BMP/BME280 over sample

Moderators: grovkillen, Stuntteam, TD-er

Post Reply
Message
Author
Carples
Normal user
Posts: 2
Joined: 03 Jan 2020, 00:30

BMP/BME280 over sample

#1 Post by Carples » 03 Jan 2020, 00:40

Hello

I am using some BMP280 sensors to monitor air pressure in conjunction with Wemos D1 - ESP Easy works very well for this task

But due to the need to measure pressure in Pascals I have trouble with erratic readings, I think because no over sample is set

Having a stable reading in Pascals is possible according the the data sheet, and external over sample via node red stabilizes the readings but is a bit clunky and slow i.e. 1 reading second X samples to smooth = lots of seconds

I would like to customize the driver for my application but don't seem to be able to locate it in github

Ton_vN
Normal user
Posts: 170
Joined: 21 Oct 2016, 15:20
Location: Hengelo (Ov)/ NL
Contact:

Re: BMP/BME280 over sample

#2 Post by Ton_vN » 03 Jan 2020, 10:07

Wondering why you care about filtering delay related to atmospheric air pressure.
Atmospheric pressure usually does not change very fast and a simple Kalman-filter can do the job of cleaning 'twitches'.

TD-er
Core team member
Posts: 3137
Joined: 01 Sep 2017, 22:13
Location: the Netherlands
Contact:

Re: BMP/BME280 over sample

#3 Post by TD-er » 04 Jan 2020, 12:26

The sensor itself can do basic filtering itself, but it will introduce some delay.
If I remember correctly there is already some filtering active, to make sure slamming of a door will not cause erratic readings.

Carples
Normal user
Posts: 2
Joined: 03 Jan 2020, 00:30

Re: BMP/BME280 over sample

#4 Post by Carples » 05 Jan 2020, 09:59

Hello

Was trying to use it to measure and record room pressure differentials of 10-20pa, for measuring Barometric pressure current settings would be fine noise is generally less than 1Hpa but that is 100pa

High over-sample puts the noise threshold below my measurement range rather than over 0.2pa for full over sample and

Was thinking sending 10110111 to 0xF4 during sensor startup would be simpler than implementing filtering on secondary system for potentially many sensors

I currently have 2 units running both with same senors and a running average of 100 readings and they maintain readings within 10PA of each other and I'm pretty sure the variance is more due to noisy readings than anything else.

I'm not sure how to implement a Kalman-filter, but I do see that node red has a kalman filter I will try

TD-er
Core team member
Posts: 3137
Joined: 01 Sep 2017, 22:13
Location: the Netherlands
Contact:

Re: BMP/BME280 over sample

#5 Post by TD-er » 05 Jan 2020, 15:28

I think it is the best if we add a setting to adjust the sensor filtering.
This is the most generic option.

One of the features I had in mind was to add a filtering module which can be used for any sensor value.
My idea was to have it being used as a specific formula statement, so it can be used very transparent on any value you would like.

TD-er
Core team member
Posts: 3137
Joined: 01 Sep 2017, 22:13
Location: the Netherlands
Contact:

Re: BMP/BME280 over sample

#6 Post by TD-er » 05 Jan 2020, 16:06

I added an issue for the BME filter settings: https://github.com/letscontrolit/ESPEasy/issues/2844

And for the filtering: https://github.com/letscontrolit/ESPEasy/issues/2845

Post Reply

Who is online

Users browsing this forum: Bing [Bot] and 17 guests