Turtle Sense/Phase Two
- User:Dave -- Electrical engineering / Mechanical engineering
- User:Sam -- Project manager / Embedded Software
- User:Tom -- Engineering support
- User:Eric -- Fundraising / Community liaison
- User:Erin -- Turtle researcher
- User:Chuck -- Sea Turtle community outreach / Materials
- User:Chris -- Web Server Admin
- User:Mark -- Web Application Development
- User:Britta -- National Park Service Biologist
- User:trebor -- Web visualizations
When nests are discovered by park personnel, they are uncovered, the eggs counted, and a DNA sample is taken. At that time, a smart sensor the size of a turtle egg is placed on top of the turtle nest before the excavated sand is filled back in to cover the site. The smart sensor measures motion and temperature in the nest, and the data collected is used to predict when hatching is imminent. The smart sensor is connected to a communications tower which sends out regular reports on the activity in the nest using M2M technology. The text file reports are FTP'd to a website directly from the communications tower. During Phase Two we created test units, one registration units, 8 communications towers and 24 smart sensors for installation in the Outer Banks of North Carolina startin in the Spring of 2014. Many of the sensors can be used in more than one nest per season, so Data can be collected from roughly 40 nests per season. The data is analyzed as it is collected to attempt to predict imminent hatching.
Before developing the system, other commercially available data recording systems were evaluated to see if anything that would meet our requirements was available. None of the options we found were not low enough power to fit our needs, and none provided the data we were hoping to collect. The closest matches we found would have required as much or almost as much custom engineering on our part to make it work, and there was no cost benefit to going that route.
The Phase One design used plastic ping-pong balls (not good ping-pong balls, but cheap plastic knock offs) and a temperature/motion sensor pre-installed on a small circuit board (Qty 1 Price=$14.95). Assembly involved cutting a small slot in a ball, putting in some silicone caulk, inserting a sensor board attached to a cable into the caulk, and then filling the rest of the ball with caulk. A ball cut in half covered the slot and makes a thorough seal with the silicone. This design was abandoned for several reasons. The silicone did not fully set, even after a few months in the field. The sensor was unreliable, which may have been because it was connected directly to the microprocessor via the long Cat 5 cable and communications were using a very slow bit-banged SPI connection. To improve communications we decided to put transceivers on both sides of the cable. By adding a microprocessor to the sensor, most of the processing could happen in the sensor assembly. The tranceivers are on for just a few seconds a day, for the upload of information from the sensor to the comm tower. This results in a Smart Sensor with very small power requirements.
The Smart Sensor has a 3-axis motion sensor (accelerometer) and temperature sensor connected to a very low power TI MSP430FR series microprocessor. There is also an RS485 transceiver chip which connects to the communications tower via a shielded Cat5 cable. The smart sensor can be programmed to take readings 12.5, 25, 50, 100, 200 or 400 times every second. Each reading is compared with the previous reading to determine the change in acceleration -- or "jolt". Rather than record each jolt, the sensor counts how many readings have reached approximately 25 different magnitude levels. These magnitude levels (which we call "bins counters") are in a logarithmic scale. Each bin is the square root of 2 larger than the previous one. We keep count for a variable amount of time (between 1 and 6 minutes) and then start a new set record with a new set of bins. This allows us to track how much motion there was at multiple magnitudes over time.
Communications tower board
Since it is not possible to make telemetry connections under wet salty sand, the smart sensors are connected to a communications tower. The design is based around creating a custom processing mother board connected to an off-the-shelf telemetry board. The communications board has a very low power TI MSP430FR series microprocessor which controls and monitors the activity of the smart sensor and controls the telemetry board. There is also a power supply, a RS485 telemetry chip and connectors for future expansion. The electronics are mounted inside a 3 inch PVC pipe and fittings anchored in a bucket filled with concrete. The concrete base and Cat 5 cable are buried in the sand. The The tower sends text reports to our web space via an FTP connection using the telemetry board. The frequency of the reports starts out a one a day when the nests are new and not very active. Later the frequency increases to 6 times a day as the nest becomes more active. Since the communications tower only makes one to 6 reports a day, The microprocessor turns the telemetry board on and off, so that it consumes no power when it is not being used.
We are using plug in Terminus boards made by Janus. One board works with AT&T, T-mobile and many international GSM networks. Another works with Verizon. The boards are interchangeable so both will plug into our motherboard. The cost is roughly $140 qty 1 and are available from the manufacturer, Janus. Janus has generously offered us a discount on their boards.
Hand-held registration unit
A communications board is mounted in a small hand-held unit so that it can be taken in the field when nests are first discovered. This allows park service personnel to carry just the light-weight sensors and the hand-held unit to the nest sites. Later, the much heavier and bulkier communications towers can be brought in. The Hand-Held units have GPS capability. They are programmed to test the sensors to make sure they are functioning properly, get a GPS reading, check for good cell reception, and get the local time and date from the cellular network. The date, time and GPS location are sent to the sensor, so that all this information will be available to the communications tower when it eventually connected.
The only other components necessary are the cable and connector for the turtle egg sensor, one antenna for the communications tower (the hand held unit has an extra antenna for GPS.) The units should operate for roughly 4 months running on 4 or 8 low self-discharge NiMH Batteries. Also, anyone creating these devices will need the hardware necessary for programming the microprocessors.
We arranged for cellular service through m2mAIR which is part of TELIT, the manufacture of the cellular modules used in the Janus plug-in boards. This service is cost effective for hosting a large number of devices, as they can all be in the same account with one monthly fee. The devices are also charged for the data used by each device, but the quota is pooled (shared) between all the devices, and the data rate charges can be modified month by month. During periods of little or no activity, the data plans can be throttled back to keep the fees to a minimum.
Web design and Database
All reports are uploaded to our web-space. The reports are comma separated text files with headings and labels, so they can be read by people as well as computers. We are beginning work on a database to store all the data. When reports are uploaded, the communications units also check the webspace for a new parameter file and download it if it is found. The parameter file has settings that effect the behavior of the devices. Here are some of the things that can be changed:
- The frequency of reporting during inactive days
- The frequency of reporting during active days
- The number of inactive days
- The number of records per report
- How frequently samples are read by the sensor (12.5, 25, 50,100, 200 or 400 times per second)
- All of the other settings possible for the sensor
- Whether reports are verbose (labels on all the fields) or just data
- Whether temperatures are calibrated
Work has started on designing a database to collect all the data in the reports, and a website to give people access to the data. This work is just in its preliminary stages in anticipation of Phase 3.
There are essentially three devices that need embedded code. The smart sensor, the communications tower units, and the hand-held registration units. All three of these devices use the same microprocessor, and two of them use the same circuit board. A good deal of the embedded code is shared in the three applications..
All the code was written in C using TI's Code Composer Studio which is free for developers of code that is smaller than 16K.
The code is fully documented and can be seen on GITHUB:
Schematics and System design
- Block Diagram of System
- Smart Sensor PCB 0.24 schematic
- Smart Sensor PCB 0.24 description
- Communications PCB 0.25a schematic
- Communications PCB 0.25a description
Circuit Board Fabrication
- Cable assembly instructions
- Casting instructions for Smart Sensor and connectors
- Communicator PVC housing assembly instructions
- Communicator housing final assembly