@davee @papajoe0418 @zander

You may be right, the pico deepsleep on the pico may be flawed at the moment.  A soon to be released mp update may sort it.  But the article on deep sleep I linked to is suggesting creating a new funtion in C and rebuilding mp to incorporate it into the package.  

But I think the solution proposed by Davee is probably the best way to go. 

@byron @davee @papjoe0418 Here is a look behind the curtains. I would NOT attempt the dormant mode yet, even though the 0.8mA is enticing. Either put up with what you have or switch to the esp32

Here is a link I found that is enlightening but ultimately disappointing HERE

@papajoe0418 @byron @davee I just looked at the techies forum trying to get the PICO sleep sorted out and as of Jan 15, they are still not successful. If you can switch to the esp32 with an RTC external module you can run for weeks on a couple of batteries. The esp32 is much older so the code is stable, PICO is still a toddler and this code is tricky.

@davee @papajoe0418 @zander

Ron, your linked doc is dated Feb21.  The doc I linked to is dated Oct21 by the same author, and they seem to have a solution to the deepsleep.   Neverthless, for my money, the external circuit solution is the one I would go for. 🙂 

@byron Indeed you are correct but when will that code be released? As of Jan 15 this year, they were still not able to get dormant working where the draw is down to 0.8mA. The lopower fix was last updated 6 months ago but the developers' updates are as of Jan 15. The 'fix' you pointed out is flagged as experimental with all the usual caveats. For a remote hands-off project that may be an issue. Again, I recommend switching to esp32 as it is stable.

@zander - I agree, its probably only worth using the deepsleep if its in a stable release.  A new stable release is coming shortly I understand, but whether that will include a better deepsleep such as that shown in the doc's (which was not anything to do with the official mp team) I do not know.  

But, personally, and bearing in mind the length of time between the 'hoots'  I would simply go with an external circuit that powers down the pico entirely and powers on again.

A high capacity lipo battery could be used such as is put in robot cars or drones, with a voltage reduction circuit to power the pico (or esp32 for that matter).  In this circumstance the battery should last for a very long time. @Davee could probably advise on a suitable circuit and the length of time that could be expected from a typical 2250mAh battery.

@byron @davee The OP does not want LIPO, only store-bought AA batteries. These devices will be maintained by non-technical people so rechargeable (with the possibility of exploding) batteries are out.

Hi Ron @zander, @byron , @papajoe0418

A quick Google on ESP32 ... treat the information sources with the usual Internet information caution!

Also, I have only glanced at this .. so I may have it all wrong!!

ESP32 ... looks to have a 'genuine' low power consumption mode ... e.g.

https://lastminuteengineers.com/esp32-deep-sleep-wakeup-sources/

says ... from 150 micro amps ... maybe down to 10 microAmps ...

but the big caution is that this is with a low accuracy clock 

That is, if you wanted to wake up roughly 1 hour intervals ... but didn't mind if they were (say) 55 mins or 65 minutes apart ... then it would be worth looking at more closely 

But you would still need a 'Real' RTC for the longer term timing, to keep it in sync with the outside world, as over a few days it could be several hours out ...

......... unless anyone has more experience to share?

Best wishes, Dave

@davee Umm, should we say 'borrowed' from RandomNerds? 

https://randomnerdtutorials.com/esp32-deep-sleep-arduino-ide-wake-up-sources/

I ran that sketch earlier today and showed it to the OP. Here is the screen print I showed him.

Hi @byron + Ron @zander & @papajoe0418,

As it happened, this was discussed on page 1 of this thread where I noted:

AA capacity is apparently Alkaline batteries from 1,700 mAh to 2,850 mAh ( https://en.wikipedia.org/wiki/AA_battery).

 which is remarkably close to the lithium battery capacity you mentioned ... so 1 lithium cell of the size you mention has roughly the same energy as two AAs.

I know nothing about these owls (so far as know, there aren't any in my part of the world), but I am guessing @papajoe0418 is considering an annual 'hooting season' of maybe 4-6 months, and on that basis, I was hoping (without knowing if it was a crazy idea or not) that we could find as solution that would hoot for the complete season on one set of AAs, assuming the recent current measurement whilst actually hooting of under 2mA is valid.

So let's do a quick back of the envelope calculation ...

Hence, it is all about running a clock/wake up mechanism for (say 6 months) on about 1 amp-hour of battery, assuming the 'battery' is a set of 3 'good quality' AAs.

6 months is 6 * 30 * 24 * 3600 = ~16 million seconds

So you need the current to be about 1/16 th of a micro Amp ... about 60 nanoAmps ...

This obviously rules out the 2 processor families in discussion above, in terms of a 'sleep' mode .. but either if they are powered down nearly all of the time!

An external RTC, with an alarm wake up function, designed to run from a coin cell, can either be provided with its own coin cell or will only require a few milliAmp-hours from the AAs ... either would work.

Does this make any sense? Have I made any obvious mistakes and put the decimal point in the wrong place?

By the way, with circuits requiring such low power consumption, it will be necessary to take extra precautions to minimise leakage currents across the board, etc, possibly encapsulating the circuit in some waterproof layer, etc. ?

Best wishes, Dave

@davee I am pretty sure the boxes will be/need to be checked more often than once per 6 months but only @papajoe0418 can tell us for sure. If needed, then more batteries can be added but I am pretty sure PapaJoe was planning on only 3 batteries see https://forum.dronebotworkshop.com/postid/37777/

We need PapaJoe to chime in with some decisions, battery type, rechargeable, visit frequency.

@davee I sent you a PM

Hi @byron + Ron @zander & @papajoe0418,

   It has just occurred to me that I made a stupid error in my calculation ... apologies ... and having logged back on I now see from Ron, that the Wikipedia link is problematic.. I usually check them, but I forgot this time ...  ideally I would like to fix the original message, but as the 'edit' time window has expired, I'll have to do it in this new message.

The Wikipedia article on AA battery is at https://en.wikipedia.org/wiki/AA_battery

The calculation was wildly out because I forgot battery capacity is in Amp-hours ... not Amp-seconds ... 🙄 Of course I know the difference, but the SI unit of time is seconds, so in most calculations you need to convert to seconds, and over the years it becomes an 'auto-pilot' operation.

-------------------

I explained my assumption that the unit should run for up to 6 months on 1 set of AAs is 100% my imagination, but until a clearer specification is supplied, I will continue using it. My consideration was that the 'real' cost of each unit should be the sum total of the unit itself, plus the cost of batteries for at least one 'hooting season'/year. I was also assuming the electronics would be purchased at the 'keenest' prices available. Of course, local pricing of batteries and components may be very different in different places, so this was only a general aim, not a precise calculation. However, it provided a strong incentive to reduce the number of sets of batteries to a minimum, preferably just 1 set.

-------------------

So the section should have read:

So let's do a quick back of the envelope calculation ...

Hence, it is all about running a clock/wake up mechanism for (say 6 months) on about 1 amp-hour of battery, assuming the 'battery' is a set of 3 'good quality' AAs.

6 months is 6 * 30 * 24 = 4320 hours

So you need the current to be less than 1/4 of a  milliAmp ... about 230 microAmps ...

This is obviously easier than the previous calculation, and to some extent it changes the conclusions!

In principle at least, this allows the ESP32 to be considered for as running in a 'Deep Sleep' role all of the time ... but I am not clear if this 'increased' current consumption will enable the available internal real time clock capabilities to be sufficiently accurate over a period of weeks or months. It may also be adversely affected by temperature. It might be possible to find published material to ascertain the answer to this question, but before committing to this approach,  I think it should be carefully checked by experiment as well.

I think the Pico can only be considered if it is powered down for most of the time, as might be achieved by using an external RTC with alarm function controlling the power to the Pico using a MOSFET.

-----------

Best wishes, Dave

@davee FYI @papjoe0418 @byron In this application we can use hibernation mode which only draws 2.5 microAMPS. It REQUIRES an RTC module to wake it up and can't save any data but in the case of this app that is not a problem. IF that were an issue, the next level up is deep_sleep at

10 microAmps.

A real quick calculation tells me these batteries will last over 10 years. They will probably need to be changed due to old age before that.

@davee  @zander

For the lowest power draw an ESP8266 may be and even better bet than the esp32.  Things on Edge produced a board based on the ESP8266  (a Cricket) that managed the following:

• Ultra-low power: 1V – 3.5V (e.g. can run directly on AAA batteries)
• 0.5uA deep sleep
• Battery monitor
• ~3 seconds latency data transmission from a sensor to client devices

I have one of these devices in an outbuilding that transmits a temperature via wifi every 15 minutes.  One AA battery lasts about 3 months

But sadly they are no more, and went to the wall during the pandemic, but I would expect the esp8266 will have a lower power requirement than the esp32.  

Both the esp8266 and the esp32 will run micropthon, though beware that some early esp8266 boards only had 0.5mb or flash.  Normally one now finds 1 or 2mb varieties that will be more suitable for mp. 

But the rpi pico, with the addition of a simple circuit as proposed by Davee that switches off the power entirely may be just a good.  For that calculation the amount of power drawn by the pico on power on followed by a short runtime will have to be measured.