I need to design a power supply that is powered from a single Li-ion cell (3-4.2V) and provides 3.3V up to 500mA and 5V up to 1A. I have three solutions in mind, one is boost converter to 5V and then a LDO for 3.3V, second is a buck/boost for 3.3V and a separate boost converter for 5V and the third is boost converter for 5V then cascade (input for 3.3V conv. is 5V) a buck converter for 3.3V.

Since I am using batteries a LDO seems inefficient (cca. 64%). So I am curious which of these solutions is better in terms of efficiency? And would I have issues if the dc-dc converters were running on different frequencies?

  • \$\begingroup\$ Change your design specs and just use the battery with LDO for logic \$\endgroup\$ Mar 15, 2020 at 16:34
  • \$\begingroup\$ Well that's not really an option since I can't change them, I'm just doing the power supply. \$\endgroup\$
    – bangaham
    Mar 15, 2020 at 16:38
  • 1
    \$\begingroup\$ why not ? ..... \$\endgroup\$ Mar 15, 2020 at 16:41
  • \$\begingroup\$ None of them are fundamentally flawed. Do the math for each and decide for yourself. \$\endgroup\$ Mar 15, 2020 at 16:47
  • \$\begingroup\$ A lithium battery that's 4.2V when fully charged is irreparably damaged when it's at 3V. Your requirements make no sense; change them. \$\endgroup\$ Mar 15, 2020 at 17:20

2 Answers 2


Since Buck/Boost is always more complex and costly, you want to avoid that in any configuration. I'd suggest the most cost effective and lowest component count would be:

  1. Boost convert the Li-Ion 3-4.2V supply to 5V. Likely 90%+ efficiency at < 7W
  2. Buck convert the 5V supply to 3.3V. Likely 75%+ efficiency at ~ 1.7W

You may need relatively high output capacitance on the 5V supply as the 3.3V Buck converter will draw pulses of current which depending on the converter IC used could be up to 4A. You will see larger pulse current for lower frequency configurations.

For example, the 3.3V supply could be based on a TPS62237 for ultra low cost and low component count. Here the switcher is at 2.7MHz and so the pulse input current is reduced to about 0.7A which significantly reduces the required input capacitance for the converter. This would result in a peak output current on the 5V supply of about 1.7A+.

Use the TI WeBench Power Designer for details:

enter image description here

I'll leave you to research the options for the Boost converter, but WeBench suggests the lowest cost/component count would be the TPS61022.


Make a spreadsheet model of each option, noting the actual loads and the efficiency of each at that load. This will help you decide.

That said I’d focus on the buck-boost + boost as probably the most efficient. It would also be possible if you could re-spec the 3.3V section to work down to 2.7V, then run it via an LDO direct to the battery. This would be even more efficient.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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