Power over... fiber?

[daqq]

I thought about this some time ago, I just wasn't aware of it being used... and apparently quite popular in niche applications.

Pretty much what you'd expect - shine a laser through a optical fiber, convert it back into electricity at the other end and you can get some fairly high power transmitted over long distances, with practically perfect isolation between the power source and the target device.

Another advantage would be that no noise gets picked up by the 'cables', nor is any noise transmitted back - which is great for high voltage switching.

Now, the efficiency is not particularly amazing, but there are applications that are OK with it.

Relevant links:
https://www.photonics.com/Articles/The_Power_of_Light_Photonic_Power_Innovations_in/a25278
http://www.mhgopower.com/mhmain.html

[Whales]

First article references a figure of 5W.  Hehehe.

Power over fibre is arguably more dangerous than electricity.  You don't need close proximity contact to be injured, and even at 5W it's dangerous (compared to eg 12V 0.5A over wires).  There are some clever ways you can detect breaches (reading reflectance, etc) but they're not guaranteed to work in all situations (optic splitters, fibres getting bent too sharply, etc).

I've always dreamed of a house full of optic fibres to each powerpoint.  2400W or more of power available at each port, tens of kilowatts reflecting in and out of the house's fibres and the grid.   Just don't accidentally nick a cable, the whole room will become a lightshow of fire and you'll need to call the LSES.

[daqq]

I don't think anyone's considering it for anything else than places where it's very advantageous.

The target applications are remote monitoring, high voltage monitoring/switching, low noise measurements. And certainly nothing close to kW.

When you want an isolation voltage of, say, 400kV, is there any other reasonable way to get it?

[Whales]

I don't like the way the article tries to sell it.

• Lightning-proof and spark-free performance based on the nonconductive nature of fiber. This feature is critical in many industries that are subjected to severe outdoor conditions, such as cellular base stations and electrical power grids, or in applications that must avoid the potential for spark-ignited explosions, such as the monitoring of fuel gauges and cabin door interlocks in aircraft.

A damaged or cut 5W fibre is a massive ignition hazard.  There's no way I'd put this in an aircraft or anywhere near flammables.

• Increased human safety based on the elimination of cable heating, which is prevalent in copper-based systems. This is of particular benefit in applications such as magnetic resonance imaging, in which heating can create a serious risk of burns if the cabling comes in contact with the patient.

Fixing one hazard and creating another is not a good solution.


If you ever use anything like this it will either need:

(1) Some really clever fail-safe method of detecting light leaks.  Optical leak detection lanes in the fibre jackets, reflectance monitoring, etc.

(2) Really low power levels.   In which case you skip this style of high-power product and even the electrical conversion step by using reflectance-style passive sensors.

This is an already existing technology.  One method: small LCD changes the amount of light being reflected back down the fibre, its opacity gets modulated by a local sensor (big thermocouples, etc).

[NiHaoMike]

When you want an isolation voltage of, say, 400kV, is there any other reasonable way to get it?

Motor spinning a generator via a long plastic rod.

[abraxa]

My understanding is that a light beam coming out of a fiber will diverge very quickly and not remain focused for any significant distance. This way, the ignition/fire hazard should be minimal. Is my understanding wrong?

[Kleinstein]

The light out of the fiber would be rather diverging (e.g. similar to a 30 degree LED). However if the open end directly touches something the heat would still be rather local and a possible fire hazard.  However 5 W would already be quite a lot of power - but still 0.5 W local on a 0.5 mm spot can get quite hot. The very high power likely not be a very thin fiber.

Another way for power over very high insulation is ultrasound through something like a rod of glass. This also known as an acoustic transformer.
Not sure how well an optical fiber is suited for this. 

There are also "conventional" type transformers with thicker insulation: I have used one with a rather large torrid core (e.g. 30 cm diameter) and only coils on 2 opposing ends, that was good for some 100 kV if I remember right. (I used it only at 10 kV).

[Siwastaja]

My understanding is that a light beam coming out of a fiber will diverge very quickly and not remain focused for any significant distance. This way, the ignition/fire hazard should be minimal. Is my understanding wrong?

>10W (for 5W delivery, assuming <50% receiving cell efficiency) is still a huge power density if the cut fiber touches or gets close to a flammable surface within a few millimeters. Think about a soldering iron tip.

I don't see why this couldn't work with a fairly simple optical communication, though. Have the power receiver communicate back the actual power measurement it gets, and compare it at the source, to calculate loss. Communication can be worked out by, for example, feeding different wavelength back on the same fiber.

[chris_leyson]

Looked into this some time ago in order to isolate machine vision sensors used in robot welders. The sensor needed 10W so it was immediately obvious that power over fiber wasn't going to happen. PoE would have provided some isolation but not 10's of kV as in the case where an arc was struck between the welder and sensor. A hybrid solution would have fixed things, supply DC for the power and transformer isolate perhaps using a resonant topology and use fiber for the high speed data links. Fiber won't transfer any sort of reasonable power but is good for high speed data whereas a transformer with 20kV isolation will transfer power but isn't much use a Gbit data.
You might get 100mW of power out of a fiber but it's still going to be class 4 if you are using a laser.

[Twoflower]

If you exceed a certain power density you must make sure that you get the power out of the fiber correctly (like termination on RF lines). If wrongly done, the connector is removed while the laser is active or the fiber gets a defect the reflection can cause a effect called fiber fuse. The fiber will melt at the point this will destroy the fiber further and the defect moves closer to the laser. Rinse and repeat. Depending on the energy that can happen extremely fast.

[In Vacuo Veritas]

page 6
https://download.tek.com/document/51W-60485-0%2520IsoVu%2520White%2520Paper%2520%2520TN%25203-25-16.pdf

[NiHaoMike]

If you exceed a certain power density you must make sure that you get the power out of the fiber correctly (like termination on RF lines). If wrongly done, the connector is removed while the laser is active or the fiber gets a defect the reflection can cause a effect called fiber fuse. The fiber will melt at the point this will destroy the fiber further and the defect moves closer to the laser. Rinse and repeat. Depending on the energy that can happen extremely fast.

As seen just after 12:30 in this video:

[Twoflower]

As seen just after 12:30 in this video:

Much better visible. Including a microscopic slow motion view of the fiber while the plasma is running through it:
https://www.youtube.com/watch?time_continue=98&v=BVmIgaafERk

[GeorgeOfTheJungle]

The spot of a loupe of 8 cm (Ø) under the sun has ~ exactly 5W: 1e3[W/m²]*3.1415*0.04[m]*0.04[m]= 5 [W] and we all know what that can do :-)

[Whales]

I don't see why this couldn't work with a fairly simple optical communication, though. Have the power receiver communicate back the actual power measurement it gets, and compare it at the source, to calculate loss. Communication can be worked out by, for example, feeding different wavelength back on the same fiber.

That's a good idea.  Any losses whatsoever can be measured.  You'll have to make sure your implementation is rock-solid and that each end is calibrated, so it would be expensive, but I think it's a pretty perfect solution if you absolutely have to go optic.

When you want an isolation voltage of, say, 400kV, is there any other reasonable way to get it?

Motor spinning a generator via a long plastic rod.

I love this.  You could probably even contain a shorter system in an inert/controlled gas chamber with lifetime lube + redundant motor/gen pairs and sell it as a sealed product.  Make it look like a HV bushing.