On the recommendation of an SO-er, I'm looking at using a TPS54295 to power a controller that uses an ATMega 2560. The regulator will be powered by a 3-cell LiPo battery, implying input voltages between 9 and 12.4V. The controller will be in standby mode for a few minutes at a time, waking up periodically to check sensor states and perform various tasks. I expect the controller part of the circuit to draw 15-20mA when active, and 15-20uA when sleeping. Looking at the data sheet for the regulator, I'm seeing two things that concern me:

First off, there seems to be some amount of instability in the regulated voltage when the current draw goes below a hundred mAs or so:

enter image description here

Does this mean I need to consider a different regulator, does it not matter, or is this something that I need to rectify with additional passive components?

The second part is this section of the efficiency curve:

enter image description here

this tells me that when in standby, efficiency will drop into the low 20s or even 10s of %. Again - is this standard for this type of regulator (meaning I don't need to worry about it), or does this mean that I'll be burning it up at those levels?

  • \$\begingroup\$ That's not at all helpful to me. Does this pertain to the first part of my question, or the second? Where are you getting 18V and 1mA from? Why are you using the input voltage and the output current to figure out available power (which, again, I didn't ask for). Wouldn't the input voltage at 3.3v/1mA output not be 1mA? \$\endgroup\$
    – kolosy
    Jul 1, 2014 at 2:21

3 Answers 3


First question- the variation in output voltage at relatively low current is probably due to transfer to a different mode of operation to increase efficiency (cycle skipping). If the circuit you have is tolerant to small changes in supply voltage, there should be no problem, if it isn't then you would have to adapt it or the regulator.

Second question- the efficiency curves indicate a small quiescent dissipation, which is normal. However, the regulator will be burning up most of the battery energy if your circuit is only drawing a couple hundred uA average at 3.3V.

It seems rather silly to be using a 2A regulator if your circuit only draws 20mA maximum. Even a linear regulator will do better (and it will be a heck of a lot cheaper and simpler). They are available with quiescent current in the 10uA range or below.

For example, a Seiko S-812C33AUA-C2NT2G would consume only about 1uA when your circuit is drawing 20uA (efficiency of 26% at 12V in/3.3V out), and 27.5% efficiency with 20mA out. The TI switcher will draw more like 1mA quiescent so the efficiency at 20uA out will be 0.5%, and at 20mA out about 80%.

So the linear regulator beats this particular switcher by maybe 4:1 assuming ~1% duty cycle, at 1/3 the price. Suggest you look at linears or more suitable switching regulators.

It's also possible to run part of the circuit on standby and fire up a big regulator when you need it, but that's probably not indicated with only 20mA draw.

  • \$\begingroup\$ this relates to the original post that suggested this part - i have a controller and a set of servos. as a result, my peak power demands can be 400mA @ 5V on the servo rail + when some of my additional peripherals on the 3.3V rail are on, they can draw another 100+mA @ 3.3V. \$\endgroup\$
    – kolosy
    Jul 1, 2014 at 3:48
  • \$\begingroup\$ @kolosy Then you can run the controller with a low Iq regulator and enable a big switching regulator or regulators just for when the peripherals are required. The TPS54295 is pretty greedy even in shutdown though-80uA typical. \$\endgroup\$ Jul 1, 2014 at 3:51
  • \$\begingroup\$ would two different regulator active on the same Vcc line conflict? i imagine this going like this - wake, enable bigger reg, disable smaller reg, do work, enable smaller reg, disable bigger reg, go to sleep. would the periods of overlap cause issues? \$\endgroup\$
    – kolosy
    Jul 1, 2014 at 3:56
  • 1
    \$\begingroup\$ @kolosy Exactly, that's why many regulators have an enable input- to allow you to do that. \$\endgroup\$ Jul 1, 2014 at 3:57
  • \$\begingroup\$ @kolosy They could conflict, but you might find it's not worth trying to save a bit of energy when the controller draws more, just have two 3.3V lines. \$\endgroup\$ Jul 1, 2014 at 4:08

You have to look at the efficiency from the standpoint of actual power in versus power out. When you have a very light load the regulator circuit itself may very well consume more power than what is delivered to the load (i.e. efficiency < 50%). If the standby / no load power consumption of your regulator is higher that you would like then you may very well want to look for a regulator solution that is more efficient than this particular part.

Since you are looking at such a low current draw at the load the use of a 2A capable regulator may be overkill. Look for a uPower regulator designed for less current rating of say 100-300 mA for example.

It may also be helpful to look at alternate topologies. If your unit spends almost all of its time in the idle state it may be that a uPower linear regulator may actually give you better overall power savings off your battery as it's standby current at low load could very well be less than the idle power of the switcher.


Nothing to be concerned about, really. Your cpu doesn't care if it gets 3.3V vs. 3.27V. Also, any regulator circuit will look bad from an efficiency standpoint at very light loads.

  • \$\begingroup\$ This would be because the actual load current approaches the quiescent current of the regulator, right? \$\endgroup\$
    – sherrellbc
    Jul 1, 2014 at 5:19

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.