For now, I consider only weight of the hive is useful to some extent in production.The 2 months of data did not really show anything useful or interesting.
Cloud service is expensive and unnecessary; my plan is to have web server on microcomputer connected to Internet by mobile (4G) network.They talked a lot about the system setup, and how they designed the system for a cloud interface,
The thermistors I am using come from the manufacture embedded in the end of a sealed stainless steel tube and a wire lead. I was assuming that would make them fairly robust, but I was wrong. I am thinking next time I replace the thermistor I am going to seal the stainless steel tube to the cable with a glue lined heat shrink.A quote - "Moisture induced failure is the greatest cause of failure in thermistors. This type of failure is often difficult to pinpoint in an application. Moisture failures can occur in almost any epoxy-coated component or any probe configuration using epoxy-coated parts (except probes using hermetic seals)."
I have always used thermocouples for short term data and simplicity although less accurate than thermistors it is within the useful accuracy range of +/- 1 C. I have procured hermetic sealed stuff but I am not aware of the hermetic, glass bead design approach for thermistors - it cannot be easy. Glass bonds and seals are tough issues. To get around the proprietary issues we accepted material by testing ( basically connectors). You might try a cheap thermocouple with a good jacket and see what happens long term in the high RH environment. It will be more susceptible to electrical noise?
I have chosen a very simplistic approach to learn how to monitor a hive but will have to increase complexity soon. I need multiple point testing to verify early conclusions. Think I will start by install thermocouples and use a plug -in display along with my cheap weather station packages. This will allow cross checking and acquire some spatial data while observing colony performance over time.
It is a slow process but it is a hobby.
The greatest challenge is multi-point, accurate humidity testing at reasonable cost; say + or - 1% or even 2%. I prefer collecting data, cheaply, from a larger group next and will be moving to a 6-9 hive data base using weather stations and installed multi-point thermocouples with hand recorded data. I have time, it is free and the approach is easily changed. I may have to record thermocouples in the rain and snow. Transmitting weather stations are neat and cheap but have high toleranced RH values. IMO, as I have learned at the colony level, environmental data is of lesser value without corresponding internal and external humidity data.
Good luck with your efforts.
I think Sparkfun has a pcb that amplifies and digitizes a thermocouple with an i2c output.My experience with termocouples are they are not very sensitive at ambient temperatures and you need a better A2d converter to measure them well. I know thermocouples can also be affected by temperature differentials at the connectors.