Go Down

### Topic: DHT11 with temperatures outside it's measuring range (Read 866 times)previous topic - next topic

#### Bringamosa

##### Nov 10, 2017, 07:39 pm
My idea is to use the temperature sensor to convert my chest style freezer into a refrigerator. So when the temperature rises to 5 degrees Celsius a relay will turn the freezer on, when it drops to 1 degree it will turn the freezer off again.

I know that there is a margin of two degrees, i calculated on that. But my question is now what if those 2 degrees will result in a "negative reading"(so in reality it's 1 degree but the sensor is 2 degrees off so it thinks is -1 degree). It is made to read temperatures betweem 0 and 50 degrees celsius.

And, what if i put the sensor in the freezer when it's freezing minus 4 degrees? will it read anything will it give an error will it break?

I hope any of you have some info before i go for the trial and error and risk breaking the sensor.

#### knut_ny

#1
##### Nov 10, 2017, 08:28 pm
Don't use DHT in that environment. Changing temps. may give condensation on the sensor. resulting in a defect.
(Below zero will not destroy it)

A better choice is a thermistor. costs less, performs better

Ny

#### Bringamosa

#2
##### Nov 10, 2017, 09:48 pm
A better choice is a thermistor. costs less, performs better

Thanks a lot for your comment. I have a thermistor in my starter kit package I see. Is there anything else I should consider?
Is there substantial quality difference in termistors? And how do I know what temperature range I can use them in?

#### wvmarle

#3
##### Nov 11, 2017, 02:12 pm
Thermistors will work in just about any temperature until it's so hot they melt.

However you may want to change your pull-up resistor to match the typical NTC resistance for your temperature range. You get most accuracy if your NTC resistance is close to that of your pull-up resistor. A regular 10k resistor with 10k nominal NTC will work fuine, but as you want to measure 1-5°C mostly, a 30k resistor (or a value near it, like 33k) will give better accuracy in that range.
Quality of answers is related to the quality of questions. Good questions will get good answers. Useless answers are a sign of a poor question.

#### Bringamosa

#4
##### Nov 11, 2017, 10:26 pm
Thank you for your help wvmarle.

but as you want to measure 1-5°C mostly, a 30k resistor (or a value near it, like 33k) will give better accuracy in that range.
Is this because the resistance of the thermistor will be around the 30k when around 1 to 5 degrees? At 23 degrees the resistance of the thermistor is aroun 9k, so i thought to use a 10k. But if i understand you correct the idea is to use 30k because around those low temperatures that will be around the thermistor?

#### wvmarle

#5
##### Nov 12, 2017, 01:41 am
What's the nominal value of your thermistor? A 10k nominal (at 25°C) would be more like 11k at 23°C.

Indeed at your target temperature a 10k nominal thermistor would be around 30k, trying to match that value for better accuracy.

Thermistors can be really accurate but you have to know the resistance of it and its pull-up resistor accurately. For that reason I prefer to use thermistors with tolerance of 1% or better and a resistor of 1% or better tolerance. Then you have just +/-0.2°C error or so.
Quality of answers is related to the quality of questions. Good questions will get good answers. Useless answers are a sign of a poor question.

#### Bringamosa

#6
##### Nov 12, 2017, 01:53 am
Thanks a lot again. So what I could do is measure the resistance when put in ice water (close to 0°Celsius), find a resistor as close to that reading as possible and then change those values in the code. Right? Nominal value at 0°

#### wvmarle

#7
##### Nov 12, 2017, 04:15 am
Yes, that should work.
You could also just keep the rated Tnominal and Rnominal and just set the pull up resistor value to the actual value.
Quality of answers is related to the quality of questions. Good questions will get good answers. Useless answers are a sign of a poor question.

#### MarkDerbyshire

#8
##### Nov 12, 2017, 02:55 pm
Use a DS18B20 probe.  They are readily available in waterproof packages and don't need calibrating

#### wvmarle

#9
##### Nov 12, 2017, 04:06 pm
Thermistors are just as waterproof and also don't need calibration.
Just need to take a little more care in circuit design, but then they're very cheap, can come in tiny packages, are nearly indestructible and need only two wires instead of three.
Quality of answers is related to the quality of questions. Good questions will get good answers. Useless answers are a sign of a poor question.

#### Bringamosa

#10
##### Nov 12, 2017, 08:29 pm
I have a thermistor allready so i'll go with that. there is not a ton of accuracy needed. got this running for now it's just a matter of figuring out what else i want to code with it. Thanks a lot all!!