# calculating total power consumption with LDO vs switching regulator

I'm just getting somewhat familiar with voltage regulators and am trying to make sure I'm understanding a bunch of parameters correctly.

Scenario:

I have a project that spends 1/3 of its time drawing 20uA, 1/3 drawing 40mA and 1/3 drawing 150mA. The project is powered by a 3 cell LiPo battery, meaning input voltages ranging between 9 and 12.4V. I'm trying to decide whether I should power the whole thing off a single LDO, a single switcher, or try to separate out the parts that push consumption to 150mA, and use an LDO for the low power part and the switcher as necessary for the high power.

I'm looking at an LDO with the following quiescent current (listed as ground current in the datasheet)

  1mA   @ 150mA output current
400uA @ 50mA output current
100uA @ 1mA output current


I'm also looking at a switching regulator with the following efficiency:

90% @ 150mA
90% @ 50mA
85% @ 1mA
60% @ 100uA


If I'm understanding the comparison correctly, then the LDO will pass along (output current + the ground current) * output voltage and waste (output current + the ground current) * voltage difference, which reduces to just input voltage * (output current + ground current), meaning:

11.1V * (100uA + 20uA) = 13.2mW in the first state, and so on (the 11.1 is the nominal voltage of the battery, so it'll spend a fair bit of time there)

For the switcher, the total power consumption as I understand it is Vout * Iout / Eff, so at that first state it would be 3.3V * 20uA / .6 = 1.1mW. My understanding was that for switchers, the quiescent current is factored into the efficiency graph you're given in the data sheet, hence steep fall-off at the start.

Is this right?

• It might be worth pointing out that, since your load seems to be a consistent 3.3 volts, you don't need an LDO at all. Almost any linear regulator will do, since the regulator will drop a minimum of 5.7 volts, and almost anything will do that. From that fact alone, it's clear that a switcher will be much more efficient than a linear, since the efficiency of a linear will be at most 3.3 / 9, or 37%, with an "expected most of the time" efficiency of 3.3 / 11.1, or 30% (ignoring quiescent current, which is reasonable as a starting point). Commented Jul 2, 2014 at 4:28
• 2nd point i have doubt...for switcher...if input is 12V, output is 3.3V, Iout = 100uA and Iquiscent = 20uA..Then if efficiency = 60% ....input power consumption due to load current is 3.3v*100uA/0.6...input power consumption due to quiscent current of IC = 12V * 20uA (Vin_max*Iq).....then u need to sum both....Quiscent is not the converted or switched current to apply efficiency to it....correct me if i am wrong.. Commented Jul 2, 2014 at 6:43