Water Is Roughly 800 Times Denser Than Air

An ROV program that starts as "our drone, but underwater" hits trouble in the first month. The vocabulary carries over. The physics does not. Water is about 800 times denser than air and 60 times more viscous, radio does not propagate more than a few inches at useful frequencies, satellite positioning stops at the surface, and ambient pressure climbs roughly 0.445 psi for every foot of depth. A housing rated to 300 ft (91 m) carries about 133 psi of crush load on every square inch of its surface. At 1,000 ft it carries 445 psi.

Density also works for you. A neutrally buoyant vehicle does not fall out of the sky when a thruster quits, so a failure is usually a recovery rather than a crash. The engineering problem shifts away from weight budget and toward pressure, sealing, corrosion, and getting data back through a medium that blocks your radio.

Sealing, Housings, and Penetrators

Most ROV failures are water ingress, and most water ingress happens where a wire crosses the pressure boundary. Budget attention proportionally.

  • Housing form. Cylinders with domed or flat end caps are the default: acrylic tubes for shallow work and camera visibility, hard-anodized 6061 aluminum for mid depth, titanium or glass spheres past a few thousand feet. Printed or cast prototype housings are for dry-bench work only.
  • Radial seals over face seals. A piston O-ring running in a bore tolerates pressure cycling far better than a face seal on a flange, and gland geometry matters more than durometer. Getting groove width, depth, and squeeze right separates a tool that runs a season from one that floods on dive four; the fundamentals are in O-ring selection and gland design.
  • Penetrators. Every thruster lead, camera cable, and sensor wire crosses the boundary somewhere. Wet-mate connectors run $150 to $900 each; potted bulkhead penetrators cost $20 to $60 and are permanent. Minimize the count by keeping motor drivers inside the main housing.
  • Proof testing. Pressure-test every housing to 1.5x rated depth in a chamber before the first wet dive, empty first and then instrumented with a humidity and leak sensor. It is the same discipline as building a waterproof prototype, scaled up an order of magnitude.

An IP68 rating is not a depth rating. IP68 as commonly tested means shallow immersion for a stated time, which says nothing about 200 ft. A supplier who answers a depth question with an IP number has not built subsea hardware; the distinction is laid out in IP ratings explained.

Buoyancy and Trim Are Design Parameters

An ROV should be slightly positively buoyant, typically 1 to 3 percent, so a total power loss floats it to the surface instead of losing it. The whole vehicle becomes a buoyancy budget: every added sensor needs matching syntactic foam, and every block sits above the center of mass to keep the vehicle self-righting. Separating the center of buoyancy from the center of gravity by 0.4 to 1 in (10 to 25 mm) buys passive roll and pitch stability without burning thrust.

Plan trim as physical hardware, not a CAD estimate: bolt-on lead and removable foam, since a vehicle trimmed in a freshwater tank floats differently at sea.

Thrusters and Six Degrees of Freedom

An aerial multirotor controls four degrees of freedom with fixed upward thrust. An ROV can control all six, and the layout decides which. A common working configuration is six thrusters: four vectored in the horizontal plane at 45 degrees for surge, sway, and yaw, plus two vertical for heave, leaving roll and pitch to passive stability. Eight thrusters buys full active control and costs power, weight, and cabling.

Thrusters are brushless outrunners in flooded, oil-filled, or magnetically coupled housings. Flooded motors with plastic bearings are cheap and rebuildable; magnetically coupled units are sealed and expensive. A subsea propeller is a low-RPM, high-torque device: expect 3 to 8 lb of thrust per unit at 100 to 350 W. The tradeoffs across motor types are in choosing a motor for your product.

The Tether Still Wins

Acoustic modems work, and they deliver something like 100 bps to 30 kbps at range with latency measured in seconds. That is enough for commands and telemetry, not for video. Any ROV whose job is looking at something keeps a tether.

A practical tether carries a twisted pair or fiber for data plus conductors for power, in a neutrally buoyant jacket with a Kevlar strength member so pull never lands on the conductors. The two live decisions are voltage and topology. Sending 300 to 400 VDC down the tether and stepping down at the vehicle keeps copper thin and losses low; sending 48 V means a heavy, draggy cable past a few hundred feet. Ethernet over a single twisted pair through an extender pair gives 20 to 90 Mbps at 300 m and is the pragmatic default. The power-over-cable logic is the same one behind tethered aerial systems.

Navigation Without Satellites

Position comes from fusing a depth sensor, an attitude and heading reference, and a Doppler velocity log measuring speed over the bottom. The DVL is the biggest accuracy purchase at $8,000 to $25,000, turning dead-reckoning drift from tens of meters per minute into roughly 0.5 to 2 percent of distance traveled. Absolute position comes from acoustic USBL on the support boat. The estimator is close to what aerial teams build for GPS-denied navigation.

Power, Corrosion, and Time in the Water

Battery vehicles carry pressure-tolerant lithium packs and give 2 to 6 hours; tethered vehicles run all day. Either way the pack needs the rigor described in drone battery systems, plus a failure mode no aerial team models: a flooded pack in salt water.

Galvanic corrosion sets service life. Never bolt bare aluminum to stainless in salt water without isolation, fit sacrificial zinc anodes, use 316 stainless or titanium fasteners, and rinse after every dive. Biofouling on optics starts within days in warm water, so plan for wipers or replaceable ports.

A Realistic Development Plan

A credible first program runs 6 to 10 weeks on architecture, thrust and buoyancy math, and depth rating, then 10 to 16 weeks on a bench-and-tank prototype with real housings, then chamber testing and open-water trials. Expect $80,000 to $250,000 to a working inspection-class prototype, more once a manipulator arm is in scope. Test in a pool, then a lake, then the sea; each stage finds a different class of failure.

Talk Through Your Subsea Program

Projects House develops underwater vehicles end to end: pressure housings, thruster and buoyancy layout, tether and power architecture, control firmware, and topside software. If you have a depth target and a payload, send the outline through our contact form and we will tell you where the hard part actually is.