I'm doing an experiment in a graduation project, where I will compare some temperature sensors with a LM35 sensor (using it as the default point of comparation), using the Arduino UNO to collect these data, and the MATLab to export and work with the readings. The experiment was planned to be made with the sensors underwater, with an electrical system to control de temperature of the water. However, I just found out that my LM35 sensor isn't waterproof (TO-92 capsule).
I'll also use and RTD pt100, a simple Thermocouple K, an industrial Thermocouple J, and a Thermistor (around 7.6 kOhms).
I would like to make a few questions:
Is there any other sensor that has around the same precision as the LM35, that I can use underwater?
Is there something I could use to seal the sensor? Like Silicone sealant, Epoxy or any other substance?
In case something fails, and the sensor is totally underwater, is there any danger to the arduino UNO board? Is there any kind of care I should take before doing the experiment to protect the hardware?
Is there any "waterproof" version of the sensors mentioned before, that I could use to make the experiment without trouble?
Sorry for so many questions, but I'm still in the early parts of practical work, and I didn't have any prior practical experience with sensors. My mentor said we can safely submerge the thermocouples and the RTD, but I'm concerned mostly about the LM35 and the Thermistor.
Thanks in advance for all the help, and sorry for the trouble!
Lots of commercial sensors designed for underwater research are available. A web search will find them.
Unless you spend some time doing serious research on the topic, anything you cobble together yourself will be prone to failure. For example, mere exposure to salt spray can quickly destroy electronics.
You can seal a sensor plus it's wiring into a metal tube filled with silicone sealant - best to use one made for aerospace/electronics manufacturing that does not give off acetic acid during curing.
Add heat shrink sleeving over the cable/tube interface.
Physics will not tell you, engineering might tell you, but almost all temperature measurements are the result of heat transfer using at least one intermediate material between the source and the sensor.
You can do the same by using a metal sealed to keep the water from entering the enclosure, but attaching your LM35 to the metal part inside the enclosure. You will still get the temperature of the outside water. Temperature changes take a bit longer to register.
It was my mentor's suggestion. He never tried working with arduino himself, but he has decades of experience in sensors and transducers. He said the LM35 would be the best as the default sensor, since it is an integrated circuit.
Since it was his suggestion, we continued the experiment using the LM35 as the standard one. But in any case, even if we use the Pt100 as the default transducer, we still need to compare with the LM35
When I used a TO-92 sensor decades ago I connected the cable and then embedded in epoxy. I used a small section of a straw to contain the epoxy until it cured. The tip of the sensor can be seen outside of the epoxy. There is a "waterproof" heat shrink tubing that could be put over the embedded sensor for additional protection.
Realize that any of the "waterproof" sensors that are directly exposed to water will leak eventually.
Sure,the LM35 produces an accurate voltage with temperature.
But how are you going to measure that.
When measuring that voltage with an Arduino, then precise temperature display will depend on the precision of the A/D referenceof the Arduino (which is bad).
it is much better to use a digital temp sensor, like the DS18B20, as reference. which is independent of the A/D of the Arduino.
Leo..
glue the LM35 to the outside of the aquarium / tank at different heights
put thermal insulation over it.
Will only work if the wall of the tank is thin enough and can transfer heat well (metal).
Another way could be to drill holes in the tank where a LM35 just fits in, so its "head" is in the water. Add a lot of epoxy to seal things of. (don't know if destructive testing is allowed)
That is not a valid reason to elevate it to "best". Either your advisor is not informed about the universe of sensors, or you misunderstood them.
PT100 and thermocouple sensors require the use of IC amplifiers + ADC converters and are among the most accurate of industry standards for temperature measurement.
@konakona3
Why don't you spend a little time explaining what you intend to compare among all the sensor types and the implications that might have for the way you address waterproofing them?
The LM35 could be a poor reference, because it depends on another device to measure it's output voltage. It would only be OK when measured wit a calibrated DMM (digital multimeter).
It could be 10% off when measured with an Arduino.
And then there is resolution.
Expect course 0.5C steps with the default 5volt Aref of an Arduino (0.1C with better code that uses 1.1volt Aref). The digital DS18B20 defaults to 16 steps per degree C, which is better than the two or 10 with an LM35 and an UNO R3.
Leo..
The plan is to test how each sensors works in the same conditions, to compare the linearity, response-time, etc. We want to do 4 experiments in total, using a Thermopar K, Thermopar J, Thermistor, RTD Pt100 (2 wire) and a LM35. In each one of the experiments, we should use the sensor chosen as the default one, paired with one of the other sensors.
The first plan was to do the experiment underwater, submerging 2 sensors at a time (the default and one of the others), at the same water level, to compare their readings. We are using a metal recipient, with a VARIAC to control the tension, and therefore, how much and how fast the water will heat up. That is the part where we stopped, since we wanted to be sure on how to make the experiments without damaging any sensor or component when submerging.
A few more information about what we just did:
Since we were using the LM35 as the defaut sensor, all codes we prepared are using the 1.1V internal reference. We are also using the MAX31856 and MAX31865 for the Thermocouples and the RTD, respectively.
The readings are great, considering the range around 23~32 degrees Celsius. But to test them underwater, I wanted to be sure first that it is safe.
We are not limited to the underwater experiment. But we are trying to test the sensors with stuff we have here already. We could try a different way to control temperature. Something that doesn't involve water.
Since response-time is one of your measured quantities, you will want to make sure that however you encapsulate the transducers affects the response-time the same.
I was going to suggest an oil bath instead of water would be less likely to cause a short circuit.
Evidently.....
Read posts 12 &14 again - twice. One can only make a vague guess about the purpose of this exercise, it seems just a waste of time. I submit that a casual browse around here would show that the obvious go to is the DS28B20, packaged as shown above.
About the only reason for not using it would be because you are working outside its range - which is not the case here.
