Skip to main page content
Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
, 18 (2)

Cave Pearl Data Logger: A Flexible Arduino-Based Logging Platform for Long-Term Monitoring in Harsh Environments

Affiliations

Cave Pearl Data Logger: A Flexible Arduino-Based Logging Platform for Long-Term Monitoring in Harsh Environments

Patricia A Beddows et al. Sensors (Basel).

Abstract

A low-cost data logging platform is presented that provides long-term operation in remote or submerged environments. Three premade "breakout boards" from the open-source Arduino ecosystem are assembled into the core of the data logger. Power optimization techniques are presented which extend the operational life of this module-based design to >1 year on three alkaline AA batteries. Robust underwater housings are constructed for these loggers using PVC fittings. Both the logging platform and the enclosures, are easy to build and modify without specialized tools or a significant background in electronics. This combination turns the Cave Pearl data logger into a generalized prototyping system and this design flexibility is demonstrated with two field studies recording drip rates in a cave and water flow in a flooded cave system. This paper describes a complete DIY solution, suitable for a wide range of challenging deployment conditions.

Keywords: Arduino; Yucatan Peninsula; cave; data logger; environmental monitoring; open source; submersible; subterranean karst estuary; underwater; vadose hydrology.

Conflict of interest statement

The authors declare no conflict of interest. The founding sponsors had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript and in the decision to publish the results.

Figures

Figure A1
Figure A1
Exploded view of the 2 inch pipe submersible housing components and their order of assembly. The casting seams on the two indicated surfaces need to be sanded smooth for a proper O-ring seal. Proper cleaning and priming procedures should be followed to assure strong joins and heavy duty PVC cement should be used for maximum bond strength. Options chosen for potting the sensors should be deep enough for 8–10 mm of epoxy over all components not designed for high pressure applications. The 1/4–20 nylon bolts shown vary slightly in diameter from one supplier to the next and sometimes the holes in the struts need to be slightly enlarged by drilled out to let the upper cap slide freely. Some PVC couplings may need slight trimming to an optimal length of 1.5 inches.
Figure 1
Figure 1
Arduino-based data logging platform connection plan, shown with the pin layout of Rocket Scream Mini Ultra, a DS3231 Real Time Clock (RTC) module with an onboard 4 kB EEPROM and a Raspberry Pi microSD card adaptor board. Unused connections 8 and 9 on the SD adapter are pulled up to prevent power loss due to floating inputs. At least one indicator LED is recommended, shown here with a 30 kΩ limiting resistor. This basic connection plan can be adapted to any 3.3 V Arduino compatible board by moving the jumpers to accommodate the physical pin locations on modules from different vendors.
Figure 2
Figure 2
Data loggers following the connection plan outlined in Figure 1, assembled on 4 inch knockout test caps. (a) Rocket Ultra board using the 328P ATmega chip, by Rocket Scream. (b) Moteino MEGA board using the 1284P ATmega chip, by LowPowerLab. Both data loggers have the I2C bus broken out with 4-pin Dean’s micro plugs for external sensor connections and include additional connectors for a 3-color LED and for 1-wire bus sensors. (Note: The 1-wire bus is not shown in Figure 1).
Figure 3
Figure 3
A typical breadboard testing setup to assess sleep current and verify code operation prior to soldering. The configuration shown includes a no-name clone of the Sparkfun Pro-Mini, a Macetech Chronodot DS3231 RTC, a stand-alone AT24C512 EEPROM module, a KEYES 5050 LED module and an Adafruit micro SD card adapter. Typical sleep currents at 3.3 V for the main components are: DS3231 module (0.08–0.09 mA), Sleeping micro SD card (0.05–0.09 mA), Pro Mini style board (0.02–0.06 mA).
Figure 4
Figure 4
Alternative physical arrangements of the same electronic components: (a) Surface logger configuration: components are mounted on a 4 inch ABS knockout cap with double sided tape. The RTC board is supported by 12 mm M2 nylon standoffs. (b) Submersible logger configuration: components are re-arranged to fit inside a 2 inch diameter pipe housing. Double-sided tape attaches the modules to an L bracket made from thin ABS plastic that is solvent welded to a 2 inch knockout cap. Wires pass through holes in the bracket to connect components on opposite sides. The battery pack shown has two banks of 3 × AA batteries in series, which are isolated from each other with Shottky diodes.
Figure 5
Figure 5
Environmental housings for Cave Pearl loggers. Surface logger: (a) Drip-sensor unit with a “Charlotte” brand translucent knock-out plug, which is solvent welded to the top of a 4 inch PVC end cap. A flexible rubber end-cap with a standard pipe clamp completes the housing with a water-tight seal. (b) Inside-view of the drip-sensor lid, showing a circular cutout of the original PVC end cap with the accelerometer mounted on the inside of the knock-out plate. A tri-color LED is mounted on the PVC so that it is visible through the translucent surface. Submersible logger: (c) Complete submersible housing made from 2 inch pipe fittings. Nylon bolts hold the body together, compressing the central O-ring seal. (d) Outside view of the submersible housing end cap. An RGB indicator LED is potted with transparent epoxy inside the grey ½ inch threaded connector. (e) Inside view of the sensor cap showing the plumber’s putty used to plug the housing penetration so that liquid epoxy could be poured around the LED. WS Deans’ 1241 micro connectors link wires from the sensor caps to the logger platform.
Figure 6
Figure 6
Flow chart of the Cave Pearl data logger software operation.
Figure 7
Figure 7
Data logger battery discharge curves for units with similar hardware powered by 3 × 1.5 V alkaline batteries in series. The 0.33 mA data logger built in early 2014 slept between readings but had no other power management. The 0.25 mA data logger used the described code-side techniques but had no hardware power optimization. The 0.1 mA data logger had software and hardware power optimizations, including an MCP1700 regulator, pin-powering of the RTC, a low current MS5803 pressure sensor and an SD card selected for its low-power sleep state. The dashed arrows indicate the estimated run-time to reach the 3.65 V shut-down limit.
Figure 8
Figure 8
The power-input leg of the DS3231 IC is disconnected from the board and soldered to a wire (purple) for connection to a digital pin on the Arduino. When powered by the VCC line on the module, the RTC is responsible for almost 50% of sleep current of the entire data logger. This technique requires advanced soldering skills but pin-powering the clock IC can reduce a data logger’s sleep current to ~0.1 mA, allowing for operation >2 years on 3 × 1.5 V AA cells in series.
Figure 9
Figure 9
Current draw from a 4.6 V supply during the three events in the drip counter’s duty-cycle: (a) Drip counting (b) Reading sensors and buffering data to EEPROM (c) Transferring that buffered data to a no-name 1 GB microSD card. The upper graph of each pair shows current before run-time power optimization, while the lower graph in each set shows the same event after optimization techniques are applied. These records were captured with an Arduino UNO measuring the voltage drop across a 12 Ω shunt resistor at ~89 kHz (ADC prescalar = 8) [37].
Figure 10
Figure 10
Current draw from a 4.6 V supply by different SD cards accessed in SPI mode: (a) Current drawn during a power optimized save to a no-name 1 GB microSD card that is compatible with 512 byte data saves. (b) Current draw on that same logger running the same code using a SanDisk 2 GB microSD card that is not compatible with small blocks sent over the SPI interface. Though the data was recorded without error, the event required 10.33 mAs which is ~4 times the power consumption of the 1 GB card. These records were captured with an Arduino UNO measuring the voltage drop across a 12 Ω shunt resistor at ~89 kHz (ADC prescalar = 8) [37].
Figure 11
Figure 11
A drip-sensor tethered by cable ties to the top of a stalagmite, at a slight incline to prevent water accumulating on the surface.
Figure 12
Figure 12
Drip data from Cave Pearl loggers (purple line) with manual drip counts (yellow circles) in units of drips/15 min from long-term monitoring stations in the Rio Secreto section of the Pool Tunich Cave System. Panels are arranged from high to low drip rates, with the top panel showing the Pulpo Showerhead that episodically exceeds the instrumental limit of ~15,000 drips/15 min and the bottom panel showing RST013 with less than 5 drips/15 min.
Figure 13
Figure 13
Correlation between manually counted drip rate (shown as drips/15 min on X-axis) against number instrument counted drips (x/15 min on Y-axis). Linear regression on normal data has an R2 of 0.99, excluding counts where either manual or instrument reading is 1 drip or less/15 min (squares). Both axes shown as log_10.
Figure 14
Figure 14
Deployments of tilt flow meter: (a) An early flow meter installation in the Casa Cenote cave, at ~300 m from the blue hole discharge. Multi-logger deployments with ~1 m spacing allow assessment of inter-unit variability. Flow sensors are ballasted to slightly negative buoyancy, allowing soft bungee cord anchors that cause no damage to the cave. (b) High surface-area flags in low-flow conditions mechanically enhance the instrument response.
Figure 15
Figure 15
Casa Cenote flow meter data comparison between Cave Pearl accelerometer-tilt flow meter (upper row) and Aanderaa RCM 7 rotor vane system (lower row). The data over 2 tidal months (56 days) shown in left panels (a,b) for representative periods without events and over 3 days in right hand panels (c,d).

Similar articles

See all similar articles

Cited by 2 articles

References

    1. Cressey D. The DIY electronics transforming research. Nature. 2017;544:125–126. doi: 10.1038/544125a. - DOI - PubMed
    1. Kushner D. The Making of Arduino: How five friends engineered a small circuit board that’s taking the DIY world by storm. [(accessed on 28 January 2018)];IEEE Spectrum—Geek Life. Posted 26 October 2011. Available online: http://spectrum.ieee.org/geek-life/hands-on/the-making-of-arduino.
    1. Naone E. Arduino Uno—How a Cheap Microcontroller Is making It Possible for Anyone to Design and Build Hardware. [(accessed on 28 January 2017)];Technology Reviews. Posted 22 February 2011. Available online: www.technologyreview.com/s/422845/Arduino-uno/
    1. Sorrel C. Just What is an Arduino, and Why Do You Want One? [(accessed on 28 January 2017)];Wired. Posted 15 April 2008. Available online: https://www.wired.com/2008/04/just-what-is-an/
    1. Public Lab Mainpage. [(accessed on 28 January 2018)]; Available online: https://publiclab.org/
Feedback