5v and 3.3 power supply - use two switch regs or one switch reg and one linear?

[graemegets]

Hi,

I need to put togeather a power supply unit for a project that has :
1x xmega at 3.3v
1x RF transmisster as 3.3v
1x rs232 level shifter at 3.3v
1 x gps at 3.3v
and 1 x iridium sat unit at 5v

None of these modules will run simultaneously so only need to supply a peak of 400ma and an average of around 10 to 50 ma during various modes when not transmitting.

im not sure what configuration to use. Im thinking a buck boost from LT (unit will run off batteries) .

The main duration of power will be 3.3v with the iridium only needing to be switch on every 2 hours at 5v?

Should i run two switch for the 3.3v and 5v or run one at 5V with a liniar reg for the 3.3.

Battery efficiency is v important.

Hope thats enough info.
Thanks

[georges80]

You can do the math yourself - since YOU know the actual duty cycle that each module is running at and the actual current each module draws and the various sleep modes you plan to implement.

Since you haven't provided enough numbers and haven't quantified HOW important battery efficiency is versus additional hardware cost or even what battery voltage/capacity you are using and what the required battery life is, I'm not sure how you can expect an intelligent response to your question...

cheers,
george.

[Ian.M]

Its obvious that using a linear regulator from 5V to get 3.3V wastes 33% of its input power as heat.  As such it only makes sense to do it that way if nearly all the power consumption is from the 5V rail.

[AndreasF]

Since none of the "modules" run simultaneously (supposedly) and only one actually uses 5V, I think you're better of having two independent switch-mode regulators. Otherwise you'll always have the losses of the 5V reg (even if those are low) plus the losses going from 5 to 3.3 for 4 out of 5 modules.

[Marco]

You could use an ultra-low dropout regulator for 3.3V and just change the feedback ratio to temporarily have the switcher do 5V when needed ... 400 mA is a reasonable amount of current, so it might make sense boardspace wise.

[graemegets]

Thanks for the feedvback - Much appriciated

[Jeroen3]

Switchmode regulators often have ENABLE pins. If you're not running them simultaneously, this can be very helpful.
Some LDO's also have enable pins, but its less common.

[richard.cs]

I quite like Marco's idea :-)

Other than that I think I'd do a switcher for the 3.3V and linear (direct off the battery) for the 5V, either picking something with an enable input or picking one with very low quiescent current and disabling the load.

If the 5V is on so rarely them is efficiency probably doesn't matter much so long as it's idle current is low. You can go with two switchers but it doesn't sound like it wins you much but don't have a 5V switcher when most of your load is on 3.3V