Curious sensor failures and cyclical measurements

Curious sensor failures and cyclical measurements

We have a station with HYDROS-21 and Turner Turbidity+ sensors. Beginning on 12/30/22 the HYDROS appears to have quit communicating, reporting only values of -9999 for all its variables. Near the same time, the Turner began reporting values with a daily cyclical pattern: higher in midday, and lower at night. Field inspection of both sensors and the Mayfly revealed no apparent cabling or other physical problems or damage. The Mayfly was also power-cycled at that time, with no change in results.

The Turner reports its turbidity measurement as a voltage, 0-5V, from which we calculate NTU using the manufacturer’s formula. Both voltage and NTU are posted to MMW. The Turner’s signal+ and signal- leads are wired to a voltage divider on an EnviroDIY Grove screw terminal adapter so that it doesn’t exceed Mayfly’s 3.3V. (This thread has a description of the circuit.) During the daily cycles, the Turner’s voltage ranges roughly from 0.012 – 0.030V, which translate to an NTU range of -1 to 10.

Given the daily oscillations in the Turner’s voltage readings, I began wondering whether the solar input is affecting reference voltage somehow: higher battery voltage correlating to higher Turner voltage. Digging further, I realized that this particular station was mis-wired when first deployed in April 2022: The 6W 6V Voltaic solar panel was connected to one of the BAT terminals on the Mayfly, rather than the SOLAR terminal, causing the reported battery voltage reached as high as 7.8V. (Discovered and resolved in this thread.) This was corrected at the next field visit (2 weeks later), and the Mayfly began reporting reasonable values for all variables, and appeared to work properly until the change on 12/30.

Is it possible that the Mayfly was damaged somehow (voltage regulation?), but only manifested some months later? Could that also affect the SDI12 circuitry, and thus prohibit communicating with the HYDROS-21? Any suggestions for additional troubleshooting or other ideas would be appreciated!

[EDIT] I’ve added some plots to show 1) the shift in both sensors’ values on December 30th, 2) Turner voltage cycling in January, and 3) Turner voltage cycling in May. Comparing January to May reveals a change in length and timing of reported Turner voltage, presumably corresponding with the seasonal shift in insolation length and timing.

Thanks,
Matt
Trout Unlimited