I did build my first RFM12 temp sensor.
So i took him outside (no case) and it takes 8 minutes to go from 10°C to 3°C?
10,81
8,12
6,62
5,31
4,62
4,12
3,68
3,37
3,12
So what will happen inside a case like this with almost no air flow:
I wonder that people put them in the wall inside switch panels to control there heading.
I've seen similar behavior -- note that the approach to equilibrium is an exponential function of time.
The plastic case of the DS18B20 is not a good conductor of heat. Most of the heat transfer is through the wire leads. If speed of equilibration is important you could consider soldering the ground lead to heavier bare wire or a small metal plate, or perhaps buy the sensor in a metal case.
I have a device that measures CO2. It is pretty slow as well, and I want it to have a stable reading after 30 seconds. So I used a cardboard tube, and put a small fan at one end. The fan moves air at only 1/2 liter a minute through the tube, and the device reacts well within the allotted 30 seconds. Perhaps a slow fan would help your situation as well.
But shouldn't it be the case that the smaller the temp sensor is the faster it goes in sync with the surrounding temp?
If i would but a heatsink on, the mass would be bigger...
If i take a big and a small peace of cold iron into a warm room then the small peace is faster warm.
@ChrisTenone, 12 fans (one for every room senor)?
What about battery runtime?
MrGlasspoole:
But shouldn't it be the case that the smaller the temp sensor is the faster it goes in sync with the surrounding temp?If i would but a heatsink on, the mass would be bigger...
True enough, but it really relates to the ratio of surface area to mass. A piece of foil would be the best to attach.
You will note that thermostats are not actually in the wall, but in a little housing with slotted vents at the bottom and top so that drafts in the room pass through them. However, the temperature in a room necessarily changes comparatively slowly, so it is not that much of a problem.
I really mean in the wall 
What you are talking about are the cases i ordered (the link i posted).
ratio of surface area to mass
Hm i already thought about something like that and was looking around for something and found stainless steel slices: 100 Metallscheiben / Metallplättchen Ø 10 mm mit Klebepunkte
They also have 15mm.
Would drilling a whole into the case and mounting the slice outside (DS18B20 attached through the whole with thermal adhesive) make it better then just the slice inside the case?
Edit
Also found some 22mm brass slices they are allot cheaper.
But brass oxidizes and would not look good on such a case.
So the options are stainless steel outside or brass inside.
Edit again
Even found now some 53mm brass slices and they are thinner.
So options are:
But shouldn't it be the case that the smaller the temp sensor is the faster it goes in sync with the surrounding temp?
True. This is why people use microscopic thermistor beads or thermocouple junctions when rapid temperature response is required.
I have a couple of Western Electric 23A thermistors, which are about 0.02 mm in diameter. They are nearly invisible, connected by even less visible wires. However, I am afraid to take them out of the box.
8 minutes is quite quick for a heating system - the time constant of a building is measure in
hours/days.
"time constant of a building" - had to google that
What we have for our house (Germany) is a energy performance certificate.
You need to have it if you want to let.
What you can see there is the heat capacity of the building (i guess that what is what it is called in english).
But that thing is a joke - it is calculated on consumption.
Trying to measure air-temperature without forced-convection is always going to be
problematical - still air has an extremely low thermal conductivity. So you either
force-convect or thermally bond the device to a largish heat-spreader with a low mass
(more surface area).
Radiant heat is significant on a black object - popping the device in a metal tube will
protect against that (unless you want to measure radiant heat too - this is important
for human perception of temperature, note).
You still haven't explained why a few minutes for a temperature sensor to reach equilibrium is a problem.
It is not sensible or practical to switch building heating/cooling on and off every few minutes.
Yeah, almost 1°C/min. I don't see a problem here with room air temperature control. Of course smaller temperature gradients will be slower, but still not too slow in practicality.
jremington:
It is not sensible or practical to switch building heating/cooling on and off every few minutes.
Maybe i have a fallacy.
I have a small bathroom 8m2 (85ft2) and want to but a infrared heater on the wall.
I want to turn it on just before i take a shower so i don't want to buy the smallest one.
If it takes for example 10 minutes before the sensor has 24°C then the heater runs 10 minutes longer and i have a sauna?
The same with water radiators where the valve only goes on and off?
I want to turn it on just before i take a shower so i don't want to buy the smallest one.
It is hard to see the need to automate that particular application.
On the other hand, you will find that the plastic DS18x20s respond very rapidly to infrared radiation.
MrGlasspoole:
Maybe i have a fallacy.
I have a small bathroom 8m2 (85ft2) and want to but a infrared heater on the wall.
I want to turn it on just before i take a shower so i don't want to buy the smallest one.
If it takes for example 10 minutes before the sensor has 24°C then the heater runs 10 minutes longer and i have a sauna?The same with water radiators where the valve only goes on and off?
A infrared heater warms objects in the room, which heat the air in turn. That makes it change the air temperature really slowly. It seems fast to you, because you are one of those objects when you are in front of it. In this case, the thermostat functions more as a limiting switch to prevent long term overheating.
Most water heaters work by convection. You can imagine how slow that is. Only forced air heating can change air temperature quickly (for obvious reasons).
I wouldn't expect massive overshoot for the reasons AARG mentions. However if it is a problem, PID control would help.
I'm thinking about this problem like this:
If the goal is to shut off IR heat when a certain temperature in a small room is attained, a high-tech sensor is probably not the best solution. Instead of your chip, consider finding an old dial-type thermostat, the ones with a bimetallic coil and a mercury switch. (don't worry, the mercury is enclosed in a glass vial, and can't get out ... unless you break it.)
The switch will be open (wire/program it to turn off) under the set temperature, and closed (programmed to turn on) above that temperature. Actually, there are a few degrees of hysteresis, so the turn off temperature is lower then the turn on temperature. Thermostats operate within a few minutes.
You will need to find the two contacts from the mercury switch and wire them just like a button or other spst switch. The Arduino can control the IR lamp, schedule, timing, stay off, etc., and use the thermostat switch to tell if the room is warm enough for you.
Thermostats operate within a few minutes.
Is "a few" significantly less than 8?
jremington:
Is "a few" significantly less than 8?
It can, but if you want it really fast, mount the bimetallic strip on a heat sink, and blow air across it with a fan. I would think that would make it react in less than a minute. I have an old electric heater that works that way, but the thermostat turns the heat on and off directly, without an Arduino.
jremington:
It is hard to see the need to automate that particular application.
Why not? If you let it run the whole time it gets warmer and warmer...
Ok i had to check what a mercury switch is in German.
First thing i thought of when i was reading mercury was this:
