Arduino Nano BLE 33

Hi guys,

For my Masters project I am using an Arduino Nano BLE 33. The idea is to power the Ardiuno Nano BLE 33 from a Thermoelectric Generator (TEG) which is capable of outputting around 2.2uW/s (30mV, 73uA). This energy is passed through a power processing unit (PPU) to step up the voltage to 5V as required by the Arduino Nano BLE 33. The Arduino will then power a temperature sensor and collect the data, it will also transmit the data wirelessly via Bluetooth to a users smart phone to view the readings. I have tested the Arduino BLE 33 and concluded that it works at 0.042W/s (4.2V, 10mA), I need to significantly reduce that value down in order to make the burst of energy provided by the PPU accommodate for the required time of data collection and transmission of about 2.5 seconds. Any suggestions on how to reduced the power consumption or perhaps has anybody dealt with low power micro controllers and transmission units as an alternative to Arduino?


Some microcontrollers have ultra low power or sleep modes. Like this one, also with BLE and apparently battery charging:

That processor can't even be run in sleep mode with the OP's power supply.

Ultra-low sleep power: 5 μA, deep sleep model

I don't think there is much chance of running a processor in an actual working mode with that amount of power.

Hi @mrmiko. I see there is some information in this thread:

It is a little bit old, but I think it should still be valid.

Very low power sensors, MCUs and transmitters can be powered by thermoelectric generators if you store sufficient energy for a short burst of activity.

The LTC3108 chip is designed for exactly such applications. Demo boards are available, and the energy storage cap can be sized according to your needs.

If OP can make it sleep when they aren't using it, it will significantly increase battery life.

Thank you for the suggestion this seems to check out for the sleep mode however operating current is 200mA which in my case is near impossible to achieve. I have posted a question to clarify whether that is accurate as it seems excessive for the board.

No, that seems about right. Of course, it generally depends on what the board is doing.

Thank you for taking your time to help, there is a lot of relevant information in there, the deep sleep mode seems to require a lot less power. My concern here would be the loss of functionality due to the low power. I want to wake the board up once it receives a reading from a sensor taken at an interval, this interval would be determined in the fully functional running board which is unfortunately not occuring in sleep mode.

Thank you for the suggestion of the LTC3108, I am currently using it in my project as you may be able to see from the image. I am capable of powering the board for an X amount of time (it's in milliseconds). Unfortunately that is not fast enough for me, I am looking to power it for 2.5 seconds. The datasheet of the LTC3108 suggests microcontrollers capable of operating in the region of single digit milliseconds. Is Bluetooth perhaps something to reconsider as a means of data transfer in order to save energy? Bluetooth takes the longest time to establish a connection which is an issue.

You could use radio. But you'd have to build a base station.

In that image a supply of 2V is applied to a Peltier module which generates a heat about similar to that of a human body. This Peltier is in contact with a TEG which outputs 30mV through a temperature difference of 16K. The TEG output is passed to the LTC3108 where I can achieve a burst of energy of about 2-3mA at 5V.

Is there any resources you could point me to in that field ? I have never used radio as a means of communication. Is radio power hungry ?

nRF24L01 is not power hungry but is impossible to get working the first time. Once it does work, though, it's amazing.

I have one connected to a Nano which is being powered off a 9V and it's lasted a fairly long time.

EDIT - I believe they use about 5mA when actively transmitting. They will need a host microcontroller, and use the SPI protocol.

Radio is power hungry. The less you ask the radio to do, the less power it takes. Anything that uses complicated, bidirectional protocols, like Bluetooth, WiFi, etc. will take a lot of power.

Transmission range is another issue, which you haven't mentioned.

What your project apparently lacks is a realistic assessment of the goals, the energy required to achieve those goals, and reconciliation with the chosen source of energy.

The Nano 33 BLE is not the place to start, for a true micropower project.

1 Like

It might have escaped your notice but the OP is not intending to use batteries.

Sorry. Force of habit for low-power stuff.

I don't see a heat sink in that photograph. For a TEG to work efficiently, you have to maintain the flow of heat, which means to get rid of it as fast as it comes in.

If you intend to transfer the exhaust heat to air, keep in mind that flat surfaces are extremely inefficient. You need a very large finned convection surface to allow air flow, or some other means of disposing of the incoming heat, like a large metal block.

Something like a CPU cooler can help. This is for a cold plate:

You might be able to get by with a 20ma burst of about 10 millisecond with an NRF24 and a low power MCU