Calibration of SI7021

Moderators: grovkillen, Stuntteam, TD-er

Post Reply
Message
Author
mackowiakp
Normal user
Posts: 527
Joined: 07 Jun 2018, 06:47
Location: Gdynia/Poland

Calibration of SI7021

#1 Post by mackowiakp » 09 Feb 2023, 22:06

I have a question for my colleagues regarding the SI7021 sensors.
I bought 3 pcs of SI7021 in the I2C version and 3 pcs in the 1-wire version. I am using these sensors for the first time.
All 6 pieces of SI7021 show similar values, but all temperature indications are always overestimated and humidity underestimated.
At the temperature outside (at the moment -2.4C indicates a "normal" thermometer) the SI7021 temperature readings are about 3C higher than the real ones and the humidity readings are about 10% lower.
A similar error occurs when the sensor is inside the house at a temperature of 20C.
Of course, using the rule in the sensor configuration, I can correct these values.
The only question is whether the addition/subtraction of these 3C of temperature or 10% of humidity is a constant error value for each value or not.
And whether such a correction of indications will be linear in the entire measuring range. That is, is this indication error for the sensor constant.
Of course, I checked the pull-up resistors, the cable length is about 15 cm and the sensor is outside the housing so that the ESP temperature does not affect the sensor readings.
What are your colleagues' experiences in this regard?

TD-er
Core team member
Posts: 8643
Joined: 01 Sep 2017, 22:13
Location: the Netherlands
Contact:

Re: Calibration of SI7021

#2 Post by TD-er » 09 Feb 2023, 22:10

Temp/hum relation is absolutely not a linear equation.

What I do in the BME280 is this:
- Compute dew point temp based on measured temp/hum
- Compensate temperature offset
- Compute compensated humidity based on earlier computed dew point temp and corrected temp.

Such a high offset is often an indicator the sensor itself is heating up.

User avatar
Ath
Normal user
Posts: 3416
Joined: 10 Jun 2018, 12:06
Location: NL

Re: Calibration of SI7021

#3 Post by Ath » 09 Feb 2023, 22:18

A quick search for SI7021 accuracy shows these IMHO relevant threads:
https://electronics.stackexchange.com/q ... 1-accuracy
https://www.kandrsmith.org/RJS/Misc/Hyg ... _many.html

Seems that the SI7021 isn't very accurate, and depending on the source, the quality might be related to price and supplier...
/Ton (PayPal.me)

User avatar
Ath
Normal user
Posts: 3416
Joined: 10 Jun 2018, 12:06
Location: NL

Re: Calibration of SI7021

#4 Post by Ath » 09 Feb 2023, 22:38

This I found in the plugin source code:

Code: Select all

//Only perform the measurements with big interval to prevent the sensor from warming up.
So, you might want to set the Interval to at least several minutes, to avoid heating up the sensor too much.
/Ton (PayPal.me)

mackowiakp
Normal user
Posts: 527
Joined: 07 Jun 2018, 06:47
Location: Gdynia/Poland

Re: Calibration of SI7021

#5 Post by mackowiakp » 09 Feb 2023, 23:12

It doesn't look good.
I set the reading interval to 2 minutes so the sensor should not be heated.
DTH22 - even quite precise but not suitable for outdoors.
So is AM 2301.
There is also a big problem with the BME280.
Most - at least the ones I bought - are marked on the board as BME/BMP (or similar). As a result, they behave like BME280, that is, they give readings of 3 measurements, but the humidity reading is always "0".
As in the screenshot below (see "Task 8").
When I ask the sellers, they don't even know what they're selling.
Tragedy...
Screenshot_20230209_231042.png
Screenshot_20230209_231042.png (111.64 KiB) Viewed 1704 times

TD-er
Core team member
Posts: 8643
Joined: 01 Sep 2017, 22:13
Location: the Netherlands
Contact:

Re: Calibration of SI7021

#6 Post by TD-er » 09 Feb 2023, 23:30

BMx280 plugin does support both the BMP280 and BME280.

It should be made clear in the logs which one is detected, but given the humidity is 0, I think it is very clear you're dealing with a BMP280.

mackowiakp
Normal user
Posts: 527
Joined: 07 Jun 2018, 06:47
Location: Gdynia/Poland

Re: Calibration of SI7021

#7 Post by mackowiakp » 10 Feb 2023, 00:14

Yep. Log shows BMP280.
But there is still problem that if You buy board labeled BMP/BME It is not clear which one is in fact.
So far I was not lucky to buy exactly BME280. Maybe somebody have good source.

TD-er
Core team member
Posts: 8643
Joined: 01 Sep 2017, 22:13
Location: the Netherlands
Contact:

Re: Calibration of SI7021

#8 Post by TD-er » 10 Feb 2023, 00:20

BMP280 is a bit more rectangular chip, compared to the more square BME280.

mackowiakp
Normal user
Posts: 527
Joined: 07 Jun 2018, 06:47
Location: Gdynia/Poland

Re: Calibration of SI7021

#9 Post by mackowiakp » 10 Feb 2023, 00:45

Have You any picture of real BME280 module?

TD-er
Core team member
Posts: 8643
Joined: 01 Sep 2017, 22:13
Location: the Netherlands
Contact:

Re: Calibration of SI7021

#10 Post by TD-er » 10 Feb 2023, 08:38

Image
From Google.

mackowiakp
Normal user
Posts: 527
Joined: 07 Jun 2018, 06:47
Location: Gdynia/Poland

Re: Calibration of SI7021

#11 Post by mackowiakp » 10 Feb 2023, 09:03

THX! But I think there is problem with rules using BMP280 driver instead BMx280 for BMP280 sensor.
It does not occurs using BMx280 driver for BMP280 sensor.
Will describe situation in "Software" section.

Post Reply

Who is online

Users browsing this forum: Bing [Bot] and 30 guests