Lower voltage for (very) small current

[Watth]

Hey,
I'm working on a project of a device powered by a 402030 3.7 V 220 mA.h battery :  LP402036.pdf (1723.75 kB - downloaded 45 times.) https://www.tme.eu/Document/a1605afd8f195ae52678a26003ecf6ac/LP402036.pdf
That drives, amongst others:
µCU :  STM32L011x3 STM32L011x4.pdf (1688.05 kB - downloaded 69 times.) https://www.farnell.com/datasheets/2602778.pdf
e-paper display (EPD):  Electronic Paper Displays - ePaper 1.54 EPD, Spectra, w.iTC.pdf (1167.17 kB - downloaded 52 times.)E2154JS0C1 https://www.pervasivedisplays.com/wp-content/uploads/2020/12/1P226-00_01_E2154JS0C1_20201019.pdf

Voltage:
Battery's voltage (Vbatt) will vary from 4.2V while charging, to what I guess about 3.4 V
Both µCU and EPD are rated with recommended max V supply of 3.6 V:
EPD has operating min/typical/max V values : 2.3V 3.0V 3.6V.
µCU's datasheet gives no typical VDD, only a range between 1.65V and 3.6V
Current:
According to its datasheet, when active, the EPD draws about 4mA; Otherwise idles with no current at all.
From what I understood, the µCU's consumption varies from less than 1 µA to less than 7mA.

The battery voltage, especially while charging at 4.2V, will be too high for the µCU and EPD, I must include a way to provide about 3V to these to devices.
What would you advise to do so, knowing that the less wasted energy the better ?

Thanks for reading and take care of yourselves !

[tunk]

Can't give you any advice, but it may help if you say
what the estimated active/inactive times/percentages are.
Typical fully discharged voltages of li-po cells are around
3.0V (from the datasheet the BMS cut-outs at 2.75V).

[Peabody]

I guess in theory a switching regulator is supposed to be the most efficient, but my understanding is that at low current levels that's not necessarily the case.  For a few milliamps, I think I would just use a good LDO linear regulator that has low quiescent current.  Like the MCP1700.  I think the 3.3V would be ok if you're not taking the battery below 3.4V.  But a 3V regulator might also work.  It looks like all your stuff will work on either, but it will probably use less current at 3V than at 3.3V.

If you run the processor at less than maximum speed, that will reduce current consumption too.

[Watth]

I guess in theory a switching regulator is supposed to be the most efficient, but my understanding is that at low current levels that's not necessarily the case.  For a few milliamps, I think I would just use a good LDO linear regulator that has low quiescent current.  Like the MCP1700.  I think the 3.3V would be ok if you're not taking the battery below 3.4V.  But a 3V regulator might also work.  It looks like all your stuff will work on either, but it will probably use less current at 3V than at 3.3V.

If you run the processor at less than maximum speed, that will reduce current consumption too.

Yes, I think that the current is so small that even with bad efficiency, I will lose much less with a LDO linear reg than with a switching one.
I found this: https://docs.rs-online.com/cc81/0900766b8170d51a.pdf , what do you think?

[Kleinstein]

The NCP115 is still rather high in current consumption at some 50 µA. The MCP170x series is at some 2 µA.
The µC would likely need less current at a lower voltage like 2.4 V. As a first approximation the current for CMOS is proportional to the supply voltage.
Things may be a bit different for flash / EEPROM writing and maybe also for EPD.

A low power switched mode regulator may help for the higher current phases, but chances are it is not worth the hassel.

[mariush]

The microcontroller may require a minimum voltage to reach certain frequencies. 

As the display works with 2.3v ... 3.6v  I'd probably pick 2.8v as a fairly safe voltage. It's also a very common voltage, and you may find switching regulators hardcoded for this output, saving you a bunch of feedback resistors.
2.5v may also work, but it could be too low in some scenarios, like maybe if the device is used in very cold weather or something like that.

The display says it will consume up to around 4 mA, your micro's gonna be under 1mA ...

See for example : https://www.digikey.com/en/products/detail/ablic-u-s-a-inc/S-85S1PC28-I8T1U/8574276 , fixed 2.8v with around 250nA quiescent current.

here's a fixed 3v torex XC9265 : https://www.digikey.com/en/products/detail/torex-semiconductor-ltd/XC9265A3014R-G/5964774

[Peabody]

I still like the MCP1700.  The NCP115 has up to 10 times the ground current at the output current levels you're looking at, and a higher dropout voltage.  But that switching regulator posted by mariush also looks very good.

You haven't said anything about the duty cycle of your circuit.  Is it going to be on all the time, or will it sleep part of the time?  Also, will it always be battery powered?  How long do you need the battery to last?

[strawberry]

ST1PS03

[Watth]

I still like the MCP1700.  The NCP115 has up to 10 times the ground current at the output current levels you're looking at, and a higher dropout voltage.  But that switching regulator posted by mariush also looks very good.

You haven't said anything about the duty cycle of your circuit.  Is it going to be on all the time, or will it sleep part of the time?  Also, will it always be battery powered?  How long do you need the battery to last?

Indeed the MCP1700 looks more interesting.
Regarding duty cycle and all that, I don't know. It's more of a first real project for me, that I do as a hobby. The thing would be like a weird watch that can be plugged at night, but keep functioning.

[Watth]

ST1PS03

Not bad at all, indeed.

[SiliconWizard]

Or a number of similar TI parts, like: https://www.ti.com/product/TPS62840

[Peabody]

If you are new to this, you need to consider how the final device will be put together.  I'm just a hobbyist, so a lot of these parts that only come in hobbyist-unfriendly packages are just out of bounds for me.  The MCP1700 actually comes in a through-hole TO-92 package, as well as the usual SMD versions.  The tiny little switching converters generally do not.  Of course these days I guess you can get JLCPCB or the like to build the device for you, but where's the fun in that?

You will find that a big opportunity for saving battery power is to arrange for your processor to sleep most of the time.  And generally it only matters what percentage of the time it sleeps.  For example, if you sleep for 9ms, and wake for 1ms, you might think that's not sleeping for very long, but in fact it's still sleeping 90% of the time, so well worth doing.

[NiHaoMike]

Perhaps consider switching to a LiFePO4 battery? The voltage range is basically a perfect match for 3.3V logic.

[thm_w]

I use these 3V regulators:
- MCP1700T-3002 250mA 1.6uA
- XC6206P302MR 100mA 1uA
- ME6208A30M3G 200mA 3uA, very low cost, 5c each.

LiFePO4 is a good idea but IMO, seems quite difficult to source cells right now, and cost is much higher than salvaged or purchased li-ion cells. When a 3 terminal regulator is only a few cents.

If you are new to this, you need to consider how the final device will be put together.  I'm just a hobbyist, so a lot of these parts that only come in hobbyist-unfriendly packages are just out of bounds for me.  The MCP1700 actually comes in a through-hole TO-92 package, as well as the usual SMD versions.  The tiny little switching converters generally do not.  Of course these days I guess you can get JLCPCB or the like to build the device for you, but where's the fun in that?

Its a tiny watch with STM32, OP will not be using through hole.
SOT-23 is quite a hobbyist friendly package once you get some smd experience.

[Watth]

I'm impressed by the quantity and quality of the replies here. Thanks a lot every one.