Ultrasonic or Radar Level Transmitter? A Field Engineer’s Honest Decision Tree

There Is No “Best” Level Transmitter. Only the One That Fits Your Tank.

There’s no universal answer to “which level transmitter should I use?” If anyone tells you otherwise, they either haven’t spent enough time inside actual plants, or they’re selling something. The truth is less comfortable: the right instrument depends on what you’re measuring, the environment around it, and whether you actually need continuous level data at all.

Before we go further, some context. I’ve been specifying and commissioning process instrumentation for eight years. In that time, I’ve personally made—and documented—11 significant selection mistakes. Roughly $18,000 in wasted budget, plus a few uncomfortable conversations with plant managers. Now I maintain our team’s pre-purchase checklist. This article is basically that checklist with scars attached.

Your situation probably fits one of four scenarios. Each has a different answer:

• Clean liquid, stable tank, no serious foam → an Endress+Hauser ultrasonic level transmitter is likely all you need.
• Dust, steam, foam, pressure, or temperature swings → go straight to a radar level transmitter (Endress+Hauser Micropilot).
• You only need presence/absence detection, not continuous level → be honest and buy an ultrasonic proximity sensor instead.
• Digital integration, encoder feedback, or verifying existing meters → learn your IO-Link encoder options and know how to read a Sensus digital water meter before you swap anything.

Scenario 1: Clean Liquid, Stable Tank → Endress+Hauser Ultrasonic Level Transmitter

If you’re measuring water, wastewater, or light chemicals in a vented or atmospheric tank, start with an ultrasonic level transmitter. It doesn’t have the jaw-dropping specs of radar—but it doesn’t need them. The principle is simple: a sensor fires sound pulses at the surface, measures the time of flight, and converts that into a level reading. No moving parts, straightforward installation, and a price that keeps project accountants happy.

For reference, the product people usually mean when they search “endress hauser ultrasonic level transmitter” is the Prosonic series. The classic workhorse is the FDU91 sensor, and Endress+Hauser quotes accuracy around ±2 mm (technical information TI00428F). That’s more than sufficient for the vast majority of water storage and chemical holding tanks I’ve worked on.

Cost is the other selling point. A basic ultrasonic setup with 4–20 mA output and relay contacts was running roughly $800–1,800 on published list prices as of January 2025. It’s the cheapest continuous level measurement option that still gives you a real measured value—not just a switch. (Should mention: prices exclude installation and hazardous-area certification, which can add significantly. Verify current pricing at endress.com before you budget.)

Real-world example from my records: in 2019, we installed a Prosonic on a 12-meter municipal water tank. Five years later, it’s still on the same calibration. The only maintenance was cleaning algae off the transducer face—which, honestly, took longer to schedule than to do. A small rain shield solved a condensation issue we had the first winter.

Where do people go wrong here? They try to stretch ultrasonic into applications that only look clean and stable. The moment you have heavy foam, high vapor concentration, or significant temperature gradients, the sound pulse scatters or travels at a speed the sensor didn’t expect. That’s exactly how I bought a radar transmitter in early 2019.

Scenario 2: Dust, Steam, or Foam → Radar Level Transmitter (Endress+Hauser Micropilot)

Here’s my strongest opinion: if your tank has visible dust, steam, foam, or operates under pressure, stop calculating the budget difference between ultrasonic and radar. The upfront savings will be quietly devoured by downtime, rework, and a very annoyed plant manager. I know because I lived it.

The request seemed simple: measure level in a 20-meter cement silo. Procurement preferred ultrasonic because it was about $800 cheaper. I opened the datasheet. Range: 20 meters. Silo: 18 meters. It fit. The numbers said this was fine. My gut said something felt off—that dust cloud during silo filling. I went with the numbers anyway.

You can guess the rest. When the silo was being filled—the one moment you actually need the reading—the cement dust absorbed the ultrasonic echo completely. The instrument went blind at exactly the worst time. We patched it with a redundant high-level switch and ordered a radar replacement with rush shipping. The damage: $800 saved, about $2,700 in unplanned costs, and a three-day delay. Classic penny-wise, pound-foolish.

The fix was an Endress+Hauser radar level transmitter—specifically, the Micropilot FMR20. It uses an 80 GHz radar pulse that cuts through dust and vapor without blinking. And here’s the surprise most engineers still haven’t absorbed: compact 80 GHz radar is no longer the exotic, budget-breaking option it was in 2015. As of January 2025, the FMR20’s list price is surprisingly close to a premium ultrasonic unit. The phrase “radar is too expensive” is, honestly, outdated.

Even after the radar arrived, I kept second-guessing. What if the dust was too thick for radar too? Didn’t relax until the first full fill cycle showed a stable reading from empty to 20 meters. It’s been running without a single echo-loss event since. If someone searches “radar level transmitter endress hauser” on a dusty brownfield project, the FMR20 is the first thing I’d tell them to evaluate.

One caveat: radar isn’t magic. On liquids with very low dielectric constants—certain hydrocarbons, for example—selection gets trickier, and you should have a vendor application engineer confirm dielectric values first. But for the common problem cases (silos, condensate tanks, foam-covered basins), modern radar is the answer I’d argue for every time.

Scenario 3: Presence Detection Instead of Level Measurement → Ultrasonic Proximity Sensor

Here’s the scenario that saves my customers the most money, and it’s about not buying a level transmitter at all.

A few years ago, a beverage customer asked for “continuous level measurement, 4–20 mA output” for what they called a buffer tank. I approved the initial spec without visiting the site. Rookie move. When I arrived, the “tank” was a 50 cm plastic hopper with bottles passing underneath. They didn’t need level. They needed to know whether a bottle was under the filling nozzle. That’s not continuous measurement—it’s a yes/no presence question.

The right tool was an ultrasonic proximity sensor: it detects objects within a short range and outputs a discrete PNP/NPN signal. Cost comparison: $1,800 transmitter vs. roughly $150 sensor. We revised the spec, saved the customer about $1,650, and the line has run fine for years since.

I retell this because it’s the classic mistake of spec’ing for the sensor category instead of the actual function. Ultrasonic proximity sensors are also the workhorse in environments where optical sensors embarrass themselves: dusty loading bays, transparent PET bottles, dark or shiny surfaces. Sound doesn’t care about color or reflectivity.

If your question is “is something there?” and not “how much of something is in there?”, price both options before committing. An industrial ultrasonic proximity sensor typically lands in the $100–350 range, based on list prices I checked in January 2025. Your maintenance team will also thank you—replacing a $150 sensor hurts a lot less than a $1,800 one.

And a broader note, because it matters: small orders and small applications are not shameful. I still remember when my own first test orders were worth about $200. The suppliers who took those seriously earned loyalty they never had to spend a dime on advertising for. Good advice should scale with the job, not with the invoice.

Scenario 4: Digital Integration, Encoders, and Reading the Meters You Already Have

If your machine spec calls for an IO-Link encoder

IO-Link is one of those standards that engineers either love or haven’t encountered yet. It’s a point-to-point digital communication protocol—imagine USB for industrial sensors. Each device connects to an IO-Link master with a standard M12 cable, and the master reads configuration and diagnostics without any of the address-switch fiddling that older fieldbuses needed.

When someone writes “io link encoder” on a spec sheet, they usually want position or speed feedback that the control system can configure automatically. The practical benefit, in my opinion, is diagnosability. I was skeptical of IO-Link initially—I thought it would just be another interface to support. That changed in September 2024, when a packaging line’s feedback fault took us four hours to find: a crushed encoder cable inside the machine frame. With an IO-Link master, it flagged the faulty port in seconds. I’m now a believer, at least for machine-level and hybrid applications.

For pure continuous process environments, you’re usually fine with standard analog or HART. But if you’re building or retrofitting packaging, filling, or material-handling machines, the IO-Link conversation is worth having.

How to read a Sensus digital water meter (before you replace it)

This feels basic, but it has cost me before. If you’re replacing an older water meter with a new flowmeter—or auditing consumption data—you have to be able to read the existing unit. I once collected ten days of consumption data from a Sensus meter, only to realize later the display was in cubic feet, not gallons. The baseline was garbage, and we re-did the entire audit.

Here’s the quick field primer, based on Sensus meters I’ve actually handled:

• The main LCD shows the totalized volume. Read the unit label carefully: gallons, cubic feet, or in some models, cubic meters. Don’t assume—an assumption can cost a day’s worth of data.
• Many Sensus digital meters display a flow or leak indicator (a small arrow, triangle, or drop icon). If it moves while everything downstream is closed, you have a leak.
• Models like the iPERL family can show current flow rate on one of their scrolling screens. The display sequence is usually controlled by a button or a magnetically actuated switch; the manual online explains the exact sequence.
• Record the serial number and the totalizer reading with a timestamp. That 30-second habit is the difference between a smooth changeover and a three-week investigation.

And if your plant manager says “just swap it, we’ll sort the numbers later”—push back. Twenty minutes of documentation now saves a week of regret later.

The 10-Minute Pre-Check: Which Scenario Is Your Tank In?

Instead of reading hundreds of pages of datasheets, run through the same routing I use with our team. Work it in order:

1. Level measurement or presence detection? If you need yes/no or object detection, pick an ultrasonic proximity sensor and stop. If you need a continuous number, keep going.
2. What’s the medium and environment? Clean liquid in an open, stable tank? → ultrasonic. Dust, steam, foam, or pressurized tank? → radar, specifically a compact 80 GHz unit like the Micropilot FMR20.
3. How does it integrate? Machine-level PLC with rich diagnostics? Ask about IO-Link options. DCS process loop? 4–20 mA/HART is still the safe default.
4. Are you replacing or verifying an existing meter? Record the final reading first. If it’s a Sensus digital water meter, check the units and look for the leak indicator.

You don’t need the most expensive instrument, and you don’t need the one with the prettiest datasheet. You need the one that fits your process—and the honesty to admit what that process actually looks like. I paid for that honesty with $18,000 in mistakes. Hopefully this list saves you some of the tuition.

And if somebody tries to steer you toward a high-end solution for a simple problem, ask them one question: “Would you put this in your plant?” A good engineer answers with reasons. A good salesperson answers with another question.