Feature request: firmware pre compiled binaries, without the interrupt driven low power sleep

[jerabaul29]

Follow up of https://github.com/sparkfun/OpenLog_Artemis/issues/126 :

Can we get some pre compiled firmware binaries that deactivate the power loss interrupt, so that in case of a transient power interruption for example due to vibrations, the OLA just stops / brows out temporarily, and then recovers / reboots when the power is back, rather than get stuck in a deep power sleep?

Use case: uses of OLA that need to log, "no matter what", but where reboots are not such a big issues, and are inaccessible for user reset button pressing.

FYI @tnn77

[adamgarbo]

@jerabaul29

In case you weren't aware, this is something you can do yourself from the Arduino IDE:

Sketch > Export compiled Binary

Cheers, Adam

[jerabaul29]

@adamgarbo thanks :) . Yes, I know, this is already what I do with the OMB firmware :) . The idea here is that I have not set up the libraries in the right versions to compile the OLA firmware, and I am not too familiar with the OLA firmware either, so would feel 'safer' if someone very familiar with this specific project tales care of the setup and configuration and compilation :) .

[jerabaul29]

Any update on this? In addition to firmware update / different versions, could an option be to 'simply' add a connection pad between the coin cell battery and the Artemis, so that the coin cell battery can be disabled altogether if wanted?

[ryanneve]

@jerabaul29 I've been disabling this behavior in firmware after our OLAs have gone to sleep and never woken up. Resetting the OLA fixes it, but my OLAs are very hard to get to. Batteries seem to always be fine. It's a mystery. See #127 Unfortunately, I've got a new PC so need to re-setup my development environment. Will try to get a change pushed that adds a configuration option to ignore low voltage sensor interrupt.

[PaulZC]

This is still on the things-to-do list. I haven't forgotten, I just have a ton of other projects to tackle first...

Sorry for the delay, Paul

[PaulZC]

Hi @jerabaul29 @ryanneve ,

Apologies again for the delay. I finally found time to look at this...

I've created a new binary for you which I hope will meet your needs. You can download it from here:

https://github.com/sparkfun/OpenLog_Artemis/blob/release_candidate/Binaries/OpenLog_Artemis-V10-v23-NoPowerLossProtection.bin

The new "NoPowerLossProtection" option does two things:

• It stops the PIN_POWER_LOSS interrupt from being generated (see here for details)
• It sets up the Watchdog Timer to reset the Artemis if the code crashes during a low voltage event (I'm reusing @adamgarbo 's WDT code here)

Some things to be aware of:

• The code still goes into deep sleep between measurements if the logging interval is > 2 seconds
  • The WDT is disabled during sleep (otherwise it would keep waking the Artemis up)
  • If the power is lost while the Artemis is asleep, the back-up battery will keep the processor running and it will stay in deep sleep until the next wake interval
  • When the Artemis wakes, it will draw all of its power from the back-up battery. Its voltage will collapse quickly. The Artemis will then restart when power is reconnected.
  • If power is reconnected before the Artemis wakes, then it should simply wake at the correct time as if nothing had happened
• If the battery voltage falls below the battery voltage threshold, and you have low battery voltage detection enabled via the power options menu, then:
  • When the Artemis wakes from deep sleep to take a new measurement, it will check the battery voltage, find it is low, will print a warning message, and will then reset itself
  • The Artemis will then keep going around this detect-low-voltage-and-reset loop until the voltage level returns to normal.
  • The new code for this is here.
  • I think this is the only safe way to avoid getting stuck in another needing-a-reset scenario... Maybe? I'm not totally sure...

I think this is probably the best I can do. I hope it does what you need it to do! Please test it and report back here.

Best wishes, Paul

[jerabaul29]

Amazing, many thanks!

I wont have the hardware to test, but maybe @tnn77 can? :)

[ryanneve]

What do you think about still checking if (digitalRead(PIN_POWER_LOSS) == LOW) powerDownOLA(); in setup() but still disabling the interrupt. I think there is much higher likelihood of this being a real power loss if it is seen at boot, rather than during operations. This would just mean moving #ifndef noPowerLossProtection down one line.

[PaulZC]

Hi Ryan,

But if we do that, the code will almost certainly need a reset to wake again. powerDownOLA puts the Artemis into a deep sleep state. If there is any charge in the back-up battery, the battery will keep it in that deep sleep state... Only a reset will wake it. Sounds ~risky to me?

Thanks! Paul

[ryanneve]

@PaulZC , you may be right. My thought was that transient power issues probably won't appear while the OLA is in setup(). Let's try it your way and see how it goes. Is the goal to change this to a configuration option eventually? Excited about the watchdog timer!

[PaulZC]

Configuration options are nice, but, for critical stuff like this, I always lean towards having it hard-coded. Then there's ~zero possibility of the WDT being mis-configured. The WDT is there to help save you from code-corruption during a brown-out. If running the WDT is selective, there's a small but non-zero chance of it being disabled when you really need it.

Please give the binary a thorough shake, whenever you can spare the time. I'm grateful for the feedback.

Very best wishes, Paul

[tnn77]

@jerabaul29 Yes, I have one at my desk. I'll try to find time to test the binary. I'll let you know when I get a chance!

[PaulZC]

This feature has been included in v2.4. Going forward, we will include noPowerLossProtection binaries for each new release.