strikesrule
Posts: 3
Joined: Mon Sep 09, 2019 1:10 pm

Sensor Accuracy - BMP180

Mon Sep 09, 2019 6:17 pm

Running many sensors:
4 - BMP180s
3 - DHT11s
1 - TMP102
1 - Digital Room Thermometer Unattached
Raspberry PI 3B+

Sensors all report different temperatures. None the same as Digital Room thermometer. They are all within 1 foot of each other. Range varies over 5 degrees between sensors and independent unit.

Needing to have correct readings on any sensors used. It's evident that all the given readings can't be correct. I see the room thermometer as a starting point in accuracy. The issue is how to set these up so the readings are accurate.

kflmiami420
Posts: 61
Joined: Thu Mar 21, 2019 6:54 pm

Re: Sensor Accuracy - BMP180

Mon Sep 09, 2019 11:44 pm

First get an accurate source to be your control device like this one

https://www.dracal.com/store/products/u ... /index.php for TEMP

https://www.dracal.com/store/products/u ... /index.php for Pressure


Second

Stop using the bmp180 move on to the next gen sensor bmp/bme280 or the even newer bmp388

DHT11 has been changed to the DHT22 other alternative for humidity is the sht31-d i2c ready ( newest is sht85 avail on mouser bu price is 30 each )

tmp102 is ok sensor a better i2c temp sensor is the mcp9808 which is also i2c ready

Once you decide what your control device is and you have a base measurement to go by , Then you will have to adjust the python code or other code you use (by adding a math formula to compensate for the difference ) to get the results that will come close to the control device

PhatFil
Posts: 1363
Joined: Thu Apr 13, 2017 3:55 pm
Location: Oxford UK

Re: Sensor Accuracy - BMP180

Tue Sep 10, 2019 2:25 am

You cant expect any temp sensor to be really accurate out of the box without some calibration. For that you pay a premium for a certified and tested device..

a 2 point calibration is good enough to validate most sensors and close to 0C and close to 100C are temperatures you can easily approximate with water at home. If you are really serious about decimal place accuracy take the specific gravity of your water and elevation above/below sea level into account too. (web lookups these days..)

crushed ice in room temp water at a 60% ice 40% water stirred and left to sit for 3-5 minutes should provide a close to 0C temperature reference point. And the surface/steam point of a simmering pot of water should be close to 100C

when taking a 0C temp reading ensure the sensor is NOT in contact with Ice and only the water..

Cling film treated gently can provide a throw away waterproofing for sensors that need it, ( blutak over sharp pcb corners etc..).

If your sensors read off the expected value as long as its a similar difference at both ends ie +2c out at 0 and 100
your OK and just Add/Subtract this constant error to correct for display/use.
(specific gravity and elevation will have a very minor impact on boiling temp of waters but if looking at 2+ decimal places this could be significant. )

If your readings are off by different values at each end of the test scale Bin the sensor its garbage.

If really serious a 3 point test would be better.. Tho I never had the need to go that far even when calibrating the class A platinum sensors for the cold side of the microbrewery (aka brewshed)

strikesrule
Posts: 3
Joined: Mon Sep 09, 2019 1:10 pm

Re: Sensor Accuracy - BMP180

Tue Sep 10, 2019 1:26 pm

Thank you.

Water is a sound method, especially for the price and simplicity. The BMP180 python script has a ton of calculations that can be made. Never thought of going back to plain ol high school science. Since the pressure measurements are in a closed system and are not relative to outside atmospheric pressure, this can be adjusted in the script also.

The sensors I have are cheap, proven, and easy to come by. In this case, newer, more expensive, does not add up to "better".

BMP180 wins.

:D 8-)

Return to “Automation, sensing and robotics”