Buck-boost for 1-4.2V input & 3.3V output. Max. power 1W

[plazma]

I'm designing a small metering device for an university embedded course.
I would like to use any AA sized battery. Primary or rechargeable.
Input voltage may be as low as 1V with LSD NiMH and as high as 4.2 with a LiIon 14500 cell.

I need 3.3V for the MCU and sensors.

I found a one chip solution LTC3106. http://www.linear.com/product/LTC3106
However this chip got a poor performance at lower voltages.

Any suggestions for a buck boost solution with good performance? Two different chips?

[tszaboo]

You are making your life very difficult with those requirements. I would suggest a stepup with 2 AA/AAA battery or a stepdown with the 14500 cell. It is easier with the 14500 if you dont want to charge it, with the two AA/AAA it is more user friendly.
The issue with complicated power systems is simple. You use two dozen components and complicated design, hard to source inductors. IMHO, inductors in the Bill of material are pain in the ass. And you spend all that time on the power system, instead of spending it on the application. Nobody will raise an eyebrow, if you need two batteries for it. You can even run the entire system on it, without any power management, if parts are selected for that.
Or use a coin cell, or oversized coin cell.

[plazma]

At the moment I'm going with the LTC3106. It's a decent buck-boost for LiIon. Using NiMH or alkaline remains an inefficient but usable option.

There is only space for one AA. Two AAA sizes is not an option. The battery is inserted into the device from the side through a round hole.

However I'm still interested in making the NiMH and alkaline option more efficient.

[DBecker]

Remember that it's easy to get lithium rechargeable cells in AA size.  You'll still need a buck-boost to get 3.3V and use most of the capacity, but if you don't need the full capacity, a micropower linear LDO may work for your application.  And it will be far quieter than a switching regulator, which can be a real advantage when you don't want to waste the power to filter samples.

[plazma]

A linear regulator + a boost is what I'm researching? atm. The buck and linear regulator are quite close for 3.3V when using LiIon cells.