New ultra-low power Arduino board with built-in RF (RFM69)

Hello Mike. Thanks for posting that, I was actually looking for something similar as I'm after some board to run my IoT projects on batteries. Your video mentioned something like "The first Arduino to run years on a single AA", and I think you might be right... I couldn't find any other similar board with RF and step-up available. I only found some step-up regulators modules at www.pololu.com, but they don't seem to be really designed for low stand-by current.

At the moment I'm running my ATMegas directly, without any regulator, but I already faced some issues:

1. Impossible with a single AA

2. When using 2xAA it works at lower frequencies ok, but as soon the battery voltage starts to drop the brown-out kicks in and the MCU starts to behave erratically

3. Tried the CR2032, even the lithium cell keeping it's voltage stable during the battery life I noticed that on current spikes, like transmitting a radio, sometimes make the whole thing to restart

4. Using 3xAA things start to get bigger, and then I face voltage issue as 4.5V is over the max voltage for my RF module... having a LDO just waste the extra energy

Could you please let me know if your board has been tested on situations similar to the above? Also, what's the stand-by current expected when everything is sleeping?

For a bit of background, I do have some projects, from temperature sensors, to swimming pool activity sensor to simple HAL switches that I really want to install and forget about... and the battery/low-power seems to be the most challenging part.

Sorry for the big message, just looking for some clarification.

Danke

Hi mmoebius,

I understand your frustration, and that's why I've started this little project to share our expertise and make it available to general public. I'll try to answer your questions - sorry for the huge reply, but I like to explain things instead of giving the direct answers.

First the board consumption will vary depending on the supplied voltage, in a practical real-life example: Let's suppose you have a non-regulated circuit running at 3.3V and consuming 10uA when sleeping. If you add a step-up regulator in front of that with, let's say 2xAA (3V), the circuit will still consume the 10uA, but the step-up will draw a bit more from the batteries, something around 15uA. That's because it needs to compensate the 0.3V difference and also use some current for the regulator itself.

Here some tests I've done with Whisper Node board. I've being using the uCurrect from Dave Jones (https://www.eevblog.com/) to measure the current, so there's very little voltage drop/loss when measuring.

Test A - Step-up consumption only

Supplied Voltage (battery voltage): 2.4V (that's to simulate 2xAA at the mid/end of it's charge)

Circuit Voltage: 3.3V (Output from the regulator)

Circuit consumption: 0uA(I actually physically disconnected everything, except the regulator)

RESULT: 2uA to 2.5uA  (quiescent current)

In other words, the 2uA current consumption will be always there. That's used by the Step-up regulator to power-up itself. Still very little and keeping the step-up regulator always ON guarantee energy promptly available for the MCU when it wakes up. Some forums discuss the possibility of shutting down the step-up regulator. This technique would bring the regulator consumption down to less than 0.6uA but in this case you would  need to plan for the regulator start-up delay (1 or 2ms). Also step-up regulators don't like to start under heavy load, which might cause unexpected issues... specially when the feed by lower voltages.

Test B - Step-up consumption + 100K resistor load

Supplied Voltage (battery voltage): 2.4V (that's to simulate 2xAA at the mid/end of it's charge)

Circuit Voltage: 3.3V (Output from the regulator)

Circuit consumption: 33uA (The same circuit above but I hooked up a 100K resistor. According to the Ohms law: Current = Voltage/Resistance, so Current = 3.3V/100000 = 0.000033A, which is the same as 33uA or 0.033mA)

RESULT: 35uA to 45uA

The idea of this test was to check the step-up regulator efficiency at very low current. A 100K resistor is actually simulating a 33uA load, which could be a MCU and any other component sleeping. You probably will be able to get your components to consume much less than 33uA when not active. Anyway, the result is pretty impressive, showing the regulator works quite well at low currents... and here is where you project should spend most of its the time!

Test C - Step-up consumption + MCU, RFM69 and Flash Memory Sleeping

Supplied Voltage (battery voltage): 2.4V (that's to simulate 2xAA at the mid/end of it's charge)

Circuit Voltage: 3.3V (Output from the regulator)

Circuit consumption: around 2uA to 5uA (This is the estimated load for all components at sleep)

RESULT: 4uA to 7uA

As you can see the consumption is very, very low with everything at sleep and ready to wake-up. Here the step-up regulator starts to drop the efficiency a bit when compared to the "Test B", but still pretty amazing. In real life, if you have 2xAA batteries with a 2500mAh capacity, this board should be able to sleep for over 8 years (http://oregonembedded.com/batterycalc.htm).

More details about real-life test results can be found at the Whisper Node documentation page: https://bitbucket.org/talk2/whisper-node-avr

Now lets estimate the power consumption of a project which wakes up every 60 seconds and run for 30ms to do some task. Let's say read a sensor, blink a LED and transmit some data over RF. To make our calculation simpler we estimate 50mA consumption for the circuit during the awake period, it should translate it to no more than 100mA being draw from the battery. Those values are very conservative and the only time it might draw 100mA from battery would be during the wireless transmission... but let's being pessimist here! Now using the same Battery Calculator link (http://oregonembedded.com/batterycalc.htm), we would get an impressive 2.14 years life span. Now believe me, you can do a lot of things in 30ms. Also, if your project allows, you could store all collected data locally and only use the transmitter every 5 minutes, saving much more energy.

That should answer many of your questions... now a specific note related to the CR2032. Although those coin cells have 3.0V, they don't tolerate very well big current spikes. When a spike happens the internal battery resistance increases and the voltage drops significantly. Have the step-up regulator in from of that will help preserving the output voltage constant, but in this case you need to be more conservative with spikes, for example, never blink a LED while transmitting RF at same time. You migh need to sleeping things for some millis between heavier tasks to allow the battery to recover before squeeze more juice from it.

For more detailed information, here a blog post showing practical study how to build a sensor node running on a CR2032: https://wisen.com.au/blog/running-a-wireless-temp-and-humidity-sensor-for-over-1-year-on-a-cr2032/

At the end there's no magic here, the components are every day more efficient but we still need to design smart solutions when running things at low-power. To have an idea, even a few extra centimeters for the battery leads can cause a significant voltage drop, increasing the overall consumption. If you're interested on this subject, please have a look on the first part of a post I start related this subject: Low Power Budget | Talk² by Wisen

Let me know if I've answered your questions.

Cheers,

Mike M.

Wow! Thank you for the full explanation Mike, now I started to understand the pitfalls and the common mistakes I've been doing... and seems that you have already thought about those.

I really believe that Internet of Things revolution will only happen from the point we start to master the "low-power". As you mentioned, the wireless communication still a big challenge here. Anyway I hope your project succeed so we can have better options when developing low-power devices.

but alkaline batteries have shelf life of ~5years i think. In that case self discharge of AA batteries might be larger effect than your current consumption, hehe

Those days good alkaline batteries can last up to 10 years. Also the battery life calculator link above already consider around 15% for the self-discharge, so still a good estimate. If necessary lithium batteries (non rechargeable ones) can last up to 15 years

Cheers,

During the Forget Me Not challenge, I ran across these.

All about Moteino | LowPowerLab

I used one for my Hot Water, but in my case it was powered, so low consumption wasn't the criteria.

eLDERmon  External Sensor

This design differs by including an onboard step up regulator and the price is slightly improved.

Good luck.

Mark

Good stuff!

It's always nice to see people being creative how to "measure" real-life events. Is this project still alive? How did you find out the RFM69 communication at all?

Cheers!

Mike M.

Design Challenge finished ... (link should take you to it)

One of the other competitors was sharing his experience with these units.

I did find a distributor in the USA that was priced right and willing to ship to NZ.

The units via Low Power lab were also very reasonably priced (for low volume sales)

Mark

In your Kickstarter pitch you don't mention any RF testing, approvals or plans to get such approvals. Is this an oversight or are you just hoping for the best.  It is very unusual for an RF design to meet FCC or EU regulations without at least one test and tweak cycle.

MK

The RFM69 have some testing, and state that they meet FCC regulations.

Not sure if it's true but there are documents here.

http://www.hoperf.com/rf_transceiver/modules/RFM69HW.html