@BoRRoZ
Yeah, the data from Adafruit is non-sense. I can go down to 10uA using several sensors, flash and atmega328p no problem at all.
TPL5110 draws more like 35nA while the rest of the project is asleep.
This is actually difficult to measure such a small current without a lab equipment. uCurrent Gold is noisy on nA and good DMM resolution is not that good. Keithley Picoammeter has to be used.
@alexsh1
the board on the picture is as simple has
Arduino Mini 8MHz 3.3V (no regulator or power led)
SI7021 direct I2C wiring (no regulator)
RFM95 direct SPI wiring with DIO0/DIO1/DIO2 connected see schematic on github above
Removed BOD from fuse (can be done by software) and no Watchdog use, that’s it but may be my measurement are not “so good”, but look like the same with my multimeter (1uA) and UCurrent (0.778uA) so I think I can trust the results
This is exactly my setup, however, there is still higher consumption.
I have uCurrent Gold and it shows me very similar readings to with my DMM (121GW), which is 2.9uA
I have implemented TPL5111 in my design. sadly dont have meter that can messure below 1uA.
Added some DIP swithes so you can choose what time interval you need.
Here is some lates pcb
Still wating for it to arrive, but similar design is in field up un running for almost 2 months. https://snapshot.raintank.io/dashboard/snapshot/QTzAqP8YY8M6YfSVmzZrmyd44AS576XJ?orgId=2
So just using one of Charles’s PCBs, a bog-stock Chinese arduino clone with only the linear reg and power LED removed this thing is pretty amazing.
I fitted a LiFeP04 cell into this node as the cell shipped to me in week 2 of April. It read 3.3v then and having woken, polled a Bosch sensor and transmitted to TTN every fifteen minutes since it still shows 3.3v! That would be 96 cycles per day or about 10k wake and transmit cycles.
I am thinking that one of the current breed of low power devices with SI7021 and TPL5110 type manager chips has power to see it well into obsolescence, no?
So this is still sitting on the shelf in my kitchen and waking up every 15 minutes to check temp/humidity and battery voltage. Battery was installed as it shipped from Ali Express, 3.3v (LiFeP04) and never charged before or since. That was in April and this is how the battery and frames looks now. I had an admirer reset the Arduino once since the battery went in but it has run continuously since April.
I am tempted to check the cell voltage in case the voltage divider is not right but I actually did this before
Impressive @Charles
Does anybody have the magical incantations to upload the OptiBoot firmware compiled by @Charles on 3.3v Arduino Pro using avrdude?
(Or any other tool which doesn’t require MS Windows…)
Great video and test, the low power consumption is amazing!
What is you preffered way to power those boards?
Lets say you’re using some rechargeable batteries. If using LI-Ion this would mean that a LDO is required which also needs a bit of power all the time (for exmaple the MCP177-33: typical 1.2uA input quiscient current), as the absolute maximum rating for the RFM95 is 3.9V according to its datasheet.
A full Li-Ion battery gives you about 4.15-4.20 volts.
Myself, I would not be too concerned about the differance between a node runing at less than 0.8uA without a regulator, versus circa 2.5uA with a low cost MCP1700 or similar.
Take an example, a node that uses a pack of AA Alkalines to power a node for 5 years. At a capacity of 2800mAhr, the battery goes flat in 5 years doing useful stuff such as reading sensors and doing transmissions. Thats an average of 1.53mAhr per day.
The ‘good’ node uses in that 5 years 0.0008 x 24 x 365 x 5 = 35mAhr doing sleep. The ‘bad’ node uses 0.0025 x 24 x 365 x 5 = 109mAhr in sleep.
The difference in mAhr used during sleep, between good node and bad node is 109mAhr - 35mAhr = 74mAhr, or around 50 days extra battery life over the 5 years, big deal.
So no need to follow the crazy advice to operate the LoRa device direct from a LiPo, just use a regulator, it has very little affect on battery life.
If the very small amount of current the regulator uses is somehow significant, and its not in a lot of cases, use a LIFePO4 to power the node. These are safe to use on LoRa devices without a regulator, safer then LiPo bombs too.
If 150mA output is enough, there is the MCP1711, quiescent current of 0.6uA.