PoE and ROV power management


#1

I recently posted on the subject of using power over Ethernet (PoE) thru the tether as a reply in the Open ROV electronics discussion. Since this topic will need research and discussion on it's own, separate from the workings of the on-board electronics, I will start a topic here. For convenience I will repeat the posting below:

At the June 9, meet up of OpenROV at Noisebridge in SF we discussed alternatives for sending power over the tether. Someone (forgot name – sorry) mentioned power over Ethernet (PoE) and we investigated that approach. See reference article here ( http://www.bb-elec.com/tech_articles/Power_Over_Ethernet.asp and http://en.wikipedia.org/wiki/Power_over_Ethernet

).

Basically you can inject 48VDC into one end of the Ethernet cable and split out regulated DC (12V, 9V or 5V) drawing 12Watts or 30Watts depending on which version of the 802.3 standard your PoE devices support (802.3af or 802.3at).

It would also eliminate the need for the Baluns (twisted pair adaptors - http://en.wikipedia.org/wiki/Balun ) and possibly one of the voltage regulators if this works out.

I purchased an inexpensive ($25) PoE kit (injector and splitter pair) that delivers 12W and supports 10/100Mbps (TP-Link PoE Adapter Kit Model No.TL-POE200).

I did a simple test and powered the BeagleBone with the 5V output from the splitter.

There is more work to do here. PoE likes to have 2 pair of UTP wires. I haven’t yet experimented with choices of cable (there are CAT 5 rated 2PR cables from Belden – 72001E Indus Cat5e 2PR AWG24 PVC).

Also I want to see if this can work with a lower voltage at the injector side (say 24V – from 2 small 12V batteries in series).

On the Splitter side, we may be able to adapt a standard unit (or buy the required center tap transformer) and step the 24V down to 12V (DC-DC converter) for the motors and draw a 5V charge for the electronics. We also discussed using a large capacitor (MegaCap) vs a 9V battery for reserve power for the electronics.

The assumption is that batteries are still needed for the motors since surge power would be more than 30W (12V@ 2.5A), but average power may be less than 30W. If we start with charged batteries on the ROV and add a trickle charge circuit from the PoE source, we could have extended missions for the ROV based on the amount of energy available on the surface.

Will keep folks posted as progress is made and if any one wants to pursue these ideas I would appreciate any help.

Jim


#2

I am next investigating rechargeable batteries for the ROV and a simple charging circuit for on-board. Here are some notes/references and comments:

NiMH batteries (1.2V) are reasonably low cost, but don’t hold the energy of lithium ion (more expensive and different voltage, 3.7V – see http://www.electronicsonsales.com/protected-25500-37v-3500mah-rechargeable-lithium-battery-p-324.html )

So first we look at NiMH

Battery Chart

http://www.powerstream.com/NiMH.htm

Tenergy C Size 5000mAh High Capacity High Rate NiMH – 8Pcs $24

http://www.google.com/products/catalog?q=nimh+c+batteries&hl=en&prmd=imvns&bav=on.2,or.r_gc.r_pw.r_qf.,cf.osb&biw=1143&bih=711&um=1&ie=UTF-8&tbm=shop&cid=5202098000679390966&sa=X&ei=PNjkT5nfK6H20gHKkpWHCg&ved=0CKMBEPMCMAA

Charging NiMH batteries

http://www.powerstream.com/NiMH.htm

Try trickle charge at C/10 (where C is Ah capacity).

Ex. C = 5000 mAh. C/10=5A for trickle charge

For 8 batteries in series (output 9.6V @1.2V each).

Charge all in series in place in ROV - 18V@5A = 90W (likely more than PoE can supply)

If we don’t exceed the max trickle charge rate, then it should be safe to go with a simple charging circuit.

Low cost solution for charging of both NiCd and NiMh batteries

http://www.extremecircuits.net/2009/10/low-cost-universal-battery-charger.html

To really make proper design choices for the ROV power management, we will need a breakdown of power usage (voltage, current – both average and peak) for each key component (motors, lights, BeagleBone, etc.). I will try to get together with Eric to take measurements on his unit to get some baseline data.

We can also investigate lithium ion as well.

Jim


#3

Thanks for investigating this Jim,

I will be interested in powering a data logger and this would be the preferred option as opposed to running off of self-contained batteries (especially if other components start getting upgraded, i.e.. lights).


#4

Tested the PoE kit I acquired (TP-Link PoE Adapter Kit Model No.TL-POE200) and found it required 4 pair cable (did not conform to 802.3af standard which provides for 2 pair operation). Well back to Fry's for this baby. I will be researching other PoE equipment that supports 2PR or alternatively transformers used for PoE with center taps.


#5

I tried another pair of units for PoE and this time I was able to use 2PR cable (data pins: 1,2 (also V+) and 3,6 (also V-). This equipment cost a bit more than what I first tried (since it supports Gigabit speeds), but it did support a 2PR wire cable, which will allow me to go with a thinner tether.

The equipment was: TRENDnet Gigabit Power over Ethernet Injector (TPE-111GI) and Gigabit Power over Ethernet Splitter (TPE-112GS). It delivers power and data up to 100 meters (328 feet) and works with 2PR (4-wire) cable. Power In: 48VDC 0.3A. Power Out: 5V(2A), 7.5V(1.33A), 9V(1.11A), or 12V(0.86A) – approx 10Watts.

Prices at Fry’s were $46.99 for the injector (TPE-111GI) and $29.99 for the splitter (TPE-112GS). While I didn’t see these in the store, TRENDnet also has a less expensive pair of devices that only go to 10/100Mbps but deliver 15.4Watts (TPE-103I and TPE-104S). This sounds attractive since the BeagleBone only supports 10/100Mbps. It isn’t clear from the datasheet however if these units work with 2PR or not – may need to contact the manufacturer to confirm.

Now for the investigation into power usage to see what 10Watts plus local batteries can do.


#6

If the goal is to have the least linear weight tether line, an unshielded twisted pair is possible.

Instead of carrying power onto data line, the idea is to carry data over powerline: CPR

Ok, CPR homeplugs means Ethernet carried by 110v! The idea is not to dive a main extension cord in deep water, of course.

I think these CPR requires 110v AC, but perhaps it is possible to use lower voltage AC.

Raising the voltage, with a galvanic isolation, is a well-known technique used to distribute audio signal to loudspeakers (ages ago).

In this case, 2 wires are sufficient (twisted or not), perhaps a galvanic separation (2 small transformers) is needed, perhaps not.

When you raise voltage without amperage, there is no danger, it is like static electricity.

You also decrease the loss (Ohm), why not.

Ethernet -------->CPR device CPR device---------> Ethernet

AC supply--[Transfo]----------------------------PAIR------------------------[Transfo]----[5V-12v DC] --->

Trying to find acceptable AC line voltage on CPR IC datasheet.

About battery choice, I would evaluate a LiFePo4 cell.

I think this kind of cell has many advantages, one being to accept recharge while discharging.


#7

Thanks Dominique.

Eric and I talked about investigating CPR as an alternative to PoE using a lower AC voltage (say 48V - the same as PoE only AC). I think this would be a good parallel investigation and at the end we could compare the alternatives. I haven't checked into CPR much yet, but Eric said he had some equipment to try out. I think he mentioned it only worked up to 10MBs if I recall.

A far as your battery suggestion, I did see a C size (not exactly) on eBay. http://www.ebay.com/itm/4-High-Power-LiFePO4-26650-Cell-3-2V-2500-mAh-40A-Rate-8-0Wh-without-Tabs-/130716879131?pt=US_Rechargeable_Batteries&hash=item1e6f55491b

I guess we should have a battery discussion as well to compare choices and associated charging circuits.


#8

Jim,


CPR runs 200 MBs, (500MBs displayed) much better!

Have a look at this review-> 500 MBs CPR

And the Homeplug Alliance announced AV2 gigabyte.

With just 2 CPR endpoints, attached to just one device, the Ethernet will be collision free, so real end-to-end throughput can probably be more than sufficient.

CPR latency to check.

You did not mention LiPo battery type, but you probably discarded it already.


#9

Looks interesting - the challenge is to run on less than 110v powerline. I wouldn't think the COTS products would work at lower voltages (they would assume 110v is available for their use for onboard electronics - could be wrong here). The chipset that Atheros/Qualcomm sells, the AR7400, has an interesting datasheet and could be adapted to work on a lower voltage AC powerline. I didn't see any appnote on their site or technical data sheet. They could be contacted to see if they can share that type of info. It would be a bit of design work to build from scratch.


#10

In the AR7400 schematic

There is a "safety coupling", a transformer (galvanic isolation and voltage).

This means that AR7400 does not "see" 110v, you can choose any AC tension.

From here, it should be possible to hack a couple of them!


#11

Right - If you can open up a commercial product and modify its power supply circuit to work with the lower voltage you would be good.


#12

Hi. This is my 1st post and I just started on the ROV path. At this point I have only been reading about ROV building to get up to speed on the technology. I wanted to build one and did not want to reinvent the wheel but rather take input from great people like yourselfes who seem to be committed to the Open ROV project. The topic of powering the ROV is where my thoughts are at this time becuase it will determin what will hang off the other end of the tether. I see that you are chatting about Power over Ethernet and the limitaions that it brings so I wanted to ask about using 300V DC or lower to the ROV with a DC-DC Converter in the ROV to drop it down to 12V. The specs I have seen show 300V DC @ 1A through the cable can produce 12V DC @ 21A in the ROV so if less current in the ROV is required then that could drop the current through the cable and also reduce the Guage of wire. I know this probably does not fall within the scope or objectives of the Open ROV project because of the cost factor but wanted to know if has been discussed or if it is in the direction I should want to go. It seems to me that the current project requires 8 Batteries and will give a bottom time in the order of 1.5Hrs. and this could be costly over time. I have been looking at these DC-DC converters: http://www.powergatellc.com/300v-bus-dc-dc-converters.html

Any feed back +/- is welcome.


#13

Generally for ROV tethers you want to use the highest voltage you are comfortable with. Work class professional ROVs use voltages as high as 7500VAC 400Hz. The observation class ROVs I have worked with generally use 120VAC or 150 VDC at the bottom of the tether so that they can use common power supplies inside the ROV. Note the voltage at the top of the tether may be 300V or greater, losing half the power in resistance loss in the tether. Yes the tether may melt if it is left on a spool and not water cooled, been there, done that, very expensive lesson learned! This also requires feedback from the ROV to the tether supply to maintain regulation.

For adult hobbyists I think using 120VAC with a simple isolation transformer is quite reasonable. Then power-line Ethernet modems can be used to send data over the same two wires.


#14

Thanks Douglas for your feedback, I did not think of the issue with the coil on the spool and heat disipation, good point. From what you are saying it would be best to use isolated 120V ac and run the data on the same wire. Not having seen the Open ROV with all the electronics mounted in it, I am not sure how much room is available for the Power & maybe a Battery to help with the surge currents from the motors. I think I will start ordering up all the ROV Electronic parts need so I can do some bench testing of my own and then I can play with Power options. Maybe someone is already working on this or has come up with a good solution for power through the Tether. I look forward to more input on this topic. I assume that I am on the correct Blog for this topic.


#15

I've used Ethernet Over Powerline (IEEE 1901 "Homeplug") systems successfully, but the parts are expensive, take up a large amount of volume, and (this is what stopped me altogether) can be very dangerous if the right safety features are not installed. Such a system could be modified to use a lower amount of voltage, but an AC source would still be required. If these factors aren't an issue for you, and you understand electronics well enough to stay safe, it's the method I'd recommend.


#16

I just ran across this tiny, 24Volt POE injector (see attached pic) that might be useable...

http://www.cdw.ca/shop/products/EnGenius-EPE-1212-power-injector-power-adapter/2466047.aspx#TS

I'll see if I can source it locally.



#17

I have been dabbling a bit with some of this gear and am proposing our local school do a rob project...anyway, that's why I'm here now.

Keeping a simple and light weight tether is a good thing for many reasons and I agree that POe does fit the bill abeit a bit limited in the wattage you can pipe down a thin conductor. Basic resistance dictates if you're going for the longest tether possible on a spool that you go with the highest voltage so the current is lower and therefore the voltage loss kept to a minimum. Once the power is onboard the ROV it would have to be transformed to a lower voltage obviously. I'm currently considering something like this since it would facilitate trickling down power at a low wattage and allow for higher current draws to be dealt with via the battery.

http://www.adafruit.com/products/390

It would be nice to have something like this that could deal with a higher current and voltage than a cell phone...ideally something that could accept a 48V POe line and provide same battery charging and load management functionality.