Tests with Many Devices

Previous Up Next

We wanted to see how well the network would work if there were many devices connected in parallel, so I arranged 19 thermometers on a breadboard. One of the thermometers stopped working, so it was removed, leaving 18 thermometers:

At a short distance, the network worked perfectly. Below is a contour graph of the average temperature at each thermometer during a test. The left end is the warmer side.

However, when I put the breadboard full of thermometers at the far end of the 300m cable, although most of the readings were successful, there were a few errors. I traced the problem back to the resistance of the cable; specifically, when about 6 - 8 devices are performing an analog-to-digital conversion (e.g. converting the temperature into a digital reading), the power consumed by the devices is high, a lot of current flows through both ground wires, and there is a voltage difference between ground at the far end and ground at the near end. This decreases the voltage difference sampled by the master and can turn ones into zeros. Fortunately, this is not likely to be a problem at Double Chooz, because the cable lengths used there will be on the order of 10 - 40 meters, and because, since the devices will probably only be read about once a minute, they can perform conversions one at a time.