I'm working on a battery-powered device using an ESP32. It needs 3.3v, and my battery is Li-Ion, so I've got 4.2-3.5ish (of useful voltage, anyway) - so max dropout has to be around 200mV. Most of the time, my device sleeps and consumes <10uA, so I need an LDO that's got a really low quiescent.

There are two other design considerations: First, sometimes I plug it in to USB to charge, so my available voltage goes up to ~5.1V with a safety margin. Second, my device sometimes spikes up to 600mA. This means I might need to dissipate up to ~1.1W.

I haven't been able to find an LDO that can meet all those requirements. Looking at block diagrams, however, the LDO itself doesn't seem terribly complex. I'm debating building one out of discrete components using a small LDO as a voltage reference so it's temperature compensated, and an op-amp driving a PMOS FET (or a few in series to spread out the heat).

Is there any reason I shouldn't do this? Or is there any other way to get the current I need? I looked into putting LDOs in parallel, but a balancing resistor always seems to add too much voltage drop, and not keep the current evenly balanced enough. I don't THINK a switching regulator is an option because of the necessary super-low quiescent current.

I would appreciate some input! Thanks.

  • 2
    \$\begingroup\$ Switching regulators are by far your best option, actually. Check out the TPS627431, for example. It can only provide 400mA but I'm sure there are other ultra-low quiescent current buck converters that can do more. Depending on the length of the 600mA spikes, capacitors might also work (together with such a buck converter). \$\endgroup\$ Nov 12, 2022 at 2:20
  • 3
    \$\begingroup\$ "Is there any reason I shouldn't do this?" Yes -- compensation of an open-drain amplifier, including at low voltages, and over wire ranges in current and load impedance, is very much nontrivial. I would say designing a good one is as much trouble as using a switching regulator here! \$\endgroup\$ Nov 12, 2022 at 2:50
  • \$\begingroup\$ See the TPS72501; its datasheet discusses heat sinking in detail. \$\endgroup\$
    – CL.
    Nov 12, 2022 at 8:56
  • \$\begingroup\$ @JonathanS. I hadn't really considered them because I thought the quiescent would - in general - be too bad. Can I get your opinion on this one? The quiescent is on the upper end of what I'd want, but the efficiency seems OK... datasheet.lcsc.com/lcsc/… \$\endgroup\$
    – Helpful
    Nov 13, 2022 at 0:07
  • \$\begingroup\$ @TimWilliams If you give that as an answer I'd accept it. The simulation made it look easy, but I didn't have to worry about oscillation and such. \$\endgroup\$
    – Helpful
    Nov 13, 2022 at 0:07

1 Answer 1


Is there any reason I shouldn't do this?

Yes -- compensation of an open-drain amplifier, including at low voltages, and over wire ranges in current and load impedance, is very much nontrivial. I would say designing a good LDO is as much trouble as using a switching regulator!

  • \$\begingroup\$ Perhaps this is worth asking as a separate question, but can you think of any other good options? I've been unable to find a sufficient LDO - are there good options to switch over once the current reaches a certain limit (and such a solution that doesn't draw much current itself, since that's important for the sleep mode)? I'm also considering buck regulating, but I haven't found one I'm happy with yet. Are there good - even active - ways to parallel multiple LDOs, as long as the circuitry doesn't draw much current at rest? \$\endgroup\$
    – Helpful
    Nov 13, 2022 at 4:34
  • 1
    \$\begingroup\$ Pretty sure I've seen LDOs rated for more current and power than that, though not sure offhand if they also had low enough dropout, and what part numbers. Search around, ought to be something out there. Well. Maybe not in stock these days... But that's another matter, sort of. Speaking of dropout, you might consider a marginally lower output, I don't remember what ESP32 is rated for but if it's 3.0-3.6V you could do 3.0-3.1 for example. Makes things slightly easier. \$\endgroup\$ Nov 13, 2022 at 4:39
  • \$\begingroup\$ It's 3.0-3.6, but I've found that the other stuff on the line can cause enough voltage sag/transient that the ESP resets - some of that can probably be addressed with more decoupling, but some lasts too long for that to be practical. One of my big limitations is my fab house - they have a limited parts selection, so I'm trying to explore alternative solutions. Any controllers to balance LDO outputs? Or current-sensing switch that could changeover to a higher-current (but also higher-quiescent) LDO at a certain mA threshold? Just exploring. \$\endgroup\$
    – Helpful
    Nov 13, 2022 at 4:50
  • 1
    \$\begingroup\$ I sense many frustrated late night hours in front of SPICE 😂 \$\endgroup\$
    – tobalt
    Nov 13, 2022 at 8:26
  • \$\begingroup\$ To be more specific it’s easy to make a discrete regulator stable. But hard to make it stable and have good transient performance. Buck converter is the way to go. \$\endgroup\$
    – asdf30
    Sep 18 at 2:15

Your Answer

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

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