ESP8266 Community Forum

- Mon Jul 20, 2020 10:34 am #87977 Hi,

I'm working on a circuit that involves a load-cell + an HX711 chip that is used to measure the weight of something every pre-defined time, report it to a web server and go to sleep. I am currently using a power-bank connected by micro-USB to power the circuit. The power-bank lasts for 4 days when set to wake up every 1 minute and send the measurement over wifi only when there's a change in weight.

So I've been thinking about alternative ways to power this circuit and I was thinking of powering it using a 18650 Li-Ion battery (or batteries).

The HX-711 requires 5V to operate properly, so I understand I either need to use one 18650 battery with a boost converter or 2 batteries with a buck converter. I would like the circuit to be able to run without recharging for at least 2 weeks, and I am fine with having the deep sleep interval to be around 15 minutes (although ideally it would be less).

Is this possible using a single battery, or do I have to use two? What buck/boost converter would you recommend that is reliable and energy-efficient.

Thanks!

BTW: The specs of the power bank read:
Input: DC 5V 2.1A
Output: DC 5V 2.1A
Battery: 3.7V 4000 mAh (14.8 Wh)

- Tue Jul 21, 2020 12:36 am #87982 Lots of questions about this!

Firstly is this your own circuit? A micro USB may not be the best power input for this and you may have more luck going direct to the board. At very least it's going to be powering the USB chip as well as your module.

If this is a module with a micro USB then what does the power regulator and circuit on the board look like? You may find that the power circuit is set up assuming you are powering via USB from a computer and so it's always on and drawing current. Also I have found a lot of the modules use common LDO regulators that are not just inefficient, but also waste any voltage difference as heat.

What's your 5V sensor current draw like? The HX711 has a powered down mode - are you using that while you're sleeping? What about the load cell current draw? Do you know what that's like?

I've spent a lot of time trying to optimise the power train on the ESP and have found a couple of things:
- a buck/boost converter is more expensive than an LDO, but in buck mode they are very efficient and significantly extend battery life compared to LDOs
- in boost mode though, they pull exponentially more current in order to supply the increased voltage, and so an awful lot depends on the current draw and design of your 5V circuit. You really need to design it so that it's powered off except for the instant you measure.
- it would be worth designing two separate power circuits so you can separate the 5V from the 3.3
- the ESP can run down to 2.8V so it may be worth designing the whole ESP circuit for 2.8V operation

I've looked at the HX711 and it can run at 2.6V so it may even be worthwhile looking at why you need 5V for this - if it's your load cell then swap that out. If you could get everything running at 2.8V or even 3.3V then you'll have a lot more luck with a buck converter going straight to the ESP unit. I would get rid of anything USB related on the board, or at the very least set it up to allow the board to be independently powered by a 2.8-3.3V cell, and only powered by USB when connected to the computer.

- Tue Jul 21, 2020 5:49 am #87987 Thanks so much for the detailed response. I'm still trying to soak it in so I may have further questions later, if you're okay with that

If I go with your approach and create a 3.3v circuit, using a single 18650 battery, will a buck converter keep outputting 3.3v even though the battery voltage lowers as it is being utilized? What happens if the voltage drops below 3.3v?

Can you recommend specific a buck/boost/buck-boost converters are of high quality and energy efficient?

Thanks again.

- Tue Jul 21, 2020 5:42 pm #87993 No most buck converters will output their specified voltage but will then output the battery voltage as the it drops below that. If you want a steady voltage you need a buck/boost, which will switch to boost mode when the voltage drops below the designated level. The ESP units will run down to 2.7 or 2.6V from memory, but quite a few of the sensors need a stable reference voltage, and just thinking about a Wheatstone Bridge type load cell I would imagine that would be one of them.

I made a bunch of modules with a buck/boost converter at 3.3V (I used the ISL91107IR units - https://www.renesas.com/sg/en/doc/datasheet/isl91107ir.pdf) and what I found was that they were really efficient until you got to 3.3V but then when they switch to boost mode, they increase their current use to make up the voltage. The increased current draw pulled the voltage lower as the battery drained faster, which increased the boost needed to make up the voltage, which increased the current, which increased the speed of battery drain, which...

In the end, your standard LiPo battery cut off voltage meant that there was almost no benefit in using a buck/boost because of this cycle - I reckon I got a few hours more out of the battery and that was it.

Then I looked at the effect of operating the whole unit and my sensors at 2.8V and that got me a much greater benefit, so I switched to a pure buck converter that output 2.8V, and switched all my sensors to 1.8V or 2.8V units. I'm still looking for a particle sensor that runs at a lower voltage - all those type of sensors have a fan and these all seem to be 5V.

If you can run your load cell at 2.8V, you can run your HX711 at 2.8V.