I'm planning to use a single cell Li-ion battery to power up a custom made circuit. I'm planning on using a ESP-12E module board as the application will need to connect every once in a while to internet to gather data, acquire sensor data, before displaying them on an e-paper screen, and then fall in a deep-sleep for a given time.

So globally the application tends to be low-powered, with currents ranging from approximately 50µA to 500mA.

Knowing this, the problem is that I'm not sure how to design the power supply. I first planned to use a LDO regulator, but my protection chip allows voltages down to 2.5V. So I was planning to allow the circuit to function with voltages down to 3V.

I looked toward Buck-Boost converters, and the LTC3440 chip seemed to fit my needs. However, the datasheet doesn't mention how it would behave in low load current situation (<1mA).

Consequently, my question is:

Should I review my application voltage range higher, meaning that it should consider voltages ranging from something like 3.5V to 4V+, or is the LTC3440 Buck-Boost converter would do a great job whatsoever and allow the application to work between 3V to 4V+ power?

The planned power stage without voltage regulator

  • \$\begingroup\$ One big problem with provided schematics is that there is no power path management. The load connected in parallel with the battery will affect charging profile. One way to deal with this is to use LDO or DC-DC with "enable" input, controlled by USB power. However this means you won't be able to use and charge device simultaneously. If you want this functionality you should look for charger IC with built-in power path support. \$\endgroup\$ – Maple Aug 26 '18 at 18:38
  • \$\begingroup\$ What is the lowest supply voltage that your ESP-12E module will operate reliably at? \$\endgroup\$ – Andy aka Aug 26 '18 at 18:39
  • \$\begingroup\$ @Maple Hum right I didn't think about path management. I successfully used this setup before in an application where I disconnected the load every time I was charging the battery, thing I don't intend to do this time. But it's needed this time, so I'll look into a different IC, thank you very much \$\endgroup\$ – Raphaël Ollando Aug 27 '18 at 0:22
  • \$\begingroup\$ @Andyaka the ESP-12E is rated for 3.0V to 3.6V supply ! \$\endgroup\$ – Raphaël Ollando Aug 27 '18 at 0:24
  • \$\begingroup\$ It looks like the question should be closed as "too broad", the answers are mostly "opinion-based". \$\endgroup\$ – Ale..chenski Aug 27 '18 at 0:34

By far the best solution is to use an LDO with low quiescent current.

What is the quiescent current (aka Iq) of a regulator? This is the current consumed by the regulator when output current is zero. In general, if you look for a linear regulator with low Iq, it will have much lower Iq than a buck converter with low Iq (compared to other buck converters). During times when output current is very low, the LDO will actually consume less power than the buck or other DC-DC converter. This means that the system level battery life will likely be better with an LDO than with a buck converter.

But aren't LDO's inefficient? Well let's look at that. What is the efficiency of an LDO? Assuming the quiescent current is low, the efficiency can be simplified to Vout/Vin. The average Vin for a lithium ion battery is 3.7V. 3.3/3.7 = 89%. So a low Iq LDO can achieve 89% average efficiency. Even if you use a 3V LDO, the average efficiency is still 81%. It is doubtful that using a buck will be worthwile even if VCC is 3V. Also, the buck will cost more (this only matters if you are going into volume production).

But lithium ion (and lithium polymer) batteries can be discharged below 3.3V. Most of the time, this is just a red herring. In reality, a lithium ion battery at 3.4V is pretty much fully discharged. You can continue to discharge it to a lower voltage, but the battery life extension gained by that is really minimal.

So it is still probably going to be better and easier to just use an LDO. Make sure it can support 500mA max, and make sure its Iq is acceptably low (maybe around 10uA or less).

| improve this answer | |
  • 1
    \$\begingroup\$ The thing is, if the circuit operates fine at a lower voltage, operating it at 3.3v or even 3.0 volt or 2.8 volt is itself wasteful. So the real efficiency is not determined by the voltage you chose to operate it at, but rather the voltage you could have operated it at. That said, switchers require care in radio circuits, as if done wrong they an substantially increase the noise seen by the receiver. \$\endgroup\$ – Chris Stratton Aug 26 '18 at 17:39
  • \$\begingroup\$ @ChrisStratton, it is true, what you say. You might be able to get better battery life by operating it at a lower voltage with a buck. But my answer applies down to 3.0, which is the lowest voltage entertained by the OP. Also, my comment about Iq still stands. State of the art LDO will have lower Iq than state of the art buck, so if the device spends long periods at a low-power state, the LDO will still likely give better life. LDO probably wins on cost and implementation footprint also. \$\endgroup\$ – mkeith Aug 26 '18 at 17:52

The ESP-12E module claims to be fully functional with Vcc=3.0 V. More, it is based on ESP8266EX IC. Per ESP8266 specifications, normal operating voltage for the IC is from 2.5 V to 3.6V. Therefore you should target the entire design for power supply for Vbatt = 3.0 V, which will give you maximum battery life.

There are hundreds of DC-DC converters that can operate from Vin = 3.1 to 4.3V (Li-ion supply range) while providing stable 2.9-3.0 V output with efficiency >90%. For example, if you try the Texas Instruments parts selector (and consequently their Webbench) with the following parameters,

Vin min = 3.1V, Vin max= 4.5V, Vout = 3.0V Iout=0.5A

you will have many dozens of options. Selecting "integrated inductor" category, my first choice led to TPS82085SIL part. You might find that you don't need any extra battery protection circuitry since the TPS82xxx has under-voltage lock-out automatically preventing the battery source from over-discharging. You can choose anything else that might suit you better.

| improve this answer | |
  • \$\begingroup\$ "Vbatt = 3.0 V, which will give you maximum battery life". Since we are talking about rechargeable batteries it should be safe to assume that the device will be taken care of and recharged as necessary. In this case the operational time gained by so deep discharge is not worth significant battery lifetime degradation. I agree with @mkeith here, 3.3-3.4V cut-off is much better. \$\endgroup\$ – Maple Aug 26 '18 at 18:19
  • \$\begingroup\$ @Maple, okay, maybe "battery life" wasn't the best argument to use the 3.0V level of power. 3-V or 3.1V is not the deep discharge, 2.2-2.5 is. The real argument is less complexity of buck switcher topology, and likely more efficient one, and less expensive. I recall there are hybrid switchers that go into LDO mode when the voltage difference is too low (or load too low) for a meaningful PWM duty cycle. Or into "burst mode". Maybe LDO is better, I didn't run the whole design cycle. BTW, LTC3440 shows input range down to 2.7 V for a Li-ion source. \$\endgroup\$ – Ale..chenski Aug 27 '18 at 0:33
  • \$\begingroup\$ Oh, and I don't know how you do your shopping, but 3x3mm 3A DC-DC with built-in inductor? Damn! And they also have TPSM84 with whopping 35A. How come I never saw those?! Oh, I know how... seem to be permanently out of stock at mouser and digikey :( \$\endgroup\$ – Maple Aug 27 '18 at 23:04
  • \$\begingroup\$ @Maple, there are many dozens of other options, theoretically and practically speaking. I just pick the top one. And the TI has 63 TPS82085SIL parts in stock if you want to try... :-) Actually, you must be brutally mistaken, the part is 2.9 x 2.9 mm :-) The TPS82698SIPR is even smaller (2.9 x 2.3), works 3.2 Vout at 3.4 Vin, up to 800 mA, and is in stock for $1.49, digikey.com/product-detail/en/texas-instruments/TPS82698SIPR/… \$\endgroup\$ – Ale..chenski Aug 27 '18 at 23:30
  • \$\begingroup\$ Yeah, I kinda dropped a ball on those dimensions. Interesting... they don't have soldering profile in the datasheet. I wonder how one can solder BGA without melting the SIP itself. Probably need IR for this, not handheld hot air \$\endgroup\$ – Maple Aug 27 '18 at 23:53

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.