Tether Decision Tree



Eric drew up this decision tree to analyze all the different tether approaches. What do you think?


This is a very nice diagram. Thanks Eric


There is another 3 wire option. USB->ethernet

The converter is dirt cheap (6$ for 150 ft version) and you only need to use 3 wires in the cable if you supply the USB power onboard.


You can do direct webcam + arduino to the computer. Saves you the expensive beagebone + custom converter (153$), and if you go for the extender with build in power, you can get HD video, full 25fps.

We did a lot of testing this evening. We learned quite a bit. Here is a description of the experiments and everything we learned:
1) We tested 100ft of standard CAT6 Ethernet cable and got a good connection
2) We submerged most of the Ethernet into the salt water off Jim's dock
(we also got a good connection, with no change in performance)
3) We tested the 100m length of Belden 1353A while it was coiled and got not connection
4) We tested a loosely coiled 50ft length of Belden 1353A and got a good connectio
5) We uncoiled and submerged most of the 50ft length of Belden 1353A in the salt water and got a good connection with no noticeable change in performance
6) We cut the 100m length of Belden 1353A in half (making two 50m lengths) and tested one of those 50m lengths uncoiled. We got a good connection which seemed to be the same as when we used the shorter length and when we used Ethernet
7) We submerged the 50m length of Belden 1353A into the salt water and got a good connection with no noticeable change in performance
8) We up the 50m length on a piece of cardboard as we pulled it out of the water until we got to the point where the wire was completely out of the water and coiled up. The connection continued to be good with no noticeable change in performance
9) We put a "waterproof" LED connector inline with the wire (between one end of the wire and the ETS Ethernet Balun) and got a good connection with no noticeable change in performance
10) We unplugged and replugged the "waterproof" connector and were able to immediately re-establish a connection automatically


With my last usage of these products, we faced significant problems with these types of usb to ethernet converters.

Assuming its the same "passive" design I had used, grounding and noise issues became a problem with the robot in the water. The design used phidgets usb boards for control of an ROV, and while the system would function perfectly on the deck, the instant the robot was placed in water, the entire system would fail.


This may be a super silly question:

Have you guys looked at forcing the beaglebone to use a 10BbT-half duplex connection without the added baluns? If so, how successful was the transmission?

Seems like it might be 60-80$ that could be creatively trimmed from the budget.



I think that would help a lot, but I don't know if that could be done using only software changes. It would definitely be worth looking into if anyone has the savvy. That could improve transfer speeds in each direction and eliminate cross-talk issues. I have just always assumed it couldn't be done with the existing hardware. Thanks for the recommendation, Chris!


The easiest way to do it is actually to just adjust the NIC settings on the host computer side (won't work if you put the rov on the other side of a router). In a MAC you can do it from the network preferences menu, similarly under windows.

You could alter it on the beaglebone side but it would be much more work.

The beaglebone will then autosense the connection change, here's a log file with some testing:

ubuntu@omap:~$ dmesg | grep PHY
[ 16.060754] PHY 0:01 not found
[ 19.051446] PHY: 0:00 - Link is Up - 100/Full
[ 1235.051560] PHY: 0:00 - Link is Down
[ 1236.051532] PHY: 0:00 - Link is Up - 10/Half
[ 1238.051631] PHY: 0:00 - Link is Down
[ 1240.051547] PHY: 0:00 - Link is Up - 10/Half
[ 1284.051549] PHY: 0:00 - Link is Down
[ 1345.021553] PHY: 0:00 - Link is Up - 10/Half
[ 1357.021523] PHY: 0:00 - Link is Down
[ 1358.021476] PHY: 0:00 - Link is Up - 100/Full
[ 5386.021499] PHY: 0:00 - Link is Down
[ 5387.021537] PHY: 0:00 - Link is Up - 10/Half
[ 5643.021558] PHY: 0:00 - Link is Down
[ 5644.021511] PHY: 0:00 - Link is Up - 10/Full

This would of course require some crimping onto the end of an ethernet jack with the two-wire cable but could provide significant cost savings if it works at distance. I don't have any two-wire handy but i'll try it over the next several weeks.

Doing some bandwidth testing with images i get approximately 400 kBytes / second with half-duplex 10bT.

Thanks for all your work!




We tried your idea out and it didn't really work for us. We soldered the solid orange and solid green wires from an ethernet cable to one of the leads in our long twisted pair tether, and soldered the white/orange and white/green wires to the other lead of the tether. (this was done on both ends of the long twisted pair tether so that we ended up with ethernet jacks on both sides). When we plugged the system into a laptop and our BeagleBone, the Ethernet connection light on the BB blinked at a regular interval, which indicates a connection error. Were you able to make it work?!. I've asked Ron Crane, a guy who helped develop Ethernet, about his thoughts on the idea, and he suggested that there would be two principle issues with this scheme.

1) Even in half-duplex mode, the receive channel of a transmitting ethernet port is still active, so it hears itself and gets confused

2) The impedance of the line becomes drastically mismatched because of the shorted connection between the terminals.

If you got this to work, please let us know how!

Blocking diods might solve issue #1, but I don't know if #2 could be fixed without just making a hybrid circuit like the ones we're using.

This is great thinking- ideas like yours are the kind of thing that may give us the elagent solutions we're looking for.

Thank you very much!



Unfortunately I was looking at it today and ran into the same issues. I had unfortunately misunderstood the role of half versus full-duplex in ethernet. I also don’t think it is possible with this setup.

Your solution does seem to be a good balance of cost vs. performance.

Thank you again for all your hard work!



What about a tether that uses two 16-18 gauge wires for 12v and ground, then two pairs of cat5 strands for ethernet connectivity? That's 6 wires, but the only ones that really count are the two large gauge ones. (This isn't necessarily for the openROV, just an ROV in general that I'm planning on building)


The point of trying to keep the wires as small/few as possible is to try and keep the tether as neutral buoyant as can be. If the tether has downward drag on the ROV then the ROV needs to compensate via more buoyancy or more propulsion.


Dan's got the idea! SpringHalo- I originally had something like what you describe for my first prototype, and I had also purchased long strings of 6mm foam tubing for buoyancy that I ran parallel to the wires (all of which was inside a mesh sheath to keep it together), and the bundle ended up being hard to build, expensive, thick (>15mm diameter) and ridged. It became apparent that more wires + more buoyancy creates a run-away performance drop in terms of drag, agility, and practicality, so staying with the thinnest, simplest wire possible would work out a lot better. Here's an image of the original tether I built.



What are your thoughts on this idea? For long distance power transmission it seems pretty solid. A single cat5, 5e or 6 cable shouldn't be nearly as heavy as speaker wire tethered in as well.


As I have noted elsewhere, the higher the tether voltage the better, limited by safety and your nerve. Most of the small commercial ROVs I have worked with use 150VDC at the ROV, so the voltage at the top of the tether is 150V + losses and sometimes approaches 250V. Large commercial ROVs use much higher voltages.

Higher voltage means less copper so less weight. If the tether is neutral density less weight means thinner. Thinner means cheaper, more flexible, and better heat conduction to the water. Better heat conduction means you can tolerate more resistive loss so you can use even less copper which makes it even thinner!

But high tether losses mean you may need some sort of regulation. Commercially this is a circuit that monitors the bottom end voltage and transmits the info topside where a variable power supply responds to the feedback. The feedback can be by two voltage wires, one voltage wire, one current wire, an oscillator tone, or an A/D reading in a data packet.

For adult hobbyists 120VAC with an isolation transformer should be OK.


This project may later be adapted for dive assistance usage (carry a tether) for depths of ~100m, so an ethernet cord of that length will be used. Would it be smart to use a 12v to 220v adapter instead, whose rectified DC voltage will end up being around 150v, plenty of extra voltage to compensate for transmission losses (30v maybe?) down to the ROV?

Regarding ROV current requirements (cold start and various power draw) would it be wise to add a capacitor connected to the output of a 120/240v -> 12v converter (4A or higher rated converter, considering the openROV's continuous draw is 3A at full load) to assist in delivering high current during motor startup? What size and rating would be adequate for such a capacitor if it is necessary?



The ROVs I work with use 150DC because most universal input AC to DC supplies take it and it does not cause hum in long analog video signals, also Tecnadyne makes nice thrusters for 150DC.

You want some capacitance in the ROV to provide local surge capability. You will find if you put capacitors on the 150V line instead of the 12V line you can store more energy in less bulk. Capacitor size is proportional to C * V while energy is C * V ^ 2, so 10 times the voltage is 10 times the size but 100 times the energy. Just make sure the 12V converter can handle to max pulse load. You may end up with some capacitance each side.


For those of us sticking with the tether for now (lots of wireless ideas - keep em coming!), I think it is important to consider not only the buoyancy or data transmission characteristics, but the actual strength of the tether itself. Having a tether strong enough to “reel in” a dead ROV is a huge safeguard.


Yes, you are right. I work in a subsea cable manufacturer and the tethers we make colud be two applications: the first one is to transmit signal and/od power to the ROV. The second one is to be able to recover the ROV like a rope, if necessary.


Yeah- Philip, you make a good point. Although when the ROV only weighs a few kilograms, that strength doesn't need to be particularly large to be effective. a single twisted pair of 26awg wire is almost impossible to break with your bare hands. To all- any thoughts on a general rule for how strong a tether aut to be? In my opinion, twice the weight of the ROV in air is more then enough. Thoughts?