Office plant monitor




  • Hardware Contributor

    @carlierd excellent, it looks very nice :+1:


  • Admin

    So nice! :thumbsup:


  • Hero Member

    Very nice indeed !
    Do you have an idea of the battery life on this sensor ? Why not change to 1MHz to allow the voltage to go down to 1.9 or 1.8 V ?


  • Mod

    @Nca78 from the description:
    Expected duration more than 2 years for the CR2032 and about 3.5 with the two AA (power consumption less than 5uA during sleep time; measured using a uCurrent Gold).



  • @Nca78 Like @mfalkvidd wrote, the expected life duration is greater than 2 years. But it's just theory. My first node (on 2*AA) was started several weeks ago and the battery level did not change. So I am confident ;)

    I measure the power consumption when working at 1 MHz and there was no difference between 1 or 8 MHz. As I have to decrease the transmission speed at 1 MHz, I got some issues. The main was probably with the signing feature.


  • Hero Member

    @carlierd you cannot compare the behavior of AA batteries and CR2032. Button cells have a high internal resistance, you can't draw much current from them else the voltage drops very quickly. The more current you draw the more you lose in heat because of this internal resistance, the solution is to use a big capacitor to provide the current, then you lose energy because capacitor has a current leak, but usually it's better than what you lose with internal resistance.

    Maybe your 100uF is doing that job if data sending is quick enough, but at 8MHz and the voltage drop when sending you might already be very close to brown out level (or crash level if you have disabled brown out), you should check the voltage just after sending and not before to have an idea.

    Look at the difference in voltage level in the second "pulse mode" graph on page 2, which I believe uses a similar pulse current draw than your board when you transmit, after only 50mAh of capacity used the voltage is down to 2.4V...
    http://data.energizer.com/PDFs/cr2032.pdf



  • Interesting !
    I will check in few weeks the result. There is several capacitors to help the stability maybe it could be enough. If it's not the case I will replace the CR2032 by 2 * AAA.


  • Hardware Contributor

    @Nca78 you can have brownout if the battery is near end of life and nothing has been designed to handle a bit the internal res. but with a fresh a coincell and some capa, you can do some chain tx but that's not the best for the battery life.

    The only case where i had issue with brownout at startup, plus it was a fresh coincell was with a crappy ali coincell batt!! very bad quality! now I'm using only good quality coincell like duracell, varta, maxell.. and that's day and night ;)

    @carlierd If you're interested this doc explains very well the capacitor calc etc... http://www.ti.com.cn/cn/lit/wp/swra349/swra349.pdf

    • capa ideally calculated/estimated (on mine I have 200uF, plus others for sensors, radio is 86uA etc..) and after multiple msg presentation on frsh coincell i don't fall under 2.85 (voltage starts 3.05V) because I can't recover voltage during this period.
      Hopefully a transmit is not 1sec! More something like says 30+ ms (depending if signing is needed, ack, retries.. etc). To not break capa benefits, it's better to sleep between tx during chains..to recover voltage and optimize battery life. Sleeping 200ms can be enough
    • For the capa leakage, it depends of capa. common good quality ceramic have not those leakage (few nano). Sure on this design it's not a ceramic.
    • You're right at runtime 1Mhz consumes less than 8Mhz; mA vs thousands uA. But at 1Mhz everything is slower, comms too (I mean code execution). During deep sleep, there is no difference in power consumption. Oscillator is stopped so mhz does not mean. Note: at 8Mhz you can wake in few uA, at 1Mhz it's slower of course.

  • Hero Member

    I wonder whether a higher capacity coincell, like maybe a CR2450 (620mah) or a CR2477 (1000mah capacity), might give you the pulse current you need without needing a capacitor, thus avoiding leakage losses? Plus maybe you wouldn't need to change it as often.


  • Hardware Contributor

    sure 2450/77 would be better. 2477 is a bit expensive.
    but not sure if those coincell would be better without capacitor (I mean in the long term), looking at coincells datasheets.. which are always given for x kohms load.
    leakage current for ceramic common quality capacitor is in nA range (hopefully), so i think it's better to have it buffering to help the coincell (still for the long term, or at a momemt it won't be enough strong). if alka, there would no need..
    that's imho.


  • Hero Member

    Stop wondering. I tried with 2477 cells and it was not much better than 2032 without capacitor ;)



  • @scalz : I missed your reply ! Interesting withe paper !

    @scalz said:

    @Nca78 you can have brownout if the battery is near end of life and nothing has been designed to handle a bit the internal res. but with a fresh a coincell and some capa, you can do some chain tx but that's not the best for the battery life.

    The only case where i had issue with brownout at startup, plus it was a fresh coincell was with a crappy ali coincell batt!! very bad quality! now I'm using only good quality coincell like duracell, varta, maxell.. and that's day and night ;)

    @carlierd If you're interested this doc explains very well the capacitor calc etc... http://www.ti.com.cn/cn/lit/wp/swra349/swra349.pdf

    • capa ideally calculated/estimated (on mine I have 200uF, plus others for sensors, radio is 86uA etc..) and after multiple msg presentation on frsh coincell i don't fall under 2.85 (voltage starts 3.05V) because I can't recover voltage during this period.
      Hopefully a transmit is not 1sec! More something like says 30+ ms (depending if signing is needed, ack, retries.. etc). To not break capa benefits, it's better to sleep between tx during chains..to recover voltage and optimize battery life. Sleeping 200ms can be enough
    • For the capa leakage, it depends of capa. common good quality ceramic have not those leakage (few nano). Sure on this design it's not a ceramic.
    • You're right at runtime 1Mhz consumes less than 8Mhz; mA vs thousands uA. But at 1Mhz everything is slower, comms too (I mean code execution). During deep sleep, there is no difference in power consumption. Oscillator is stopped so mhz does not mean. Note: at 8Mhz you can wake in few uA, at 1Mhz it's slower of course.

  • Hardware Contributor

    @carlierd yep, this doc inspired me too, plus coincell datasheets,. I use mega-pile for sourcing batt and they have lot of battery datasheet.
    I have also recently designed few small sensors I need to put on a panel now :)

    • moisture+temp
    • moisture+salinity/conductivity+temp


  • How did you source the probe portion of the sensor?



  • Could this be used with a capacitative soil sensor to avoid corrosion of the sensor over time ?


  • Hero Member

    @tomtastic said in Office plant monitor:

    Could this be used with a capacitative soil sensor to avoid corrosion of the sensor over time ?

    Yes. Do you have one in mind?


Log in to reply
 

Looks like your connection to MySensors Forum was lost, please wait while we try to reconnect.