Efficient low-power buck (step-down) regulator IC for Li-Ion - 3.3V conversion?

Question

It's not easy for electronics novice to find good element base, so I hope this suitable question.

Requirement: Efficiently power a lean, 3.3V-based MCU design from a Li-Ion battery. "Efficient" means no more than 20uA quiescent current, 10uA is better.
Ground current for 1mA output current <50uA.
Maximum required output current >= 50mA.
Having a SOT package (vs those <2x2mm SON packages) is also a plus.
Higher frequency and smaller inductor is also good (especially if doesn't conflict with package requirement ;-) ).

Good availability and price matters too. That's definitely subjective, but providing 2-3 names is better than telling that X is the best.

Thanks!

Answer 1

You don't want a switcher. Your input-output difference is so low (for most of the discharge curve the voltage is less than 3.8 V) that even at 90 % efficiency it won't do better much than an LDO.

Have a look at the Seiko S1167:

• Ground current typical 9 µA
• Available in 100 mV steps from 1.5 V to 5.5 V
• 1 % output voltage accuracy
• 150 mA output current
• 150 mV drop-out at 100 mA
• Shutdown input
• SOT23-5 package

edit
Found an even better one in the S1313:

• Ground current 0.9 µA
• 200 mA output current

Seiko doesn't give data for ground current under load, but in my experience you should count on 1 % of load current, so at 1 mA that would be around 10 µA, most likely less than 50 µA.

Answer 2

You should start by using parametric search, provided by the various distributors (e.g. http://digikey.com has a fantastic parametric search) and manufacturers. Usually you can search for the parameters you have given in your question.

After you have found some parts that meet you requirements, you should consult the datasheet and verify that the part is realy what you want.

Answer 3

The best buck may be the best choice:

The very best available buck converters will be superior to a linear regulator.

Consider the battery range to be 4.0V down to 3.4V (ignores initial brief period above 4.0V and allows 0.1 V output headroom).
Vin = 3.4 min, 3.7 mean, 4.0 max.
For 3.3V out a linear regulator gives efficiencies of
3.3/3.4 = 97%, 3.3/3.7 = 89% and 3.3/4.0 = 83%

Actual mean efficiency will depend on cell used and load level etc but say 3.7 Volt in is typical (which it probably is) for 89% efficiency.
A very good buck regulator with synchronous switching and careful attention to detail should be able to reach around 95% for much of this range. The difference between the linear and buck supplies is small, but notable.

Ground current can be a VERY bad performance measure: Note that using ground current as a figure of merit for a linear regulator can be VERY misleading. eg imagine a linear regulator with zero ground current and Vin = 3.6V and load = 1 mA. The efficiency will be 3.3/3.6 = 91.7% so about 8% of the input energy will be lost. This is equivalent to a regulator with 100% conversion efficiency but a ground current of 8% of the load. ie here 8% x 1 mA = 80 uA. So even if the regulator has 10 uA ground current at 1 mA load this will be swamped by unavoidable conversion losses across most of the Vin range,

Finding an IC to meet the above spec will take care. Buck regulators usually give peak conversion efficincies for limited combinations of Vin Vout, load and more. Outside optimum ranges the efficiency will fall off - sometimes badly.

I looked at the Digikey parametric selection guide to see if a suitable part could easily be identified. It's not an instant task as eg setting Iout_max to 100 mA may mislead as the most efficient IC may have a switch capable of switching higher current than is needed.

Answer 4

Just to show that I'm doing my homework, here's what I found so far:

Linear has lots of that magic. Caveat: very expensive, usually more than $2 in vendor prices. Besides being just expensive, it also means availability problems, because few parties want to buy such golden stuff and local distributors don't bother to carry it (being hobbyist, I'm interested in single quantities and DHLing from Digikey and friends is not a good option). Anyway, LTC3525 is even a boost/buck with Iq as low as 7uA (will be likely rather more in buck mode). LTC3620 is buck with Iq=18uA and Imax=15mA which is probably too low, packaged into that 2x2mm DFN.

LTC1474/LTC1475 is 10uA no-load Iq MSOP/SO8 package. Sourceable from eBay.

TI's TPS62230 is good at 22uA and even sourceable locally, but complete curse at 1x1.5mm.

So it seems that chips I found so far are "too new" (based on packaging), it would be nice to find "previous generation" in SOT23, so hints welcome.