Getting Torex boost LDO to work

Thread Starter

mmcginty

Joined Feb 4, 2010
57
My circuit is a sensor logger to WebAPI based on ESP32. Throughout the evolution of my ESP projects I've been using LDOs like the REG1117 to get 3V3 for the SoC. This worked for the most part, but became problematic when I started using LiPo batteries like the 18650. When the LiPo discharges down to the lower range of its output voltage, it's lower than the minimum required by the LDO in order to maintain 3V3. So at 3.9V battery the LDO output is 2.9, which is below the SoC's minimum, but enough for it to try to boot up. It goes into a continuous brownout error->reboot cycle, that over-discharges the battery.

I used a voltage divider and the ADC to detect the battery voltage, and at a certain level force it into a deep sleep until the PV recharges the battery. But even so, it underutilizes the full charge capacity of the LiPo, and though it greatly delays/reduces chances of any brownout death cycle, it doesn't preclude it.

My search for a better power solution led me to the Torex boost LDO, XC9261B33DER-G, which claims to be able to output 3V3 from a single AAA cell. It seems to be perfect for LiPo output voltage range and meets all my other requirements except for one: I can't get the damn thing to bloody work! The output voltage is the same as input. I tested for continuity between nets on a bare board, there is none.

The first image is my old tried and true power circuit. #2 is the schematic, #3 is the board CAD.

sch1.jpg

sch2.jpg

brd3.jpg

One glaring problem is the tiny wire going from the Schottky diode to the inductor, I forgot to make it wider again after a rip-up. But I soldered a piece of wire in to correct it, even tried bypassing the diode, didn't change anything. I'm about ready to give up and go back to my inadequate but working power component... a downside of going all SMD is it makes breadboard tests all but impossible.

If anyone has any insight, knowledge -- hell, even wild guesses at this point would be appreciated.

TIA.
 

Thread Starter

mmcginty

Joined Feb 4, 2010
57
hi,
Checking the d/s for theXC9, shows the Vinp is 2.7V to 5.5V.??
Not 1.5V ie: a AAA
E
Still well within the range of a LiPo, and not what I'm trying to do (maybe it was one of its cousins I'm thinking about. TI makes some that claim to work with 1V5 in, but not enough amperage at that input... and their example layout was for a 3 layer board that I can't do and couldn't interpolate.)

My input voltage will range from 3.7 (ideally) and 4.8 (with PV at max output.)

Thanks
 

Thread Starter

mmcginty

Joined Feb 4, 2010
57
Are you sure you have the right pinout? There are two different pinouts, for two different packages.
The part is SOT-25, I just checked it over, XC926 has a different pinout, I'm using XC9141, it's a different part, different package.
 

Thread Starter

mmcginty

Joined Feb 4, 2010
57
That one's a flyback, so it can only increase the output voltage. The other was a buck, so it can only decrease it. If you want to keep the output stable whilst the supply varies from above the output to below the output, you need a SEPIC.
Thanks for the reply. I actually only want to decrease it, 3.7 is fully depleted for a LiPo, target output is 3.3.
 

Ian0

Joined Aug 7, 2020
9,749
Is it really worth the extra complexity of a switched-mode? Will an LP2951 do the job? Maximum dropout voltage @100mA is 0.38V
 

Thread Starter

mmcginty

Joined Feb 4, 2010
57
Is it really worth the extra complexity of a switched-mode? Will an LP2951 do the job? Maximum dropout voltage @100mA is 0.38V
Looking at the complexity and all the support components for the SEPIC parts, my strong feeling is no! Unfortunately LP2951 doesn't supply enough current for ESP32 peak load (when the radio is powered up.)

Thanks for the suggestion.
 
Top