New HTML template for the wiki

Microfermentation Platform

Microfermentation platform

When taking your synbio project to real applications, fast growth is crucial for viability. We took the concept of a DIY spectrophotometer and expanded it to a device that carries out three small scale fermentations simultaneously. With the files provided below you can build your own microfermentation platform, to compare the performance of our modified strain against a wild type or evaluate different growth media.


"When it comes to software, I much prefer free software, because I have very seldom seen a program that has worked well enough for my needs, and having sources available can be a life-saver."

Linus Torvalds

An important part of our project was the screening for substrates All this data was generated using a Hamilton robot for liquid handling. This robot enabled us to compare the growth characteristics Yarrowia lipolytica and Saccharomyces cerevisae. Such experiments are generally a great tool, to validate the growth on particular substrates or the degradation of toxic compounds. Furthermore, they can be used to compare different strains. However, we realized that it is a technology not everyone has access to.

To challenge this, we started a hardware project aiming to develop a cheap alternative to the Hamilton robot. A small device, that would enable hackerspaces and high schools to easily acquicire and compare growth data for their projects. To make this device as cheap as possible, we reduced the Hamilton to its key features:

Requirements for our robot

In order to carry out growth experiments automatically, our minimum viable product should have the following features:

  • OD-measurements
  • Aeriation
  • Stirring
  • Automatation
  • Data-logging

To provide a tool for higschools and hackerspaces, we imposed these additional requirements:

  • Simply reproducable from online material
  • Development using only free soft- and hardware
  • Modular and expandable design

Figure 1: The reference: Hamilton robot for liquid handling


We managed to create a working prototype over the summer. It is stable and has all the features we aimed for. This wiki page will introduce you to the development process and enable you to recreate your own version.

The following points are the key achievements we have

Figure 2: Our microfermentation platform


The Microfermentation Platform is a simplistic device that can run up to three batch cultures, while automatically measuring OD\(_{600}\), where custom settings such as duration of experiment are determined by the user and the program is controlled by a microcontroller.

Figure 3: Side view of the robot.

For fermentation, the device is equipped with 3 chambers each capable of holding a cuvette with a liquid culture, exposed to stirring and aeration by a pump.

For management, it is furthermore equipped with:

  • A liquid crystal display (LCD), displaying status and menus.
  • A 6-button keypad to navigate through the menus.
  • A real time clock (RTC) to keep track of time.
  • An SD-card reader for storing configuration data and saving measurements.

For OD\(_{600}\) measurement, each chamber is equipped with

  • An LED emitting the light to be measured.
  • A sensor recording the light from the LED after passing through the chamber (with culture)

The menu system is simple and user-friendly, as the main menu only has two options, either:

Start culture, which immediately starts a run, or

Edit settings, which has the following options:

  • View date and time.
  • Select number of chambers used.
  • Interval between measurements (1-99 minutes).
  • Duration of run in hours (1-99 hours).
  • Check the name of the output file.
  • Calibrate the light emission for correctly interpretable measurements.
  • Test the pump in order to be able to adjust the air pressure properly.

Whenever 'Start culture’ is selected from the main menu, the Microfermentation Platform carries out a full run corresponding to the settings and continuously saves all readings in a CSV file on the SD card. Built-in warnings are implemented to make sure the user is aware if the SD card reader or the RTC aren’t working properly, as deficiencies on these modules are devastating for the experiment, and are not easy to discover otherwise.



Growth rates and spectrophotometry

Basic Concept - Measuring Optical Density

When assessing growth of microorganisms, it is necessary to be able to obtain and interpret information about population size, or at least concentration at a given time. Optical Density (OD\(_{600}\)) measurement is a widely used, suitable method known by everyone familiar with biotechnology. This principle is based on spectrophotometry, where a light transmission of 600 nm passes through biological material and the resulting transmittance, \(\Phi\), is measured at the receiver end. Because biological material reflects light, higher biological density reflects more light, giving lower transmittance readings. Thus, at full transmittance, \(\Phi_{\text{ref}}\), OD\(_{600}\) should be equal to 0.

For the Microfermentation Platform, OD\(_{600}\) at time t is calculated from the Beer-Lambert law1,2,3,4: $$ \text{OD}_{600,t} = -\log_{10}\Bigg(\frac{\Phi_t}{\Phi_{\text{ref}}}\Bigg) $$ Where the light source is an LED, with its intensity controlled by some input voltage, and the transmittance is measured by a sensor and recorded as a voltage.

Figure 4: Brief outline on how spectophotometric measurements for calculation of OD\(_{600}\) are carried out by the Microfermentation Platform.

Maximal Saturation - Calibration of Light Emission

The sensor recording the transmittance has a maximal saturation level. Transmittance of higher intensity than this level will yield the same constant reading. Thus, the input voltage must be calibrated, such that the emission from the LED matches the maximal saturation level of the sensor exactly when tested under conditions considered as reference.

Figure 5: The maximal saturation level phenomenon. The graph shows sensor reading as a function of input voltage. Only in a limited range of input voltage gives linearly increasing sensor readings. The lowest input voltage yielding the maximal sensor reading is denoted maxpwm (for ‘maximal PWM’) and the corresponding sensor reading is called refmsm (for ‘reference measurement’).

Growth Curves of Microorganisms

Growth course across species of microorganisms exhibit some common characteristic behaviour. Typically, the growth curve of a population has an initial lag phase, with limited growth, and exponential phase, where the doubling time is constant, and finally, a stationary phase, where the population capacity has been reached, either because of limitation on resources or physical space.

Figure 6: A typical growth curve with lag, exponential and stationary phase.

User Testing

This device is meant to go out to fellow iGEMers and Hackerspaces. To validate the assumptions we had when building the prototype, we brought it to our friends from the local hackerspace Biologigaragen and our highschool student Tobias.

Higschool Student: Tobias

The purpose of this experiment was to test whether the instructions provided in our protocol were sufficient for someone who had never before worked with the device. Tobias volunteered to carry out the experiment. He had the task to compare two different Y. lipolytica strains growing on YPD. He was provided with the robot, all necessary material and the protocol and then left on his own device.

Tobias had no problems following the protocol we wrote. He found the menu very intuitive and liked the overall design. However, he mentioned that it was difficult to take the cuvettes out again from the robot because the holder is embedded very deeply in the housing. He helped himself by removing a panel and accessing the chambers from the side. A future prototype should be build slightly larger in order to increase comfort.

Despite the positive feedback, the experiment failed. This was due to a major flaw in the protocol given to him. It did not include calibration of the airflow. Therefore Tobias left the valve fully open. Most of the liquid was driven out of the cuvettes and spread out inside the housing. It short-circuited several pins of the Arduino, causing the program to crash. Consequently we adapted the protocol and the program so that the airflow could be regulated correctly prior to fermentation.

Figure 7: Tobias inoculating Y. lipolytica in the cuvettes.
Figure 8: Tobias fitting the top on the measuring chambers before starting the run.

Hackerspace: Biologigaragen

Being quite confident about our prototype, we took it to Biologigaragen a DIY Biohackerspace in Copenhagen. We showed our result to Keenan, who just build a photobioreactor for their lab. After introducing him to the main features and our considerations, we gave him a chance to take it apart and ask questions.

He was very positive about the interface and agreed that all the materials used would be available for them. He agreed with Tobias that we should make the next version bigger, so that the cuvettes can be moved in and out more easily. Besides he had a lot of suggestions, e.g. on upgrades for fed-batch fermentations and temperature control. Therefore he was happy to hear that all source code is available.

Figure 9: Keenan from Biologigaragen taking apart the robot, literally.

Proof of Concept

The figure shows two distinct exponential phases, representing the diauxic growth of S. cerevisiae. Based on the data recorded by the Arduino, we were able to compute the growth rates, shown in Fig. 4.

Figure 10: Growth curve of S. cerevisiae. The green and the blue line show the slope of the two exponential phases. The values are given next to the lines.

During the development, we gradually expanded our prototype, testing all the changes on the way. After going through the protocol with Tobias, we ran a final test to prove that the overall concept is working now. The final prototype was successful at measuring growth of Escherichia. coli, as evident from the results shown in Figure 11.

Figure 11: OD\(_{600}\) measurements of 2 simultaneous samples over 7 hours by the Microfermentation Platform. The red graph shows the sample of Escherichia coli in LB media inoculated at time 0, while the blue graph represents the negative control of LB only.


The prototype was build using laser cutting, 3D-printing and printed custom boards (PCB). It is controlled by an Arduino UNO R3 that was programmed using the Arduino IDE. All the files are created in the native formats of open source programs. We have used the programs listed below and recomend them, as the source files are in the respective native formats.

In addition, all these programs have great communities to help, if you get stuck.

Figure 12: Before using a printed circuit board, there were even more cables.

  • Inkscape to create vectorgraphics for laser cutting
  • FreeCAD to make CAD drawings for 3D-printing
  • ArduinoIDE for programming the microcontroler
  • KiCAD for PCB design

This is a brief overview of the development history. For a chronological order, see the notebook

Starting point

Figure 13: First experimental set-up

Together with Docent Erik Vilain Thomsen and Associate professor Martin Dufva, we developed a first prototype. In this version, a single cuvette was stirried with a small magnetic pellet in a black measuring chamber. The light intensity was regulated manualy using a very precise geared potentiometer. The microcontroler received the voltage level from the photodiode, converted it into a 10bit number and send this via USB to the laptop which wrote the output to a file.

We learned the following:

  • The LED remains constant during the whole duration of the experiment, giving constant light output.
  • The Arduino requires an external power supply larger than 5 V to supply stable voltages
  • Magnetic stirring in a cuvette is insufficient

Martin changed the container to a small glas flask which was giving sufficient stirring and demonstrated that the measuring unit was functional. However, the measuring cell grew even bigger. In order to reach our initial goal, we continued working on our own. We wanted to keep cuvettes as containers because they are cheap and easily available. Also our design needed the following changes

  • No magnetic stirring
  • No manually controlled potentiometer
  • Data logging on µSD-card

New concepts

We could demonstrate that aeriation via common disposable syringes and needles was possible and sufficient to homogenize a S. cerevisiae culture at a tolarable amount of foaming.

In addition, we managed to replace the potentiometer by a set of low-pass filters combined with the 8bit PMW output-pins of the Arduino. This cut down the cost for the potentiometer and enabled us to use an automatic calibration routine to find the level of saturation for the photodiode.

We designed our own measuring cell. Compared to Martin's version it has a simpler locking mechanism to hold LEDs and photodiodes in place. It is also much smaler, takes three cuvettes and has connectors for aeration.

After debugging our circuit on breadboards, we designed printed circuit boards. Ten boards only cost US$ 10, plus shipping.

Figure 14: Our freshly lasercut wood housing.
Figure 15: The three placeholders for the cuvettes.
Figure 16: Soldering connections on the Printed Circuit Board (shield for the Arduino).
Figure 17: Cuvettes after an overnight run. The two containers to the right were not sealed with parafilm. Therefore we included it as a requirement in the protocol.

Make Your Own

iGEM 2016 is over. We have received both, positive feedback about our creation and many suggestions on how to improve it. We created this prototype with the purpose to share it with hackerspaces, high schools and other iGEM teams world wide. Therefore, we have decided to move the project to github now to have a transparent platform for developments. The files for the robot we brought to Boston are still linked above as a documentation.

If you want to build your own or contribute, please wait until the our github page is up and running or contact bensc[ot]dtu,dk.

Figure 18: Do your own growth experiments.


We have worked very hard this summer to create the robot and achieved something that seemed totally out of reach only a few months ago. This would not have been possible without the help of other people. We would especially like to thank our supervisor Chris, Martin and Erik to get us started in the right direction. To see all credits, pleas go to the Attributions page.

Facebook Twitter
  • 2800 KGS. LYNGBY

  • E-mail:
Lundbeck fundation DTU blue dot Lundbeck fundation Lundbeck fundation