BMP280 Accuracy – Four sensors, four values – How to calibrate?

[Wookbert]

Firmware Build: ESP_Easy_mega_20230306_normal_ESP8266_4M1M Mar 6 2023

I’ve set up four ESP8266 each with a HW-611 BMP280 board (and three of them also with a BH1750) in order to test accuracy. As you can see from the attached screenshots, there’s a huge difference in Temperature between the four boards, despite all being set to the same altitude (121 m).

Am I doing something wrong here? Are the sensors quirky or are there known bad Chinese knock-offs, which are used on this board? Or is this caused by cross-talk of the wires?

I mean: Is this normal? If so, how to calibrate them? Using a premium-quality thermometer for reference and using the offset option in ESPEasy?

4x BMP280 compared.png (413.38 KiB) Viewed 6941 times

4x BMP280 compared Setup.jpg (2.17 MiB) Viewed 6941 times

[Wookbert]

Another observation: As you can see in the following graph, the sensors are wildly jumping. Only Sensor A (which is bread-board mounted) is rather steady and calm. So could this indeed be a cross-talk issue, caused by my wild enamelled copper wiring?

Bildschirm­foto 2023-03-11 um 20.46.49.png (143.45 KiB) Viewed 6926 times

[Ath]

The wiring is I2C, and that's a digital protocol, so there shouldn't be much interference, unless you are using very long wires, or very careless wire-routing. That doesn't really seem the case, according to the photo you shared, so I would expect the sensors just being of poor quality.
You might consider the BME680 sensor (that uses a different plugin, and is a bit more expensive), or obtain your sensors from a more reliable source.

[Wookbert]

This really comes as surprise. I mean yes, this is China stuff from AliExpress, but how can these sensors be of bad quality unless they are either using a BMP280 knock-offs or Bosch is selling off B or C grade versions of its sensors (which I seriously doubt)? There isn’t much on the boards expect for the sensor and some caps/resistors.

My fear is that if I trash these and purchase a couple of BME680 (which are around USD 6.50 on AliExpress), the outcome might be the same.

[Ath]

That's why I suggest to use a reliable (European) source to obtain your sensors That may be a bit more expensive, but quality comes with a cost...

[Wookbert]

I hear a good friend praying the same words. Well, from looking at the pictures, they are selling the exact same China stuff, only more expensive (e.g. here or here; surprisingly Reichelt — which usually is a bit pricey — is pretty cheap on this item).

Advantage is that you can hold them accountable ... which is also the case with AliExpress sellers ... but it doesn’t help if you have the sensors laying around for 18 months before finding the time to use them.

[TD-er]

A few things to check:
Try to see if the enclosure of the sensor is dented or bumped.
I have had one just last week that was obviously not flat anymore on the top.
Well, wasn't so obvious, only after looking at it when I saw the readings of 180C and 1300+ mbar.

Another thing you could try is to add some 100 nF capacitor as close as possible to the 3V3 and GND pins of the board.
If they "forgot" to add real capacitors, you can also get really strange readings.

Those BMP280 boards are extremely cheap now.
A few weeks ago I bought them for like 25-ish cent a piece.
So I wouldn't be surprised if those were from some batch which may have been mishandled and perhaps had a lot of customer complaints.
Or... they are simply fake.
I have not yet compared those ultra-cheap units as I only needed the air pressure part of it for some test setup.

[chemmex]

This is very common with all BMx sensors regardless of where they were purchased. Basically, they all exhibit positive offset of up to 3 degrees.
You may calibrate them by two points, the large the distance the better. The resulting coefficients could then be put into the formula field of the task

[max100]

Good day,
is there a table of the quality of individual temperature, pressure, humidity sensors from the best to the best?