This circuit pulls energy from thin air

Most of us are familiar with Texas Instruments (TI) as the company that made the graphing calculator we used in school.

Calculators made be the company’s bread and butter, but that’s not all they’ve been up to. TI recently introduced the industry’s lowest power DC/DC step-down converter, a tiny circuit that could allow future electronics to power themselves without batteries or wires.

The tiny power circuit allows mobile accessories and wireless sensors to manage microwatts generated from solar, thermoelectric, magnetic and vibration energy, which could eliminate the need for bulky, toxic batteries in the future.


sort by: active | newest | oldest
Orngrimm4 years ago
BTW: 1:1-Copy of:
Interesting. Not exactly copies, but rather parallel postings by exactly the same originator, Beth Buczynski. I suspect yours was first, and "TGDaily" picked it up from there. Iceng properly provided his source for the content.
Your two posts are parallel however :-D
iceng (author)  Goodhart4 years ago
Parallel is good to infinity  :-)
Goodhart iceng4 years ago
Unless one speaks of a parallelogram :-)
iceng (author)  Goodhart4 years ago
The space bounded by two sets of parallel lines, very good ;-)
Goodhart iceng4 years ago
Bound but not gagged :-D
iceng (author)  Goodhart4 years ago
Can't really comment on that............................
Goodhart iceng4 years ago just did ! :-P
iceng (author)  Goodhart4 years ago
Mmmmfff ..... mmfff .............
Goodhart iceng4 years ago
Weird. I've deleted one of them; thanks!
iceng (author)  Goodhart4 years ago
Thank you

Thank you
Orngrimm4 years ago
The lowest power only goes for the power the chip needs by itself.
The linked NanoBuck is by any means not the chip with the capability of using the lowest powersource to get usable Outputs.
If you are really low-voltage and low-current, i suggest you take a look at Linears LTC3108, LTC3108-1 and LTC3109 which start operating at 20mV! But it should be noted that they only go up to 0.5V input...
If you want 0.25-5V input you could take a look at LTC3105 which features a nice Maximum Power Point Control and can also be used to charge a LiIon-cell as backup. I used that one in the past quite successfully.

But yes: the TPS62736 holds the flag on the lowest self-consumption but fails in other parameters if you ask me... Also efficiency isnt this great with the TPS62736...

Got any cuter ideas than Linear's TEG "booster" ? I'd like to boost a bunch of metal-metal junction TCs to something usable, like 5V, starting with, say, 250mV@20A
iceng (author)  steveastrouk4 years ago

Did you see Randofo's flame candle
running a peltier module
running a tea LED light shaped as a candle :-)

That is a supercalafragilisticexpialidocious application in my book !!
Orngrimm iceng4 years ago
Do you know tPOD1?
"make light from a candle" is just mindblowing. Especially if the produced light is brighter than the candle in the first place.
iceng (author)  Orngrimm4 years ago
Thanks for the neat link.

I guess that a Candle puts out a whole bunch of invisible  IR Energy.

While the LED only puts out the little pinch of the Visible light spectrum
that we humans are able to perceive.

This is the best IR to visible light Filter Converter :-)

Orngrimm iceng4 years ago
Yes. And to clarify: IR is nothing more than heat.
And a candle "wastes" almost 100% of its energy as heat and only a fraction of it as visible (to humans) light.
According to wikipedia, a candle has around 80W as heat.
And if i look at my LEDs, i think about the same lightammount comes off from a white LED driven with 10-15mA.
So the LED has 3.6V * 15mA = 0.054 watts.
So a LED is around 1480 times more efficient in terms of converting energy to visible light.
See, a TEG is not this efficient... Again, according to Wiki their efficiency is around 5-8%.
Slash that in half by "escaping" heat (You cannot channel all the heat produced trough the TEG).
So lets round it down to 2%... Heck! I go for 1% efficiency of a TEG.
Now lets see:
1% of 80W = 0.8W
And if i check a 1W white LED thats a LOT more light than a candle still.
Also for comparison notes on how wastefull a candle is.
Wiki tells:
"a uniform 1 candela source emits 12.6 lumens."

Also wiki tells:
Type Overall luminous efficiency Overall luminous efficacy (lm/W)
glass halogen 2.3% 16
So the normal halogen-Bulp puts out more light per watt than a 80watt-Heater called candle-flame with its full 80w combined.
So as a rule of thumb, a halogen-Lamp is roughly 80x more efficient in terms of generating light than a candle.

A cool comparison also can be found @
Overall luminous efficiency:
candle: 0.04%
white LED (raw, without power supply): 0.66–22.0%
In the past employment, I have measured eff's of around 10-15% including powersupply.
Yes, but the trouble with TEGs is they REALLY don't like being hotter than ~160 C. A metal-junction can GLOW.

Admittedly TEGS ARE more efficient though.
Well... Whats the output-power? You only say that you have 20Amps available at 250mV as input...

What wattage is the output @ 5V?
Also does it have to be constant voltage or constant current? Voltage i suppose...

The problem with 0.25V is that is below silicone-Forwardvoltage. Also the proportions of 0.25V and 20A are very high (factor 80 is HUGE).
neither Linear, Maxim came out flat (They dont have stuff with this voltage AND current).
Texas has something close, but no real hitter:
TPS61202 (0.3V in, 1.3A Switches)

But i think it should be possible to make a little booster yourself (Simple Boost-circuit with fixed freqnency and so on) and quadruple the voltage to 1V @ 5A. THEN you could use a LT1308A and B as an example.

Or you go back to a simple self-made DC/AC-converter and use a small transformator. That would be one of the better solutions i think...
Output power is a "high as possible" The input could be configured differently at a pinch, but I was looking at it for powering phone chargers in parts of the world where a TEG is too delicate, and KISS is a great maxim. The Linear front end where they have a transformer was damned clever, but, as you say, with the current proportions as it stands the transformer would need to be built like a welder !
I dont think 5W will require big transformators... If you chop your DC to a FAST AC, then your transformator becomes tiny. I am sure not more than 100g are needed if chopped at 10kHz or similar...
And 20A dont require 1cm^2 cables...

In fact, i found @ that AWG 20 is enough for 20A. and thats just 0.8128mm diameter... Make 1.2mm off that and you are golden...
The secondary coil is even thinner (by the thickness of the wire).
iceng (author)  Orngrimm4 years ago
Interesting but true.
Orngrimm iceng4 years ago
Yes. But it is important to keep the goal in mind and what you have and what you want.
If you need usable output from ultralow-input, then the linears are king.
If you happen to have usable inputvoltages for the texas and you have to count every electron in your efficiency, then the texas may be the better choice.

As often in electronics, there are many horses. Not one is the best, but one for each discipline and version. :)
iceng (author)  Orngrimm4 years ago
Good of you to question the nearly absurd TI instilled postings.

An EE buddy sent me that article asking my opinion.

My response was i would be afraid to wave my hand near it
for fear of destroying the IC.

You see it is a dry winter in the high desert now and a short
three steps from my PC to the door-knob will raise a healthy
25 mm spark :-)

You speak of our four hoofed termites.  Regrettably Nevada range lands
have too many wild horses ( Mustangs ).
More that can be adopted takes about $2500 to properly maintain
just a single horse ( feed, vet bills and stable ).
There is a great controversy that overflow stallions might be slaughtered 
to feed the French. !