Alright so maybe im searching wrong, or using the wrong keywords. But google digikey and mouser are not cooperating.

I am looking for a 6pin or less QFN or sqfn? (The IC with either the small ball pads, or pins bent under) to work with 13mV up to 60mV and at most 50mA input, and output enough to power 4leds. I wanted to use 0201 if I can find them, otherwise 0402 at 3v each. I am also looking for better color options for the 5mA leds but no luck.
I need the ic to use as minimum external components as possible, because I will be using the smallest option for the components as possible. Maybe a rf choke or capacitor external to boost the power?

The power source is two nextreme eTEG mini peltier units with a delta of about 30 degrees, 40 max.

Im open to lots of suggestions! Any help is very much appreciated

 

60mV times 50mA equals 3mW. How much power do your three LEDs need? I bet it is much more than what you have.
Even lousy 1mA at 2V that would barely be visible is 2mW. Now account for losses in the conversion chip, which will be huge.

 

Well its 4leds. They run at 20mA/3V. Like I said I will look for ultra low power leds, but there aren't many options for those 5mA lights.

But If its a switching supply, couldn't it charge the inductor or capacitor and discharge to power the leds?
Like a joule thief is a basic switching induction circuit. And can power an led from less than 1V out of a dead AAA. I would use this but the inductor has to be too big for my space requirements, even smd versions.

And losses I have seen for the chips I looked at where less than 100uA

 

Conversation of energy says that you can´t get more out than you put in. So you really can´t continuously power something that needs 4*0.005*3=60mW with a supply that is barely capable of putting out 3mW, no matter what technology you use. What you could do is periodically charge a capacitor or battery for about 95% of the period and then have the LEDs on for 5% of the period.

Or you could just get more cells.

 

Well maybe im messing up info somewhere.... the data sheet specifies 36mW output at 50K delta. I was using two so that could scratch enough power out for just under 60mW...maybe 50mW. Am I wrong switching supplies that pulse their output, draw less power constantly, then directly powering from the source? Wouldn't it otherwise be pointless to make drivers?

 

You mean datasheet of the peltier? Does it have some curve for delta T versus output power?

Switching supplies are something different. They usually store a bit of energy from the input in an inductor and then put it out into the output, usually along with changing the voltage up or down. The key is that they are more efficient than analog regulators where you just waste a lot of power in heat, but they still cant put out more than they get in.
Also linear regulators cannot step voltage up.
As an example, if you have 1V 1A input, then you can get for example 10V at 100mA output, if your converter is 100% efficient.

 

http://www.mouser.com/new/nextreme/nextremeeTEG/

That's the link to mouser, the data sheet is at the bottom, it is the 3.3X3.1mm module.

Yes so I'm looking for suggestions for that driver IC that uses a very small inductor or capacitor to boost the voltage at cost of amperage.

 

Still no suggestions for my IC??

 

Maybe a rf choke or capacitor external to boost the power?

You want a free lunch to go with that booster?

Nope, not without adding even more power then the boost. So your joule thief will work just fine if you say add a battery or a wall plug.