= Speed and Accuracy of Temperature Sensor =
== Speed of Temperature Measurements ==
I have 3 Dallas Semiconductor's DS18B20 connected to RS232 port through the IOM #142A board made by P.H. Anderson http://www.phanderson.com/iom142/iom142.html. Here are my first Speed Measurements.
The highest measurement frequencies I was able to achieve with DS18B20 (parasite power) is
||1 sensor||2 seconds||
||2 sensors||3 seconds for both||
||3 sensors||4 seconds for all 3||
Vadim Tkachenko added:
There is a possibility of optimization - currently, all the sensors are doing temperature conversion one at a time. However, it is possible (after making some code changes that will somewhat break abstraction levels) to make them do the temperature conversion all at the same time, per branch. The necessary condition for that is that 1-Wire network *must* be powered, so please plan ahead.
I will check with P.H.Anderson if there is a way how to use his card to do the temperature conversion all at the same time.
== Accuracy of DS18B20 ==
I was runing all 3 sensors close together all night and there are visible systematic measurement errors (see image 1 - [http://smartHVAC.com/graphs/night.gif night run] and [http://smartHVAC.com/graphs/NightDiffs10hrs.gif differences]). Beside showing different temperatures, sensors differ in the speed and sensitivity. To see more details I used hairdryer to expose sensors to hot air and to let them cool after (see image 2 - [http://smartHVAC.com/graphs/bigWave.gif wave] and [http://smartHVAC.com/graphs/diffBigWave.gif wave differences]).
Are those differences caused by different orientation of each sensor? I taped all 3 sensors to a flat surface of hard wood and measured another wave: [http://smartHVAC.com/graphs/Wave2raw.gif wave on wood]. On the [http://smartHVAC.com/graphs/wave2peak.gif detail of the the peak] you can see that the first sensor reacts faster (5 seconds?!). The third sensor "overreacts", as can be seen on the [http://smartHVAC.com/graphs/wave2arma7.gif same wave smoothed] by moving average (7 values = 1/2 minute).
Before I use those sensors to measure variances of temperature throughout the room I have to come with reasonable calibration, so values could be compared!