I could need some advice from anyone with experience designing low-power Internet of Things devices because I’m working on a project on it right now. The objective is to design a gadget that can run on battery power for prolonged periods of time with dependable performance and communication.
The following details pertain to the project:
Power Requirements: A single battery charge must be sufficient for the gadget to function for at least six months. Although I’m thinking about utilising a Li-ion battery, I’m open to suggestions for other batteries if they provide higher efficiency or performance.
Connectivity Protocols: Wi-Fi, Bluetooth Low Energy (BLE), and LoRa are a few of the connectivity choices I’m researching. Understanding the trade-offs between these alternatives in terms of range and power usage is something that really interests me.
Strategies for Power Management: To save energy, I intend to use sleep modes and low-power components. Please offer any advice you may have on power management techniques or particular parts that have proved successful for you.
Design Considerations: When creating a low-power Internet of Things device, are there any design factors or typical traps that I should be aware of? I would be grateful for any guidance on choosing components with the least amount of power usage or optimising PCB layout.
If you have no RF requirements, here are the major players in order from least power to most power:
New WiFi Backscatter
BLE (modern Bluetooth, ANT), Matter, Zigbee, 802.15.4
WiFi
LoRa
Cell modem
Satellite
Disclosure: I am the creator of Joulescope which you should definitely buy so that you can accurately quantity your device’s energy performance during development.
I don’t think anyone here can really answer your questions easily. You are close to asking everyone to design it for you. I will offer a few tips however.
We have design really low power sensors 20yr life with broad temp ranges as well as BLE for multi year applications. These were generally done with primary cells not rechargeable batteries. watch self discharge on your battery and charging circuit!
Buy a joulescope.
hitting the datasheet power specs is hard. leakage currents due to dirty boards, dielectric leakage, I/O pin settings all impact power a lot. charging and discharging capacitors when sleeping and waking can even kill you.
your first several prototypes should be big and have liberal use of zero ohm resistors. You want to be able to individually measure as many components as you can. Deconvolving what components is causing excessive power loss is tough without individual measurement.
accurate power analysis of wireless protocols is non trivial though many vendors have calculators these days. Your power will often be dominated by radio use Tx and Rx and heavily dependent upon duty cycle. BLE can be great for low data but has some hard boundaries depending on your need for compatibility with other devices (like phones)
Good luck, most importantly let us know how it goes so everyone can learn from your design experience.
It would be useful to more fully define the product specs. A “single battery” for a lipo pack can be anywhere from 20mAH to 10,000mAH (or more) and span quite a bit in size.
Beyond designing the hardware for low power, you need to trade off sleep versus active time duty cycle. An outdoor temperature sensor can last a few years on a pair of AA batteries, sending temperature updates every ~15 minutes. If it reported data every minute, it’d only last a few months.
The first step is to outline your requirements and target specs – and then work to meet them or to realize that it might not be achievable and to negotiate some tradeoffs.
A few things that might help (in no particular order)
1.) Buy a Joulescope. Must have for serious low power dev.
2.) Here is a paper on power optimization of the DSP, raw driver code, etc. for the nRF9160 (Cellular / CAT-M). I profiled/optimized everything with the Joulescope
Being able to see high resolution / high sample rate current waveforms will reveal a lot of things about your firmware.
3.) Self discharge / leakage of Li-Ion is often greater than the sleep current of a modern MCU. For example the sleep current of nRF9160 or nRF52840 is on the order of uA’s.
For Industrial IOT products for multi-year life, a popular option is lithium thionyl chloride (Li-SOCl2) primary cell. Low self discharge, long shelf life (nearly forever as they self passivate with no load)
Rechargeable almost always have poor leakage/self discharge.
4.) Once you can power profile your code, operational states, etc, a simple spreadsheet can make accurate predictions about battery life. You need to capture your desired duty cycles, etc. (how often you will transmit, time spent doing sensor capture, etc).
Basically, the tool to power profile and you operational model is everything. Without it, the chances of optimizing the wrong thing is near 100%. Sometimes it is not obvious which things you should be optimizing until you see what is actually happening in your circuit.
This is an iterative process. You will probably find along the way, small adjustments to your operational model can yield huge improvements in battery life.
Just for completeness (which bit us once) is that the passivation can work against you as well. If you don’t have a minimum load on the lithium thionyl chloride it that can ruin them (the passivation can get past a point of practical no return). Otherwise they are an amazing battery for broad environmental condtions (that’s what we use for military 20yr life sensors)
,
I would be grateful for any guidance on choosing components with the least amount of power usage or optimising PCB layout.
in advance.