Hello,
I connected an NTC temperature sensor like the circuit provided here. I don´t know which sensor I have exactly.
What can I do to assign a temperature (°C) to the analogue measured value?
In a room around 17°C I get an analogue value of 125. If I touch it with the hand around 150, and with contact to material from the freezer around 60.
Is it a linear relationship? Then I would take to reference temperatures (0°C and room temperature) and carry out a linear regression
If it´s easier, please provide an example sensor which I can buy, because then I would have the datasheet. It should be supplied by 3.3V of Arduino MKR 1400
The resolution and accuracy are different things.
The resolution of the DS18B20 is 0.06K. The accuracy of the measure depend on calibration. You will never get the accuracy even 1K with NTC without the calibration.
Is it for scientific research? Then only use calibrated and wel certified products.
There are not so much things that need a temperature at 0,1°K correct. If you measure all times with the same system, you will have always the same fault but it make no sense for conclusions. You can only calibrate with calibrated instruments that are better spect then what you use. Calibrating at 0,1°K needs instruments that where spect at 0,01°K. Melting water or boiling water are pressure independent. At sea level water boil at 100°C. At 2000m water boil at 93°C.
No, for agriculture purpose. I need to recognize exactly when the temperature goes under 0°C. The range where I need the highest accuracy is -6 to +2 °C
You focus on 0°C. But that isn't correct. My car is always outside. At 2°C he is totally frozen. The roof of the house is then frozen too. If it is to prevent fruit or something to be frozen take a temperature a little higher.
The resolution of the temperature sensor is user-configurable to 9, 10, 11, or 12 bits,
corresponding to increments of 0.5°C, 0.25°C, 0.125°C,
and 0.0625°C, respectively. The default resolution at
power-up is 12-bit.
It sounds like you are maybe doing a wet bulb, dry bulb measurement. If this is what you are after I would use two DS18B20 temperature sensors and just use a wetted wick on one. Then have a small fan pass airflow over both. My experience with the DS18B20 temperature sensors is the ones that are bulb style as linked to meet their specifications.
The only way to actually calibrate any temperature sensor is using comparison to a higher standard of accuracy. I use an old set of mercury in glass thermometers for comparison which meet my purposes. Resolution and accuracy are two different things. Resolving for example three places to the right of a decimal means absolutely nothing unless the uncertainty of the sensor can support it. Looks good and has no significance.
You also want to consider a sensor's response time. Using the DS18B20 it should be apparent the sheath encapsulated version will have a slower response to temperature change than the standard TO92 case version.
What is the advantage over programming the alarm with the Arduino (if-conditions)?
Yes, but why a small fan? In order to meassure the "real" temperature the natural wind should be enough?
Just ice water and boiled water and linear regression is not exactly enough, I assume?
This one is not available for me. Do I have to look for some special specifications or producer?
How is the DS18B20 calibrated? For example calibration points out, that the temperature is 1.5K to high. Is the difference only taken into account in the programme (Arduino) code?
Wow, have to give you credit. Some really good questions you have asked. OK as I mentioned when I see wet bulb I assume someone is after getting a RH (Relative Humidity) measurement. Today's really nice and accurate temperature sensors were not always available. We were pretty much limited to liquid in glass thermometers. For humidity we measured a wet bulb / dry bulb temperature and then went to a nomograph to get the RH.
So why do I mention a fan? The idea here is the wet bulb / dry bulb is pretty much based on evaporation. Pretty much how if I wet my finger and hold it in a breeze the wet spot feels cooler. The ambient air is the same temperature be my finger wet or dry but the breeze causes evaporation resulting in a cooling effect. If I construct a wet bulb / dry bulb less any breeze both thermometers or temperature sensors will just read ambient temperature which will be the same. I need the breeze to get the evaporation process going. A good example is a Sling Psychrometer, a hygrometer consisting of a wet-bulb and a dry-bulb thermometer, the difference in the two thermometer readings being used to determine atmospheric humidity. You sling it around (the breeze) and read the two thermometers. Here is an example. Here is an example where a fan is used.
So why not just rely on the wind? What happens on a windless day? The fan simply guarantees airflow over the wet wick.
No actually making an ice bath was how 0 degrees C (32 F) was done. Crushed ice was placed in a Dewars Flask. It was called a slurry. You can make your own which is good enough. Crushed ice and distilled water and make a slurry mix. Sort of like slush.
Next bring distilled water to a boil which is your 100 C (212 F) at sea level. If not at sea level just look up the boiling point of water at a giver altitude. For example at 2,438.4 metres (8,000 ft) in elevation, water boils at just 92 °C (198 °F).
If a encapsulated DS18B20 sensor is not available to you then you should be able to just get some off the shelf versions in what is called a TO92 case design. A little epoxy and you can easily make your own.
Finally if measuring relative humidity is what you are after there are plenty of RH sensors available like the DHT 11 and DHT 22 just as a few examples. There are plenty of RH sensors to be found using Google and just about all of them compatible with Arduino micro-controllers. RH calibration does get a little more complicated.
Amazon proposed this for me instead. Does the manufacter matter, respectively could I buy these sensors?
Measuring the RH is not the plan, so far.
I calibrated the NTC sensor which I have already with ice and boiled water, but it was not exactly enough. In my room the sensor measures minimum 4 degrees too less... But I will restart with DS18B20. Maybe I can ask a friend from a laboratory for a good reference thermometer.
Well figure it this way as Wawa pointed out. An actual Dallas Semiconductor DS18B20 cost about $12.00 USD last time I looked. The stuff off the boat from China for the most part works. Not always meeting specifications of an actual Dallas Semiconductor data sheet but they work overall OK among the ones I have played with and compared to accurate temperature standards. So what it comes down to is you get what you pay for. Especially for hobby applications. A high end PRT sensor meeting ITS 90 specifications can cost a few thousand US dollars and that is just the sensor. Calibrated across several points the price goes up.
So what exactly is your project and just how accurate do you actually need?
OK, I have a DS18B20 sitting here just like the one you linked to. Comparing to a liquid in glass mercury thermometer the DS18B20 off the boat cheapie is reading 20.87 C (69.57 F. The standard thermometer is reading 20.56 C (69.00 F) so really not bad considering I paid about $10.00 at most for 5 of them through Amazon.
Again knowing exactly what your project is would go a long way in you getting good advice here.
I want to get a notification when the (wet) temperature goes under 0°C or some degrees less, depending on the vegetation status of fruit trees, in order to start frost protection measures. The accuracy around 0°C is really important, otherwise the whole project is mostly for nothing.
What means not fake in this context? Produced by Dallas?
Unfortunately Adafruit has to high shipping costs to Germany. Can you propose something else?
Well go figure? I never would have thought of that, especially using a wetted bulb. Here in the US when a freeze is predicted the citrus growers start taking corrective action. Not sure why you want wet bulb but you're the grower. I figure at ice point 0.0 degrees C is easily tested for using a ice bath slurry as I mentioned before.
As to counterfeit verse legitimate sensors. My experience with the inexpensive versions off the boat from China has actually been good. The one I have been playing with since the thread began is within 0.5 F of nominal. Anyway easily calibrated so you can include any offset in your code.
I would just go with the bulb design we mentioned earlier. That or as covered roll your own using the TO 92 case style.
In your design if you want to add a nice feature just add a display so you know what is going on at a glance. Add an alarm function so it can give you an audible and also display on your display. I use one of these in projects. Not knowing your exact location I have no idea what is or is not available to you. OK, I see Germany. You have Amazon in Germany? Looks like you do and I really hope you guys have a mild winter this year.
OK, I see Germany. You have Amazon in Germany? Looks like you do and I really hope you guys have a mild winter this year.