- Fri Oct 11, 2019 7:28 am
#84130
... and fourth: bmp280. I bought a few some months ago. Could not get them working. Air pressure was always some random fixed (wrong) value, all of them. But the bmp280's I already had worked fine with the same code. So I ordered some more and they do actually work like expected. Anyone an idea? Fakes?
Moderator: eriksl
- Sun Oct 13, 2019 5:30 pm
#84153
There was a post on a semi related topic and it referenced a thread that looks interesting:
https://github.com/arendst/Sonoff-Tasmota/issues/3364#issuecomment-475992268
Check out the CCS811 data graph - it's exactly as you describe. The problem was tracked to an I2C clock stretch that was too low, set by the code on a different sensor.
I've never seen the problem you describe, mainly because I deep sleep a lot, but also because I have clock stretch set very high
eriksl wrote:First one: ccs118 gives max value continuously after a few days on. I have it running in "continuous" mode and there are no errors. After a reset, the readings go back to normal.
I am not using the baseline save or temperature/humidity correction, because both temperature and humidity are quite constant here.
There was a post on a semi related topic and it referenced a thread that looks interesting:
https://github.com/arendst/Sonoff-Tasmota/issues/3364#issuecomment-475992268
Check out the CCS811 data graph - it's exactly as you describe. The problem was tracked to an I2C clock stretch that was too low, set by the code on a different sensor.
I've never seen the problem you describe, mainly because I deep sleep a lot, but also because I have clock stretch set very high
- Mon Oct 14, 2019 9:32 am
#84158
Thanks for your interest. I read the link, but since they're talking about "Core" and "Wire" it suggests they're using arduino - I am definitely not. I have my own i2c implementation that is very, very resillient and manages to reliably talk to even the most broken (i2c-wise) sensors. There are no i2c errors, at all. It's just that the value returned is nonsense. So I guess it's another issue altogether.
- Mon Nov 11, 2019 9:37 pm
#84518
I've been running a BME280 continuously in 'forced' mode since July, one sample every 30 minutes. No problems after I filtered out the bad boards. Stock Wemos D1 Mini and the Arduino IDE.
I had problems early on: some of the boards out of China may have 'clone' parts on them that either didn't work, or didn't follow the Bosch specs. 3 boards wouldn't even ACK their address, so I kept testing until I found a few boards that worked OK. The metal can on the sensor doesn't look like the Bosch pictures, which is why I presume they're no-name clone parts. Initially I thought the three non-functional boards had the parts soldered on backwards, as the hole in the can was on the other end of the can compared to the Bosch datasheet.
I never did get my BME680 board: the US Post Office lost it. I didn't bother trying to get a replacement.
I had problems early on: some of the boards out of China may have 'clone' parts on them that either didn't work, or didn't follow the Bosch specs. 3 boards wouldn't even ACK their address, so I kept testing until I found a few boards that worked OK. The metal can on the sensor doesn't look like the Bosch pictures, which is why I presume they're no-name clone parts. Initially I thought the three non-functional boards had the parts soldered on backwards, as the hole in the can was on the other end of the can compared to the Bosch datasheet.
I never did get my BME680 board: the US Post Office lost it. I didn't bother trying to get a replacement.
