Continuing my geeking-out of my aquarium, I wanted to have temperature monitoring. I have a pile of Dallas-Maxim DS18S20 1-wire temperature sensors left over from when I added temperature monitoring to my house, but the bare DS18S20 is not going to like being submerged. Since this is an aquarium with fish I had to be careful with what I use to waterproof it, anything that leeched metal or chemicals would be ‘bad’. So, the obvious solution was to use aquarium epoxy (normally used to stick rocks together, or attach coral to rocks.
The DS18S20 is pretty neat because all you have to do is wire it directly into the 1-wire sensor network. While the network is called ‘1-wire’, it actually needs at least two wires (data, and ground), and a thrid optional line for power. Since I have a large sensor network I use the power line (it helps reliability with large networks). The standard way of hooking these things up is to use an RJ-11 ports/connector, so I attach the sensor to a piece of Cat 5E cable, and crimp an RJ-11 connector on the other. Here is a sensor badly soldered onto Cat 5E:
To prevent the wires from shorting when, I coated them in liquid electrical tape. This stuff is supposed to be waterproof, but I don’t know if it is fish safe. I wouldn’t recommend just using it to waterproof the DS18S20.
The probe was encased in Marineland Holdfast aquarium safe epoxy. This stuff is really easy to work with, it comes as a stick that you break a piece off from, then knead it with yours hands till it is a consistent grey-white. Once that is done, you have 7 minutes to work with it. I rolled it between my hands to make a rope, then flattened the rope on my workbench, put the probe in the middle of it, wrapped the flattened epoxy around the probe, that rolled the probe between my hands again to turn the epoxy+probe back into a rope. Care was taken to ensure that of the probe with the sensors was a nicely rounded, and thick enough that the water shouldn’t get through it once cured. Curing was only supposed to take an hour, but the probe still was slightly flexible 12 hours later (but the surface of the epoxy seemed to be pretty hard), maybe it took longer because of the thickness of the epoxy, or the temperature in my basement (only about68-70F). Here is what the probe looked like after curing:
I mounted the probe in my tank by using a twist tie to fasten the Cat 5E tail to the output of the filter. Since I already pulled a 1-wire network run to behind the tank for the heater powre monitoring, I just plugged the temperature sensor into a second jack I added. The probe tip is about 3-4 inches under water now, and I am hoping the filter output is circulating the water well in that area. You can see the probe in the upper left corner of this picture:
Reading of the probe is done using digitemp in Linux, under normal use logging once a minute. For testing I briefly had samples being taken once a second. The probe is surprisingly responsive, with a 17 degree Fahrenheit temperature change settling in less then 90 seconds:
The probe has been in the water for about 2 days now and is still functioning perfectly, hopefully it will stay that way for years.