ESP32 (And ESP8266) general stability status?

[rnyqvist]

Hi,

I've done several experimental Arduino sketches to evaluate if the ESP32 or ESP8266
would be suitable for a long-term home-automation sensor node purpose.
i.e. the ESP would have DHT22 or similar attached, would publish the date via MQTT, powered via USB power supply, etc...

But, with several ESP32 variants and also ESP8266 (wemos D1 mini) I end up in the same situation. Code runs ok for a few days or even a week and the systems halts and stops communicating.

I've seen at least that the I2C bus simply does not respond anymore. But the freezing has happened in a device where I had just the DHT22 (one-wire) equipped with ESP32.

Before posting any code examples just wanted to get some basic idea if the ESP family (using Arduino, not ESP idf) is mature enough that it could be considered for this kind long-term use?

br
Rami

[sellonoid]

I've had a Sparkfun ESP8266 Thing Dev board (https://www.sparkfun.com/products/13711) running on solar charged batteries since December 2015. Here's the first set of readings it pushed out to my Raspberry Pi based server and MYSQL database.

RSSI -79.00 dBm 2015-12-23 12:29:50
OAT 41.00 degreesF 2015-12-23 12:15:16
IAT 73.40 degreesF 2015-12-23 12:15:16
battVolts 14.24 voltsDC 2015-12-23 12:15:16

It's been down only 2 days in all that time, and that was because the batteries died. I have a second one that's been running just about 2 months now.

The Espressif ESP8266 chip, and Sparkfun's supporting chips and circuit board, and the Aruino based libraries seem to be rock solid. I'm just now trying to build my next project using the Sparkfun ESP32 so I can do OTA updates.

Jim

[rnyqvist]

thanks for the reply sellonoid!

Did you have a temperature sensor or something else hooked up to the sparkfun chip's gpio's
And if you did, did you power them from the chip or with own power supply?

I'm starting to think that the problem is with powering the used bus with the ESP chip's 3v3 pin...
I might give a shot with that sparkfun version, though the price is three times higher than the ones I've used

[rin67630]

Hi Jim,
how much current does your board and periphery draw?
Here during the German winter I must be able to run a full month quasi without solar charging.
My system draws ~20mA, since it measures 24/24/365 every second.
I have got a 10Ah Battery block and it was not enough to pass an ugly cloudy december.

Please note, you may have a system with deep-sleep, but if you want OTA you will not be able to run low power any more...

[sellonoid]

Sorry for the delay in replying to you guys.

rnyqvist- I just went and looked at my installation on my gate opener and saw that I used the ESP8266 Thing NON-Dev model (instead of the Dev model), but I don't think it matters regarding your questions. Yes, I have two DS18B20 temperature sensors connected. I ran a +3.3 volt wire, a ground wire, and a signal wire to each sensor. Each signal wire has a 5.1K Ohm pullup resistor. I also run two instances of OneWire in my program to keep it simpler (in my mind). Since I have the solar charged 12 volt batteries available, I don't use a LiPo battery. I have the ESP8266 Thing board on a small perfboard, and on that perfboard I made a 3.3 volt regulator using a LM317 (TO-220) three terminal regulator, which supplies the Thing board on one of its 3V3 pins. In theory, the regulator is good for at least an amp, if not two or three.


rin67630 - The ESP8266 Thing NON-Dev board is in deep sleep for 15 minutes, wakes up, takes the RSSI reading, takes the battery voltage reading with the ADC, reads the two DS18B20 sensors, connects to the WiFi, then sends the data. During transmit it draws around 170 to 180 milliamps. During deep sleep the current is super low as advertised. Since it runs only once every 15 minutes, it does not impact the two solar charged 7 amp hour batteries much. Running 24/7 like you are doing with no recharging is tough.

Now, my most recent project runs from commercial power so I will try OTA with the ESP32 and the ESP-IDF tools, which is what brought me here to begin with.


Jim

[rnyqvist]

Seems my problems with I2C stability pretty much disappeared when I switched to use ESP32 arduino branch from 'stickbreaker':
https://github.com/stickbreaker/arduino-esp32

I guess the problem would have disappered with switching from Arduino to ESP-IDF...