Microamp-level boost converter (1.4V->2.5V, 60uA)?

[Artlav]

Are there any boost converters or charge pump style devices that could provide 60 uA of power at around 2.5V from a single AAA cell, while themselves consuming less than that?

I'm looking for ways to increase the lifespan of my little LED night lights (running at 50-60uA) from a month on LIR2032 to something on the order of several years.
An easy way to do so is to use 3 AAA cells, but that makes it huge.
One AAA would be about as large as i want to go, but the voltage is wrong.
There are some exotic lithium primary cells at 3.6V that would work just fine, but they are obscenely expensive.

One more possibility is to find a really low power step-up converter, and use small sized alkaline/lithium primary cells.
So, is there such a beast?

[Marco]

Not worth boosting, just use a low frequency charge pump.

Here's a regulated one for instance. There might be single IC solutions as well, didn't google much.

PS. another one.

[matseng]

Or even simpler - just parallel two regular cr2032. They have each about 6 times the mAh of the lir2032 used today. That would give about one year runtime.

[Fungus]

+1 for CR2450 cells.

I've made gadgets with an ATTiny85 + 6 LEDs that have been running for 3.5 years on a CR2450 and the voltage has barely dropped. I just checked one and it measures 2.85V, there's probably another 1-2 years left in it.

(nb. The LEDs aren't continually lit up... they wake up and flash every few minutes or so)

For a night light you could add a light sensor and check it every five minutes, switch the LED off during the day. An AVR chip sleeps with <10uA of current so that would be your consumption during the day. At night it would be <10uA plus whatever the LED uses. It's a big win if your LED is using 50-60uA.

I haven't actually tried it, but: A Tiny85 could probably run the LED in PWM mode and save a lot of the power being wasted as heat in your dropper resistor. You'd select the 128kHz low power internal clock, output a PWM signal using Timer0, disable all the other peripherals, put the CPU to sleep. The LED would be pulsed at a high current but the average consumption should drop quite a bit for the same perceived LED brightness.

Wake the CPU up every now and again to see if daylight has arrived then stop everything during the day. 

[Artlav]

change the battery in the light when it dies...

Well, the whole idea was to avoid ever changing the batteries again, since i got sick of recharging them every month.
Once a year is kinda neither here, nor there in that regard...

Ideally, i would have gone for nuclear power - they make tritium lights of that brightness and size that would last for decades, but it would cost $150 apiece.
As a matter of fact, trying to reproduce such a light for cheap was what started the whole project years ago.

so we're going to need a bigger battery

Sounds like it.
As a matter of fact, these LiSOCl2 cells are actually rather cheap (1/100th of the nuclear option), have minimal self-discharge, and the right kind of capacity.
I should be able to get 4-5 years out of an AA one even without the uC tricks.
And a 1/2AA one makes the thing only double the size of LIR2032 version, without sides of the disc sticking out - a cute little brick.
That with some tricks Fungus suggested might get me the same 4-5 years.

All in all, thanks for pointing these out.

[PartialDischarge]

What kind of led does light on only 50-60uA??  or even 200uA.. Id be interested to know

[matseng]

More or less related... Keeping a led lit for over two years on a single CR2032. https://hackaday.io/project/12874-automated-ledlaser-diode-analysis-and-modeling/log/45017-its-not-just-for-diodes-anymore

[Fungus]

What kind of led does light on only 50-60uA??  or even 200uA.. Id be interested to know

All of them, I think.

If any current at all goes through a LED then you'll get photons.

[wraper]

What kind of led does light on only 50-60uA??  or even 200uA.. Id be interested to know

at 200 uA, a good led will be already quiet bright.

[PartialDischarge]

What kind of led does light on only 50-60uA??  or even 200uA.. Id be interested to know

at 200 uA, a good led will be already quiet bright.

I have a very good LED with me, the last one in the attached pic ( http://www.vishay.com/docs/81351/vlmd31.pdf ) and at 200uA the light is very dim.
No point in taking a pic since the auto brightness of the camera will fake it, but most leds I have will start to have a decent light at 1mA at least
Maybe Im missing something

[Marco]

Hmmmmm  We are looking at a C-Size battery from Tadiran -- about US$13....  Too big I think for the OP.

Also for the cost difference with Alkaline you can build a charge pump and a detector to turn on at low light (if your time is free of course).

[Artlav]

Maybe Im missing something

Human's night-mode vision.
In daylight they are as good as off.
In the indoors light they glow slightly.
For dark-adapted eyes these are very bright.

The goal is to avoid the need to turn on the lights while going somewhere at night, and for that a few 60uA LEDs provide ample light.

[wraper]

Maybe Im missing something

Maybe 

[PartialDischarge]

Maybe Im missing something

Maybe 

So I dont have any InGaN leds , 
Will give them a try 

[wraper]

So I dont have any InGaN leds , 
Will give them a try 

Different color leds can be very bright as well.

The thing is, your leds are made with a technology from 80's.

[PartialDischarge]

So I dont have any InGaN leds , 
Will give them a try 

Different color leds can be very bright as well.

The thing is, your leds are made with a technology from 80's.

I see, however the VLHW is still brighter at low currents. At 1mA the luminuos intensity gets divided by 10 vs 100 for the TLCR

[wraper]

At 1mA the luminuos intensity gets divided by 10 vs 100 for the TLCR

Divided by 50, not 100.
Particular white leds will be 2x brighter. However not necessarily for your eyes. White leds are actually blue leds with luminophore, therefore often have a huge peak at the blue spectrum where human eye sensitivity is very low, especially cheap ones. And human eye sensitivity for blue light essentially sucks.

[PartialDischarge]

At 1mA the luminuos intensity gets divided by 10 vs 100 for the TLCR

Divided by 50, not 100

Nope, 100. the current-mcd is not linear, look at the specs and youll see a less than 0.01 factor that needs to be applied at 1mA

[Fungus]

22mcd?

Ah, yes, I remember those. 1982 wasn't a great year for LEDs.

[Kalvin]

Silicon Labs has some DC/DC-converters which has around 150nA quiescent current:

http://www.silabs.com/products/analog/dc-dc-converter/Pages/default.aspx

I haven't used those devices, so I cannot help any further.

[Fungus]

The goal is to avoid the need to turn on the lights while going somewhere at night, and for that a few 60uA LEDs provide ample light.

I've got some little plug-in LED lights for the passageways and bathroom - no batteries!

I think they say "0.5W" on the back so it's no big deal to leave them on 24/7. Plenty of light to sneak around at night.

[wraper]

the current-mcd is not linear, look at the specs and youll see a less than 0.01 factor that needs to be applied at 1mA

You are right about that. However not relevant if we compare different color leds in general. Those are just higher power leds from the first random datasheet I found.

[David Hess]

There are old style "high efficiency" LEDs (GaP?) which are optimized for low currents.

If you can live with some complexity, then pulsing an LED at a higher current will allow greater luminous efficiency.

[mark03]

This "LEDs less efficient at low currents" thing is completely new to me.  Are there any rules of thumb for how much current you need to get into the linear region (w.r.t. the datasheet quoted output at some test current)?  When I look at LED datasheets I usually don't see anything about this loss of proportionality at low currents.

Are modern lighting-class LEDs like this?  E.g. Cree XP-E.

[David Hess]

I am not aware of any general rules other than LED efficiency dropping off above and below a specific current which depends on the device and the type of device.  Something could be put together from a sampling of datasheets.  This web page includes real world data.

That page mentions the Cree CP41B-GFS-CN0P0674 working well at 10s of microamps so some modern LEDs will work but otherwise pulsing the LED is definitely the way to go.