Fiberoptic Tether


Looking at the tether decision tree, I really hate the idea of a electrical connection underwater. So after a little brain storming a I designed a Cape for the Beaglebone that transmits and receives over a pair of TOSlink cables to a USB Transceiver.

Based on the data sheets for the for the AM3359(Beaglebone CPU) the UARTs of the beaglebone support data rates of up to ~3 Mbit/s. The USB Transceiver supports data rates of up to 6Mbits/s. The system will transfer data at 3Mbit/s which is more than enough for a control signals and video. The actual tether is currently 2x 2.2mm OD POF cables.

I have a few developer boards, usb transceivers, TOSlink jacks for sale. If there is interest I will do a production run.

Also I understand that the AM3359 supports s/pdif as an IO, this is not an implementation of s/pdif

BeagleBone Bark Cape

USB Transceiver

Cape and Transceiver Connected by Coiled FOP Cable

Update 1: Let me see if I can address some of these questions.

Why I chose TOSLink? There are fiberoptic standards out there with both incredible range and data rates. But the cost of these systems are very high and in almost every imaginable use case, overkill. The cost of TOSLink is several orders of magnitude lower. This is because of it market penetration in digital home audio. Additionally, TOSLink connector are designed to accept 3.3v and 5v TTL signals.

USB Transceiver model - This is something I designed and manufactured for this purpose. The Whistle USB Transceiver is USB to TTL converter IC connected to a TOSLink Transmitter and a TOSLink Receiver. Inside of a case that is I printed on a Thing-o-matic. More information on the design is available in it git repository

Range and Attenuation

Here is where I am still operating theory here as I just manufactured the boards this past week and haven't tested them with long cables yet. The industry has seemed to settle on 30m as being the longest usable TOSlink cable so we will see what maximum length is in the coming weeks. If the max length is 30M then I already have a repeater system on the drawing board.

Cable cost

Short cables are cheap, super cheap. Like $7 for two 25' cables. The longest commercially available cable is ~30M (100FT) and costs about $60 for two.

The cost of a repeated 100m cable could be as high as $200 in the early stages of development.


This is awesome! I'm really curious to see how this performs.

Especially curious about attenuation and range as well as durability. Please keep us posted!

I'd definitely be into getting a few if you decide to do a run.


Electrical connections in salt water are the bane of commercial multi-km depth-rating ROVs too. So ditching electrical lines is likely to go a long way towards improving reliability.

As David Lang says, attenuation, range, durability? (The latter can, to an extent, be controlled by the users' choice of FO cable.)
I'll ask the ROV technicians (I know several on the rig and in the Real World) what problems they find with FO links. They certainly use them, but also electrical power and hydraulic power lines too. I can't say that I've investigated too much into their techniques and technologies, but I'll pay more attention in future.



Nice job- that thing rocks! This is why developing a project with a community is so great. I can't wait to see how it works out!



Any information about the transceiver models? Specs?


I think optics is the way to go, but I have a few concerns: how much would 100m of such cables cost? Another concern is weight and maneuverability of the ROV when attached to optics cables. It looks like they are more rigid than twisted pairs.

But yeah, interested to know how it performs in real water :)


@Simon : rigidity isn't a simple question. Having done cave diving in an earlier life (and cave reconnaissance was one of the original drivers of the concept, then I've had unpleasant experiences with nice flexible line coming off my line spool then forming a cloud of (potentially-) lethal spaghetti. After that, and after hosing out the inside of my wetsuit (see "earthy" caver's songs ad nauseam) I replaced the 3mm spaghetti with 4mm line, precisely in favour of increased line rigidity.

When using base-fed line (as appears to be the implicit mode for this generation of models, low stiffness is good, because it allows the line to flex around entrapments. But low stiffness is also bad, because if tension comes off the line (current, a steering search, line-handler's inattention), loops can more easily form. which is when you end up with a dead diver or a trapped ROV. Neither of which is particularly good.

I would anticipate the line stiffness being chosen for a specific task. Possibly the data line would be backed up with (threaded into) a mechanical line too, to form an umbilical. If FO line is available by, say the 100m drum, then the cost of different line weights could be amortised between several group. (I've seen this done with different load ratings of heptacore wireline cable costing hundreds of thousands of dollars, and weighing a couple of tonnes per 5km reel. That's the industrial big brother relative, of course)

I'm thinking ... how would you manage tension and re-coil on an ROV-fed line reel. That might be a separate water-proof module bolted onto the main body. But for complex penetrations, it would be necessary. People with rock-climbing experience will have relevant input about how many corners you can drag line around - it's harder than you think until you try it.


@Aidan: good points indeed... I'm not a diver myself, and I'm thinking of driving the ROV in a cave so i might never encounter all that problems.

I think a fed-recoil system on the ground station should be developed sometime in the future and otherwise it becomes a bit complicate to steer to ROV on the monitor and then recover it.

I do rock-climbing, and definitely you usually try to avoid dragging ropes over corners :) for many reasons: after a few corners the rope becomes almost impossible to pull, and friction destroys it. And in case of fall, corners act as brakes and they "hard-stop" the fall instead of the usual "soft-stop" where the rope slides a bit reducing the force of the impact. I don't remember exactly how many corners you can have rope go through, but practically, after 2-3 90deg corner the rope becomes blocked


This is great. It's really important that there are multiple tether approaches happening simultaneously. Each strategy has its benefits and drawbacks. The more interchangeable and modular, the better for the project.

@aiden Good points. A few things...

Rigidity isn't as big of a deal as is mass. When dealing with small, observation class ROVs the mass of the tether can quickly add up to be equal to the ROV. The biggest issue we've always come up against is managing the thrust:drag ratio. More tether means more mass. More mass means more drag.

In our experience with the two wire, we did find ourselves with too much line out in a few different occasions. It was really easy to reel in the tether topside. The diver safety risk isn't an issue when you have a relatively low cost ROV down there. BUT many folks want to dive with their ROV so that would become an issue then.

@Simon - Yes! Tether reel is needed. Depends on the tether that people choose to use. You need to be very careful with fiber, as it breaks easily (which becomes an expensive mistake). We've had some issues with cross-talk on the two-wire tether so we need to make sure the reel that we develop solves that issue.

More tether talk on the dev call tomorrow. 12pm PDT on my G+ page:

Maybe we can start off with Eric talking about his tether philosophy - it will be a great jumping off point for the discussion.


More mass means more drag ? If "flying", then yes (because you need forward motion to provide lift), but for a neutrally-buoyant device ... are you sure on that?

As Simone points out from his rock-climbing experience (also commonly indicated from procedures for "open water" SCUBA diving when the diver is on a base-fed line, be it a guide line, or an umbilical for air/ mixed gas/ communications/ hot water supply ; I've not done that sort of diving myself, so my knowledge is purely from RTFM), more than a (surprisingly small) number of corners can make it impossible for a climber on rock (let alone a battery-powered set of props) to make progress. More importantly, before you get to that point, the "tender" who is spooling out and reeling in the guide line / umbilical can't "feel" the action at the sharp end of the rope. Which is followed by entanglements as day follows night.

In the cave-diving world, the limitations of "base-fed" line were realised in the mid-1930s, though little was successfully done about it until the re-formation of the Cave Diving Group after WW2. Various contraptions were tried (consult Martyn Farr's "The Darkness Beckons" for historical detail - under "AFLOLAUN"), ending with the present system of carrying line spooled onto a frame with a carrying handle. As you explore into new territory, you spool out the line along and behind you ; if you retreat, then you spool the line back onto the reel as you come out, keeping the line under a small tension to prevent spaghetti-fication. And you pop back out onto the surface, or where you tied you line onto existing line. (In this ROV context, you'd be back at base station, on "surface".)

David and Simone's comments suggest the different arenas of interest for the ROV - exploring spaces too compact for a human (or too unsafe, or too unhealthy - I don't believe Stumpy's tales of concrete diving, but I know people who have had to do multiple-corpse recovery jobs where the water got unbelievably noxious after the first couple of months), as opposed to co-diving with people. I would suggest that base-fed might well be appropriate for the latter case, but spooled-line (with the spool on the ROV) for the former cases.

Driving the ROV and managing the tether in one set of hands? As David indicates, that's a recipe for trouble.

Base-fed tether is conceptually and constructionally simple - breed a humanoid handler and hit it until it stops making mistakes - maybe provide a couple of tent-pegs to assist line management on surface.

What is passing through my mind for ROV-deployed line would be a bolt-on module for the main ROV, which contains it's own batteries, motors and control circuitry. (That reduces the need for co-development between spool and main ROV.)

- When the ROV is progressing forward or stationary, the line spools OUT against a brake (some motors will do that when connected across a resistance? IANAEE - I Am Not An Electrical Engineer.).

- When the ROV is reversing (how does that information get into the spool module? Coloured LEDS through the transparent pressure vessel walls perhaps?) the motor turns to spool IN the line and attempts to maintain a constant tension (a few newtons only would suffice) on the line. If the thrusters are being effective, the motor will run faster ; if the machine experiences drag (walls, current), then the spool motor (and it's batteries) is assisting the ROV thrusters (and their batteries) to bring the ROV out.

- If the spool module fails (flat batteries ; other major failure mode), it should fail "safe", so that the machine can still be hauled in as if on a base-fed line.

I'm trying to work out how to keep the cable count down. Are the FO lines rotatable? Can they be terminated to the inside of the line spool so that the signal gets through without imparting twist to the cables or connectors? If you can do that, then the well-established trick of threading a communication line through the middle of the navigation line (reducing cable-in-the-water count and protecting the comms line from the worst of abuses)) would work. But with a 100m line reel (of approx 4mm line ; not incomparable to the FO lines that I've seen), you're looking at some 200 turns on a 15cm radius by 10cm wide spool (remembering from my caving equipment). That's unfeasable for a twisting connection. I think.
(Why 100m? That's the nominal depth limit for the current OpenROV designs. My brief searches indicate GBP 41 for a 30m TOSLINK cable from a company called "LINDY." But I know next to nothing about audio, so that means very little. These are audio cables?)

I'm going to show my unfamiliarity with FO equipment now : what I've seen in the past has been twin pairs of FO lines, but I don't know why (and never needed to ask). Are separate TX and RX lines essential, or can you TX and RX on a single line by some sort of interleaving of packets/ commands?

As a plus ... given a design for a line-management module that can go into the water ... by tying it to the boat, you've got a line-management unit for a single person at surface to both steer AND line-manage the ROV. I think that sounds like two birds with one stone, which as a geologist I like!

Oh, it needs to be said (people have died for forgetting this!) : tie off the end of the line on the spool, otherwise you'll run off the spool at some point. And people have forgotten to tie the base-fed line to something solid too, in the past. Ask at the rock-climbing club.


To clarify, tether mass (actually, tether weight in water) is a separate issue from tether drag, but both are a major concern in regard to ROV performance. In fact, tether drag and weight are generally inversely related since displacement (cross sectional area) can be increased to make a given tether have less heavy in the water, but that will cause there to be more drag and vice versa. The only way to improve one without damaging the other is to use less material that is denser then water (i.e. less metal which equates to thinner wire) ...or as Beau is attempting, ditch using metal to communicate all together.

I have gone the rout of using thin copper wire instead of fiber because it is cheaper, easier to integrate with, and more available, while still offering the potential to communicate broadband data as well as possibly power. That being said, if thin copper has the potential to give 95% of optimal performance, fiber has the potential to give 99% (100% being some magical wireless broadband option, I suppose). Fiber is close to neutrally buoyant, can be made very thin, and (to answer Aidan's question) can relay data in both directions over a single fiber (using different wavelengths in each direction).

De-spooling from the ROV like a spider spinning a web is a great capability to have, however my view it is that such a system should be simple and would not involve automatic re spooling. I picture the tether being wound in an on-board container like an inverse figure-eight the same way Ethernet cable is sometimes wound up in boxes. This winding technique does not twist the wire, so there is no need for a slip ring. For use in caves or ship wrecks, enough tether could be available in this system for the ROV to pay out tether all the way to its destination and back again, so if the tether gets stuck, the ROV can still be recovered.

If you want to see a great example of a system like this, I really recommend checking out "Ghosts of the Abyss" by James Cameron. He uses two ROVs named Jake and Elwood to explore the Titanic, and they use thin fiber-optic strands that pay out from the ROVs to avoid accumulating drag effects as they go through the ships inner structure.


I appreciate the point of simplicity in a spool system. Do less ; have less to fail.

However, if the ROV achieves a penetration, then steers itself out with the line hooked around a projection (a speleothem, a piece of a strut in a wreck, boulder on the floor, whatever), then retrieving the line post-dive is going to be really difficult. Unspooled line in an enclosed dive is a spaghetti death-trap for the next entry attempt, which is probably bad enough for a lump of plastic and ironmongery, but ... well you're looking at using the ROVs for cave exploration ; how would you feel about entering a passage with 3 or 4 ROV tethers in it which you'd dropped in your earlier mechanised explorations?

I'd say that spooling the line back into the ROV spool module is essential. You could mitigate it to a degree by trying to steer back along your entry line, but you'll know yourself that you're often doing the exit phase of your dive in heavily silted water. Very difficult. (Most of what I say below applies to wreck penetrations as much as caves. But I know caves better.)

OK, I accept that systems can (and will) fail ; but dealing with a guide line that has been left in a sump because of someone (or some machine) having a "bit of an epic" is a different thing to dealing with something that is left in the sump as a part of the basic design.

By definition, these machines are going to start from and go to places where it is difficult or excessively dangerous for a human diver to go. So the routes are going to be one or more of narrow (50cm and smaller), deep, rugose-walled, contorted, heavily silted ... otherwise the first human diver or caver to get there would just have kitted up the side-mounts and moved on in with standard kit. (I know that Americans often dive on back-mount systems, but side-mounts seem to be becoming common in open water diving, so I assume they're well-known in the cave diving "grotto" too.)

Line weight : UK practice is to lay permanent line in sumps as being slightly inherently buoyant (material : polypropylene , tensile properties are relatively unimportant here) so that it floats loosely tethered in the sump and sheds silt with the current flow. Negatively buoyant line tends to get buried in the silt. For a retrieveable line ... well to be honest, I never noticed any change in buoyancy spooling out or retrieving a 100m line reel in line-laying and retrieval exercises. The line reel changes (obviously), but only by a hundred or so grammes for the whole spool - it was positively buoyant at all times. That can be trimmed with a hot-glue gun and penny coins.

Are the "sweet spots" of the FO systems under discussion sufficiently "broad" that the FO cable and it's connector could be brought into axial contact with the data pickup from the ROV, so that the cable can rotate on it's axis which is co-linear with the axis of the pickup?

If you're using electrical contacts ... single wire with the water to complete the circuit, or two wire? Do you try to keep the contacts dry (implying a need for pressure-tight rotating seal), or have them wet but in an insulating fluid? (I had an amusing mail a few months ago from an engineer at Poseidon recommending that I use fine whisky to fill anti-freeze caps on my gags, "because some will get on your fingers, and it might as well leave a nice taste in your mouth!" Which begs what sort of insulating displacement oil would work best and be easily biodegradable.)

I'll stick that Cameron book on my wishlist. I've not actually seen any of his Titanic work (is that what he's famous for? Apart from that Avatar shoot'em'up?), even when I was working on the Grand Banks. It was just a boat full of a lot of poor people, and that's never really sparked my interest. The iceberg on the other hand ... fascinating.


I saw a documentary on Nereus the other day, they used a single, 25km long (!!!), fiberoptics cable to control the entire ROV, including all peripherals. That cable was only about 0.6mm thick, if I remember correctly. Of course they only had to deal with open water, wouldn't work in caves, etc...

In my opinion the choice of tether, if twisted pair, Cat5 (Cat6), or FO, really depends on the specific use. When you're in a cave, you want a strong tether that doesn't break easily and that you can pull out -or dive after- in an emergency. In open water, a lightweight tether with as little drag as possible is more desirable, so FO should be better there and so on. I will be doing a lot of filming in the Everglades, so I'll have a special tether with a 2mm stainless steel cable in it, just in case a gator wants to take a nibble and take off with it..... j.k.

Anyway, there are a few safeguards that you can put on you OpenROV so you can find it, should it somehow get separated.

Re. going around corners, the NEC (National Electric Code) forbids anything more than three 90 Degree turns in a cable raceway (PVC pipe, etc.) for the simple reason that you can't pull around more than a certain amount of degrees in a circle before the cable starts tightening around the bend, pulling itself tight and getting 'locked' in position.

This is without putting into consideration any kind of friction, not to mention jams in cracks etc. of course. Thinking further about that, I guess the tether profile would also make a difference in cave diving, round vs. twisted wires, etc...I am looking at a single strand, shielded cable (used in fire alarm systems), I don't have the exact name, just a sample at the moment.

I'm working on the design for a lightweight, motorized, tether feed system that can be remotely operated and will hold around 900ft Cat5 cable ( which is about 4mm dia.) in a plastic case. And of course it will fit nicely on the (remote controlled) support barge ("OpenBase"?) that I am planning for it....

Re. tether strength, there is a simple solution for that called "Spiderwire". it's a supertough, woven fishing line that weighs next to nothing but unfortunately can seriously cut you, too if not handled carefully. One of my first dives will be in a rock-lined canal with about 15-20 knot current, so I'll tape one of these to my tether every foot or so to make sure I don't lose physical connection. Or I'll try the shielded cable I mentioned earlier.


Nereus ? Ah, Nereus (underwater vehicle) !
The cave diver's rule book (in the UK at least) is that you never, ever, ever pull on the line. Pulling will almost always result in the line being pulled into a non-optimal position in the cave. So you're at best storing up problems for yourself on the outwards leg of the dive. If you actually manage to break or saw through your lifeline ... doubleplus ungood. That's with UK practice of 4mm polypropylene (slightly) buoyant line with a safe working load of some hundreds of kilos ; the thinner lines used in "continental" practice are even more delicate. Between the likelihood of getting the ROV wedged into a slot (Dead Man's Handshake and all that jazz) and sawing the line across a sharp edge and cutting it ... well I doubt you're going to accomplish anything much. About the best that you could hope for from tugging the line in a penetration scenario would be to jam the ROV in place so that the line is well anchored, and having to send in either a recovery ROV or human diver to follow the line.

there are a few safeguards that you can put on you OpenROV so you can find it, should it somehow get separated.

Apart from making it in high-visibility colours, using reflective tape on all faces ... and of course, keeping the guideline attached? (Incidentally, UK cave practice is, if available, to use bright orange line for visibility.)

Spiderwire ... sounds interesting. But at 0.4mm thick, this does conflict with the desire for visibility (which depends on the particular task). But that does sound an affordable option. Incidentally one procedure that I've heard of being used (though not actually seen myself - I'd stopped cave diving by the time Nott-II was opened) was to thread 2-strand telephone wire in between the 3 strands of a standard piece of guideline (cable-laid rope, not plaited) so that a telephone line could be run through a sump for use in cave rescue. That might be worth adopting or adapting - three reels of this "Spiderwire" stuff plaited around Cat5 might be an interesting material. It'd be a bit of a PITA to make ... but not terribly daunting. 80lbs is the highest rated "Spiderwire" I can find online - which is useful, but not adequate in itself, I'd think.


Yup, that one, I'll look for the title of the documentary, it was awesome!

Makes sense what you're saying and I really like the idea of the phone wires together with a nylon rope! That could be done with the twisted pair as well, then; especially if it's stranded wire. Your local boat supply house might have a machine to do that, otherwise find yourself a rope maker or check with a sail maker. It would be cool to pull the wires inside the sleeve of a paracord (a.k.a. survival rope), not sure how to do that over 300ft, though...

Speaking of combined wire, in the Service we had field-telephone-wire that had a few strands of steel wire in together with the copper wires, that stuff was nearly indestructible!

Regardless of the stuck tether scenario, I was contemplating putting a small webcam (like the ones in a laptop lid nowadays) on the rear of my OpenROV, maybe even with a few tiny lights, so I can see what's going on behind me. I just need to figure out how to best run multiple camera signals, preferable in full HD from the ROV to topside....

I don't even know if the beaglebone could even run two cameras at once, total noob with these things, so far. Any suggestions are welcome.

So, there is your elegant answer to the "getting in a tight spot" situation, be it a cave, wreck, sewer, or reef: backup camera!!! ;-)

@Beau, I like your cape an would like to get my hand on a couple as well, whenever you do a production run let me know, please!