The WORKBENCH part 3

We had a cold night (for Belgium that is) with temperatures down to -8 degrees Celsius. The Lion 3.7V 18650 batteries of the outdoor meteo sensors are starting to fail occasionally when the temperature went below -5 degrees Celsius last night :frowning:

The brownouts occur when the device wants to make a Wifi connection (which requires lots of power); I think it is a good time to replace Wifi with LoRa assuming that drains less power from the battery. For now the measurements are stored on the device until a good connection can be made (probably when I take the sensor boxes inside).


  • The sensor box is 1 meter above the ground. If I put the sensor box on the soil then the battery dies even quicker. That is logical I think.

  • I hope the battery will recover by itself when the weather gets better.

  • Do you know other battery chemistries that work better (in the battery size 18650 and having 3.7V nominal voltage)?

yes that’s a known 'problem with that type batteries.
they 'recover however.

I would suggest 2x CR123A’s in series followed by a DC/DC but then you still won’t have the same capacity.
but they can stand lower temps, also 3 x AA in series (duracell long life) works but not rechargable.

if possible some insulation inside the enclosure can help ?

1 Like

A chemistry that ‘works better’ is Lithium-Thionyl chloride (Li-SOCl2).
These are primary (non-rechargeable) batteries however and relatively expensive.
To save battery power try to optimize your device for low-power (even power usage for Wifi can be optimized).


Features & Benefits

  • High operating voltage, stable during most of the application lifetime
  • Highest energy density among primary lithium chemistries
  • Lowest self-discharge for extended shelf and operating life
  • High pulse capability, thanks to well-controlled passivation with limited voltage delay
  • Excellent reliability over extended operating life
  • 20 year+ design with best quality components and materials
  • Wide operating temperature range, matching all environment conditions
  • Low magnetic signature

Technical Specifications

  • Nominal voltage: 3.6 V
  • Cylindrical format: from 1/2AA to D
  • Capacity range: 1.2 Ah to 17.0 Ah
  • Max pulse discharge rate up to 4 A
  • Operating temperature: - 60°C to + 150°C
  • Non-flammable electrolyte
  • Compliant with IEC60086-4, IEC60079-11 and UL1642 certified
  • Compliant with the European directive RoHS and REACH

Thanks for the info. Good to know they will recover.

I also noticed that the performance of batteries vary; now only 2 out of 3 sensor boxes have failed occasionally.

“3x AA in series” is a good tip!

These Li-SOCl2 are not powerful enough (mA) when using Wifi; LoRa should work, no idea for NB-IoT.

I was told that LiFePO4 batteries are a little bit better in cold environments so I will try that next night. If they survive up to -10 degrees Celsius then that is fine for me. The downside is that their capacity is half that of Lion 18650.

Another thing is that some humidity sensors do not work anymore when humidity gets close to 90%-100% (it gets stuck at that percentage; the only solution is to cook them dry in an oven). I’m going to try to replace the BME280 with an Sensirion SHTx but I think that will have the same problem.

And tomorrow they predict -12 (i take that with a grain of salt :wink: )

been there :rofl:

1 Like

Yes, but not the same voltage either – Nominal voltage is 3.2V.
(I used them quite a lot, but I power my nodes directly in 3.3V)

I normaly never have this problem with my SHT15’s. My F.A.R.S. has a fan in it that forces air past the sensor. The only problem I have is corrosion of the PCB it is mounted on.


Good to know.
I assume that system runs on mains power instead of batteries (fans are power hungry)?

I use acryllic spray for everything except the nozzle of the sensors (the sensor has a small piece of PTFE sheet on it).

Yes, only the FARS runs on mains. The rest are “Jeenodes” on batteries.

I even use the SHT15’s without membrane. And waterproofing is way to professional for my DIY-setup.

1 Like

I have thrown a node with a LiFePO4 battery outside (green line).
Let’s see how it survives the night…


1 Like

lol… in an enclosure or naked :smile:

No enclosure yet :roll_eyes:
So I left it on the floor in the garden shed…
Same temperature as outside, a bit less moisture (hopefully)

for sure

1 Like

LTH5200 series is a LoRa based temperature and humidity sensor.
Soon on the workbench for a review, the very first, hope to find the right ‘style’


Here it is…


Temperature dropped only to -3, but the LiFePO4 battery voltage has been stable throughout the night, no drop with the temperature.

I need to look more in details to the data, because I have 2 gaps in the graph. The device didn’t crash as the packet counter didn’t reset. It might be unrelated as I also have gaps with other nodes (e.g. the orange line) and this is very unusual…


select max(temperature),min(temperature),last(temperature) from environ where dev_id = 'bulles-minilora-01' and time >= now()-24h

time                           max  min  last
----                           ---  ---  ----
2019-01-21T08:31:20.274882581Z 22.5 -2.8 -1.3


select max(voltage),min(voltage),last(voltage) from environ where dev_id = 'bulles-minilora-01' and time >= now()-24h

time                           max  min  last
----                           ---  ---  ----
2019-01-21T08:35:21.574417311Z 3.28 3.27 3.27

Missing points:

select * from environ where dev_id = 'bulles-minilora-01' and time >= now()-24h

time                           analog_input counter data_rate dev_id             humidity pressure rssi snr   temperature voltage
----                           ------------ ------- --------- ------             -------- -------- ---- ---   ----------- -------
2019-01-21T22:21:28.504029864Z              2322    SF7BW125  bulles-minilora-01 80       1009.8   -84  9.25  -2.1        3.27
2019-01-21T22:31:36.201477425Z              2324    SF7BW125  bulles-minilora-01 79.5     1009.7   -84  8.25  -2.2        3.28
2019-01-22T06:17:32.845840515Z              2416    SF7BW125  bulles-minilora-01 77.5     1001     -83  9.25  -1.8        3.28
2019-01-22T06:37:48.487811162Z              2420    SF7BW125  bulles-minilora-01 77.5     1000.6   -83  7.75  -1.8        3.27

So I am missing 1 packet at 22:21Z and 3 at 06:17Z.
I don’t have gateway logs to cross check, but given the fact that frame counter was incremented, it looks like the node sent them but got lost…

1 Like

My test with LiFePO4 failed during the night. Temperature was -5 degrees Celsius. So this battery type is not better than Lion in these cold environments :frowning: Note that I enabled brownout detection on the MCU; it would have stayed alive longer if I had disabled that feature.

Strange – temperature in Brussels was a couple of degrees higher, but I don’t observe any voltage drop…

I did also place the sensor boxes on the cold soil in the garden outside.