Sensor is offline for some time

[gohan]

I had similar problems but it was mainly radio interference. Are you using nrf24?

[parachutesj]

yes, I am using NRF24 on my nodes.
I was thinking of creating a 2nd gateway with a different channel and connect the weather station to this in order to rule out interference - however it is still strange that I get around all 3 minutes a connection and then for 1-2 hours not even one.
I honestly suspect a hardware issue which is caused by temperature/humidity/sun...

[sundberg84]

@parachutesj and you have controll over the voltage? I mean - wintertime, its just not getting enough sun so the voltage drops?

[parachutesj]

@sundberg84 that's a good point. The battery was full when I first noted it and checked it. So it seems that it charges enough in order to get it running.
I am no expert but the coldest temp is not around noon, so basically it fails when it is getting warmer. Is this a typical scenario or should the voltage drop when it is coldest?

[sundberg84]

@parachutesj - I dont think the temperature has anything to do with solar charger? Im thinking that now during winter the sun is very low so its a long period without incoming power. Maybe even all night and the first hours during the day (untill noon when its at the lowest voltage?). Then at noon when the sun is in the right angel again it starts to charge.

[gohan]

If it can disharge the battery overnight, it means the node is drawing quite some power

[parachutesj]

@sundberg84
it does not use a lot power at all. It sleeps for most of the time. I do not remember the exact figures but it would be ok without solar for at least one year.
So the only correlation I could see is that when it starts charging and there is enough power via solar available, it is causing some side effects lately.
I do not have the most efficient solar panel and mabye aging caused it to be weaker. It is quite cloudy today - maybe it is running then. Let's see

[parachutesj]

just a quick update: it failed again. I just looked at the monitor and last update was at 12:42pm - no update in last 45 minutes. All other sensors still report. And there is no sun today.
oh boy, I have no idea. Maybe I replace parts and see what happens.

[sundberg84]

@parachutesj - could you measure the voltage when it was "down" ? Is it going inline directly again if you reset it?
Do you have repeaters?

[parachutesj]

@sundberg84 it is on my roof so measuring in real world conditions will be difficult. I will try what I can find out. No access at the moment due to boring day job ;-)
When I took it down (on Sunday), I had it inside (with no sun exposure) and opened the case. I found that all still was up and battery voltage was ok. As soon as I hit reset it was back ok. It took me about 1 hour more until it was back on the roof. So I am not so sure if it was the reset, the missing sun exposure or whatever.
So I monitored it yesterday. Again at around noon it stopped and got back online after 1.5h. Will see what it does today.
But I think first will be adding a new gateway on different channel connected to my PC in order to do some tracing.
To your question: I am not sure if it is communicating directly with the gateway or via a repeater node. As I am using MQTT there is no information when I receive the message at my controller. I think I could debug the gateway in order to find out. Or ist there some other possibility?

[gohan]

If you are going to change the sketch, take the chance to add the voltage measurement, just in case.

Also try both lowest and highest NRF24 channels to see if you get any changes.

[sundberg84]

@parachutesj - a hard nut to crack this one!
I have a similar case - thats why I asked about the repeater. Its my rainsensor (2xAA + reed switch, sleeping) which very seldom but sometimes disconnects. I know when I launched this node it was living on the radio range edge so I added a repeater. The repeater isnt located 100% good for the rain node so sometimes its like it gets disconnected from this repeater and tries to reconnect according to protocol. I dont know if the sensor sometimes connects directly to GW and sometimes through a repeater but the symptoms are the same as you. The good thing is that if it goes through the repeater you might be able to debug some through the repeater.

[parachutesj]

@gohan @sundberg84 thank you both for your inputs.

I think I have a design issue here.
I originally had changed the weather station a few times in the beginning and the brain was a Nano. At some point I figured that I wanted to rebuilt it, make it "cleaner". Eventually I changed the DHT22 to a BME and the Nano to a Pro Mini 3.3V

My solar panels deliver 5V and the charging circuit feeds a 3.6V 18650 and charge cut off is 4.2V

I looked at my battery percentage and any time I get an interrupt in transfer is when it reaches 116% (which is based on 3.6V =100% => 4.2V )
That also fits to current weather situation: Anytime around noon it has enough sun rays to charge the battery.

The question is, what is wrong? Either the charging circuit is behaving strange when it reaches cut off or 4.2V are too much for the Arduino?

That of course was no issue with the Nano and explains a bit.
If I remember correctly, I desoldered the power regulator on the Arduino (need to check next time I get it down).
Does anyone know what is the max Voltage until it fails?

At least I have some glues what to change: I think I can put in a new Pro Mini with power regulator and go via RAW pin, the consumption will be higher but still ok I guess.
And second option would be working on the charging side, however this is a bit more complicated during winter time as I don't know what is wrong.

The solar panels are currently disconnected, since then it runs as expected and updates every 3 minutes, voltage still at 4.2V

Here's a pic of it - not very nice but (mostly) functional

It fell down during a storm until I found a better solution and the solar panels are getting old. New one is already here but needs to be attached...

[gohan]

Just get a ldo le33 or mcp1700 to have a stable 3.3v output and see from there. The raw pin would not work well with 4v

[parachutesj]

@gohan
quick update:
I already had a LDxxx33 in order to have a stable output for my NRF24 radio. Connected the VCC to this source too and since then it works. So it really seems that it overpowered the Arduino.
Also switched the SolarPanel on but need to wait a few more days with good weather in order to have a "final" conclusion.

[parachutesj]

and on a side note: my solar panels are total overkill, it charges for only a very short time during day until the battery is charged completely.
Thinking of a heater for my rain sensor now...

[gohan]

those solar panels are quite huge, while arduino doesn't use that much power especially when sleeping

[parachutesj]

All:
thank you for your contributions. I think I have finally fixed the problem. It was due to the input side from solar and I had overvoltage shutting down my charge control board. This is quite strange as it has been running for a longer period and then stopped. Maybe it altered and at some point the tolerance went down.
Anyhow it is logical: When I had good sun, I measured 6.2V on the input side which seems to be too much for the board - even the solar panels are all rated at 5V.
I added a voltage regulator making sure the solar output is max 5V. Since then it seems to be stable.
Of course this is not optimal as this is causing some loss.
Does anyone have a potentially better solar charger control board for an input of around 4-9 V to charge a 3.7 LiPo?
Currently using those: LiPoCharger AliExpress

Cheers
SJ

[Winston Evans]

Your takeaway is that there’s a low probability that you will need to buy a replacement for your solar panel before the warranty ends.
You can compute the degradation rate if you want the exact figure. If you’re not happy with it, you can always add upgrades to your panels. You just have to make sure that the new parts match with your solar panels.