Playing with sensors:

I've been goofing around lately with some one-wire sensors.  I have RJ45 jacks all over the house, and I've been wiring them into the one-wire hub, and getting temperature data on all the rooms.  The problem is, I purchased a bag of 25 raw DS18B20 chips, so I've been soldering the things onto little stub wires, and then crimping them onto rj45's and it takes forever.

Then I had the idea of just crimping the chip right into the RJ45.  It actually fits perfectly.  But I was worried that the data would be messed up, because it would basically be measuring the temperature of the wall, not the room.

So I crimped one stub, one 4" wire, and put them right next to eachother for awhile, and recorded the results:

 

The long wire is much more sensitive.  The way it jumps up and down, is actually the air conditioner going off repeatedly.  The stub seems to smooth the results out more.  It does seem however, that the stub is consistently about .5 degrees higher on average, so I guess that's the effect of the wall holding heat.  Based on these results, unless I need actual detail, these stubs will be fine, and are much easier to make.