Full Arduino Mini LoraWAN below 1uA Sleep Mode

(Amedee) #264

Q1. There is no good reason to use the watchdog if you have external interrupt available
Q2. As already said, if you cut the power you loose the state, so you would have to store it “somewhere” (e.g. FRAM) to restore it. It is something I am doing on boards where I can’t get the power down to an acceptable level in sleep mode, but it is a pain.
Q3. If you are using @Charles design you can’t use @rocketscream low power library as the 5110 is not wired to an external interrupt (you need to use pin change interrupt)

(Bialy) #265

Could you clarify what you have on your mind? These two approaches are not mutually exclusive.

(Amedee) #266

Well, you need to take my answer in the scope of the question, that is using both a TPL5110 and the watchdog timer.

  • If you rely on TPL5110 to wake-up, why would you need to wake-up additionally on each watchdog cycle? The whole purpose of using a TPL5110 is to keep the Atmega in sleep until the timer is expired.
  • On the other hand if you have to make some housekeeping on each watchdog interrupt, then you don’t need a TPL5110 as the watchdog will keep you awake anyway…

So I really don’t see a use case for doing both.

(LoRaTracker) #267

My own view, for nodes where you can get the sleep current under say 5uA, is that the TPL5010 is a better choice.

You get the ‘wakeup’ interrupt you want, and if you dont respond to it in a timely manner (if for instance your node is locked) then the TPL5010 obliges with a reset to recover the node.

(Basse) #268

This clarifies it for me, thanks!

Too bad about the power library, I guess I need to read up on how to sleep/wake with “pin change interrupt”, thanks for this feedback, probably saved me hours of headache trying to get the library to work…

I guess I manged to choose a kind of advanced project as my first arduino and lora-project :stuck_out_tongue:
At least I got the BME280/lora/ttn to work, but it seems like the sleep will be the hardest part to figure out.

(M2mlorawan) #269

Will it be possible to get around .50uA ? Why is it so many difference between LoRa and LoRaWAN power consumption in sleep mode?

(LoRaTracker) #270

Its not difficult to modify a Pro Mini to get the sleep current down to 0.5uA, you remove the regulator and some resistors.

But then you have the issue of powering it.

AA Alkalines are not a good choice in my view, when new they are arounf 1.55V so you cannot use 3 since 4.65V is too much for the RFM95 without a regulator. AAs for around half their life are 1.2V, so 2 would give you 2.4V or lower, which can be an issue for some sensors.

As for needing to get as low as 0.5uA ?

Well at 0.5uA sleep current, assuming the board was only in sleep the battery would in theory last 639 years. At 25uA sleep current (which might sound a awful lot) the battery would last 13 years. The shelf life of the battery, due to self discharge issues, will be around 5 years. So there will be no practical differance in battery life between 0.5uA sleep and 25uA.

(Amedee) #271

I am using LiFePo4 batteries, they are relatively stable around 3.2v for most of their capacity (when they reach 3v, I have typically a week to swap them before they drop under 2.7v)

(LoRaTracker) #272

Yep, the AA LiFePo4 are almost ideal, safer than your average LiPo incendiary too.

3 x AA Alkalines is fine, and they are cheap at places like Ikea. Just use them with a LDO regulator with a low IQ such as MCP1700. With one of them you can get sleep current to just under 2uA.

(Luiz H. Cassettari) #273

I like the idea to use the LiFePo4 battery, but is expensive and hard to find.
MCP1700 regulator is great but I like the challenge to use 2 x AA Alkalines, for some projects works perfectly.