Design of a Real Time Reconfigurable Bioreactor

Amol Vengurlekar

Ruoshi Zhang

Taylor Johnson

Design of Real Time

Reconfigurable Bioreactor

The University of Texas at

Arlington

TI Innovation Challenge 2015 Project Report

Team Leader:

Amol Vengurlekar <amol.vengurlekar@mavs.uta.edu>

Team Members: Team Ruoshi Zhang <ruoshi.zhang@mavs.uta.edu>

Team
Teail
Advising Professor:

Dr. Taylor Johnson <taylor.johnson@uta.edu>

Video

We had difficulty uploading a video to www.ti.com/videos. So we have

shared a youtube link.
https://www.youtube.com/watch?v=ZT3BtsZP1sU&feature=youtu.be

Texas Instruments
Mentor (if applicable):
Date:
Qty.

Ex: 1

5/24/2015
List all TI analog IC
and TI processor part
number and URL

1) Explain where it was used in the project?

2) What specific features or performance made
this component well-suited to the design?

We use the BeagleBone Black hardware platform that uses the TI AM335x
AM335x SoC
(BeagleBone Black board) system on chip. We used the processor and peripherals like ADC, SPI bus,
GPIO, PRU (Programmable Real-time Unit).

Ex : 2 OPA4350 (op-amp)

The OPA4350 was selected for its single supply operation and quad channel. It
is used to amplify the sensor voltage. Together with SPI bus controlled digital
potentiometers, the OPA4350 constitute the configurable sensor gain circuit.

Ex : 3 LM358 (dual operational

amplifier)

The ADC of the BeagleBone (AM335x) senses the sensor voltage. But the ADC cannot
tolerate input voltages 1.8V. The sensor amplifier can produce voltages greater than
1.8V. Therefore we need an over voltage protection circuit between the amplifier and
the ADC. The dual op-amps in the LM358 makeup the over voltage protection circuit
that protects the BeagleBone from permanent damage.

Ex : 4 LM2940 (Low Dropout

Regulator)

We used the TI LM2940 +5V voltage regulator to generate a +5V power rail. It
was selected for its low ripple voltage.

Abstract:
In this project we present the design of a reconfigurable continuous culture bioreactor that allows
researchers to grow microorganisms in a controlled environment for long periods of time, as long as several
months. The bioreactor monitors the experiment and measures the growth of microorganisms. A continuous
culture bioreactor is an important tool that biologists use for research, most notably to study the response
of bacteria towards drugs. This helps in the fight against the evolution of drug resistance in bacteria and
other pathogens.
The bioreactor acts like an isolated ecosystem. It periodically exposes the microorganisms to drugs
that hamper their growth. One must understand that in each successive generation of the microorganisms
in question, some random mutations occur in their gene pool. Some mutations enable the microorganisms
to survive the drugs. Over the course of several thousand generations only the microorganisms that evolved
drug resistance survive in the bioreactor. This process of natural selection is the same mechanism by which
pathogens evolve drug resistance in the real world. Therefore studying this process helps us in developing
drugs to counteract the threat of drug resistant pathogens.
In this project we are redesigning a bioreactor for our colleagues in UT Southwestern Medical Center
at Dallas, TX. They have built a bioreactor, however it has technical problems. The problems stem from
crucial errors in design choice regarding the software and the underlying hardware. We redesigned the
system and solved their problems.

Table of Contents
Introduction:................................................................................................................................................ 4
Bioreactor as designed by Toprak et al. and implemented by our colleagues ................................... 4
Design of the bioreactor using the BeagleBone Black.......................................................................... 5
Software Design: ......................................................................................................................................... 6
Hardware design: ........................................................................................................................................ 7
Sensor Board: Optical Density Measurement ...................................................................................... 7
Relay Board: Pumps Control ................................................................................................................. 8
Results .......................................................................................................................................................... 8
References:................................................................................................................................................. 11
Appendix A ................................................................................................................................................. 12
Appendix B ................................................................................................................................................ 21
Appendix C ................................................................................................................................................. 26

Table of Figures
Figure 1 System block diagram for the bioreactor ........................................................................................ 5
Figure 2 Bioreactor in action ........................................................................................................................ 8
Figure 3 Sensor board and relay board with the BeagleBone Black............................................................. 8
Figure 4 Uncontrolled cyanobacteria growth................................................................................................ 9
Figure 5 Controlled cyanobacteria growth.................................................................................................. 10
Figure 6 Controlled E.coli growth .............................................................................................................. 10

Introduction:
Since the 1960s human kind has tamed many of the infectious diseases that terrorized generations of
people. Widespread use of antibiotics, vaccines and pesticides resulted in a dramatic reduction in fatalities
due to diseases such as pneumonia, tuberculosis, meningitis, influenza, whooping cough and diphtheria [2].
So much so that due to global vaccination campaigns smallpox was declared to be eradicated in 1980 by
the WHO and polio has been eradicated from most of the world.
But these great victories against diseases have come at a cost. On December 12, 2014 the Washington
Post carried an article [4] that says superbugs can one day kill more people than cancer. The article refers
to infections caused by drug resistant bacteria that can easily kill thousands. This is a disturbing claim. In
places where a variety of factors like lack of regulations, widespread use of antibiotics and high population
densities create conditions that are perfect for drug resistant superbugs to evolve and thrive. For example,
bacterial infections that are resistant to most known antibiotics were responsible for more than 58000 infant
deaths in 2013 in India [3]. Besides developing new drugs, there is not much we can do to fight infections
that are resistant to all known drugs.
A bioreactor is an essential tool for biologists and researchers to develop new drugs. Using a
bioreactor, biologists can regularly expose microorganisms to drugs and examine their response to drugs.
Over very long periods of time, due to genetic mutations and changes, the microorganisms can evolve
resistance to the drugs. The study of this process is essential to develop new drugs. This is why a bioreactor
is a crucial tool for developing new drugs and continuing the fight against multi drug resistant bacteria.
We are working on improving/replacing an existing system (bioreactor) to grow and monitor bacteria.
The bioreactor is described by Toprak et al. in their paper [5]. This bioreactor was being used by our
colleagues at UT Southwestern medical Center. But it had some functional problems. So in order to correct
those problems we redesigned the system. A culture of bacteria (or any other microorganism) is grown in
a beaker. The bioreactor supplies the organism everything that they need for growing and thriving (nutrition,
air, light). The bioreactor periodically measures the microorganism concentration. When the concentration
exceeds some preset threshold parameters the system takes control action. This control action may change
from experiment to experiment. It may either dilute the culture by pumping in more media or reduce the
microorganism concentration by injecting some drugs in the culture to kill some microorganisms. The drugs
may kill the organisms or inhibit their growth.
The bioreactor operates on a set of well-defined system parameters (e.g. Maximum microorganism
concentration, motor/pump on/off timings etc.). The bioreactor has a central controller which takes all
control decisions. It measures microorganism concentration using sensors and actuates motors/pumps
which control the flow of culture media, drugs and air into the culture solution. Waste may also be removed
from the bioreactor. Light intensity may also be controlled for photosynthetic microorganisms using LEDs.
This setup stays operational and uninterrupted for long periods of time allowing the biologists to observe
some change in the microorganisms. The ultimate goal is to predict and observe genetic changes in the
microorganisms. By studying these genetic changes (which may be a response to the drugs injected by the
bioreactor), one can draw more conclusions about the evolution of drug resistance in microorganisms.
Bioreactor as designed by Toprak et al. and implemented by our colleagues
The bioreactor as described by Toprak et al. was implemented by some of our colleagues at University
of Texas Southwestern Medical Center. The bioreactor conducts optical measurements using an infrared
emitter and a photodiode as a detector. The detector voltage is sensed by a data acquisition unit (in this case
USB-1616FS by Measurement Computing Corporation). The control algorithm was written in Matlab on a
Windows computer. The algorithm measures the optical density periodically and dilutes the microorganism
culture or injects drugs periodically (or upon exceeding threshold values) to achieve growth inhibition.

When culture media or drugs are to be injected, the computer actuates pumps using USB-ERB24, a
24 channel electromechanical relay interface by Measurement Computing Corporation. The pumps are
peristaltic pumps frequently used in biological experiments. They can inject controlled amounts of fluid
drop by drop into the culture.
However they faced problems regarding reliability of their system. Their bioreactor would not work
properly or stall after a few days. A common problem was that the pumps would fail to turn off resulting
in beakers to overflow. A closer inspection of the underlying software reveals a critical flaw. The software
is written in Matlab. To implement all timings and synchronization, the biologists used the in-built timer
class of Matlab. However the documentation of timer class clearly mentions that it is not advisable to use
the timer class in any real time application [1]:
The timer object is subject to the limitations of your hardware, operating system, and software.
Avoid using timer objects for real-time applications.
Furthermore, even while the bioreactor is operational, we cannot be certain about the data that is
collected because the timer class gives no real time guarantees. Matlab is implemented in JAVA which is
subject to garbage collection events which are non-deterministic in nature. The underlying operating system
(Windows) provides no real time guarantees.

Design of the bioreactor using the BeagleBone Black

Figure 1 System block diagram for the bioreactor

Figure 1 is the system block diagram of the bioreactor as required by our colleagues at UT
Southwestern Medical Center. The bioreactor has the capability to add more media and drugs to the culture
solution. The bioreactor continuously adds air and provides lights to sustain bacteria growth. When new
media or drugs is added, it also removed an equal volume of solution from the culture to keep the volume
constant and prevent the culture from overflowing. Drugs and media are added only when their respective
conditions are satisfied. If bacteria growth rate increases more than a specified limit, then the system will
add a predetermined amount of drugs to kill some bacteria and inhibit their growth. If bacteria concentration
increases beyond a predetermined limit then the system will add a specific amount of media to dilute the
culture solution. This functionality was implemented by our colleagues but as described before, they faced
technical difficulties.
To overcome the shortcomings of the bioreactor designed by our colleagues at UT Southwestern
Medical Center, we redesigned the whole system again from the bottom up. Instead of a computer, we used
an embedded Linux platform. We built the system using a BeagleBone Black running Debian Linux. The
BeagleBone Black has a Texas Instruments AM335x 1GHx ARM Cortex-A8 CPU.
Software Design:
We get multitasking abilities because of the underlying Linux operating system. However, Linux is
a general purpose operating system. It cannot provide us the real time guarantees that we need. So we
patched the Linux operating system with the Xenomai real time framework. The Xenomai user space allows
us to write programs that deliver real time performance. The Xenomai patch appends the existing Linux
kernel with its own real time scheduler. It also makes use of priorities in the Linux OS making sure that
deadlines of the Xenomai tasks are always met. Also, we have designed the system such that one
BeagleBone Black can run three instances of the software. This enables us to control three different
bioreactors, each with potentially different system parameters. Our software consists of 5 Xenomai tasks,
each managing a particular component of the system. Very briefly, they can be described as follows:
Sense Task: Manages the bacteria concentration measurement timings and requests for measurement
whenever needed.
Sample Task: Continuously measures the microorganism concentration and provides the sense task
with a block of data whenever requested. All measured data is sent to the log task for logging.
Control Task: Takes a control decision based upon measurements obtained from the sense task. It
injects drug/media depending upon system state.
Log Task: All tasks send data to be logged to the log task. The log task logs whatever data it receives.
Config Task: It manages the system configuration. It reads the system config file and defines the
configuration. If the file is updated during system execution then it updates the system configuration at run
time.
The system configuration is defined by the system config file at startup. It contains variable
definitions like the optical density threshold values, amplifier gain values which are used to set the amplifier
gains on the hardware which amplify the signal from the sensors. To operate the system, we have to set
appropriate values to the configuration variables and start the program. The program will read the variable
values from the configuration file and behave accordingly as specified.
The entire system is connected to the internet. So the user can login from a remote location and
control it, modify the configuration settings and change the system behavior as per the requirements.
The onboard AM335x SoC by Texas Instruments controls the pumps that add drugs, culture media
and remove waste in addition to supplying air and lights for photosynthetic microbes using its GPIO pins
that are connected to pump driver circuits. This is managed by the control task of the software system.

The TI AM335x SoC also has ADC inputs. The system uses three of the ADC inputs to acquire sensor
data for the three bioreactors that it can control. This is managed by the sample task in the software.
Different microbes respond differently to the infrared sensor. So we need different variable gain
amplifiers on our setup so our sensors can detect a wide range of microbes, cyanobacteria (needing small
amp gain) to E.coli (needing large gain). Software can set gains of each individual amplifier in OPA4350
by changing feedback resistor value. We use a digital potentiometer (MCP4361) as the feedback resistor,
which receives commands through SPI bus from Beaglebone. The config task in the software system reads
the desired gain values from the config file and sets the amp gains in hardware.
Time stamping all acquired data is a crucial for data management. However under certain
circumstance acquire time through Internet (NTP) is not available, a battery powered real-time clock (RTC)
chip is necessary. We choose SPI bus based DS1306 to serve this purpose, it helps the system maintain
time information even when the system totally cut off from power.
One request of our biologist colleagues was the ability to generate PWM on all control signals to
change the flow rate of drugs or light intensity. This is achieved using the on chip PRU on the AM335x
SoC. The PRU executes PRU code written in assembly that emulated PWM hardware and generates PWM
signals on all 15 output control lines (5 for each bioreactor).
Another feature is that all data that is generated is backed up online on a cloud. We have written a
python script that used Dropbox API to connect to a Dropbox account and upload all data to the Dropbox
account.
Hardware design:
We measure the microbe concentration in the bioreactor by measuring the optical density (turbidity)
of the culture solution. The most intuitive way to measure a liquids optical density is to project a beam of
light through it, and measure the effects of the liquid on light strength on the other side. To be specific, in
our project we choose infrared (IR) LED as the light source, an IR sensor to receive the light that is scattered
by the microbes in the bioreactor; output signal of the IR sensor is amplified (including current amplification
and voltage amplification) and regulated (clamp down to 1.8V to protect latter stage circuits) to accurately
and safely feed into the analog to digital converter (ADC) of the AM335x SoC. According to this reading
a series of control actions can be made. These above all functionality form one printed circuit board (PCB)
which will be addressed as the sensor board in this paper. The media, drugs(antibiotics) and humidified air
are added by dedicated pumps, the wastes or excessive liquid are removed by built in pressure of the beaker.
Two pumps are used to control the flow of media and antibiotics and an AC aquarium pump provides air.
The pumps are controlled by another PCB called the relay board. We developed two versions of the relay
board, one for AC pumps and another for DC pumps. It should be noted that the air pump is always an AC
aquarium pump. For AC pumps, the media, drug and waste pumps are controlled by solid state relays that
also provide optical isolation between the high power AC pump circuit and the low power digital circuit of
the AM335x SoC. For the DC pumps, the BeagleBone controls the pumps using a ULN2003A Darlington
array chip.
Sensor Board: Optical Density Measurement
In order to acquire an accurate measurement, we choose TSAL5100 from VISHAY semiconductor as
the IR emitter. This light will be scattered by microbes in the culture media contained in the bioreactor. The
scattered light is detected by the infrared detector BPV10NF also from VISHAY semiconductor. As the
microbe concentration increases, they scatter more light. This light is picked up by the infrared detector and
it shows up as a higher output current through the BPV10NF detector.
However this current is very tiny (around 60 micro-Ampere). In order to amplify the current and give
a robust detection in later stages, we hook up it with a BJT KSD1616. We can use operation amplifiers

(OPA4350 by Texas Instruments) to manipulate the magnitude of the sensor reading. Figure A-4 in
Appendix A is the schematic of this part of circuit.
The BeagleBone Black ADC cannot tolerate input voltages greater than 1.8 Volts. However under
many circumstances the sensing circuit may produce voltages much greater than this limit. This can
potentially damage the BeagleBone Black permanently. Therefore to prevent damage we have incorporated
an ADC protection circuit that protects the BeagleBone Black from high voltages. If sensor voltages are in
the tolerance limit of the BeagleBone Black (0 1.8V), then the circuit ensures that the same voltage is
passed to the ADC pin of the BeagleBone Black. However when the sensor voltage increases above the
threshold of 1.8V the protection circuit pulls the voltage that is passed to the ADC down to ground in order
to protect the BeagleBone Black. LM358 dual-opamp from Texas Instruments is selected to work as the
comparator, the other op-amp inside the same package would acting as a buffer amplifier to segregate input
signal and the drain end of the FET (MMBF170). Figure A-8 in Appendix A shows this configuration.
Signal flow is from left to right, AMP_3 end is the input to the buffer amplifier and ADC_3 end feeds into
the ADC channel.

Relay Board: Pumps Control

The control of all the DC pumps are carried out through the Texas Instrument Darlington transistor
array IC ULN2003A. This Darlington transistor array serves as a relay between the BeagleBone and the
actual load it controls. For the air pump which is driven by AC power special care needed to be take. In this
case the solid state relays (SSR) is a good choice. By eliminate the mechanical moving parts like a physical
relay, the life-expectancy of the system is extended. G3MC series from OMRON is our finial selection due
to its slim shape and simple connection. As the connector which interconnects with real load and SSR and
Darlington transistor array have to have the capability to safely handle AC power and meanwhile provide
a simply operation and space saving, Phoenix 1770995 PCB terminal block become our choice on the board.
Results
Overall, this task that we undertook of designing a real-time reconfigurable bioreactor was successfully
accomplished. The bioreactor what we designed and built conforms to all requirements of the system. It
has performed very well in the tests that we conducted. We were able to grow sampled of cyanobacteria
and E.coli bacteria using our bioreactor and the results are as expected. The main problem of long term
reliability was also solved. Our long term tests showed that the system remains stable and operational for
months on end without any problems, which was the main objective behind this project. We hope that our
bioreactor would prove useful to more biologists and that they would use it to conduct their experiments.

Figure 2 The Bioreactor in action. Here we are growing

a culture of green photosynthetic cyanobacteria (Hence
the bright lights).

Figure 3 The sensor board (left), the AC relay board

(right) along with the BeagleBone Black (center).

Figure 2 is an image of our bioreactor system using the BeagleBone Black. It is growing and monitoring
cyanobacteria. It provides the cyanobacteria with lights and air for growth. The culture media has ample
nutrients to support cyanobacteria. When the bacteria concentration exceeds the limits defined in the system
configuration it dilutes the culture thus, controlling the bacteria concentration. Figure 3 is an image of both
our boards, the sensor board and the relay board along with the main BeagleBone Black.

Figure 4 Uncontrolled growth of cyanobacteria.

Figure 4 is a plot of a growth experiment in which we were growing a strain of cyanobacteria. Cyanobacteria
typically have a slow growth rate. This growth rate is subject to various conditions like availability of light,
air, nutrients and the type of culture media used. In this experiment we are not controlling the bacteria
growth, but merely observing it. This is to demonstrate uncontrolled cyanobacteria growth. The measure
optical density (OD) is roughly proportional to the cyanobacteria concentration. We fit the raw data for an
exponential growth model of cyanobacteria growth. The growth equation is () = 0 where, OD
is the exponential curve fit for the optical density that models the Cyanobacteria growth, x is the time in
hours, 0 = 470.4 ADC units and B = 0.05181 hours-1. This equation results in a doubling time of 13 hours
22 mins (approximately).

Figure 5 Controlled growth of cyanobacteria. Here their growth is caped at OD of 600.

Figure 5 is a plot of a growth experiment in which we were control the growth of cyanobacteria. In this
experiment we cap the OD to a value of 600 units. We can see that as soon as the bioreactor measures the
OD grow above this limit, it pumps some media and dilutes the solution. We can see that it successfully
keeps the bacteria concentration under control.

Figure 6 Controlled growth of E.coli bacteria

Figure 6 is a growth plot that we obtained when we were growing E.coli bacteria using our bioreactor.
The bioreactor is limiting the bacteria concentration to 200 ADC units. We had to set higher op-amp gains
in order to detect E.coli bacteria. This plot is of an execution duration of 2.5 hours. We have successfully
tested our setup to run continuously for very long periods of time; over 3 months.

10

References:
[1]

Matlab timer class, 2015. http://www.mathworks.com/help/matlab/ref/timer-class.html.

[2]

Patrick Guilfoile and I Edward Alcamo. Antibiotic-resistant bacteria. Infobase Publishing, 2009.

[3] Gardiner Harris. superbugs kill indias babies and pose an overseas threat, Dec 2014. News
article at http://www.nytimes.com/2014/12/04/world/asia/superbugs-kill-indias-babies-and-posean-overseas-threat.html?_r=0.
[4] Jason Millman. Superbugs could eventually kill more people than cancer, 2014. News article at
http://www.washingtonpost.com/blogs/wonkblog/wp/2014/12/12/superbugs-could-soon-killmore-people-than-cancer/.
[5] Erdal Toprak, Adrian Veres, Sadik Yildiz, Juan M Pedraza, Remy Chait, Johan Paulsson, and Roy
Kishony. Building a morbidostat: an automated continuous-culture device for studying bacterial
drug resistance under dynamically sustained drug inhibition. Nature protocols, 8(3):555567,
2013.

11

Appendix A
Schematics of the Sensor Board

12

The circuit is divided in two boards, namely the sensor board and the motor control board. The
sensor board has circuits for measuring the sensor voltage, amplifiers and overvoltage protection circuits.
Other than sensing electronics, it also has the circuits for the real time clock, LED control, DC power
distribution (+5V and +3.3V). These are the schematics used for the sensor board. Some portions of the
schematic are shown only for sensor 1. It should be noted that those circuits are exactly identical for sensor
2 and 3.
1. Figure A-1 is an overview of the schematic and shows how different modules are connected.
2. Figure A-2 shows the sensing circuit for sensor 1.
3. Figure A-3 shows the LED control circuit for sensor 1.
4. Figure A-4 shows the sensor voltage amplifier circuits.
5. Figure A-5 details the amplifier gain control circuits.
6. Figure A-6 shows the circuit for the on board real time clock.
7. Figure A-7 describes the SPI bus arbitration circuit.
8. Figure A-8 details the overvoltage protection circuit for sensor 1.
9. Figure A-9 describes the circuit that generates +5V power supply for the board.
10. Figure A-10 describes the circuit that generates +3.3V power supply for the board.
11. Figure A-11 shows the connectors that share electrical power with the pump control board which is
mounted on top of the sensor board.
12. Figure A-12 shows the connectors used to mount the board on the BeagleBone Black.
13. Figure A-13 shows the connectors used to connect the sensor cable which carries power, sensor voltage
and LED power from the sensor board to the experiment setup.

13

Figure A-1 Overview of the schematic.

14

Figure A-2 Sensing circuit for sensor 1.

Figure A-3 LED lights control for sensor 1.

15

Figure A-4 Op-Amps for amplifying the sensor voltage.

16

Figure A-5 Digital Potentiometer for controlling the gain of the op amps.

Figure A-6 RTC chip.

17

Figure A-7 SPI bus arbitration circuit.

Figure A-8 Overvoltage protection circuit for sensor 1.

18

Figure A-9 +5V regulator circuit.

Figure A-10 +3.3V regulator circuit.

Figure A-11 Pins to share power with the pump control board.

19

Figure A-12 Connectors for the P8 and P9 headers on the BeagleBone Black.

Figure A-13 Connector for the sensor cable for sensor 1.

20

Appendix B
Schematics of the Relay Board for AC pumps

21

The second part of the schematics of the system are the motor/pump control boards. We designed
two versions of the board, one with AC pumps and another with DC pumps. These are schematics for the
pump control board supporting AC pumps.
1. Figure B-1 shows the overview of the schematic
2. Figure B-2 shows the connectors used to mount the board on the sensor board and thus, establish
connection with the BeagleBone Black.
3. Figure B-3 shows the connector for AC power input as well as a connector that mates with a
corresponding header on the sensor board. This is used to share +5V power generated by the sensor
board with the pump control board.
4. Figure B-4 details the circuit for the relays and SSRs used to control the pumps corresponding to sensor
1.

22

Figure B-1 Overview of the schematic.

23

Figure B-2 Connectors for the P8 and P9 headers on the BeagleBone Black.

Figure B-3 Power Connectors.

24

Figure B-4 Relays and SSRs control circuit along with the connector socket for pumps
corresponding to sensor 1.

25

Appendix C
Schematics of the Relay Board for DC pumps

26

In addition to designing PCBs for AC pumps, we also developed a pump control board that drives
DC pumps. It must be noted that the air pump still operates on AC voltage. These are the schematics for
the motor control board with DC pumps.
1. Figure C-1 shows the overview of the schematic.
2. Figure C-2 shows the connectors used to mount the board on the sensor board and thus, establish
connection with the BeagleBone Black.
3. Figure C-3 shows the connectors for the AC power cable as well as +12V DC power required by the
DC pumps.
4. Figure C-4 details the Darlington array driver used to control the DC pumps.
5. Figure C-5 shows the SSR drivers needed to control AC air pumps.
6. Figure C-6 shows schematic of the connector used to connect the pump control board to the pumps.

Figure C-1 Overview of the schematic.

27

Figure C-2 Connectors for the P8 and P9 headers on the BeagleBone Black.

Figure C-3 Power connectors.

28

Figure C-4 Darlington array driver for all DC pumps.

Figure C-5 SSR drivers for air pumps corresponding to sensor 1.

29

Figure C-6 Connectors for the pump cables.

30