehicle that vibrates and slosh-loads the sensor.

Tech	Typical accuracy	Moving parts	Best for	Watch out for
Capacitive rod	±0.5–1 % FS	None	Vehicle / genset / IBC	Mixed-fuel tanks; conductive sludge
Reed-switch float	±5–10 mm steps	Magnetic float	Cheap aftermarket; small tanks	Float sticks on biofilm; resolution = magnet pitch
Ultrasonic (top)	±0.25 % FS	None	Above-ground bulk diesel	Foam / vapour / dome echoes
Magnetostrictive	±0.01 % FS	External float ring	Custody-transfer fuel terminal	Cost: 4–8× capacitive

For most fleet-fuel and standby-power applications, capacitive hits the sweet spot — better accuracy than a reed-switch float, immune to the tank-dome echoes that plague ultrasonic in mobile applications, and a fraction of the cost of a magnetostrictive setup. We cover the magnetostrictive trade-offs in our magnetostrictive tank level measurement guide.

Where Capacitive Wins — and Where It Doesn’t

Capacitive fuel sensors are the default for vehicle tanks, gensets, day tanks, and IBCs because they have no moving parts to seize, no top-of-tank vapour cone to degrade the signal, and they handle the −40 °C to +85 °C ambient range that reed-switch floats cannot.

• Wins: no moving parts, ATEX-able for petrol vapour, immune to slosh and roll on vehicles, ±1 mm resolution, life expectancy 10+ years.
• Loses: requires recalibration per fuel type, sensitive to thick sludge or wax buildup on the rod, accuracy degrades if free water sits at the bottom, longer probes need careful centring.

For tanks where free water and biological sludge are unavoidable — old underground storage, marine bilge — we usually recommend a non-contact alternative. See how to check level in underground tanks for the comparison.

Calibration and Field Diagnostics

Two-point wet calibration is the reliable method. Drain the tank, record the empty-pF reading, fill to a known volume against a calibrated dipstick, record the full-pF reading. The transmitter linearises between the two endpoints. A two-point cal beats a single-point dry-cal by 3–5 % in real fuel because it captures the actual ε_r of the fuel batch.

Common field faults and how we diagnose them in 5 minutes:

1. Reading drifts upward over weeks. Sludge / biofilm on the rod increases the apparent ε_r. Pull the probe, wipe with a lint-free cloth.
2. Reading suddenly jumps to 100 %. Short between rod and outer tube. Check the PTFE jacket for cracks at the bottom.
3. Reading reads zero with fuel in the tank. Open circuit on the rod cable, or a failed AC excitation oscillator on the transmitter board.
4. Reading is 7 % off after a fuel change. ε_r mismatch. Re-run the wet calibration.
5. Reading bounces ±20 mm at idle. Vehicle vibration moving the rod inside the outer tube. Add a centring spider.

Where We See It Installed

In our field installations, capacitive fuel level sensors show up most often in five settings. Each has its own probe length, output format, and certification requirement, and the transmitter SKU usually changes between them.

• Truck and bus diesel tanks — 800–1500 mm coaxial probe, CAN J1939 output, ATEX zone 2 if any vapour space.
• Standby diesel genset day tanks — 600–1000 mm rod-to-wall probe, 4–20 mA into the genset controller, with a high-high alarm at 95 %.
• Construction equipment fuel tanks — short 400 mm probes, 0–5 V output for telematics displays.
• Marine fuel polishing systems — 1–2 m PTFE-jacketed probe, 4–20 mA, IP68 head.
• Petrol forecourt UST monitoring — 3–4 m probe, intrinsically safe barrier, RS-485 to the SIR / SIRA controller. For new-build USTs we now recommend pairing capacitive with an independent diesel fuel tank level gauge for redundancy.

If your application is closer to truck-mounted oil level sensing rather than diesel — for example hydraulic reservoirs or lubricant day tanks — the rod material and gland rating change, and the calibration shifts to ε_r≈2.5.

Featured Capacitive Level Products

FAQ

Can a capacitive fuel level sensor work in a plastic tank?

Yes — but only with a coaxial probe. The outer tube provides the second electrode and fixes the gap, so the tank wall material does not enter the measurement. A rod-to-wall probe needs a conductive metal tank to work.

Will it read correctly with water at the bottom of the tank?

Free water (ε_r≈80) on top of the wetted length will read as 100 % full because water has 40× the dielectric of diesel. A capacitive probe cannot tell water from fuel — pair it with a water-bottom sensor or drain the tank periodically.

How often does a capacitive fuel level sensor need recalibration?

For a single-fuel tank, once at commissioning and once after the first 500 hours of service. After that, recalibrate only when fuel grade changes or when the reading drifts more than 2 % against a manual dipstick check.

Is capacitive sensing safe for petrol vapour?

Yes, with the appropriate certification. The probe itself is passive; the transmitter electronics need to be ATEX/IECEx Ex ia or housed in an Ex d enclosure with an intrinsically safe barrier on the cable. Excitation voltages are kept below 5 V to stay under the ignition energy threshold.

What is the highest temperature a capacitive fuel sensor can handle?

Standard PTFE-jacketed probes are rated to +85 °C process. High-temperature versions with PFA jacket and a remote-mounted electronics head extend that to +200 °C, suitable for heated heavy-fuel-oil tanks on marine bunkers.

Capacitive vs ultrasonic for a diesel day tank — which one?

For a sealed indoor genset day tank under 1.5 m, capacitive is the safer pick — ultrasonic struggles with vapour cones and the small dome geometry creates spurious echoes. Above 2 m or in an open-vented bulk diesel tank, top-mounted ultrasonic catches up on accuracy.

Request a Quote

A transformer oil level indicator monitors mineral or ester insulating oil inside a power transformer’s main tank, conservator, or on-load tap changer (OLTC). It is the single most important piece of low-voltage instrumentation on the asset — a 10 % drop in conservator oil exposes the active part to atmospheric moisture, which raises moisture-in-paper above the IEEE C57.106 limit of 2 % and starts the insulation-degradation clock years before the planned replacement date.

This guide is structured the way oil-level indicators are actually selected on substation projects: by location on the transformer first, then by indicator type. Each location has a different set of constraints — the conservator wants a wide-angle dial, the OLTC wants a remote alarm, the radiator headers want a binary switch — and a single sensor type does not fit all three.

Contents

• Why Transformer Oil Level Matters
• Three Indicator Locations on a Power Transformer
• Magnetic Liquid-Level Indicator (MLI)
• Prismatic Sight Glass
• Capacitive and Magnetostrictive Indicators
• Standards: IEEE C57, IEC 60076, GOST
• Alarm Wiring and SCADA Integration
• Featured Indicators
• FAQ

Why Transformer Oil Level Matters

The insulating oil in a power transformer does three jobs: dielectric insulation between windings, heat removal from the core, and physical protection of the cellulose paper from oxygen and water. When oil level falls — through a leaking gasket, an over-temperature event that opened the pressure-relief valve, or simple seasonal contraction below design minimum — all three protections degrade simultaneously.

• Dielectric strength loss. Air gaps in the upper main tank reduce phase-to-tank withstand by 30–50 % per IEC 60076-3 — a slow-burn fault path waits for the next switching surge.
• Cooling loss. A 10 % drop in radiator-header oil raises winding hotspot temperature by 8–12 K under full load — paper aging accelerates per the Arrhenius law (doubling for every 6 K above 98 °C).
• Moisture ingress. Conservator under-fill draws atmospheric humidity through the silica-gel breather; the silica becomes saturated, and free water enters the oil at 30–60 ppm — well above the 20 ppm IEEE C57.106 alarm.

The oil level indicator is a $200–$2000 device protecting a $500k–$50M asset. Spending an extra 10 % on a contact-output indicator that wires into SCADA — instead of a pure visual dial — pays for itself the first time it catches a slow gasket leak before maintenance walks the yard.

Three Indicator Locations on a Power Transformer

A typical oil-immersed power transformer has up to five oil compartments needing level supervision. Three of them carry an indicator on every modern unit; the other two are on units above 200 MVA or with multi-radiator banks. Each compartment has different geometry, oil expansion behaviour, and alarm logic — so the indicator type changes with location, not with manufacturer preference.

Location	Geometry	Indicator type (typical)	Output	Critical alarm
Conservator	Horizontal cylinder	Magnetic liquid-level (MLI) dial	2 SPDT contacts	Min & max
Main tank	Rectangular box	Prismatic sight glass + thermometer pocket	Visual + 1 contact	Min only
OLTC compartment	Small cylinder	Compact MLI or magnetic float switch	1 SPDT	Min
Radiator header	Pipe	Binary float switch	1 SPDT	Min
Bushing turret (oil-filled bushing)	Vertical small reservoir	Capacitive or float	1 SPDT or 4–20 mA	Min

The conservator indicator drives day-to-day decisions because the conservator absorbs all the thermal expansion of the entire oil mass — its level swings ±15 % between 0 °C ambient empty load and 40 °C ambient full load. The dial therefore needs a wide angular range and clear MIN / MAX bands sized to the design oil volume, not a generic 0–100 %.

Magnetic Liquid-Level Indicator (MLI)

The magnetic liquid-level indicator (also called magnetic oil-level gauge, MOG) is the workhorse on transformer conservators. A toroidal float inside the conservator carries a magnet; outside the tank wall, an isolated dial pointer follows the float magnetically through the steel wall. There is no penetration of the tank, no gland to leak, and the dial is hermetic to weather.

Two SPDT (single-pole-double-throw) reed-switch contacts sit at MIN and MAX positions on the dial scale. The MIN contact wires into the substation alarm panel and SCADA — typically as “Transformer T1 Conservator Oil Low”. The MAX contact catches over-fill during oil top-up and is normally a maintenance alarm only. The contacts are rated 230 VAC / 1 A typical, which is fine for direct relay coil drive.

• Strengths. Hermetic dial, no electronics, 30+ year service life, ±2 % accuracy on the dial scale, immune to oil dielectric variation, ATEX-compatible variants for marine duty.
• Weaknesses. Mechanical — the float pivot can stick after long stagnation; reed contacts can weld on inductive loads; no continuous output without an additional 4–20 mA transmitter module.

For continuous remote monitoring, look for an MLI with an integrated 4–20 mA potentiometer transmitter — the dial pointer drives a precision pot, and the loop signal is brought out on a separate gland. Cost premium is typically 30–40 % over a contact-only MLI, and it eliminates the need for a separate magnetostrictive transmitter.

Prismatic Sight Glass

On the main tank, a prismatic sight glass is still standard — even on units with full SCADA. The glass is a wedge-shaped sodium-silicate prism that totally internally reflects light when oil is in contact (looks dark), and transmits light when it sees vapour space (looks bright). No moving parts, no electronics, and a quick visual check from the substation walk-down platform is enough to confirm oil is above the gasketed bushing flanges.

Sight glasses are usually 50–100 mm wide and 200–400 mm tall, mounted at the level of the upper bushing flanges. They are not sized for the full oil swing; they only show whether oil is above the safety minimum during a walk-down. Pair every sight glass with a discrete-contact float switch wired to SCADA — the visual check is the secondary diagnostic, not the primary alarm.

Capacitive and Magnetostrictive Indicators

For OLTC compartments and bushing turrets where space is tight and the contact dial of an MLI does not fit, capacitive and magnetostrictive transmitters are the alternatives.

Capacitive. A short rod (200–400 mm) inserted through a top flange. Mineral oil ε_r≈2.2; ester oil (Midel 7131, FR3) ε_r≈3.2 — the transmitter must be wet-calibrated to the actual fluid. Best for OLTC compartments because the rod is small, robust, and survives the small but frequent oil flushes during tap-change events.

Magnetostrictive. An external float ring rides up and down a stainless tube; a magnetostrictive wire inside reports the float position with ±0.05 % resolution. The accuracy is overkill for a conservator alarm, but on a shunt reactor or HVDC converter transformer where oil-level trending is part of asset-health analytics, magnetostrictive feeds into PI / OSI-Soft historians and supports leak-detection algorithms by correlating level rate-of-change with ambient temperature.

The deeper trade-offs between magnetostrictive and float technologies are covered in our magnetostrictive tank level measurement guide; for general oil-level engineering across storage, transformer, and lubrication contexts, see the oil level measurement overview.

Standards: IEEE C57, IEC 60076, GOST

Three standards govern oil-level indicator specification on power transformers, and the spec changes with the asset’s destination market.

• IEEE C57.12.10 (US, Canada). Requires a magnetic liquid-level gauge on the conservator with two contacts (MIN, MAX) and a visual dial visible from ground level. Sight glass on the main tank optional but typical.
• IEC 60076-1 (Europe, Asia, Australia). Same intent; differences are mostly mechanical (DIN flange dimensions vs ANSI, M-thread vs NPT). IEC 60076-22-1 covers prismatic gauges for distribution-class units.
• GOST 11677 (CIS countries). Adds a third “AVR” (high-high level) contact and stricter shock-test requirements (50 g for 11 ms) for seismic-zone substations.

For OEM transformer manufacturers building for multi-region export, an MLI specified for IEEE C57.12.10 with a third optional contact is a common compromise — the third contact is left unwired for IEC delivery and wired for CIS delivery without a dial change.

Alarm Wiring and SCADA Integration

Three wiring patterns cover 95 % of substation oil-level integrations. The right one depends on whether the gauge has a continuous transmitter or only contacts, and how many compartments share a marshalling box.

1. Pure contact (MLI). Two SPDT contacts wired to the substation alarm-panel relay, then to SCADA via a digital input (potential-free dry contact). Each contact gets its own SCADA point (“T1 Conservator Oil Low”, “T1 Conservator Oil High”) for clear diagnostics.
2. Contact + 4–20 mA. Both signals brought to the marshalling box on the transformer; the analog goes to a SCADA RTU analog input, the contacts feed the alarm panel. Allows trending in the EMS historian and threshold change without re-mounting.
3. Smart transmitter (HART / Modbus). Magnetostrictive or capacitive transmitter with HART. Brings position, status, and self-diagnostic in a single 4–20 mA loop. Pairs well with our LP-series magnetostrictive tank level indicator on HVDC converter transformers.

Featured Indicators

FAQ

What is the most common transformer oil level indicator?

The magnetic liquid-level indicator (MLI) on the conservator. It is required by IEEE C57.12.10 and IEC 60076 for utility-class transformers and provides a hermetic dial plus two SPDT alarm contacts. A prismatic sight glass on the main tank usually accompanies it for visual confirmation.

How much can transformer oil level vary with temperature?

Mineral oil expands roughly 0.07 %/°C. Across a 60 °C swing (cold start at −10 °C, full load at +50 °C top oil), conservator oil level changes by about 4.2 %. The conservator is sized to absorb the swing without ever falling below the air-cell connection or rising above the breather.

Can I retrofit a 4–20 mA transmitter onto an old MLI?

Often yes. Many MLI manufacturers offer a bolt-on retrofit kit that adds a precision potentiometer and a 4–20 mA loop converter behind the existing dial. Retrofit takes 30 minutes and does not require a transformer outage.

Is the same indicator used on a Buchholz-protected transformer?

Yes. The Buchholz relay sits on the conservator inlet pipe and detects gas accumulation or oil surge — it is independent of the MLI. The MLI catches slow leaks (gasket weep, valve drip) that don’t generate gas; the Buchholz catches sudden faults. Both are required on transformers above 1 MVA per IEC.

What level indicator works on natural-ester (FR3, Midel 7131) transformers?

Magnetic / float-based indicators work without modification — the indicator reads geometry, not fluid properties. Capacitive transmitters need a recalibration because ester ε_r≈3.2 vs mineral ε_r≈2.2. Magnetostrictive floats use the same hardware but with a slightly heavier float to match ester’s lower density.

How do I check if a transformer oil level alarm is genuinely low or just a stuck float?

Three checks before a costly outage. First, compare the conservator MLI dial reading with the prismatic sight glass on the main tank — agreement rules out a stuck dial. Second, check ambient temperature against last 24-hour load; a real low alarm tracks load drop. Third, look for fresh oil weep at the bushing flanges and tap-changer base — the most common slow-leak path.

Does the OLTC need its own oil-level indicator?

Yes — the OLTC oil is separate from the main tank oil for contamination reasons (arc by-products from tap operations stay isolated). A small dedicated MLI or float switch on the OLTC compartment is mandatory under IEEE C57.131 and IEC 60214.

Request a Quote