Can You Power a Phone With a Capacitor? | Hacker News Reader

Can You Power a Phone With a Capacitor?

wired.com

56 pointsColinWright52 comments

Here's her actual submission [1]. She created a novel dielectric and achieved an energy density of 20.1 Wh/kg which is about half of that of a lead-acid battery, and something like 6%-10% of a Lithium Polymer battery. She lists a capacitance per volume, but nothing about breakdown voltages, so I can't say what the size of her capacitor would theoretically be for 1200F rated at 6V. Nor can I say if it'd even be feasible without knowing how difficult/expensive the dielectric is to manufacture.

Also, here's a page [2] with a list of common battery chemistries and ballpark figures for their energy density values (about halfway down in green). Newer Li Poly tech is somewhere in the 180Wh/kg range.

The sad thing about this is that we're missing a bigger point. Capacitors are useful because of their low internal impedance. They're able to dump (or receive) a tremendous amount of power incredibly quickly. It's useful when you have a circuit which needs a high-current burst which is powered from a low current (but high capacity) power source.

1: http://www.usc.edu/CSSF/Current/Projects/S0912.pdf

2: http://www.allaboutbatteries.com/Battery-Energy.html

It's useful when you have a circuit which needs a high-current burst which is powered from a low current (but high capacity) power source.

One application for this is lasers. I'm part of a team that works on a converter that uses wall power as an input to continuously charge a few dozens ultracaps then empties them all at once to fire very short pulses in the hundreds of kW.

What I find the most impressive about this is that the mechanical engineers have managed to fit this in a 2U rack.

Holy crap, what can you do with kW lasers besides instantly cut almost anything?

Popcorn?

kW isn't that high for pulsed lasers. Particularly since with Q-switching you can get very short pulses and mode-locking ridiculously short pulses. Total energy for a 1ps pulse at 100kW is negligible.

You get some nasty ionization effects with shorter pulses, right? I did some work trying to model a fs pulse laser in college.

Exactly. Since the output is not continuous, the average power remains quite low. This is why you can still power it from regular domestic power :)

A much cooler (but obsoleted by improving capacitor technology) was flywheel based pulsed energy: http://www.bevoware.com/research/cem/IEEE/PR%2071%20Werst%20...

Ah, I knew about these but never found an actual paper. Thanks for posting that!

I think it's much better to think of a capacitor as a spring:

* they store energy;

* they can release energy quickly;

* when put in parallel they get stronger;

* when put in series they get weaker;

* the RCL circuit differential equation is the same differential equation as the damped harmonic oscillator.

That is better given one is more comfortable with mechanical systems. Since my field is EE, I tend to do the opposite: translate a problem to the electrical domain to be more intuitive about it. You can use Bond graphs[1] to easily translate between different domains (Mechanical translation,Mechanical rotation, Electromagnetic, Hydraulic and Thermic in a more limited way). This technique generalises force and voltage to "Effort" and velocity and current to "Flow".

[1]:https://en.wikipedia.org/wiki/Bond_graph

Amidst the general decline in quality of publications on the Web, it's great to see Wired, which put out lots of fluffy, inaccurate stuff in the past, publishing so many solid, interesting articles. This was fun to read, had actual equations and graphs, and made sense. I remember reading the headline about charging a phone in 20 seconds and not bothering to look at the details, because, no way.

Interesting, certainly. That said, I found this article frustrating to read because of the many spelling and grammatical errors; I don't want to start on stylistic issues. Does Wired not have a copy editor on staff?

For example, two that quickly caught my eye:

* "... just a plain boring capacitor." should have a comma between coordinate adjectives.

* "... the case were stuff is in the middle ..." should be "where" instead of "were".

Is my sarcasm detector broken? This is a fluffy inaccurate post masked by a bunch of equations any apt high school physics student could have gotten from their textbook.

Nobody uses mica capacitors for high energy-density applications, electrochemical EDLC's have much higher energy densities.

Furthermore switching regulators that can operate the phone relatively independent of input voltage are cheap and efficient.

Yes the other articles about this were bizarre, no, cell phone battery replacement (where nobody seems to mind plugging them in at night, and ED is king) are not a good application for any capacitor in the foreseeable future, but this article is hardly better.

More nitpicks: "The news reports don’t actually state how much energy the storage device can store."

when the story showed up on HN, I searched google news for a few words from the headline and found the ED in one of the top 5 links. I don't remember the exact number, but it was on the order of a SLA starter battery (so lower than a deep-cycle SLA battery). Clearly not in cell-phone territory.

I was not being sarcastic, so don't worry about your detector. This was the way a physicist would take a first look at the problem, so I could relate and I enjoyed it. You and some other critics here are taking the author to task for not writing a detailed electrical engineering analysis, but I don't think that's entirely fair. Although the article is not completely accurate in that sense, it's also not "fluffy", and much better than the stuff that caused me to write Wired off until recently. Scoff if you must, but physicists are fond of their spherical cows and mica capacitors.

To take the spherical-cow analogy further, this article is a bit like responding to a headline of "New non-spherical cow shape way more efficient than spherical shape, promises free milk for everyone!" by doing an approximation with a spherical cow.

Particularly when a tiny bit of googling turns up actual numbers claimed for the cow in the experiment.

> You and some other critics here...

You're probably referring to me, since I've been posting all over this thread and I did nothing to preface my criticisms. I really didn't mean to "take the author to task," and I was as astonished and happy as you were to see this piece in Wired. This was just something I became interested in when the news first hit, and for whatever reason I've enjoyed talking about it and breaking it down more than usual.

The only thing I'd criticize him for was that he's basing his back-of-the-envelope on some arbitrary things (the 6V rating and choice of a ceramic/mica dielectric, for instance) rather than trying to base it on an appropriate lower bound. Had he done the latter he would've reached the same conclusion but with a far less outlandish example. It's only bad because relatively small supercaps which would have worked for his example do exist (much to the chagrin of the news stories he's criticizing), and his example makes it look like he's suggesting otherwise.

"You're probably referring to me"

No, your comments were informative while politely avoiding direct criticism of the author. And you make another good point here.

This guy doesn’t seem to know much about really existing supercapacitors. Assuming that 1200F is enough, you can get a 1200F supercap (Maxwell BCAP1200 P270 K04) rated for 2.7V that’s 60.7 mm in diameter and 74.30 mm tall. He’s assuming 6V, so let’s say you need two of them. That’s still way too large for a phone, but nowhere near as huge as he suggests.

You'd actually need 9 of them. Capacitor networks don't behave like resistor networks. Placing caps in series increases their voltage tolerance but reduces their overall capacitance. You'd need three parallel sets of three of these caps in series to get the voltage and capacitance you're after.

Edits: It turns out that 3*3=9 these days, not 6.

You can use a DC/DC to boost the voltage.

Sure. In fact you'd want to use a high-efficiency buck/boost converter to take advantage of the entire discharge curve. However you're only storing about 20% of the energy at 2.7V.

Edit: I'll show my work.

Farads are Joules (energy) per Volt (electric potential) squared. That means that the formula for energy stored (E) in a capacitor is E=CV^2.

A 1200F cap charged to 2.7V stores 8478 Joules of energy. A 1200F cap charged to 6V stores 43,200 Joules of energy. That means that at a charge of 2.7V there's ~80% less energy stored in the capacitor than at 6V. Incidentally this ratio holds no matter value of the capacitor.

The point is that you don't need to round up. 2.7 volts is pretty close to 3, so you need pretty close to 2x2=4. Bump it to five to compensate for the rounding and look it matches the actual math. Half of your initial suggestion.

Caps tend to explode when they fail, and high energy caps explode violently. You always, always, want to round down when spec'ing caps, not up.

But forget all that. Lets do it differently and fix our math toward the author's original comparison, the 1500mAh battery, not his weird, somewhat arbitrary choice, of a 6V-rated cap which drains to 2.5V after ~10 hours. And instead of talking about electricity, let's talk about what we're really after: energy.

A 1.5Ah battery stores about 22.68kJ of energy when fully charged to 4.2V (1.5Ah×3600C×4.2V=22,680J). From above, At 2.7V, a 1200F cap stores 8.478KJ of energy. 22.68kJ/8.478kJ = 2.675. Meaning we need at least 3 of these caps to meet or exceed the energy storage of a 1500mAh battery.

So we're both wrong in the context of the original comparison. And more importantly, so is the author.

Edit: This 3000F supercap rated for 2.7V would get us damn close to the 22.68kJ requirement, coming in at 21.87kJ of total storage: http://www.mouser.com/ProductDetail/Maxwell-Technologies/BCA...

Edit again:

I messed up again! The battery energy capacity is actually less than 22.68kJ, because the energy equation I used assumes that the voltage stays constant during the battery's discharge cycle. It doesn't. For a constant-current discharge of a LiPoly battery, the mean voltage is something like 3.4-3.6V (estimating from the discharge voltage curve, assuming discharge stops at 2.7V). So really we're talking something closer to 19kJ of energy stored within the battery. By the figures in her submission, her cap would weigh 262.6g and would occupy 30.8m^2 (no figures on thickness to calculate volume). Titanium density is 4.5g/cm^3. Assuming titanium dominates the density of her material, it'd be about 58cm^3 in volume - something like 10cm by 5.8cm by 1cm. If the density matched that of copper, it'd be about half that size.

Whoa, whoa, nowhere did I suggest applying extra voltage. I was saying you could use them at 2.7, or 2.6 if you want a safety margin. You don't have to use pure RLC and drop down to 2 volts.

It is amusing that the author screwed up.

>Sure. In fact you'd want to use a high-efficiency buck/boost converter to take advantage of the entire discharge curve. However you're only storing about 20% of the energy at 2.7V.

That doesn't have to be an issue. There are high efficiency joule thief IC's that'll take a voltage input around 0.3v.

At the cost of power which will increase your necessary total capacitance.

Digital circuits require some minimal voltage in order for their internal transistors to switch. This is because semiconductors only conduct in the presence of an electric field with a high enough potential to excite and free up the electrons hanging out near their junctions. That means that if your voltage drops below, say 1.8V (common for most low-power digital circuits these days), you won't be able to use your device any more.

There are some very efficient buck/boost converters out there. Like 95% and above efficient. Typically the cost of the power to use a buck/boost dc-dc converter is far outweighed by the ability to convert more of the stored charge into a potential that's suitable to operate the circuit.

Put differently, picture a barrel with a tap on it's wall somewhere near the bottom, and a different barrel with a conical-shaped bottom which tapers into a drain-like tap. Which is more useful? The latter is what you get from a good buck/boost converter.

I've learned something from your comments on this thread, and thanks for taking the time, but you've lost me with the barrels. Pressure depends on height of the water and is independent of the container shape.

benjamincburns was referring to a barrel with a tap "somewhere near the bottom". The point being that it's very difficult to get stuff out near the flat bottom of the barrel, especially if you can't move the barrel itself. The other option with buck/boost converters is if the barrel's bottom tapers into your tap, so that there's no way for any liquid to avoid gravity's call and leave through the tap.

That makes sense. I was over-thinking it.

Thanks for the compliment! As for the question, see what Elessar said.

Also:

In the typical water/electric circuit analogy the volume of water is synonymous to the amount of electrical energy stored (Joules). The pressure of the water is synonymous to the electric potential (Volts). Pipe diameter can be thought of as resistance (ohms), flow rate through a pipe (meters per second) can be thought of as current, and flow volume (cubic meters per second or litres per second) as power (watts).

There most certainly are capacitors available that meet his specs without being unreasonably large. For example this one has a capacitance of 1200F and is only 8 cm long: http://www.mouser.com/ProductDetail/Ioxus/RSC2R7128LR/?qs=cf...

Modeling it as a parallel plate capacitor with a mica dielectric is basically totally ignoring the part about it being a supercapacitor. He may as well have been seeing if it would be feasible to get to the moon using a wood fired rocket.

Most supercaps are rated for very low voltages (this one is only rated for 2.4V), as dictated by the breakdown voltages of the dielectric and its manufacturing tolerances. A 6V-rated 1200F supercap made from the same materials will be considerably larger. Further, most electrical engineers will enforce a factor of safety beyond the manufacturing tolerances . This is doubly true for capacitors, since dielectric failure on high energy caps can be somewhat catastrophic (read: explosive).

High efficiency DC/DC converters are now commonplace. Energy density (J/m^3 and J/kg) and internal resistance are the really important quantities in energy storage capacitors.

Agreed. I discussed this elsewhere:

https://news.ycombinator.com/item?id=5799723

https://news.ycombinator.com/item?id=5799713

Why would you need 6v for a phone? I got feeling this was pulled out of his ass. Everything works on sub 2V there if I am not gravely wrong (not sure for the light source for the display)

My old HTC Sensation drew 3.7V.

http://www.amazon.com/Anker%C2%AE-1900mAh-Li-ion-Battery-Sen...

You don't. I'd bet most of the voltage rails on a modern phone are 1.8V and 3.3V. Then the display backlight usually uses a constant-current regulator, and you'll probably have some oddball low voltages for the CPU core and other various things. Given Ohm's law you can generate all that using DC-to-DC converters.

This in reference to the recent stream of posts about charging a phone in 20 seconds. Or 30 seconds. Or something.

Here's the most recent, with a collection of links to others: https://news.ycombinator.com/item?id=5798189

Can someone remind/explain to me what happens if I were to use both my phone's current battery but put a super cap in front of it?

ie When I plug my phone in to charge, the super cap is "instantly" charged. Then a smart charging circuit would use the super cap to charge my battery.

Situation and my expectation: I'm headed home from work, drop off my laptop and leave for some social gathering, but my phone is about to die. Trying to charge me phone for the 1 min I'm in my house in futile. However, my expectation is that the super cap would fully charge. Then while I'm walking to my destination, it could charge my battery to some reasonable percent.

I'd be really happy even if I could get ~20% or so from the fully charged super cap. Would definitely be better than nothing.

It's kinda like you have a bucket you carry around you for drinking water. And you get back home, your bucket is about empty, and you have 3 seconds to get water before you need to leave and you're like "Let me fill up my glass instead and then slowly pour it into my bucket". Your glass will be full, but it wouldn't make a dent in you bucket, and frankly if you were to carry around a glass of neccesairy size to fill up your bucket to even 20%, you could just carry another bucket around.

It's a really bad metaphor for supercapacitors having worse energy densities than for example lithium ion batteries, and that having a supercapacitor large enough to recharge your battery to 10% would esentially entail carrying around another battery.

Honestly, if you have a phone you can actually open, you'd be better advised with an exchange battery, and if you can't open it, an external battery pack you carry around for emergencies like this.

For example, I think this external charger might do what you want it to do http://www.amazon.com/Energizer-Energi-To-Go-Battery-Operate... but your requirements might differ.

In 10-15 years, you will almost certainly be able to charge your cellphone in 30 seconds.

And a supercapacitor will probably be what's inside to allow that.

The incredible charge rates of supercaps will someday (when the energy density improves enough) become a nice feature.

It's possible some version of LiPo batteries will improve the charge rate dramatically, but I believe the rate is limited by chemical reactions.

I can totally imagine phone that has two huge metal contacts on sides that get exposed and connected as you shove whole phone into the charger. Such connection could transfer 1kW easily.

how about a flux capacitor? 1.21 gigawatts!

I used to TA undergraduate science courses. I successfully convinced at least one generation of young minds that the SI prefix G is pronounced hard, except for power, where it is pronounced soft like J. I blamed the French for this historical oddity. They bought it.

Giga actually has an acceptable pronunciation with a soft g. The American National Bureau of Standards actually has this pronunciation formalized.

gigga what? gigga who? Switcha flow, getcha dough

EDIT: apparently downvoters don't like Jay-Z.

You're being downvoted because your comment is inane, off topic, and adds nothing of value to the conversation. Even the grandparent quotation from Back to the Future suffers from the same problem and is being downvoted.

message clear - hacker news: NOT the place for idle joking

You used your powers for evil.

With the right voltages, you can get 1.21 gigawatts out of a capacitor no problem! Just not for very long ;-)

Reply on news.ycombinator.com