During approximately one year I've been doing tests with Tricorduino 3.0 and I would like to talk a little about these tests in a qualitative way. Initially, I would like to comment about some aspects of the sensors used. The Clima Tempo function proved to be extremely useful for predicting frost in the orchard where I live, with some accuracy. The thyndall effect analysis of the atmosphere was very promising. It was possible to observe the variation of dust in suspension for different times of the year. With an initial calibration of the sensor inside a dry Nitrogen chamber it will be possible to better quantify the effect. The gamma radiation sensor is very promising, but I haven't been able to test it with radioactive samples yet. However it is possible to observe variations of radiation reading at different altitudes, sea level and above 3000 meters due to cosmic rays. The TSL2591 light sensor easily saturates with the sun which requires a neutral filter of approximately 30% to operate in daylight. Ultra violet light sensors work very well and are very reliable. The air quality sensor is also very efficient at detecting changes in the concentration of CO2 and volatile gases. However, the response time is a bit slow if there is not good air circulation. On the other hand, the gas sensor is not very reliable as the sensitivity is far from being specific for each type of gas. Besides, I didn't particularly have much use for this type of measure. The inertial sensors were completely useless as I never really used them. Perhaps it could be useful inside an aircraft or boat. Electric field and magnetic field sensors look promising in metal mining areas. There is a lot to be explored by these sensors, but a previous calibration process is necessary for more quantitative investigations. The spectrograph is extremely promising, but needs a constantly available calibration system. Due to the lack of a calibration system, its use is quite limited and the repeatability of measurements is greatly affected. As I said before, it was not possible to use this system due to the lack of pins available on the processor. To make calibration available, it would be necessary to dispense with some other sensors. The lightning sensor can be fine-tuned to avoid interference, but the processor is very susceptible to shutdowns when lightning strikes less than 1 km away. The combination between the MLX90614 and AMG88xx sensors works relatively well, however it would be desirable in a next version to use the interpolation library for the thermal camera.

Finally, it was quite evident to me, after this period of testing, the need to further miniaturize the device. A small box that could be carried in a pants or jacket pocket would be very desirable. In addition, a printed circuit board is essential to avoid frequent bad contact with wire up connections.