Drones will fly for days with new photovoltaic engine | Hacker News Reader

Drones will fly for days with new photovoltaic engine

techxplore.com

256 pointsdnetesn95 comments

"For the past 15 years, the efficiency of converting heat into electricity with thermovoltaics has been stalled at 23 percent. But a groundbreaking physical insight has allowed researchers to raise this efficiency to 29 percent. Using a novel design, the researchers are now aiming to reach 50 percent efficiency in the near future by applying well-established scientific concepts."

^interesting part, the drone stuff is just click-bait

I wonder if this would be enough to create the ultimate flex-fuel vehicle?

Does this mean we can make lighter/smaller RTGs?

left out the actual science part of this interesting part:

> According to Yablonovitch, this finding builds on work that he and students published in 2011, which found that the key to boosting solar cell efficiency was not by absorbing more photons (light) but emitting them. By adding a highly reflective mirror on the back of a photovoltaic cell, they broke efficiency records at the time and have continued to do so with subsequent research.

Perhaps a stupid question:

Is this simply a matter of getting a second chance to capture the photon (by the same mechanism) as it makes a second pass through the cell after reflection?

We have multijunction cells with 47.1% efficiency (as of June 20th).

Other than the price, what's the significance of single-junction 50% efficiency?

https://en.wikipedia.org/wiki/File:CellPVeff(rev190702).pdf

The solar panel market is dominated by cheap silicon wafer based production from China. This research involves uses of higher efficiency PV in other markets. Thermophotovoltaics have applications in hybrid cars, drones, and cogeneration of electricity. In the conclusion of this talk (https://www.youtube.com/watch?v=lDxJsa8miNQ&feature=youtu.be...) Eli Yablonovitch says the automotive market is ten times larger than the solar panel market and is something that higher efficiency PV can target in the near term. Drones are a niche market that Alta Devices is already working in.

I was unclear how this differed from what I would traditionally consider as solar power (photovoltaic power) as I had never heard of thermophotovoltaic power. After a quick search I found the following helpful:

“The basic principle [of thermophotovoltaics] is similar to that of traditional photovoltaics (PV) where a p-n junction is used to absorb optical energy, generate and separate electron/hole pairs, and in doing so convert that energy into electricity. The difference is that the optical energy is not directly generated by the Sun, but instead by a material at high temperature (termed the emitter), that causes it to emit light. In this way thermal energy is converted to electrical energy.” [1]

[1] https://en.m.wikipedia.org/wiki/Thermophotovoltaic

Basically a highly-inefficient method of converting energy. I've looked at the article and the pictures, and what I see is basically slapping a solar cell with a gold mirror backing right near a hot ribbon of graphite - eg we're trying to use highly inefficient incandescence to generate light.

Want to know what works better? Just the simple mix of gold into the p-n junction matrix and finger lines. Takes a silicon solar cell up to about 35% efficiency when properly doped into the junction material.

This technology? Well, given incandescence isn't going to break ~30% efficiency (excepting in the IR-glass-sandwiched tungsten filament) you're literally just wasting energy. Best we've seen in general incan is ~12% efficiency, so enjoy your 30-ish percent of that (4%).

You're better off with other technologies. This is neat, but you're looking at pretty much over-simulating the conditions of a solar cell welding line, which means that solar cell is going to likely bust over time under the heat stress. I foresee tons of problems with this tech, from the incan ribbon breaking to the cell harvesting energy breaking from the heat/cool cycling.

People seem to be misinterpreting this a lot.

This is not about solar cells. This is about thermophotovoltalics, and the application here is using a burning fuel source to generate electricity.

The "drone" the paper talks about seems to be a reference to another paper on unmanned sea vehicles though. Edit - to be fair the flying drone part is from the corresponding author.

> This is not about solar cells.

There's wasted thermal energy when using solar cells. If you can capture some of it you'd be able to increase the efficiency of the cell.

Exactly. The paper is here: https://www.pnas.org/content/early/2019/07/15/1903001116

The context of thermophotovoltalics is very interesting - conventional heat-to-electricity engines are limited by Carnot efficiency, which relies on a "cold side" and generally has low theoretical maximum efficiencies. This approach takes any hot emitter heated by any process - fuel burning, solar concentration, nuclear - and extracts electricity directly from the mostly IR photons it emits.

> Here, we present experimental results on a thermophotovoltaic cell with 29.1 ± 0.4% power conversion efficiency at an emitter temperature of 1,207 °C. This is a record for thermophotovoltaic efficiency.

I sense a new kind of nuclear power plant?!

That would be amazing to get rid of the traditional boiling water part of nuclear power plants. Is this what you are thinking about?

Yes, and especially for my pet daydream nuclear reactor in space, or on Mars, etc.

Too bad, solar cell don't like much radiation. Add shielding and all mass benefit will be negated.

What about collecting the "waste" IR emissions from the heated water vapor?

No, the whole trick with thermal PV is that you harvest the entire range from thermal infrared to near. A vapour will only glow so so in lower reaches of thermal IR range.

So (one of) the advantages compared to a Peltier element is that you no longer need a cold side?

You still need to keep the thermophotovoltaic cells cool, it doesn't violate thermodynamics (energy can only be extracted from a temperature differential).

While it looks like you still do, the other advantage is that these are far more efficient than Peltier devices, which tend to be single-digit percentage efficiency.

Isn't Carnot efficiency higher than 29%?

Photovoltaic’s are also limited by Carnot efficiency.

The theory is you can increase the hot side more as you are less limited by mechanical stresses and moving parts. However, the sun’s surface is 5,505°C, so these are always going to have lower efficiency than traditional solar panels. Which are still less efficient than the bets thermal engines.

THERMOphotovoltaics are limited by Carnot's efficiency. This does not apply to basic pure-photon non-thermal photovoltaics.

This also relies on a "cold side" - the thermophotovoltaic cell has to be colder than the IR source or it will emit photons faster than it can absorb them.

> "What the mirror does is create a dense infrared luminescent photon gas within the solar cell, a phenomenon that adds voltage," said Yablonovitch

> Recently, his team recognized that this mirror could serve double duty. In fact, it solves one of the biggest challenges in thermophotovoltaics: how to exploit the thermal (heat) photons that have too little energy to produce electricity. It turns out that the mirror can reflect those small photons to reheat the thermal source, providing a second chance for a high energy photon to be created and generate electricity. This phenomenon leads to unprecedented efficiency.

Given these paragraphs and the name "thermophotovoltalics", that this is about solar power seems like a reasonable conclusion to draw. Also, googling around I am seeing figures for steam turbine efficiency of between 40 and 80%. How sure are you this is not about solar cells?

Your quotes seem to very clearly support the idea its a an onboard heat source as claimed by gp. Specifically

>>the mirror can reflect those small photons to reheat the thermal source

I'm pretty sure holding a mirror up to the sun is not going to change the sun's temperature much.

> I am seeing figures for steam turbine efficiency of between 40 and 80%

There's more that goes into choosing an engine for a drone than just efficiency. Particularly important would be power/weight. A Carnot engine would get much higher efficiency for sure, but would not put out nearly enough power per weight for a drone to fly. (It's so low power we don't even use it for cars!)

> Given these paragraphs and the name "thermophotovoltalics", that this is about solar power seems like a reasonable conclusion to draw. Also, googling around I am seeing figures for steam turbine efficiency of between 40 and 80%. How sure are you this is not about solar cells?

That it's not about the solar cells people in the comments mean? Very, the paper and citations talk about furnaces, and the temperature this operates at is 1200C. This isn't something you'd stick on your roof.

Why couldn't this be used in cars -- 50% seems better than traditional internal combustion engines?

Hybrid cars are potentially their biggest market. Ten times solar PV. https://www.youtube.com/watch?v=lDxJsa8miNQ&feature=youtu.be...

Surprisingly never heard of Thermophotovoltaics.

Here's a simple diagram https://proxy.duckduckgo.com/iu/?u=http%3A%2F%2Fanilyuksel.f...

It seems like 'impedance matched' photovoltaics.. very interesting.

Hah, now I have a vision of one of these things hooked up to every air conditioner in the world to claw back some of the electricity that the air conditioner uses. I imagine the heat would need to be a bit more than what an air conditioner generates though, but I know nothing about the topic, so who knows.

It reads a bit overhyped, considering they claim to power a house with a generator the size of an envelope; that's a hard no since there is a limit on how much energy hits every square meter and it's not THAT much.

To elaborate, solar energy hitting the Earth’s surface is about a kilowatt/m^2 if conditions are perfect. If you were able to convert solar power at the maximum theoretical PV efficiency (the Shockley-Queisser limit, 33.7%), you’d need about 3 square meters to just to make toast.

For those confused by that oddly low theoretical limit: it applies to single junction cells. Panels that achieve higher efficiencies do it by stacking multiple cells.

Your broad point is right, but the SQ limit only applies to conventional PV cells, it's not a theoretical limit on PV efficiency and in fact has been exceeded by other cell architectures.

Is this supposed to sound bad or good? Well, 3m² to make toast intuitively sounds bad I guess. Well, consider this: It's enough to run 2 small office PCs with all peripherals. Considering that it sounds much more positive. Really just anything producing heat would sound bad because of power dissipation. A toaster is probably one of the most extreme ... (Imagine putting 1kW into a slice of bread without power dissipation!)

Someone else wrote the comment I meant to, essentially doing the math to show that one would need an awkwardly large solar panel to power something like a quadrotor drone.

The idea behind individual solar energy is not consuming it real time but associating it with a storage solution to allow bigger throughput. You're not going to make toasts all day

>You're not going to make toasts all day

I bet you're doing something throughout the day. Maybe not making toast, but cooking, ironing, heating, cooling, watching TV etc. I'm not sure you can rely on any excess energy being available for storage if you're not connected to the grid.

>The idea behind individual solar energy is not consuming it real time but associating it with a storage solution to allow bigger throughput.

If that's the idea then you are exposing a major issue with solar. Namely: in order to rely solely on solar (+battery), you need a solar deployment that covers your energy needs now + energy to charge battery for when the sun isn't shining (taking into account the associated loss of storing and retrieving power from battery). This means that you need to oversubscribe/over capitalize solar to charge up the battery (which then lowers your total efficiency) in order to bridge daily and seasonal variability in solar output. This means that you need generation capacity that probably exceeds the surface area available to what a typical house or apartment can provide and explodes your costs.

This has major implications for large scale solar+battery deployments because at that scale, there is no grid to fall back on (which is why solar ALWAYS needs reliable backup generation - which is typically gas or biofuels). This is why solar+battery is never going to be cost-effective because you're always going to be forced to over-build infrastructure that will sit idly doing nothing most of the time.

This doesn't even touch on the fact that there is no actual battery technology that keep enough load to power a modern city overnight, much less to bridge seasonal variability at that scale (where the battery would be expected to keep enough energy for weeks at a time).

“Forcing over building of” solar is still cheaper than building a new coal or gas plant in some areas. I am positive there are plenty of people who live off grid and do not always need gas or biofuel backup.

>there are plenty of people who live off grid and do not always need gas or biofuel backup.

I'm sure there are plenty of such people, but those people don't really factor when you take into account the society as a whole. The vast majority of people and businesses cannot live off grid.

Absolutely agree and vast majority are not economically justified to live off grid in North America. Parent comment used all caps so just pointing out it is not absolute.

Yes, the max. energy from sunlight on earth hitting an envelope should be around 20-30 Watts (at 100% efficiency).

Also, the reported 29.1% efficiency have been reached at a temperature of 1,207 °C.

This isn't about sunlight, but about putting this with a furnace or something else generating heat.

Yes, the original article is about thermophotovoltaics.

Regarding the article posted here (and since a drone with a furnace is not feasible) I went with solar.

The corresponding author to be fair did say unmanned aerial vehicles.

True, though it is combined (in the article) with a maximum power consumption: "those that require as little as 100 watts, [such as] a lightweight unmanned aerial vehicle"

The original paper says "unmanned vehicles" (no aerial) and has a reference [0] to "unmanned undersea vehicles".

[0] https://www.pnas.org/content/early/2019/07/15/1903001116#ref...

The article says "a wide range of applications, from those that require as little as 100 watts, [such as] a lightweight unmanned aerial vehicle, to 100 megawatts"

I read the 100 watts as referring to lightweight unmanned aerial vehicle.

100 megawatts is followed by "[providing] electricity for 36,000 homes."

The original paper does not mention lightweight aerial vehicles as an application, but a drone is mentioned in the title (and shown in the top image) of the article.

Applications mentioned in the paper are: hybrid cars, unmanned vehicles, deep-space probes, energy storage, enabling efficient cogeneration systems for heat and electricity.

(A less-misleading title might even be: "New thermophotovoltaic engine for deep-space probes")

How about one of the authors talking about a credit-card sized device supplying 100 watts to quad-coptors using liquid fuel and staying airborn for 16+ hours. (https://youtu.be/lDxJsa8miNQ?t=3202)

Is this something you could use a laser for too? To bring heat to the flying drone? Or would that be super inefficient?

Beamed energy always has attenuation problems that make it inefficient.

Also, the reported 29.1% efficiency have been reached at a temperature of 1,207 °C.

Another 12% and they can compete with a standard steam turbogenerator.

Wouldn't you slap this on top of a steam boiler to get the best of both worlds, though?

Steam turbines have row after row of blades, each successive blade wheel larger and lower pressure than the last. Exit temperatures from modern turbogenerators are around 120° C, where the steam is about to turn into water, and pressure is usually below atmospheric. Almost all the energy which thermodynamics permits to be extracted has been extracted.

I probably should have phrased it better - I was picturing [Furnace] | [Thermophotovoltaics] | [Boiler], so the PV part gets first crack at the light/IR from the furnace and then the boiler cools the PV gear as it boils the water. This is also how it's suggested to be used in a paper linked elsewhere in the discussion above: https://www.sciencedirect.com/science/article/pii/S030626191...

If you make it harder to cool the turbine by putting those neat the radiator, efficiency will go down; if you make it harder to keep the heater hot, by putting those near the fire, efficiency will go down.

As animats already said, turbine efficiency on power plants is very near the theoretical optimum. The only way left to improve it is by increasing the temperature.

The furnace burns at over 900C but the superheated steam (which is your ‘hot’ temperature for your heat engine) is only at ~400C, you might lose some power output but you won’t be losing efficiency.

Perhaps for small scale applications this could be cheaper and more reliable? (if they can also get to 50% and if the materials used are cheap enough)

That is talking about heat and power cogeneration, not solar power if that's what you're thinking. The paper they site [0] says the density can 300-400kW/m^2 and so at 300kW, 50% efficiency and a small C5 envelope you're looking at over 5kW.

[0] https://scholar.google.com/scholar_lookup?author=M.+Bianchi&...

Solid-state cogen that is simple to include in heating furnaces would be very interesting for old stock high density residential in cold regions. Mire modern heating options than gas furnaces (various variations of heat pumps, usually coupled with solar) don't scale down to individual flats and anything more central (ideally district heating) is effectively impossible to retrofit in a free market economy. But the very high efficiency electricity conversion component of cogen is just too good to skip, so a smallest-scale cogen option would be a welcome addition to the energy mix.

The device is mentioned in this talk (https://youtu.be/lDxJsa8miNQ?t=3105) and provides 1kW of electricity and about 1kW worth of "waste" heat used for hot water. The key idea is that the amount of light would be the equivalent of 280 suns so you don't need a large panel.

Terrible article. "Scientists have invented a new thing which could (i.e. doesn't yet) power drones, and houses and space ships and cars, and computers and ..... (10 paragraphs later). Oh btw it works by adding a mirror or something. The end."

Researchers from Rice University have an approach they think can get to 80% efficiency using a carbon nanotube film:

https://news.rice.edu/2019/07/12/rice-device-channels-heat-i...

I really dislike headlines like this on HN that imply that lab research mixed with theory are somehow practical. It would be great if there were a "Theoretically" in there along with removing the hype-included "drone" reference.

Eli Yablonovitch's company, Alta Devices, manufactures a thin film high efficiency PV cell but they can't compete right now against cheap Chinese silicon wafer cells in the solar panel market, so they are focusing on niche markets, primarily drones at first.

His 2017 talk to the MIT Energy Initiative (https://www.youtube.com/watch?v=u_K1URyarE0) goes into the economics of solar and why his company is using this strategy. Basically, they need to find niche markets until they can work their way down the learning curve and achieve economies of scale.

He also discusses thermophotovoltaics. Very efficient light-weight PV allows for longer drone flights but night is an obstacle to multi-day flights. Given the choice between batteries and fuel to keep aloft until dawn this does seem like a possible breakthrough, deserving of the article's title, although the text could be a lot more clear.

https://www.youtube.com/watch?v=lDxJsa8miNQ has a lot more information on thermophotovoltaics, including drone applications, and some more info on solar panel economics.

For the record, to sustain a modern small drone at 50% conversion, you still need at least a 50cm x 50cm panel with optimistic estimates. This is completely ignoring the fact that such panels would need to be mounted without destroying the aerodynamics of the propellers

So what does this mean, we will be able to have drones flying continuously and getting energy from the sun?

There's been an efficiency break through for solar cells.

> ... an achievement that could lead to an ultralight engine that can power drones for days.

No such thing exists yet, but that's one of the hype ideas being thrown out there to try and get lay people excited about the breakthrough.

DJI Phantom 3 is 0.3m wide and needs 120W to fly. Solar radiation at surface is 1000W/m2. 0.3m0.3m1000 = 90W at 100% efficiency. They claim 29% eficiency* (compared to regular 23%) which is only 26W out of 120W in ideal circumstances, realistically it would be maybe 1/3 of it. It will not fly.

You can safely ignore ALL articles about breakthroughs in solar cell efficiency and look only at what is sold in shops because that's the only thing that matters. For years this has been 21-23% despite 5 or more articles about solar panel efficiency breakthrough every month for the last 20 years.

It's the same as the "OMG this new battery will power your phone for years!!!" bullshit articles that come around every six months.

I essentially second everything that the person above is saying.

There are an endless number of press releases and public relations bullshit articles put out by people who have some new "breakthrough" in photovoltaics. Whether it's special weird cells, or flower shaped ground mount things, or whatever.

What I believe in, is what I can pull out my visa card and buy right now. And at the moment, here's what that looks like:

One pallet load (22 panels), of high quality 156mm monocrystalline silicon cells, assembled into a 1.99 x 0.99 meter sized panel. Rated at 370W STC (standard test conditions) per panel. Under $0.60/watt.

STC: https://www.altestore.com/blog/2016/04/how-do-i-read-specifi...

Everything else is either so high priced that you have to contact a sales person to buy it (weird fresnel lens concentrator and triple junction GaAs cells intended for use on spacecraft), or is not manufactured in sufficient quantity to gain even 1% of market share, and therefore is not stocked by major photovoltaic equipment dealers.

The article is not about solar cells though. It's about thermophotovoltalics, where you'd use a high temperature heat source.

Quadcopter "drones" are much less efficient than winged plane "drones". Additionally, they have the large surface area of the wings to collect more solar energy.

Though the image does show a consumer quadcopter device...

Yeah the Airbus Zephyr solar powered drone was up for 25 days.

No, the article is about efficiently converting heat to electricity, it's not about solar panels.

We already have drones with wings that are capable of that, or close to it. This would make them more efficient.

https://www.youtube.com/watch?v=8m4_NpTQn0E

For quadcopters and similar I doubt it will be enough.

I’ve seen reports of thin film organic solar cells that produce >1 watts per gram of material. A quadcopter motor can generate multiple grams per watt of power. You can today go and buy 100W from a 3.3lb panel, which is like half way there and not even trying. I’m guessing it’s probably possible to make a quadcopter that can hover indefinitely in the sun.

No, not really. The math for watts consumed in total just to hover, for a very large quad, hex or octocopter (with 24 to 28 inch size props, so optimal grams/watt thrust efficiency for something you can fit in a van), doesn't work out.

The weight and size of a 400W solar panel made from the very best cells you can buy commercially today is considerably more than the wattage you would need to hover a 10+ kilogram coaxial octocopter capable of carrying such a large panel on top of it.

For a large octo, a single Tmotor U8 motor with a 28 inch prop requires anywhere from 310W to 850W while in flight, and there's eight motors + eight props on a coaxial medium/heavy lift octo.

I showed this article to a few people in the professional arducopter/ardupilot/arduplane community and it has been quite thoroughly chuckled at.

For reference, here's the datasheet for a prototypical motor, a tmotor u8 170kv with 28 inch prop. You'll see the table for watts consumed and grams/watt efficiency with motor+prop mounted on a thrust stand.

http://store-en.tmotor.com/goods.php?id=324

If you wanted to put photovoltaics on a quad/hex/octocopter, you have multiple issues with not enough top deck surface area, not being able to introduce a lot of "sail area" (something big and light that will be grabbed by the wind) sticking out of the top, and the weight of the pv itself.

There is no PV solution anywhere near small enough to fit on the top deck of a really big commercial coaxial octo that is also capable of producing 2000W. Not even 100W in sunlight. Somewhere from 18 to 30 times less watts than is required just to hover in one place.

Hey! I love this problem :) The motor you shared shows >20g/W efficiency at ~40W of power and net lift of 500grams with those 28x9.2 (100g CF props). This implies that we need to produce roughly 40 Watts with under 500grams -- hard to assume negligible structural/electronics mass so more conservatively 40 Watts under 250g or 320Watts under 2kg to make like an octocopter with some power to spare. this is roughly 1 to 4 sq meters of solar area at 32% to 8% efficiency, and the swept area of eight 28" props is about 12 square meters it's might not even look totally ridiculous!

I can find super-light 100Watt panes under 2kg on Amazon (probably actually 60 Watts) but it's definitely more like 2 - 4 times more mass efficiency than what is commercially available and not 18-30.

Silicon mass is under 4 grams per Watt of solar so I think it should be possible to get there with silicon with a focused engineering effort. Organic photo-voltaic materials have shown 6 watts per gram in the lab (imagine for space applications!) -- though they would need even larger area due to lower efficiency and act as more of a sail .... not saying it is a practical machine to make, but again I definitely think it is in the realm of possible.

It doesn't scale down like that. Small motors and small props under 7 inch size have less than 5g/watt thrust efficiency. If you look at the motor and propeller setup that would go on an Armattan Chameleon Ti 7" for instance (a freestyle/gopro hero7 carrying sized FPV quad frame you add your own parts to).

On the large scale you would never actually be applying only 40W to a tmotor u8 with a 28" prop on it, because the total weight of the motors and the whole craft will mean it would just sit on the ground. You'd be moving air around like a fan and not lifting off. Ideally you would design a large octo for a 1:1 thrust ratio (hover) at about 48 to 50% throttle per motor, which is why you see the thrust tables starting from 50% and going upwards to 100% throttle.

Bigger and slower (lower kV) props and motors have higher efficiencies. If you google a bit about the few successful human powered helicopters that have been built in the past ten years you'll see the ultimate extent of using very big, very slow props to maximize lift per watt.

I would dearly love to be proven wrong on my above post. If you can find me a super efficiency triple junction GaAs panel to mount on top of a 20 kilogram sized coaxial filming octo, I'll find a way to get it into the hands of XM2 and see if they want to try flying it.

I've got an immediate idea reference from the movie Interstellar, which shows a drone that supposed to be flying non stop around the globe for X number of years until it had its navigation system hacked by people and then landed.

"Fly for Days", Wondering what are the trade-offs here? Can those drones carry a load or cover certain distances to be of any functional use.

Surveillance, spying, communications. Such drones could encroach a lot into roles currently assigned to satellites.

I wonder if it is feasible to make PV panels that concentrate the IR part of the light spectrum and guide it to a TVP unit through fiber optics.

I don't see why you'd need fibre optics for that. Mirrors would work just fine, as in a concentrated solar plant.

I'm having trouble envisaging how this would work?

What might work though is what I'm going to call Pink Floyd Concentrated Solar: collimate light from the sun over a large area into a beam, then run that beam through a prism. That would allow the use of solar cells with different efficiency bands.

(I doubt this would make overall sense given the cost and size of optics, but it's fun to think about)

Just increasing the efficiency to 30pct is a pretty big deal. If these could be mass produced cheaply to market, that would be incredible.

In 6th grade I connected a 9V battery to my mechanical pencil lead and lit my science folder on fire. Basically the same thing...

Not really, as it didn't use any energy not present in the 9v battery. This is about using solar energy in a more efficient arrangement, through reflection of waste solar energy to heat a medium and create a higher efficiency thermo-voltaic cell based on a photo-voltaic cell (I'm probably very wrong with my terminology in the last sentence, but, either way, it ain't no burning graphite connected to a battery.)

So did they show an actual drone flying for days?

No, but they could... someday... in the future!

I heard the raw cocaine price of a kg dropped by 20k last 10 years. I bet they are already flying.

Reply on news.ycombinator.com