Side-scan, multibeam or 3D: which sonar does a small survey USV need?

The three options, and what each one actually produces

Side-scan sonar sends a narrow, tall fan of sound out to one side and records what comes back. Move the boat forward, stack the returns, and you get a photograph-like image of the seabed. It is excellent at showing you that something is there and roughly what shape it is. It does not measure depth. A single unit images one side only, so covering port and starboard on the same pass means two units.

Multibeam echosounder (MBES) fans a set of beams across the track and measures the range to the bottom on each one. You get depths, not pictures - a point cloud you can turn into a chart. The Cerulean Surveyor 240-16 uses 16 channels over an 80° swath at 240 kHz. It is the right tool if the deliverable is a bathymetric surface, and the wrong one if the deliverable is “find the thing”.

Real-time 3D side-scan sits between them. The Cerulean Omniscan 3D 450 SS uses a 16-channel receive array to work out the vertical angle of every return, so it builds a dense 3D point cloud while still producing the conventional side-scan waterfall. One pass, both products, no offline processing step.

Why 450 kHz, and when you would want something else

Frequency sets the trade between range and detail, and it is the first thing to check when comparing two side-scan sonars that look similar on price.

Higher frequencies resolve finer detail and lose energy to the water faster, so they see less far. Lower frequencies carry further and show you less. 450 kHz sits in the middle and is the common choice for small-vessel search work - enough range to cover ground, enough resolution to pick out an anomaly worth going back to.

For comparison, the DeepVision BR-ROV, built for the BlueROV2, runs at 680 kHz and is published at 50 m per side. That is a higher-resolution, shorter-range instrument, and for close inspection work it is arguably the better trade. The Tritech StarFish range also uses 450 kHz but in towed and OEM forms, which is a different operational model - deploying a towfish behind a boat rather than bolting a transducer to a hull. Neither publishes a price; both are quote-only, so we cannot put them in the table below honestly.

The point is that “side-scan sonar” is not one product category. Check the frequency and the range together before you compare two prices.

What each one costs

The €4,550 and the ~300 m swath are arithmetic - two units at the single-unit price, two 150 m ranges back to back. Everything else is published.

Add an Ethernet switch to any of these. On a Blue Robotics BlueBoat it is required and not included.

The specification that decides your survey speed

Every one of these sonars is quoted at a 20 Hz maximum ping rate. That figure is true and it is also the least useful number on the datasheet, because it only holds at short range.

Sound travels at roughly 1,500 m/s in seawater. A ping set to 150 m has to travel out and back - 300 m - before the next one can go out. That takes a fifth of a second. You cannot ping faster than the water allows, and no manufacturer can engineer around it.

Blue Robotics publish the actual curve for the Omniscan 450 SS, which Cerulean’s own datasheet does not: 20 pings per second up to about 30 m, 10 per second at 50 m, 5 per second at 140 m.

Here is why that matters more than the headline. At 1 m/s - a realistic survey speed for a small USV - 20 pings per second puts a ping every 5 cm along the track. At 5 pings per second it is every 20 cm. Set a long range to cover more ground per line and you thin out your along-track sampling by the same factor. Push the boat faster and you thin it again.

So the real trade is between swath width, boat speed and along-track detail, and you cannot have all three. For searching, long range and a quicker boat is usually right - you are trying to notice anomalies, not identify them. For documenting something you have already found, shorten the range, slow down, and let the ping rate climb.

This is physics rather than a product limitation. It applies to every side-scan sonar ever made. It is worth understanding before you choose a range setting, not after.

The honest caveats

The 300 m depth rating may not be the number that applies to you. Cerulean rate the Omniscan 450 SS transducer to 300 m. Blue Robotics’ technical documentation for the same unit states 50 m when the M14 bulkhead adapter is fitted - and the M14 adapter is the BlueBoat hull fitting. For a surface vessel towing nothing, the distinction is academic. For anyone planning to put this transducer deep on an ROV, it is not. Until Cerulean publish a figure for that configuration, plan against 50 m if you are using the M14 adapter.

There is no direct ArduPilot integration. Cerulean’s hardware talks to SonarView, and SonarView runs inside BlueOS on a Raspberry Pi. Tony White of Blue Robotics made this point plainly on the community forum in June 2026 when a custom-AUV builder asked. On a BlueBoat or a BlueROV2 this costs you nothing, because BlueOS is already there. On a bespoke vehicle it is a real integration task and you should budget for it.

The modular design is why it is cheap and also why installations go wrong. Unlike the forward-scan Omniscan, the SS arrives as a transducer plus a separate circuit board that you house in your own watertight enclosure and wire yourself. That is what keeps the price where it is. It is also, consistently, the thing that goes wrong: the recurring pattern on the Blue Robotics forum is an image full of noise with no discernible signal, diagnosed as a transducer connection or wiring-harness fault. Those threads end in resolution, not returns. But if your first log looks like static, check the harness before you conclude the sonar is faulty.

The transducer is physically delicate. Cerulean flag this at the highest warning level in their documentation: the array contains long, thin ceramic elements, and lifting or stressing the housing can damage them. Do not use it as a handle. Separately, do not leave it pinging in air for extended periods - it heats up and can be damaged.

The question people ask next: do I need a DVL?

If the vessel is on the surface, no. GPS and a heading give SonarView everything it needs to lay each ping down in the right place and build a georeferenced mosaic.

Underwater, you need something. A side-scan image is assembled from vehicle motion, so without a position source you get a conventional waterfall image rather than a map - usable for spotting a target, not for telling anyone where it is. The options are a DVL, a USBL, or Omniscan’s own Doppler motion tracking, which derives translational motion from the sonar returns themselves and needs two units at right angles for full 2D tracking.

This is the cost people miss when budgeting. A €2,275 sonar on an ROV with no positioning is a €2,275 sonar producing data you cannot georeference.

Three lines

If you are searching for objects on the seabed from a surface vessel, buy two Omniscan 450 SS and get on with it.

If the deliverable is a depth chart, you want the Surveyor 240-16, and side-scan will not substitute.

If you need to measure the height and shape of what you find as well as see it, the Omniscan 3D 450 SS does both jobs in one pass and costs about as much as the other two together.

Frequently asked questions

Not sure which of these fits your vessel?

Tell us what you are trying to find and what you are flying it from, and we will name the exact model and part number - including what we ruled out and why. We supply all three, so there is no version of this where we gain from talking you up a tier.

We are an authorised Cerulean Sonar distributor: the hardware, the warranty and the documentation are Cerulean’s; the stock, the EUR quote with no hidden import costs, and support in your time zone are ours.