Sap Flow Meter
Current Project Leads: Ryan Bohl, Jackson Miller, Ryan Christensen, Stephen Good
*Past Project Leads: Marshal Horn, Joshua Barksdale, Vladimir Vesely
This project has been officially wound up. Conclusions drawn here is yes, we can make probes significantly cheaper than commercial price points. However, these require a great deal of craftsmanship, time, and technique to produce by hand, with additional rigorous testing of each probe. After calculating time to train, assemble, and test, we posit this is not a significantly stronger approach compared to commercial devices.
Measuring the amount of water consumed by a plant is crucial for targeted irrigation and fruit yield calculations. A very precise method of determining water intake is sap flow measurement.
At the present time, there are few options for measuring tree sap flow that are both cost effective and easy to construct / deploy. Davis et al [1] showed that custom made probes can approach the accuracy of commercial probes for a fraction of the cost. However, the standard method for creating DIY probes involves custom wrapping thermocouples. Previous iterations of the Sap Flow project found that even with an experienced assembler, 30% of thermocouples would be dysfunctional immediately upon creation. Once a working probe was inserted, it would be nearly impossible to remove without damage. Sap Flow intends to change this, and provide a durable and affordable device with comparable accuracy to commercial systems.
The Talking Tree is a sap flow meter that utilizes the Heat Ratio Method (HRM) for measuring the rate that sap is flowing in a tree. It is designed to be easy to build by anyone with basic soldering skills. Every component used for measuring the sap flow has been assembled and tested successfully. It is our hope that in the future more features such as transmitting data and attaching multiple probes to a single logger become implemented since the hardware has the capability of introducing these features. There is still work to be done in developing the software for logging, transmitting, and displaying data.
System Diagram
- Relatively Low cost (less than $300 per module)
- Long battery life (continuous four week operation on single battery charge)
- High accuracy (Greater than 0.8 R to commercial sap flow meter)
| Property | Customer Requirement | Engineering requirement | Met? |
|---|---|---|---|
| Accuracy | The system must have accurate readings | Measurements must be with 0.8 R value of a commercial probe | Yes: 0.88 R value |
| Affordable | The system must be affordable | The system shall cost less than $200 | No: ~$310 for all tools and materials |
| Documentation | The system must be well-documented | The documentation conforms to customer-provided style guide | Yes: Hardwarex Article |
| Easily Serviceable | The system must be easily serviceable | The SD card can be removed within 60 seconds | Yes: about 10-15 seconds |
| Low Power | system must have low power consumption | Average power consumption must be under 500mW | No: over 1W |
| Protrusion | The probes should not stick out too far | Probes should not protrude more than 3cm from the tree. | No: Housing is 1 1/2" in diameter |
| On a stake | The enclosure must be high off the ground | mountable at least 50cm from the ground | Yes |
| Standard Batteries | The system must accept 12V batteries | The system must accept power input of at least 12V | Yes: Accepts 12v |
| Transportable | The system must be compact | Can fit in a box of 50cm x 50cm x 30cm size | Yes |
| Twitter updates | The system shall post live updates to Twitter | post updates at intervals of 2 hours or less | No twitter support |
| Waterproof enclosure | The system must be weather resistant | can be submerged under 5cm of water for 1 minute | Yes |
After several field tests, the team collected data that met our objective. The software used in calculating the sap flow needs more development time, as the data values are very low. When the data is processed off of the micro sd card and multiplied by a constant, it within 0.8 R value to a commercial probe (Dynamax) that is installed in the same tree. In its current state, the only way to read the sap flow data is by removing and reading the micro sd card after turning the system off. We hope to improve this with LoRa communications in the future.
Probes
Show More
There are several circuit boards that are used in this setup, three of which have been custom designed.The first of these custom PCBs is used for inserting the probe into the tree. Last year a set of 1.5 mm PCB probes was created to allow users to reuse probes. While they allowed for multiple installations, they would often stick out from smaller tree trunks, increasing the risk of damage from passerbys.
We experimented with flex PCBs. Using flexible probes allows a single set of probes to be easily reused in trees of multiple sizes without brittle protrusions. Small steel rods are attached to allow for easy insertion and removal. After much experimentation, we finalized our design for the probe that is inserted into the tree. The Below is a document on how to assemble theses:
Final Flex probe design with shielding
The second of the custom PCBs is the Smart Probe, which is connected to the flex probe inside the probe housing. It contains a filter and an ADC for reading the temperature values of the flex probes. The silkscreen at the top of the PCB outlines 8 through hole connections--these connect to the flex probe within the Probe housing:
The smart probe connects to a Sparkfun I2C breakout which allows the smart probe to send its readings over a long distance CAT5 (Ethernet) cable where they will be received by the datalogger. There are 2 breakouts: one is encased inside the probe housing and the other is located inside the enclosure with the datalogger:
The final custom PCB is the Sapflow wing board which has circuitry for voltage regulation, heater control, a real time clock, as well as a micro SD card slot for logging the data. the 12v battery is connects to this PCB, which is how the user will power the system on and off.
Lastly, The sapflow wing connects to the top of the Adafruit Feather M0 microcontroller. It helps with controlling the system, and it has capabilities of broadcasting data over a LoRa radio frequency (covered below)
In the following year, the team designed the housing for the flex probe and Smart probe. It is assembled with various PVC pipes. Three holes are drilled for the flex probe. The smart probe and an I2C Breakout are located inside the housing, with the CAT5 cable being fed out the bottom. There is a adjustable cord grip that the user can twist on the bottom to hold the cable in place and prevent strain.
The sparkfun I2C driver, Smart probe, and flex probe connect to each other in the top half of the probe enclosure as seen below. The CAT5 cable going into the probe connects to the I2C driver
Power
Show More
To ensure power efficiency, we used a switching topology. Switching power supplies have efficiencies that can be up to 10% more efficient than linear regulators. This is of chief importance in maintaining a long operating time without battery change. We chose the MAX5033 buck converter as our voltage regulator because of its high voltage input range, efficiency, and simple application circuit. With these design considerations, the 12V batteries have been able to power the Sapflow meter for over 3 weeks without recharging in our field testing. Battery life varies with weather conditions, with colder temperatures reducing operation time.Measurement
Show More
After phasing out the thermocouple last Spring, Sapflow transitioned to using platinum RTDs (Resistance Temperature Detectors). We have been experimenting Adafruit RTD amplifier and Texas Instruments INA125 instrumentation amplifier. The first set of comparisons between a Fluke Industrial Thermometer and the two amplifiers revealed that a simple four-wire method of resistance compensation would not lead to accurate enough readings from the INA125. PT100 readings are from the Adafruit amplifier. The INA125 readings were measured using a Feather M0 ADC, and so had to be mathematically converted to temperature readings. Error bars on this graph are 1.0%
Our second round of testing compared a wheatstone bridge measurement circuit for the INA125 with a Fluke industrial thermometer. We found that the data showed high amounts of linearity, allowing a simple slope and intercept adjustment to produce the following graphs:
Even after adjusting the offset, the wheatstone bridge INA125 configuration resulted in better results than the original PT100 data. Error bars on this graph are 0.5%
.png)
Our next round of tests will involve a wider temperature range to more accurately reflect the operating conditions of the system.
Data Transmission
Show More
Once sap flow has been measured, it is transmitted back to the Feather M0 via an I2C line. Using I2C drivers allows for the connection of multiple long lines to the microcontroller, decreasing the number of microcontrollers that must be purchased.Once data is at the Feather, it is stored in a CSV file on its SD card. We had hoped that at the same time, the Feather could convert the data into the JSON data format, and transmit it over LoRa, but this has not been very effective to this point and it is a point of emphasis for this project going forward.
LoRa is a wireless transmission protocol similar to wifi. However, it is optimized for lower power, longer range communications. With direct line of sight between the transmitter and receiver, LoRa systems have been documented to be able to transmit over a 3 Km distance. Our experiments have shown possible transmission through orchards up to 650 meters, but this has been inconsistent in testing. With directional antennas this range can be extended.
Currently, the only way to access the Sapflow data is to turn off the system, remove the SD card from the sapflow wing and read it on a computer. There will be 3 files--A syslog.txt file that describes the system behavior with timestamps. This is primarily useful for debugging purposes. There is also a tree#_log.csv file in the image below which shows the rtd temperature readings (Celsius) at each second. Column B Shows the instantaneous temperature at the upper rtd, Column C shows the bottom RTD temperature, and Column D shows the temperature of the Middle RTD next to the heater resistor.
The temperature log values are used in the formula below to calculate the Sapflow values every 15 minutes. Upper and lower RTD temperature is averaged every 40 seconds during measurement period. These average temperatures are logged in columns B and C of the Sapflow log seen above. The current temperature averages in the logging process are compared with the previous average temperature values, and the difference in these values become the V1 and V2 variable in the formula below. The formula ultimately calculates the Sapflow in cm/hr. The sapflow values are logged to column D in the sapflow log (seen above)
7/2021: There's a lot of potential within this project and long term goals include the use of several SmartProbes within one SapFlow Sensing System to help reduce cost when scaling to larger measuring systems. But before those are implemented there are a few fundamental systems need to be put in place first. Namely battery monitoring and brownout protection circuit. Changing I2C differential Drivers to RS485 to increase the maximum signal distance, while still having a large number of potential nodes available. Using better connectors between the FlexProbe to SmartProbe and SapflowWing to FeatherM0.
2/2021: In the project's current state, we have achieved good data that has a 0.88 r value when compared to the commercial Dynamax probe. but there is still work to be done in amplifying and increasing the accuracy of our sap flow measurements. Increasing the manufacturability of our overall setup is a major focus for our team, as with so many parts to the design, hardware failures were common and difficult to diagnose. We also hope to improve the LoRa transmission functionality as well as introducing software that can display the received data. Additionally we hope to improve the system's capabilities for attaching multiple probes to a single logger to help reduce the cost/probe setup. With our data from the previous deployment, the door is open to future RD with new team members.
- SdFat
- OPEnS RTC
- Low-Power
- Protothreads
- DSPlite
- RadioHead
- ArduinoJSON
- FeatherFault
- ASF core commit f6ffa8b
- Plog
https://www.youtube.com/watch?v=Hv_XhXh8ykQ
[1] T. Davis et al. “Sap Flow Sensors: Construction, Quality Control and Comparison,“ Sensors (Basel). vol. 12, no. 1, p 954-971, Jan. 2012. [Accessed November 24, 2019]
[2] Burgess, Stephen S. O., Mark A. Adams, Neil C. Turner, Craig R. Beverly, Chin K. Ong, Ahmed A. H. Khan, and Tim M. Bleby. "An Improved Heat Pulse Method to Measure Low and Reverse Rates of Sap Flow in Woody Plants †." Tree Physiology 21.9 (2001): 589-98. Web.