The big thing to remember is that thermocouple signals are very small - microvolts to millivolts - in that temperature range. This means that you must be very careful about spurious signals of all types.
One in the ubiquitous power line noise. If you have any pumps or solenoid valves in your system the voltage and power levels for those devices is 3 to 6 orders of magnitude larger than those of the thermocouples. Very small amounts of capacitive or inductive coupling between the power lines and your thermocouples can overwhelm the signals. Shielded thermocoulple wires may help.
Another is unintended thermocouple effect. This can occur when splices or connections in the thermocouple lines are at different temperatures. You must use thermocouple wire or thremocouple extension wire of the same type as your thermocouples. Solder is generally not a good idea with thermocouples because you are adding at least two new metals (= 3 or more new thermocouples) into the circuit. The size of the wires is mostly dictated by mechanical considerations. The currents are so small that the voltage drop due to the resistance of the wires is almost never significant.
The type of probe head should be driven by how the thermocouples are installed in the tank. Industrial systems usually use thermowells, but they can be expensive. For a home project I would probably select a stainless steel sheathed probe. The cost is similar to other types and they hold up well. You may want to put soem extra silicone sealant around the transition joint where the wire exits the sheath to help with the waterproofing.
Have you considered that your thermocouples will look like lightning rods to any thunderstorms which come along? Even if you do not get a direct lightning strike the static discharges will destroy your DAQ devices, your computer, and may injure you. Even if the tank is protected by lightning rods and is well grounded, your equipment is at considerable risk. I have seen the insides of equipment cases copper plated by wires vaporized by a nearby lightning strike. This was a radio receiver connected to a rooftop antenna, but the lightning does not care what gets in the way of its path to ground. Completely shielded thermocouples with the shields connected to a lightinig-protection-rated grounding system might work. Maybe. I would probably add transient voltage protection at all the thermocouple input connections to the DAQ devices. The details are beyond what is feasible in a brief posting on the Forum.
Signal conditioning at the tank can avoid a lot of the noise problems. Sending the data to the computer optically or via a radio link can avoid the lightning issues (except for power to the signal conditioners/transmitters).
Lynn
