Team:Grenoble-Alpes/DryLab
First of all, what does EWOD mean? EWOD stands for Electrowetting On Dielectric. This is a way to move small drops over a dielectric (insulating) plate. If you want to learn theory more about it let check over there. #TODO(link) To sum up, EWOD is a way to move small volumes such as tears. By applying a high voltage on pads under the drops the drop is distorted toward the electrified up to a point it easier for it to move onto the pad than to stay on it’s own pad. We chose EWOD because it is a lossless solution of moving droplets. Indeed there is always a loss while pipetting a solution and, as we are working with small quantities, we cannot afford these losses. Moreover we wanted to develop an automated devices which is possible thanks to EWOD.
In order to have a fully controllable electrowetting plate, a component capable of activating (enabling or disabling) pads one by one is needed. It is the HV507 of Microchip, or it’s soft surname according to it’s datasheet, 64-Channel Serial-to-Parallel Converter with High-Voltage Push-Pull Outputs, that have been chosen for several reasons :
Which in substance mean it can deliver a high tension over 64 pads, can be piloted thanks to an arduino, can memorise it’s outputs, has an easy way on blanking it’s outputs and can be chained. The last point is very important in our case because the design of the electrowetting plate is this one :
Each color represent a different HV507. Indeed, to automate as many processes as possible, a big number of pads is needed. Blue and Red zone correspond to the zone that are temperature controlled. The orange one is for light detection and finally green and yellow one are for storage of the drops when they do not need to be at a set temperature nor need their brightness to be measured.
As you may see, each pad has a number that unique and constructed as so : “x.y” x (ranging from 1 to 5) is the identifier of the chip in the chain and y (ranging from 1 to 64) correspond to the output and thus the pad on the chip.
As the functional block diagram from the datasheet shows it, data flows in from DIOA to DIOB through the 64bits static shift register. The data are then shifted from the static register to the latches (mini memory) on a rising edge of *LATCH ENABLE*. Finally the outputs are *calculated* based on the content of the latches and depending of the *BLANKING* and *POLARITY* inputs. The first one is used to reset all the outputs and the *POLARITY* input is used in order to say if a *0* correspond to a high or a low voltage. As the functional block diagram from the datasheet shows it, data flows in from DIOA to DIOB through the 64bits static shift register. The data are then shifted from the static register to the latches (mini memory) on a rising edge of *LATCH ENABLE*. Finally the outputs are *calculated* based on the content of the latches and depending of the *BLANKING* and *POLARITY* inputs. The first one is used to reset all the outputs and the *POLARITY* input is used in order to say if a *0* correspond to a high or a low voltage. Finally, all theses component have been routed onto Kicad in order to give the following result : As you may now, water is most of the time conductive, which mean that if a drop overlap 2 pads, current would flow through as the drop would close the circuit which is not the effect wanted. Dielectric is used as a protective layer over copper pads. However, this protective layer must have some particular properties:
These are the reasons why we chose cellophane paper, which is really thin and cheap, as a dielectric. Moreover, we coated it with some water repellent such as Rain-X. Finally a thin layer of nut oil is used to easen the gliding of the drop. One of the most time-consuming tasks in the lab is to pipette samples. As this action is also paramount for the success of any operation, we have decided to automate it. We proceeded in a similar fashion than the 2018 iGEM Grenoble team, but we went a little further by separating the command of the pipette from the pipetting module. As said in the introduction, we based the automation of the pipette on last year’s iGEM Grenoble team. We took the same device as they did: a BioHit eLine pipette. Then we just added an electronic device that could simulate the press of a button. A button is basically a wire that can be connected or disconnected by pressing it down. When we press the button, the difference of potential before and after the button is pressed, is zero (V2-V1=0), but when the button is not pressed the difference of potential is not zero (its value depends on the rest of the circuit). To simulate the press of a button, the idea is then to replace the buttons by an electronic component that could be switched on and off with a command board (the Arduino). By measuring the potentials on the pipette, we saw that for each button, one of the terminals was at ground potential. There are then two ways to command the pipette: either we shortcut the two terminals, or we just command the tension on the button’s terminal that is not to ground. Last year’s team decided to go with the first option. The main downside of this solution is that we need two times more cables (two cables for every button). The solution of imposing a voltage only requires one cable per button, and this is the main reason why we went with this option. The tension we need to impose is either the ground’s tension, or another tension (provided it is superior to half of the power tension of the pipette). In our case we just had to impose either 0V or 5V compared to GND. Those two tensions are natively accessible on the Arduino board. We decided to control the tensions sent to control the buttons through an electronic device called an optocoupler with numerical output: the H11L1. This piece of technology is commanded via the arduino and returns either 5V or 0V. Here is one H11L1: On the final board, we are using seven of them to control the seven buttons of the pipette: Each chip is connected to two resistors. One of those resistors is meant to protect the device (the 1kΩ resistor) and the other one is called a “pull-up resistor” (the 270Ω resistor). A pull-up resistor (or pull-down resistor, depending on what it is connected to) is a resistor connected between a cable which tension can vary and a point at a fixed tension. In our case, it connects the output of the optocoupler to 5V. When the output is not commanded, no current flows in the resistor and the output of the optocoupler is therefore 5V. The pipette is a device with two main parts: an electronic board and a motor. The electronic board controls the pipette and acts as the main interface between the user and the pipetting mechanism. It is composed of a screen and 8 buttons (one of them being behind the pipette, and therefore invisible on figure 1). The user starts by reinitializing the pipette by pressing one of the two “tip” buttons. Then he selects the volume he wants to pipette using the buttons, and then he presses the start button (in blue), the same button is used to pipette and release the liquide pipetted. At the end of the experiment, the user ejects the tip by pressing once again on the “tip” button. The command board detects each time a button is pressed and activates the motor to pipette or eject the right amount of fluid. In order to control the pipetted volume, the pipette uses a crenellated wheel. This wheel is placed over a LED fixed to the motor, and turns in front of two phototransistors on the command board. The signal received is then transcribed into steps for the command board so that the pipette knows precisely the amount of solution it has pipetted. As we mentioned earlier, we can replace all the buttons by optocouplers. This allows us to automate the pressing of the buttons. We also decided to separate the electronic board of the pipette from the motor so it would be easier for the motor block to move. In order to do this, we had to desolder the phototransistors behind the electronic board of the pipette and create a secondary board where to put the phototransistors, that would then sit right above the crenned wheel and the LED. This being done, we designed a new case both to protect the motor and to fix it securely to a rail that moves vertically. In order to move the pipette, we will use a mechanism that will transform the movement of a motor into a linear movement. The movement is transformed from a circular movement to a linear one thanks to a screw-nut mechanism. In order to make the movement steadier, we added two bars to guide it. The pipette will be fixed both to the rails and the screw via a nut, thus allowing it to move only in one direction. More information on guiding the pipette will be given on the “structure” part of this wiki. In order to control the pipette, we have first to identify the different actions performed to pipette a sample. This protocol will then be translated into a sequence of actions that the hardware will perform when the order is given by the software. This method is functional but shows one major downside: we don’t really know what is happening inside the pipette. Therefore we can’t really offer any ways to repair the device if anything happens to it. This is why we decided to implement a second device, using a different pipetting method that is a little less precise but much easier to repair in case of failure: peristaltic pump. This second device, that we finally adopted for the final dispositive, is presented on another page of this wiki. We have seen on another page of this wiki how we could use an automatic pipette in our device. However this solution shows one major downside: we don’t know how the pipette really works. This means that in case of failure, nothing can be done. Another downside of repurposing a used automatic pipette is that it is not made for this use and although it works, the way we make it work is far from being perfect and would never suit a classic laboratory device. For all those reasons, we had to develop another solution for pipetting the solutions. This is why we created a pipetting device using a peristaltic pump. A peristaltic pump is a device that can pump fluids using a motor that will press a tube. The idea is that by moving the pressure point, the pressure inside the pump will vary and entail a movement in the fluid. Those pumps are common on laboratory devices, mainly to inject precise amounts of solution into other solutions automatically. They are also used in cardiopulmonary bypass. In our case we want to create a system that will pump a precise amount of solution. Before anything let’s model our solution. Before developing this solution, we created a theoretical model of the device. The idea of this model is to get a good idea of the precision we expect from this device, and to validate its use in our system. The first step is to fix the different parameters of the pump. Let’s therefore create a schematic of this device: The different interesting parameters in this model are:
The physical parameters to take into account are:
With those parameters taken into account, the idea is quite simple:
In our case we used a tube with an inner volume of 3mm. The length of this tube is 4cm. The pump is connected to a step-motor (NEMA 17) that will be driven in microstep mode. This means that we’ll be able to control the motor with a precision of 1/6400 of turn. In terms of degrees this is is equal to 0,05625°. The length of the tube between the pumping mechanism and the solution is of 4cm. To get to the compressibility of air, we usually approximate the air to an ideal gas. This means that we will consider that the pressure, the temperature, the quantity of gas and the volume of the air will be linked by this formula: \(pV=nRT\) This model is close to reality at atmospheric pressure and normal temperatures. With this model, we get to the following expression of the compressibility of air (isothermic):
$$\chi_T = \frac{1}{V}\left(\frac{\partial V}{\partial P}\right)_T = \frac{P}{nRT} \frac{nRT}{P^2} = \frac{1}{P} $$
Now, when we consider our device, the maximum we might need to pipette is 20μL, which. This means that the increase of pressure due to the presence of the solution will be of \(20mg*\pi*(3mm)^2=0,05Pa\). This means that the variation of pressure in our tube due to the presence of the solution will be 6 orders of magnitude lower than the atmospheric pressure. It is therefore neglectable. In fact, the error we make with this approximation is of the order - in our application - to (at most) \(210^{-5}\mu L\). Now let’s see how low we can pipette with our device. We said that we had a tube 4cm long, of an inner diameter of 3mm. This means that the volume of the tube at the pipetting position is \(1130,94 mm^3=1130,94 \mu L\). Now, we know that we pipette all this volume in half a turn (it’s obvious when referring to the schematic). This means that we can pipette with a precision of 1/3200 of this volume. This leads to a precision of 0,35μL. As we need to pipette solutions which volume varies from 10 to 20μL, we are precise enough. Now let’s look at the pump itself. In the case where we had to pump 20μL at once, we would need to exert a force of around \(20mg*9,81m*s^{-2}=0,1962N\). The drivers that we are using needs around 1μs to perform a step. This means that to pipette 20μL, we’ll need around 57 steps, that is 57μs leading to a needed power of \(0,196N*0,7mm/57µs = 2,4W\). This is way under the power that we can deliver to our motor (we tested the device with powers ranging from 5 to 24W). What we can conclude from this model is that our device allows us to pipette the volume of solution we want with an error of around 0,35μL. If we want to pipette 10μL, this represents a relative error of 3,5%, which is more than acceptable for what we want to do. One simple way to increase the precision in our device is by adding a reductor between the motor and the peristaltic pump. With a simple 1:10 reduction, we would achieve precisions of 35nL. We described on the software part of the wiki, how we were making the Arduino pipette a solution. We chose not to modify the software to develop this solution. Changing the software on the computer was one simple solution, that would have only asked to replace a few lines, but we decided to test the adaptability of the command board’s program. We therefore only modified the code written on the Arduino. This modification asked us to replace around 10 lines. This proved to us that we could easily adapt our program. This was one of the goals of this project: create an open-source project that could be re-used as a complete basis for any new autonomous machine. In terms of pipetting, here are the tests we did. We can clearly see that the volume of solution being pipetted by a controlled pipette and by our device are indistinguishable. This proves that our system works, and represents a real alternative to other pipetting systems. To ensure that the bacteria do their job in detecting the biomarker, they need to be in a temperature controlled medium. The protocols we are using, require a constant pre-fixed temperature, therefore, we are going to control the temperature of the EWOD plate on which the experiments take place. So that the temperature of a zone is maintained, we are using a loop control system. The user enters a temperature as a set point or uses the pre-defined protocols that contain the temperature set points. The system then adjusts its variables automatically to maintain the set point through the feedback loop. The heating and cooling of the zones is done using peltier modules. The actuator used to heat up or cool down the zones is a Peltier module. It is a thermoelectric heat pump; it transfers heat from one side to the other depending on the direction of the current passing through it, using electrons as the heat carrier. It works based on an effect called, Peltier effect, which was discovered by the French physician Jean-Charles Peltier in 1834. The module consists of two different conductors linked by two junctions, when a current passes through one conductor, a thermic flux appears and starts to heat up one electrode and cools down the other; this phenomenon is related to the flow of entropy. Peltier modules are small, compact and can deliver stable temperatures, which makes them perfect for our application. To measure the exact temperature of the zone, a temperature sensor is used. The chosen sensor is TMP1075. It is an integrated circuit that uses a thermo-resistor, which delivers a digital data that can be interpreted as the surface temperature by an Arduino code. The sensor is placed on the temperature-regulated zone, to have a maximum accuracy. The integrated circuit has an on-chip 12 bits analog-to-digital converter that provides a resolution of 0.0625°C. The data is delivered to the Arduino card via the I2C bus, where it can be processed. Once the temperature of the surface is measured, it is then compared to the set point. The output voltage to be delivered to the Peltier module is computed according to that temperature difference. The calculations are done by a digital PID controller (proportional-integral-derivative action). The coefficients of the PID are tweaked to have an optimal result in regards to the stability, rapidity and precision of the system. The PID delivers an integer that can be positive or negative to indicate in which direction the current must flow and the power to deliver. These instructions are given to an intermediate circuit that manages the voltage and the current’s direction. This circuit is called an H bridge (in our case we used the L6206). It is an integrated circuit that from a voltage ranging from 8V to 27V, delivers an output voltage that will be proportional to the instruction given by the Arduino with the corresponding current direction. This allows us to control the power sent to the peltier modules, on average, and during short periods of time. This way the peltier module will heat up or cool down the zones to match the set temperature. These instructions are given to an intermediate circuit that manages the voltage and the current’s direction. This circuit is called an H bridge (in our case we used the L6206). It is an integrated circuit that from a voltage ranging from 8V to 27V, delivers an output voltage that will be proportional to the instruction given by the Arduino with the corresponding current direction. This allows us to control the power sent to the peltier modules, on average, and during short periods of time. This way the peltier module will heat up or cool down the zones to match the set temperature. Photomultipliers used to be very expensive devices which required thousands of Volts to amplify photons. The evolution of semiconductor physics has led to the creation of similar devices in their characteristics, without the need of very high voltage, and with more reasonable prices. Built on silicon, theses are called Silicon Photomultipliers (SiPM). Those devices are usually much cheaper than the former tube photomultipliers, as the one we used costs around $60, which is actually cheaper than some cameras. Another advantage of using this kind of device is their size: a SiPM usually presents itself as a square of a few millimeters side length. A Silicon Photomultiplier converts light into an electrical current. This is done by creating electrons with incident photons. Those electrons will then move in one direction, imposed by the tension applied to the SiPM. The movement of the electrons will lead to collisions with other electrons, thus creating larger flux of electrons. This phenomenon is called quantum quench. The conversion of the light into a current is done with a specific type of material called a semiconductor. A semiconductor is a material that alone acts as an insulator but that is close to being a conductor and that can effectively become one if put in the right electromagnetic conditions. Using the right configuration, light can be used as an activator to modify the semiconductor so it becomes conductor. This leads to a flux of photons inducing a flux of electrons. This phenomenon is used in phototransistors. With photomultipliers something similar is done: a flux of photons is converted into a flux of electrons and then amplified using quantum quench. The SiPM creates an electrical current proportional to the light received. Therefore measuring this current is directly equivalent to measuring the brightness of the bacteria. The component used to perform this measurement is a current sensor from TI, the INA226. It has been chosen because :
The current emitted by the SiPM goes through the Shunt resistor (Rshunt) that has a known value. As the current measurement need to be very precise, very precise resistor have been chosen (with values accurate to 1% when usual resistors have 10% accuracy especially on low value). The INA226 measures the voltage drop due to Rshunt and stores it in its voltage register. As it has been initialised, it knows the value of Rshunt and thank to Ohm’s law, it can compute the current passing through : \(I_{SiPM} = \frac{U}{R_{shunt}}\) Finally, the data is requested by the arduino through the I2C connection and thanks to the arduino library INA2XX. Data can then be stored for further analysis. As it has been shown, in order to detect the signal emitted by the bacteria, a very sensible sensor called a Silicon Photomultiplier (SiPM) needs to be used. The signal it returns is a current that will be measured thanks to a current sensor : the INA226. Finally, the data can be treated on your computer using NeuroDrop’s software.
NeuroDrop
EWOD Plate
Why EWOD?
How to make an electrowetting plate?
How to pilot the HV 507?
What about the dielectric?
The Pipette - reusing a pipetting device
What we started with
How the pipette works
What we did to automatize the pipetting
Algorithm to automate a pipette
BEGIN
The user enters the solution he wants to pipet from
The user chooses a predefined protocol
The user enters the volume he wants to pipet
The pipet is reinitialized (by pressing on the “reinitialize” button and then the “tip” button)
The volume that the user defines is applied to the pipet (by pressing a certain number of times on the “+” or “-” button, and then by pressing the “enter” button)
The pipette is placed over the tip used to pipet into the desired solution
The tip is placed on the pipette (by lowering the pipette and then raising it again)
The pipet is placed over the solution we want to pipet from
The solution is pipetted (by lowering the pipette, pressing on the “start” button, and then raising the pipette again)
The pipette is moved to the EWOD plate
The solution is dropped on the EWOD plate (by lowering the pipette, pressing on the “start” button, and then raising the pipette again)
The pipette is moved over to where it first took the tip
The tip is ejected (by lowering the pipette, pressing on the “tip” button, and then raising it again)
The pipette is moved over to its first position (over the EWOD plate)
The initial volume (5µL) is applied to the pipette (by pressing a certain number of times on the “+” or “-” button, and then by pressing the “enter” button)
END
Conclusion on the method
The Pipette - using a peristaltic pump
Peristaltic pumps
Model
Method
Parameters of our device and physical considerations
Conclusions on the model
Possible improvements
Final result
Temperature control
Control process
Peltier modules
Temperature sensor
PID controller
Peltier control : H bridge
Overall conclusion
Light detector
Photomultipliers
Current detection
How does this work?
Conclusion and overall strategy