 Notice. New forum software under development. It's going to miss a few functions and look a bit ugly for a while, but I'm working on it full time now as the old forum was too unstable. Couple days, all good. If you notice any issues, please contact me. |
Forum Index : Microcontroller and PC projects : A question of Temperature Sensors
| Author | Message | ||||
bigmik Guru  Joined: 20/06/2011 Location: AustraliaPosts: 2950 |
Hi All, I have been working on a weather station project (Yes I know this topic has been broached many times before) And I have 3 sensors that will measure Temperature BMP180 +- 2 degrees C (or 1 degree depending on who you believe) DHT22 +- 0.5 degrees C DS18B20 +- 0.5 degrees C Now I have regular reports, via an HC-12, from my remote unit (up the pole) and I thought that it would be a good idea to simply average these 3 sensors. It is probably pedantic but I am ALWAYS seeing the BMP180 with the highest read, the DS18B20 with the mid range and the DHT22 with the lowest read. Typical readings are (like this current one) BMP180. 26.1 DS18B20. 25.25 DHT22. 25.1 From the vast experience of members who have been this path before Do you think I should discard the average and only use the DS18B20 as this is designed as a temperature sensor where as the other 2 are primarily for Pressure (BMP180) and Humidity (DHT22).? The DHT22 is often lower than the DS18B20 by 0.25-0.5 degrees C. This current reading is closer than the norm. The DS18B20 I chose was a genuine part bought from MOUSER so I feel that using it ONLY would be fairly accurate to within its rated accuracy. The BMP180 being almost always nearly a degree higher than the DS18B20 does tend to skew them average calculations a bit. Anyway what are your opinions.? Kind Regards, Mick Mick's uMite Stuff can be found >>> HERE (Kindly hosted by Dontronics) <<< |
||||
TassyJim Guru Joined: 07/08/2011 Location: AustraliaPosts: 6283 |
I have seen similar variations. Some can be due to self heating depending on layout of the sensors. I have to make a similar decision for my WX station. I expect that the DS18B20 will end up winning. The location and construction of the weather instrument enclosure will be the biggest determining factor for accuracy. Jim VK7JH MMedit |
||||
phil99 Guru Joined: 11/02/2018 Location: AustraliaPosts: 2642 |
It is probably a bit late for this, as they are up the pole, but they can be calibrated against an accurate thermometer at two temps. The readings are then corrected before being displayed. This has worked quite well for a DHT11 - an inferior cousin of the DHT22. For a normal winter / summer range of 0 to 40C your calibration points could be 5 and 35C. If interested I will dig up the equation and calibration procedure. |
||||
| bigmik Guru Joined: 20/06/2011 Location: AustraliaPosts: 2950 |
Hi All, @Phil, Whilst it is 'up the pole' it is not difficult to bring it down again, The pole I purchased is a very nice unit that can rotate down 90 degrees to make the unit accessible with a small step (see attached)  All 3 sensors are within 10mm of each other and are located in the `bee hive' section under the rain collector in the above picture. Incidentally the pole was very cheap from element14 (it was $18AU) but they have almost doubled the price but is still great value at about $34AU. Element14 Pole I still probably wouldnt go to the effort of pulling it down (then stripping the unit down to access the sensors) and I don't have access to a know good thermometer either. I am leaning towards the same as Jim's thoughts of going with the DS18B20 solely and ignoring the other 2 sensors Temperature readings. With the benefit of hindsight I should have done some calibrations using a cheap mercury thermometer from the pharmacy. I have other DS18B20 from Mouser in the same batch (Presumably) I could test them to see what they are like. Thanks All, I think I will now discard the other sensors readings especially as I cant be sure that they are genuine anyway as they were purchased from that big "Chinese Candy Shop". I can be sure of the Mouser part and I did run the test that someone wrote to test for fakes on the DS18B20 and they passed. Take care all, Kind Regards, Mick EDIT ** I don't know why the image is rotated but You can make out what I mean about the pivoting pole. MBG Edited 2022-12-11 13:57 by bigmik Mick's uMite Stuff can be found >>> HERE (Kindly hosted by Dontronics) <<< |
||||
| PeterB Guru Joined: 05/02/2015 Location: AustraliaPosts: 655 |
G'Day All What bothers me is that in the Australian summer, that "bee hive" will heat up to who knows what? and as noted by Jim you will get some pretty funny results. I found the best place was under a rainwater tank mounted a couple of feet above ground and protected from direct sun light but open to air movement. I could then get believable results using a 18B20. Good luck  Peter |
||||
| TassyJim Guru Joined: 07/08/2011 Location: AustraliaPosts: 6283 |
This is what you should be using https://en.wikipedia.org/wiki/Stevenson_screen VK7JH MMedit |
||||
| PeterB Guru Joined: 05/02/2015 Location: AustraliaPosts: 655 |
G'Day Jim. That would be very nice but I'm not that fair dinkum about measuring temperature. Under the tank gave readings close enough to the official temps for my area and I suspect that the 18B20 is more than accurate for my situation i.e. other errors swamped the 18B20. Peter |
||||
| bigmik Guru Joined: 20/06/2011 Location: AustraliaPosts: 2950 |
Hi All, @Peter, I was wondering how things would be on the 45 degree C days and also how the plastic shell, especially being majority of White, would handle the sun. Whilst the bee hive is no where near as good as the stevensons screen Jim mentioned , the way it is designed gives a reasonable amount of protection due to the overlapping effect of each shell. My water tank sits on a concrete slab so unless I bury the sensors there is no going there. I also doubt that it is a great place for a temperature sensor due to the mass of water that will tend to impede the temperature due to its large mass of water.. Maybe the bee hive mounted under the eaves would be a better solution for them. At this stage the unit is just in the trialling stage as AndrewG is writing the proper code for it. I am mainly trying to calibrate wind speed (against the Bureau of Meteorology) and rain fall (against my standard tube gauges). If all goes well I will have to pull it down so I can conformally coat the PCB and put dielectric grease around possible areas of water ingress. I have also thought of conformally spraying the outside of the unit but I would have to avoid the anemometer cups and the weather vane as I don’t want to impede their actions. Similarly the rain water funnel so it would only be the white areas. @Jim, The stevensons screen might be OK in a rural environment but living in suburbia there is NO where I could install one (the unit as is is 2m from the neighbors fence) Thank you all for your informative input, it is all interesting. I think the bottom line is all measurements are indicative and I have to make some compromises (one would then ask why the point of my original question then, quite rightly). Take care All, Regards, Mick Mick's uMite Stuff can be found >>> HERE (Kindly hosted by Dontronics) <<< |
||||
CaptainBoing Guru Joined: 07/09/2016 Location: United KingdomPosts: 2170 |
you only have a 1 degree spread. When I display a temperature that is for "human consumption" I don't even bother with the decimals, I round up or down to the nearest degree. you could always do both - if you have room. if you trust the 1820, use that as your main temperature and then show the average as a secondary in brackets. so like 25.25C (25.48) I always like to have all the info and then make a choice Edited 2022-12-11 20:33 by CaptainBoing |
||||
| bigmik Guru Joined: 20/06/2011 Location: AustraliaPosts: 2950 |
Hi Captain, That is an interesting idea, but probably complicates the display of data), my spread across the 3 sensors is often nearly 1.5 degrees C with the BMP180 the highest by nearly 1 full degree. As the BMP180 is only +- 2 degrees I think it best I ignore it anyway. The other 2 sensors have +- 0.5 degree accuracy and are almost always within 0.5 of each other it is down to either averaging these 2 or going with the dedicated temperature sensor (DS18B20) I think it is now a moot point as the average will always be no more than 0.25 degrees C from just the DS18B20 and I don’t think that 0.25 degrees is going to impact me at all. Regards, Mick Mick's uMite Stuff can be found >>> HERE (Kindly hosted by Dontronics) <<< |
||||
 Print this page |
 |
The Back Shed's forum code is written, and hosted, in Australia. | © JAQ Software 2025 |
