594770

research-article2015
JLAXXX10.1177/2211068215594770Journal of Laboratory AutomationHusain et al.

Original Report
Journal of Laboratory Automation

Development of Low-Cost Microcontroller-

2016, Vol. 21(5) 660­–670
© 2015 Society for Laboratory
Automation and Screening
Based Interface for Data Acquisition and DOI: 10.1177/2211068215594770
jala.sagepub.com

Control of Microbioreactor Operation

Abdul Rashid Husain1, Yaser Hadad1, and Muhd Nazrul Hisham Zainal Alam2

Abstract
This article presents the development of a low-cost microcontroller-based interface for a microbioreactor operation.
An Arduino MEGA 2560 board with 54 digital input/outputs, including 15 pulse-width-modulation outputs, has been
chosen to perform the acquisition and control of the microbioreactor. The microbioreactor (volume = 800 µL) was
made of poly(dimethylsiloxane) and poly(methylmethacrylate) polymers. The reactor was built to be equipped with
sensors and actuators for the control of reactor temperature and the mixing speed. The article discusses the circuit of
the microcontroller-based platform, describes the signal conditioning steps, and evaluates the capacity of the proposed
low-cost microcontroller-based interface in terms of control accuracy and system responses. It is demonstrated that
the proposed microcontroller-based platform is able to operate parallel microbioreactor operation with satisfactory
performances. Control accuracy at a deviation less than 5% of the set-point values and responses in the range of few
seconds have been recorded.

Keywords
microbioreactor, data acquisition, microcontroller, Arduino MEGA 2560, process control

Introduction base solution, antifoam solution, reactor medium, etc.), set-

Microbial fermentation process development generally
demands a high number of well-controlled laboratory-scale
cultivations. These experimental works are necessary as they
are usually needed for media optimization, cell line selection,
and for maximizing process yield and productivity.1,2 To date,
most of these experimental works are done in shake flasks
and lab-scale bioreactors.3 In a shake flask, fermentation
experiments can be initiated simply by adding the fermenta-
tion medium and the necessary culture inoculum at the right
volume proportion. When running experiments in shake
flasks, smaller volumes (i.e., between 100 mL and 1 L) are
normally used with the possibility of parallel experimenta-
tion.3 Indeed, a very handy experimental tool. However,
shake flasks do not offer a similar capacity in controlling
essential process conditions compared to lab-scale bioreac-
tors. By contrast, lab-scale bioreactors are normally equipped
with sensors and actuators for a tight control over important
reactor variables such as temperature, pH, dissolved oxygen
concentration, and mixing speed.4,5 Clearly, bioreactor sys-
tems present better control circumstances for studying fer-
mentation experiments. On the contrary, handling bench-scale
bioreactors is not so straightforward. In practice, it usually
involves inoculum development (preculture phase), the prep-
aration and sterilization of various solutions (i.e., the acid/
ting up the necessary fluidics connections, and extensive
cleaning of the reactor at the end of each operation. In addi-
tion, relatively larger volume consumption (minimally
around 1 L) has hampered its exploitation for multiple reac-
tor operations.
Considering the circumstances, a microbioreactor plat-
form would provide an attractive solution to these short-
comings. A microbioreactor is a very small-scale bioreactor
system, fabricated to support biological experiments in sub-
milliliter range volume (typically less than 1 mL range).6 At
this reduced scale of operation, (1) medium consumption is
significantly lowered, (2) transport processes (i.e., heat and
mass transfer rates) are comparatively faster due to a high
1
Department of Control and Mechatronic Engineering, Faculty of
Electrical Engineering, Universiti Teknologi Malaysia, Johor Bahru,
Malaysia
2
Process Systems Engineering Centre, Faculty of Chemical Engineering,
Universiti Teknologi Malaysia, Johor Bahru, Malaysia
Received March 29, 2015.
Corresponding Author:
Muhd Nazrul Hisham Zainal Alam, Process Systems Engineering Centre,
Faculty of Chemical Engineering, Universiti Teknologi Malaysia, Johor
Bahru, 81310, Malaysia.
Email: nazrul@cheme.utm.my
Husain et al. 661

Table 1.  Typical Measured and Controlled Variables and Common Data Acquisition Interface for Various Microbioreactor Systems.

Working Measured Controlled Data Acquisition

References Volume (µL) Variable Variable Parallelism Interface
Kostov et al.17 1000 OD, pH, DO T 1 NI I/O card and LabVIEW
Boccazzi et al.,13 Zanzotto et al.22 5–50 OD, pH, DO T 1 NI I/O card and LabVIEW
Maharbiz et al.20 250 OD, T, pH T 8 µC and NI I/O card
Szita et al.,10 Zhang et al.23 150 OD, pH, DO T 4 NI I/O card and LabVIEW
Harms et al.16 1000 OD, pH, DO, N N, DO 24 µC and LabVIEW
Lee et al.9 100 OD, pH, DO T, DO, pH 8 NI I/O card and LabVIEW
Krommenhoek et al.18 100-3000 Cells, T, pH T 96 NI I/O card and LabVIEW
Schäpper et al.21 100 OD, T, pH, DO T 1 NI I/O card and LabVIEW
Buchenauer et al.15 500 Cells, DO, pH T, pH 4 NI I/O card and LabVIEW
Bower et al.,14 Lee et al.19 1000 OD, T, pH, DO T, pH, DO 1 NI I/O card and LabVIEW
This work 800 N, T N, T 2 µC and LabVIEW

DO, dissolved oxygen; I/O, input/output; N, stirring speed; NI, National Instruments; OD, optical density; T, temperature.

surface area to volume (S/V) ratio, and (3) parallel opera-
tion for higher throughput is feasible because of a much
smaller reactor footprint.6,7 Based on the current data,
microbioreactor technology for microbial cultivation appli-
cations has progressed to a state where process conditions
in lab-scale bioreactors can be reproduced and accurately
mimicked in such a microbioreactor platform.8–10
In microbial fermentation processes, growth and produc-
tion rates of the cells are strongly dependent on the process
physical parameters—namely, the dissolved oxygen (DO)
level, pH, and temperature.11 Therefore, in order for the
microbioreactor to sustain the fermentation process, auto-
mated sensing and control of these parameters are impera-
tive.6,12 Table 1 summarizes the typical measured and
controlled variables and common data acquisition interface
for various microbioreactor systems.9,10,13–23 In microbiore-
actors, resistance temperature detectors (RTDs) such as
thermistors, Pt 100, and so on are generally used for tempera-
ture measurement. They are available in very small sizes and
can easily be integrated into the reactor.18,20,21 To regulate the
temperature, miniature heaters can be employed.18–21 Since
microbioreactors are normally not thermally insulated, they
are inevitably exposed to a significant heat loss to the sur-
roundings (due to the reactor’s high S/V ratio).6 This has
made temperature control relatively challenging as energy
input is needed to balance the heat loss as well as to provide
sufficient heat to maintain temperature at the desired set-
point values. Sampling is not an option in microbioreactors
because of the low working volumes (<1 mL). Thus, moni-
toring of the fermentation progress is often facilitated by
using optical probes. For cells’ optical density, measurement
is based on the Beer-Lambert law with optical fibers used as
waveguides for transmission and collection of light sig-
nals.9,10,13,14,16,17,20–23 As for pH and DO measurements, an
optical sensing method is usually applied with the use of a
fluorescent sensor spot or “optodes.”9,10,13–17,21–23 pH optode
readings are rather linear between pH 5.5 and pH 8 but
become insensitive beyond this range.24 Presently, pH in
microbioreactors is maintained either by a buffered med-
ium10,13,16–18,21 or through an intermittent addition of acid or
base.9,14,15,19 The latter has been reported to be limited by
the microbioreactor volume and acid/base molarity.6,9 On
contrary, a DO sensor spot can work for the entire saturation
values (i.e., from 0% to 100%), and their sensitivity is opti-
mal at low DO levels. Moreover, DO levels can be main-
tained above the desired level simply by membrane
aeration.9,10,13–16,19,20–23 Details on measurement principle
and sensing instruments can be found in a review article on
microbioreactor technology for fermentation processes by
Schäpper et al.6
The monitoring and process control tasks in microbiore-
actor system are normally complex as they involve more than
one feedback control loop.6,12 In microbioreactor operation,
often sensors and actuators from the microbioreactor system
are connected to a data acquisition (DAQ) device (also
known as an input/output [I/O] card) for input and output of
signals.9,10,13–23 Truly, process data are also readable via oscil-
loscopes or multimeters, but the use of computers (i.e.,
through a DAQ device/card) is an added advantage as data
can be stored and converted to a format that can be used by
spreadsheets (e.g., Excel) or other software packages such as
MATLAB for more extensive analysis.25 Another advantage
in particular is the obtainment of real-time data for monitor-
ing and control purposes with possible realization of a fully
automated microbioreactor setup. Based on the collected lit-
erature (Table 1), it seemed that most microbioreactor opera-
tors opted for a readily available data acquisition device (i.e.,
off-the-shelf boards) to execute the above-mentioned tasks.
One of the most widely used data acquisition devices is the
National Instruments (NI) DAQ device/card (National
Instruments, Austin, TX).9,10,13–23 Program codes for mea-
surement and control can be written in LabVIEW (National
Instruments) software and implemented by interfacing the
LabVIEW software with a suitable NI DAQ device. Indeed,
662 Journal of Laboratory Automation 21(5)
this is a viable solution and obviously beneficial for process
automation and multiplexing of the microbioreactor opera-
tion. Another interesting choice that is worth looking at is the
use of microcontrollers, where this embedded-based technol-
ogy offers a more cost-effective solution with flexible design.
A microcontroller is miniaturized computer on a single
integrated circuit with a processor core, memory, and con-
figurable peripherals that form the main components of the
system. The memory consists of read-only memory (ROM)
to store the program code and ready-access memory (RAM)
that serves as the temporary data storage such as measure-
ment data or adjustable system parameters. The clock rate
running at typically up to 100 MHz in modern microcon-
troller systems determines the speed of the processing to be
executed, which then dictates the sampling time needed for
the system under control. Digital I/O serves to interface the
microcontroller to the real application in terms of bit control
while analog interfacing is done through analog-to-digital
(ADC) channels. The number of digital I/O and ADC depends
on the type of the microcontrollers used, and for optimal
design, the number of I/O and ADC used must be configured
according to the need of the application for cost efficiency.
Among the most popular microcontrollers are 8-bit, 16-bit,
and 32-bit architecture families of programmable interface
controllers (PIC) from Microchip (Arizona, USA), Alf and
Vegard RISC (AVR)- and Advanced RISC Machines (ARM)-
based microcontrollers by ATMEL (San Jose California,
USA), and Freescales (Austin, Texas, USA). However, with
the increasing trend of adapting the use of microcontrollers as
an embedded system in many applications, many manufac-
turers have created a development board to speed up interfac-
ing microcontrollers to the system. Among the popular is
the Arduino board (an open source hardware created at
Interaction Design Institute Ivrea, Italy in 2005), which runs
on an advantageous open-source platform and features easily
configured hardware and software settings. With the ATMEL
microcontroller running as its core, it has all the digital I/Os
with pulse-width-modulation (PWM) outputs that are neces-
sary for acquisition and control purposes.
In this work, an Arduino MEGA 2560 with 54 digital I/Os,
including 15 PWM outputs, was chosen to replace the off-
the-shelf DAQ card to perform the acquisition and control of
a microbioreactor. The 245 kBytes of flash memory, 8 kBytes
of static RAM (SRAM), and 4 kBytes of EEPROM (electri-
cally erasable programmable ROM) have proven to be more
than adequate for the designed microbioreactor system, even
when interfaced with LabVIEW software residing on the ter-
minal PC. The article first describes the details of the micro-
bioreactor prototype used, the circuit (hardware) of the
microcontroller-based interface, and the signal conditioning
steps. Second, measurement and control routines written in
LabVIEW software and its implementation are discussed.
The calculated output of the control algorithm was converted
to PWM signals, which were then used to regulate the
temperature and mixing speed. Finally, the performance of
the proposed low-cost microcontroller-based interface was
evaluated in terms of control accuracy and system responses.
Mixing quality of the microbioreactor system was also com-
pleted. Throughout the work, two microbioreactor prototypes
were tested running in parallel where each reactor had been
equipped with independent mixing speed and temperature
control capacity.
Materials and Methods
Microbioreactor Design and Fabrication
The scheme of the experimental setup for the microbioreac-
tor is illustrated in Figure 1, which illustrates only the gen-
eral I/O between the setup and the digital computer (which
in this case is the microcontroller). The microbioreactor
prototype presented in this work was fabricated with inte-
grated temperature and mixing speed control capacities
(i.e., minimum features needed to perform biocatalyst
experiments under buffered medium). The prototype was
designed to work with constant working volume, with no
head space and under bubble-free conditions. It contained a
single poly(dimethylsiloxane) (PDMS) layer pressed
between two poly(methylmethacrylate) (PMMA) layers.
The top PMMA layer consisted of space for insertion of the
Pt 100 temperature sensor and through-holes for the fluidics
connections where the middle PDMS layer functioned as a
gasket layer to ensure water-tight sealing between the mid-
dle PDMS layer and the PMMA layers. The bottom PMMA
layers consisted of a reaction chamber with a total working
volume of approximately 800 µL. A small magnetic stirrer
(7 mm [length] × 2 mm [width]) was also placed inside the
reaction chamber to provide necessary mixing for the
microbioreactor operation. The prototype was fixed on top
of an operating platform that contained a microheater and a
12-V direct current (DC) PC mechanical fan for tempera-
ture and stirring control, respectively. In this microbioreac-
tor setup, a voltage regulator IC (i.e., a 5-VDC, 1.5-A
transistor) was used as the microheater, and small ferrite
magnets were glued on top of the PC mechanical fan to give
the north and the south pole arrangements for the stirring
speed detection by the Hall effect switch. Details of this
operating platform have been described elsewhere.26
Prototyping of the microbioreactor was done with three-
dimensional (3D) computer-aided design (CAD) software,
SolidWorks (Dassault Systèmes SolidWorks Corporation,
Waltham, MA), and micromachining. Fabrication was all
done by using PMMA and PDMS (Dow Corning Sylgard
184 Silicone Elastomer; Dow Corning, Midland, MI) poly-
mers as materials for the fabrication.27 The microbioreactor
setup was duplicated to facilitate parallel reactor operation.
Sensors and actuators from both microbioreactors were all
linked to a microcontroller-based interface.
Programs for measurement and control routines were
written in LabVIEW Academic Premium Suite software
Husain et al. 663

Figure 1.  Schematic of the microbioreactor experimental setup. The microbioreactor prototype was fixed on top of an operating
platform that contained a microheater and a 12-V PC mechanical fan for temperature and stirring control, respectively. Also, a
standard medical fluid infusion set was used to facilitate liquid filling into the microbioreactor where a 1-mL syringe was used as a feed
reservoir. The figure illustrates only the general input/output between the setup and the digital computer (which in this case is the
microcontroller).

(National Instruments) and were implemented by interfac-
ing the LabVIEW program with a microcontroller-based
interface for input and output of signals. In the microbiore-
actor prototype presented here, only the mixing speed and
the temperature of the microbioreactor content were mea-
sured and controlled. Further analysis of the online mea-
surement data (i.e., the stirring speed frequency and
temperature measurement values) was done using MATLAB
software v7.0 (The MathWorks, Natick, MA).
Microbioreactor Operation and Experimental
Work
In this prototype, a common medical fluid infusion set was
used to facilitate liquid filling into the microbioreactor. The
tip of the fluid infusion set (i.e., the needle) was inserted
into one of the fluidic ports, and water filling was controlled
via the roller clamp valve. In addition, a 1-mL syringe was
used as a water reservoir for replenishment of evaporated
water vapor during the experimentation. The water evapo-
ration rates were measured to be on the order of 170 µL·h–1
at 50 °C. During startup, the reactor content was first filled
with water by injecting a known amount of desired working
solution. Next, the Pt 100 temperature sensor was con-
nected to the reactor, and both the temperature and the stir-
ring control routines were initiated. In this work, water was
used in every experimental work.
Experimentation carried out in this work was to objec-
tively assess the reliability and the performance of the
microcontroller-based interface. For this reason, set-point
tracking experiments for the proportional (P) – Integral (I)
temperature and the mixing speed controllers were per-
formed. The set-point tracking experiments were performed
simultaneously for both microbioreactors and were repeated
twice to check data reproducibility. Moreover, the mixing
quality of the microbioreactor prototype was also assessed
by using a simple tracer method. This method was per-
formed by adding a very small amount of concentrated dye
using a syringe into the microbioreactor that already had
been filled with water. Mixing effects were observed visu-
ally, and images of dispersion of the concentrated dye in the
reactor were recorded by a digital camera (Nikon D90;
Nikon Corporation, Tokyo, Japan). Mixing time was esti-
mated at stirring speed, N of 300 rpm (NRE = 245). Duplicates
were made to compare results attained, and the mixing time
was recorded as the time needed for the concentrated dye to
be completely mixed.
Results and Discussion
Interfacing Microbioreactors to the
Microcontroller Platform
Arduino MEGA 2560 is the embedded microcontroller that
serves as the platform to interface the electronic system and
personal computer to the microbioreactor. For this specific
system, temperature and motor speeds are the input vari-
ables captured by the microcontroller while the heaters and
DC motors are the actuators to be controlled. Since the
LabVIEW program is used as the graphical user interface
(GUI) for the system, a firmware—a specialized driver
written in Arduino MEGA 2560—is also developed to per-
form the initialization of the input/output system. The inter-
facing of Pt 100 used for the temperature sensor to the
Arduino MEGA 2560 requires three steps: (1) detection
of voltage change, (2) voltage amplification, and (3) high-
frequency noise removal. The resistance of Pt 100 varies
with temperature where at a temperature of 0 °C and 100
664 Journal of Laboratory Automation 21(5)

Figure 2. (a) Circuit diagram of the Wheatstone bridge used for the temperature measurement via the Pt100 sensor. The
Wheatstone bridge is used to detect the difference in voltage, VR, as the resistance Rx in the bridge is varied. (b) Circuit diagram of the
difference amplifier used in interfacing for the Pt 100. The difference amplifier is used as the bridge to capture the voltage difference
produced by the two terminals in the Wheatstone bridge. (c) The configuration of the noninverting amplifier. The noninverting
amplifier is used to amplify the output voltage produced from the difference amplifier to accommodate the voltage level required to
be interfaced to the Arduino input.

°C, the resistances would be 100.00 Ω and 138.5 Ω, respec-
tively. Although this Pt 100 is capable of operating in a wide
temperature range, which is –200 °C to 850 °C, the opera-
tional aspect of this microbioreactor only requires a normal
0 °C to 100 °C temperature range, thus avoiding extra sig-
nal conditional circuitry for the nonlinear range of the Pt
100. The change in resistance of the Pt 100 is converted into
voltage difference through the use of a Wheatstone bridge
circuit, as shown in Figure 2a, and the voltage change VR is
given by equation (1). The voltage VR is obtained by the dif-
ference in voltage at point 1 and point 2, producing Vout,
given by equation (2).
 Rx R2  integral (PI) controller used as the stabilizing algorithm. The
VR = Vexc  − (1)
 ( R + R ) ( R + R ) 
. Hall effect sensor is used to provide the speed information of
 x 3 2 2 
Vout = (V2 − V1 ) .
The full range of voltage value produced by the differ-
ence amplifier, Vout (Fig. 2b), is only 7.64 mV, which is
rather small to be channeled into the ADC input of the
Arduino MEGA 2560. To fully use the 10-bit available
ADC, the voltage Vout needs further amplification with a
gain of 654. This can be achieved by using a noninverting
operational amplifier with configuration given in Figure
2c. The input voltage, Vin, coming from the output of the
difference voltage will be amplified to produce Vout2, which
is given in equation (3). When R3 = 220K Ω and R4 = 6.8 Ω
+ 330 Ω are selected from the available discrete resistor val-
ues, the amplification gain of 654 is obtained.
 R 
Vout 2 = Vin 1 + 4  .
 R3 
To remove unnecessary noise such that a more stable
signal is fed into the ADC of the microcontroller, a simple
resistor-capacitor low-pass filter is attached at the ADC
input pin. The complete interfacing circuit for the Pt 100 is
shown in Figure 3a. The operational amplifier model
OPA2277, which features low offset voltage and low drift,
contains two op-amps in a single chip that can be used to
obtain the difference voltage at the Wheatstone bridge and
signal amplification. All the calculated discrete values of
the resistors and capacitor used in the design are included in
the figure.
In this microbioreactor, the mixing speed control is per-
formed in a closed-loop configuration with the proportional-
the DC motor to the microcontroller system. This sensor works
by detecting the presence of magnetic flux where the output
(2)
voltage will change alternately from high to low and so on.
These alternate changes in the output voltage will produce a
series of pulses per unit time in which it can be used to deter-
mine the current speed of the motor. The sensor is placed
around 2 mm from the propeller magnet to perform the activa-
tion. The speed measurement will be used for the real-time
calculation of the PI controller, and thus the reading should be
sampled at a very fast rate. To achieve this requirement, the
so-called mini Arduino 328P is used to interface the Hall effect
sensor, as shown in Figure 3b. This mini Arduino will capture
the pulse produced by the Hall sensor where an interrupt signal
is set such that the pulses are counted between the given set
time. The number produced will represent the rotational speed
of the motor where the calculated value of speed is converted
to a 10-bit PWM pulse at pin 15 of the mini Arduino. The ana-
log value of the speed (rpm) is then obtained by averaging the
(3) PWM pulses by using a resistor-capacitor low-pass filter in
which the average speed values are fed to the ADC channel of
Husain et al. 665

Figure 3. (a) The complete interfacing circuit for the temperature sensor (Pt 100) from the sensing point to the input pin of the
Arduino. (b) Complete interfacing circuit for speed measurement based on the Hall effect sensor to the input pin of the Arduino
with the ATMEGA microcontroller is used to accurately capture and count the number of pulses produced by the Hall effect sensor.
(c) The required interfacing circuit from the microcontroller to the heater or motor where the command from the microcontroller
forms as the required control signals for each of the actuators. (d) Configuration of the voltage regulator circuit. A standard voltage
regulator circuit is used to regulate and stabilize the voltage at the required voltage level.

the Arduino 2650, which acts as the DAQ board. It should be
noted that the mini Arduino is added in this design such that the
real-time timeliness of the closed-loop control of the motor can
be guaranteed.
For the actuation of the stirrer bar in the microbioreactor,
the commonly available 12 VDC PC fan is used with the fan
blade removed. PWM signals representing the desired
speed of the motor are obtained from the output of the PI
controller in which the PWM pulses are averaged by a low-
pass filter and amplified by an LM358 operational amplifier
to match the 12-V supply voltage. In similar line of design,
the driving circuitry for activation of the heater uses the
same conditioning and amplification approach, where the
circuit is as shown in Figure 3c. Figure 3d, in addition,
shows the schematic of the voltage regulator to supply a
stable 5-V voltage to the electronic circuit in the system.
LabVIEW Programming
The entire microbioreactor setup was controlled by the
LabVIEW programming language software and implemented
by interfacing the software to the microcontroller-based inter-
face for the input and output of signals. In the LabVIEW pro-
gramming environment (VI), data are passed from one node/
sub-VIs to another through designated input terminals (usually
to the left of each node), and the resulting values are obtained
through the node’s output terminal, which is normally to the
right side of the node. Wires in the block diagram window rep-
resent the flow of data within the VI.25 For this particular
microbioreactor setup, the whole operating scheme was writ-
ten in series. The VI featured two main aspects: (1) the mea-
surement and control of microbioreactor temperature and
mixing speed and (2) subdiagrams to facilitate parallel micro-
bioreactor operation. This is presented in Figure 4a.
In the program written, first, a connection to an Arduino
I/O card running the LabVIEW interface was initiated. It is
important to note here that the baud rate specified must
match with the default baud rate defined in the Arduino
firmware. For our application, the baud rate has been set to
a maximum value of 115,200 bits per second (bps). The
Arduino Initiate sub-VI is also linked to a visa resources
element. The visa resources element enables the program to
666 Journal of Laboratory Automation 21(5)

Figure 4. (a) Overview of the LabVIEW program for the entire reactor operation where data flow from left to the right for the
entire program. The VI features a routine for establishing a connection with the Arduino board, subdiagrams to facilitate parallel
microbioreactor operation, and the termination sequence. (b) Screenshot of the LabVIEW block diagram for a single microbioreactor
operation, which consists of the sensing and control routines for the microbioreactor temperature and mixing speed variables.

automatically identify the resources on which the VI oper-
ates (i.e., the I/O physical channels) and distinguish between
different sessions of the microbioreactor operation. This is
necessary as more than one I/O physical channel was used
during the entire microbioreactor operation. Each Arduino
sub-VI was linked together through the task in/out terminal
to ensure continuous data transfer once the program initi-
ated. Since the error in/out terminal is also wired from one
Arduino sub-VI to another, the entire program will auto-
matically stop if an error occurs anywhere within the
Arduino sub-VI. In every cycle, all necessary inputs (i.e.,
the desired physical I/O channels, baud rates, and type of
Arduino board used) will be passed through the series of
subdiagrams controlling the microbioreactor operation. The
entire operation functions repeatedly until the conditional
Boolean operator in the while loop is met (i.e., when the
stop button is pressed). At the termination sequence (Fig.
5a), all PWM signals that were specified to specific digital
output pins are reset to zero before the Arduino Close sub-
VI deactivates the LabVIEW interface and Arduino I/O
card connection. This is essential to avoid any undesirable
current flow from any of the digital pins when ending the
operation.
Figure 4b illustrates the image of the screenshot of the
LabVIEW block diagram for one of the microbioreactor
operations, which consists of the sensing and control rou-
tines for the microbioreactor temperature and mixing speed
variables. For the temperature control, the Arduino Read
sub-VI first read the analog voltage signal from the Pt 100
temperature sensor on the selected analog input pin (AI1).
The temperature input signal was then fed to the PI control-
ler sub-VI. Next, the PI controller sub-VI computed the
deviation (error) between the desired temperature set point
value and the measured signal. By using a math function,
the temperature set point value (°C) was first converted to
voltage signal values before being fed to the PI controller
sub-VI. Based on the data specification of the sensing
instrument, the voltage signal range of the Pt 100 sensor
(i.e., 0–5 VDC) used was found to be in linear relation to the
temperature values between the 0 and 100 °C range. With
Husain et al. 667

Figure 5.  Response of the PI mixing speed (top) and temperature (bottom) controllers for a series of step changes in the agitation
and temperature set point values, respectively. Responses of the first microbioreactors are on the left, and the second one is on
the right. Duplicates were made to check reproducibility, and data points presented are the average values attained from the two
experiments.

the given reference inputs (i.e., the PI high/low limits and
gains), the PID controller sub-VI subsequently estimated
the control output values. To minimize heat loss to electron-
ics, the microheater was driven by PWM signals for heat-
ing. For the ArduinoAnalogWrite sub-VI, the PWM duty
cycle signal ranged between 0 and 255 (i.e., between always
off [0 V] and fully on [5 VDC]). This means that a call to
analogWrite (255) would give out a 100% duty cycle,
analogWrite (127) is a 50% duty cycle, and so on. Therefore,
to calculate the proportional PWM duty cycle signals, the
formula Duty cycle = [control output values (volts) × 255]
/ 5 volts was used. The resulting output values were then fed
to the digital pin (DIG7) of the Arduino I/O card that was
linked to the microheater.
As for the mixing speed control, a similar sensing and
control routine was written except that the ArduinoRead
sub-VI read the analog voltage signal produced by the Hall
effect sensor, and a different output digital pin (DIG6) was
used to drive the DC magnetic motor. The Hall effect sensor
produces an analog voltage signal that is proportional to the
number of pulses for every revolution per second (rps)
made by the DC magnetic motor. The analog signals regis-
tered range from 0 to 5 VDC, and the corresponding digital
values (i.e., the corresponding 10-bit PWM duty cycle
signals) range between 0 and 1023. A similar mathematical
formula was used to calculate the PWM duty cycle signals
for the DC magnetic motor control. It is essential to note
here that the DC magnetic motor rotational speed is directly
proportional to the value of the applied PWM duty cycle
signals. For example, a high duty cycle signal will produce
faster rotational speeds and a low duty cycle will produce
slower rotational speeds. All measurement signals and set-
point values are displayed on the VI front panel and written
(saved) to a file together with the current time stamp.
PI Mixing Speed and Temperature Controller
Performance
Figure 5 shows the response of the PI mixing speed and
temperature controllers for a series of step changes in the
agitation and temperature set-point values for both micro-
bioreactors. Both controllers were tuned based on the
Ziegler-Nichols tuning method (data not shown), and the
tuning parameters attained (i.e., the proportional gain, Kc
and the integral time, τi constants) are included in Figure 5.
The minimal and the maximum PI output values were set
such that the actuators (i.e., the microheater and the DC
magnetic motor) would operate within the desired range of
668 Journal of Laboratory Automation 21(5)
interest. In our applications, the desired agitation speed
limit ranged between 150 and 500 rpm. At this rate, the tur-
bulence condition in the microbioreactor normally varies
between the laminar flow condition and the transitional
flow regime where the Reynolds number, NRe, is less than
1000 (NRe < 1000). It appears that for this type of microbio-
reactor design, the mixing time gradually increased at
higher agitation rates (indicated by a shorter mixing time,
tm).21,26–28 Furthermore, it is seemingly irrelevant to stir the
reactor content at a much higher rate than 500 rpm since it
may not improve the mixing efficiency by much. Schäpper
et al.21 and Zainal Alam et al.28 reported that a well-mixed
state can be achieved in approximately a second (tm = 1.2
s) at an agitation rate of 500 rpm. In this work, it took about
2.5 s to reach a complete mixing (indicated by a uniform
color distribution) at an agitation rate of 300 rpm.
Significant concentration difference facilitated the diffu-
sion of the dye while vigorous agitation created a larger
molecular interfacial area for mixing. This is shown in
Figure 6. As for the temperature limits, the operating range
of interest is between 30 °C and 50 °C. This is because
fermentation processes using mesophilic microbes (e.g.,
Escherichia coli, Saccharomyces cerevisiae, etc.11,29) and
biocatalytic processes using industrial enzymes (e.g., amy-
lase,30 pectinase,31 etc.) are usually performed within this
temperature range.
For the mixing speed control responses, it can be seen
from Figure 6 that there were some obvious overshoots,
and the measured mixing speed fluctuated within ±10 rpm
of the set-point value at an agitation rate of 200 rpm. On
contrary, at higher rates (i.e., at N  > 300 rpm), much steadier
responses (i.e., no apparent overshoot) with settling time on
the order of a few seconds were obtained. The controller
accuracy at this range was recorded to be about ±3 rpm of
the set-point value. The results indicated that under these
conditions (i.e., at Kc = 0.1 and τi = 0.008), better responses
are achievable at agitation rates higher than 200 rpm. It
should be emphasized here that although the controller did
not respond favorably for the entire range tested, the opti-
mal settings and/or the controller constants can always be
manually adjusted or determined with a couple of short
Figure 6.  Images of mixing of the
concentrated dye solution in the
microbioreactor at a mixing speed,
N = 300 rpm (NRE = 80). Blue arrow
indicates the point of injection.
experimentations to give a better response at the agitation
rate of interest.
As for the temperature control, it was observed that there
was a slight overshoot (around 3%–4% off the set-point
value), and it took only a couple of seconds to reach a new
steady state when the set-point values were adjusted from
one set point to another. The accuracy of the PI temperature
controller was calculated to be within ±0.2 °C of the set-
point value. It can also be seen that there was no delay when
the microbioreactor temperature was regulated. This is due
to the high S/V ratio of the microbioreactor prototype,
which facilitated relatively fast heat transfer rates.6 This
behavior was consistent each time the temperature was reg-
ulated from one set-point value to another. In our microbio-
reactor setup, there was no cooling element installed. This
was deliberate, and cooling was merely achieved by taking
advantage of the significant heat dissipation to the sur-
rounding (natural convection). The cooling step may be
slow (i.e., between 0.05 and 0.1 °C·s–1) but still within the
acceptable length. This is because in our applications, the
control objective is to keep the measured variable as close
as possible to the desired set-point value, and it is not neces-
sary to regulate (i.e., heat/cool) the reactor temperature at a
very fast rate. The controller capacity in tracking various
mixing speeds and the temperature set-point values accu-
rately suggested that the system has a good control perfor-
mance and is stable.
As the next process to prototype the system, once the
system (proof of concept) is working as expected, the elec-
tronic circuit with customized printed circuit board (PCB)
design and fabrication with dedicated discrete electronic com-
ponents will be developed. This will enable us to employ a
microcontroller-based interface with a very small footprint.
In conclusion, in this work, a low-cost microcontroller-
based interface for microbioreactor operation was designed
and realized. As proof of concept, two polymer-based
microbioreactor prototypes (working volume ~800 µL)
were fabricated and tested in parallel. The total cost for the
presented system/rig—that is, the microcontroller interface
(including all the discrete electronic components and the
sensors) and the materials used for the fabrication of
Husain et al. 669

a single microbioreactor—was approximately €80. The

proposed microcontroller-based interface demonstrated the
capacity to (1) acquire real-time experimental data (in this
case, online data of reactor temperature and mixing speed)
and (2) produce PWM signals to regulate the actuators for
the control of the above-mentioned variables. Online data
are indeed crucial since they will enable the users to keep
track of the dynamic progress of the experiments performed.
The sampling rate applied for real-time sampling of experi-
mental data was satisfactory as the actual state of the physi-
cal conditions of the reactor content has been accurately
translated. The controller accuracy is also within the accept-
able limit in which measured variables were very comparable
to the set-point values and process responses were on the order
of few seconds. Simple yet reliable interfacing circuits for the
Pt 100 temperature sensors, Hall effect sensors, heater ele-
ments, and DC magnetic motors were designed such that the
input and output signals to the periphery of the microcon- Figure 7.  A compact microcontroller-based interface with
custom-made user interface for a parallel microbioreactor
troller were matched.
operation.
The aim of the work in interfacing a microcontroller-
based platform for data acquisition and control of the micro-
bioreactor operation has indeed been successful. The Acknowledgments
success of the project has motivated us, i.e., as future work We express our gratitude to the Malaysia Ministry of Higher
to further expand the automated measurement and control Education (MOHE) and Universiti Teknologi Malaysia for fund-
features of the microbioreactor. The reactor at its current ing the project.
state (i.e., equipped only with temperature, evaporation,
and mixing speed control) may have a limited application, Declaration of Conflicting Interests
e.g., may only be suitable to carry out studies on biocataly- The authors declared no potential conflicts of interest with respect
sis processes. In order for the microbioreactor system to to the research, authorship, and/or publication of this article.
cater to different bioprocessing operations, e.g., microbial
fermentation experiments and/or cell culture applications, Funding
additional features such as cell density measurements, DO The authors disclosed receipt of the following financial support
level, and pH control are imperative. These types of measure- for the research, authorship, and/or publication of this article: The
ments are normally realized by using optical measurement project was funded by Universiti Teknologi Malaysia Research
techniques. For example, optical cell density measurement University Grant (vote number, 06J83). It is a general grant for
can be done by comparing the amplitude of the sine wave development of an engineering-based prototype that covers mainly
signal (alternating-current signal) at a frequency of 47 Hz. materials and fabrication cost. Contributors who received the sup-
In contrast, pH and DO level measurements use a sine wave port include Abdul Rashid Husain, Yaser Hadad, and Muhd
(or square wave) with higher frequencies (pH measurement Nazrul Hisham Zainal Alam, who are staff and student (Yaser
[44 kHz] and DO level [5 kHz]), and measurements are Hadad) of Universiti Teknologi Malaysia, Malaysia. Successful
based on the phase difference produced by optical sensor research work that has to be published was the main interest of the
supporting body.
spots (optodes). In practice, higher sampling rates are often
needed for measurements (e.g., four to five times higher
References
than the transmitting signal, according to the Nyquist theo-
rem). Nevertheless, the available microcontrollers with a 1. Bhambure, R.; Kumar, K.; Rathore, A. S. High-Throughput
clock speed up to 100 MHz can easily accommodate the Process Development for Biopharmaceutical Drug Substances.
sampling rate of these measurement methods. Another Trends Biotechnol. 2011, 29, 127–135.
2. Micheletti, M.; Lye, G. J. Microscale Bioprocess Optimisation.
added advantage is that the microcontroller-based interface
Curr. Opin. Biotechnol. 2006, 17, 611–618.
also offers the flexibility for a compact module design for 3. Betts, J. I.; Baganz, F. Miniature Bioreactors: Current Practices and
data acquisition and control. All the necessary interfacing Future Opportunities. Microb. Cell Fac. 2006, 5, 1–14.
circuits, power units, sensors, and actuators, for instance, 4. Chisti, Y.; Moo-Young, M. Encyclopedia of Physical Science
can be embedded into a confined module, as illustrated in and Technology; 2003; 3rd ed., pp. 247–271. Academic Press,
Figure 7, in modular design. San Diego.
670 Journal of Laboratory Automation 21(5)
5. Zhong, J. J. 2.14—Bioreactor Engineering, Engineering
Fundamentals of Biotechnology; 2011; 2nd ed., Vol. 2,
pp. 165–177. Academic Press, Burlington.
6. Schäpper, D.; Zainal Alam, M. N. H.; Szita, N.; et al. Application
of Microbioreactors in Fermentation Process Development: A
Review. Anal. Bioanal. Chem. 2000, 395, 679–695.
7. Kirk, T. V.; Szita, N. Oxygen Transfer Characteristics of
Miniaturized Bioreactor Systems. Biotechnol. Bioeng. 2012,
110, 1005–1019.
8. Edlich, A.; Magdanz, V.; Rasch, D.; et al. Microfluidic
Reactor for Continuous Cultivation of Saccharomyces cere-
visiae. Biotechnol. Prog. 2010, 26, 1259–1270.
9. Lee, H. L.; Bocazzi, P.; Ram, R. J.; et al. Microbioreactor
Arrays with Integrated Mixers and Fluid Injectors for High
Throughput Experimentation with pH and Dissolved Oxygen
Control. Lab Chip. 2006, 6, 1229–1235.
10. Szita, N.; Bocazzi, P.; Zhang, Z.; et al. Development of a
Multiplexed Microbioreactor System for High-Throughput
Bioprocessing. Lab Chip. 2005, 5, 819–826.
11. Shuler, M. L.; Kargi, F. Bioprocess Engineering: Basic
Concepts; Prentice Hall: Englewood Cliffs, NJ, 2002; 2nd ed.
12. Gernaey, K. V.; Baganz, F.; Franco-Lara, E.; et al. Monitoring
and Control of Microbioreactors: An Expert Opinion on
Development Needs. Biotechnol. J. 2012, 7, 1308–1314.
13. Boccazzi, P.; Zhang, Z.; Kurosawa, K.; et al. Differential
Gene Expression Profiles and Real-Time Measurements of
Growth Parameters in Saccharomyces cerevisiae Grown in
Microliter-Scale Bioreactors Equipped with Internal Stirring.
Biotechnol. Prog. 2006, 22, 710–717.
14. Bower, D. M.; Lee, K. S.; Ram, R. J.; et al. Fed-Batch
Microbioreactor Platform for Scale Down and Analysis of a
Plasmid DNA Production Process. Biotechnol. Bioeng. 2012,
109, 1976–1986.
15. Buchenauer, A.; Hofmann, M. C.; Funke, M.; et al.
Microbioreactors for Fed-Batch Fermentations with
Integrated Online Monitoring and Microfluidic Devices.
Biosens. Bioelectron. 2009, 24, 1411–1416.
16. Harms, P.; Kostov, Y.; French, J. A.; et al. Design
and Performance of a 24-Station High Throughput
Microbioreactor. Biotechnol. Bioeng. 2006, 93, 6–13.
17. Kostov, Y.; Harms, P.; Randers-Eichhorn, L.; et al. Low-
Cost Microbioreactor for High-Throughput Bioprocessing.
Biotechnol. Bioeng. 2001, 72, 346–352.
18. Krommenhoek, E. E.; van Leeuwen, M.; Gardeniers, H.; et al.
Lab-Scale Fermentation Tests of Microchip with Integrated
Electrochemical Sensors for pH, Temperature, Dissolved
Oxygen and Viable Biomass Concentration. Biotechnol.
Bioeng. 2008, 99, 884–892.
19. Lee, K. S.; Boccazzi, P.; Sinskey, A. J.; et al. Microfluidic
Chemostat and Turbidostat with Flow Rate, Oxygen, and
Temperature Control for Dynamic Continuous Culture. Lab
Chip. 2011, 11, 1730–1739.
20. Maharbiz, M. M.; Holtz, W. J.; Howe, R. T.; et al.
Microbioreactor Arrays with Parametric Control for High-
Throughput Experimentation. Biotechnol. Bioeng. 2004, 85,
376–381.
21. Schäpper, D.; Stocks, S. M.; Szita, N.; et al. Development
of a Single-Use Microbioreactor for Cultivation of
Microorganisms. ChemEng. J. 2010, 160, 891–898.
22. Zanzotto, A.; Szita, N.; Boccazzi, P.; et al. Membrane-
Aerated Microbioreactor for High-Throughput Bioprocessing.
Biotechnol. Bioeng. 2004, 87, 243–254.
23. Zhang, Z.; Perozziello, G.; Boccazzi, P.; et al. Microbioreactors
for Bioprocess Development. J Assoc. Lab Autom. 2007, 12,
143–151.
24. John, G. T.; Goelling, D.; Klimant, I.; et al. pH-Sensing 96-Well
Microtitre Plates for Characterization of Acid Production by
Dairy Starter Cultures. Dairy Res. 2003, 70, 327–333.
25. Morris, A. S.; Langari, R. Chap. 5. Data Acquisition with
LabVIEW, Measurement and Instrumentation, 2012, 115-
133, Academic Press, San Diego.
26. Halimoon, H.; Zainal Alam, M. N. H. A Low Cost Stirring
Platform with Integrated Temperature Control Scheme for
Microbioreactor Operation. J. Tek. (SciEng). 2013, 62, 1–8.
27. Zainal Alam, M. N. H.; Pinelo, M.; Samantha, K; et al.
Continuous Membrane Microbioreactors for Development
of Integrated Pectin Modification and Separation Processes.
ChemEng. J. 2011, 167, 418–426.
28. Zainal Alam, M. N. H.; Schäpper, D.; Gernaey, K. V.
Embedded Resistance Wire as Heating Element for
Temperature Control in Microbioreactors. Micromech.
Microeng. 2010, 20, 055014.
29. Arroyo-Lopez, F. N.; Orlic, S.; Querol, A.; et al. Effects of
Temperature, pH and Sugar Concentration on the Growth
Parameters of Saccharomyces cerevisiae, S. kudriavzevii and
Their Interspecific Hybrid. J. Food Microb. 2009, 131, 120–127.
30. Wee, L. L.; Mohamad Annuar, M. S.; Ibrahim, S. Short
Communication: Energetics of glucoamylase-catalyzed
hydrolysis of commercial sago starch. Asia-Pacific J. Mol.
Biol. Biotechnol. 2011, 19, 117–120.
31. Yadav, S.; Yadav, P. K.; Yadav, D.; et al. Pectin Lyase: A
Review. Pro. Biochem. 2009, 44, 1–10.