LED Array PCB Design
The LEDs used are high power and 5-by-5 mm, featuring a lens that converges most of the light within 24 degrees (SMBB690D-1100-03, Ushio). Each LED is supplied with 2.4 V DC when powered on, with forward current of 600 mA and output power of 520 mW26. The overall size of the array is 33 x 33 mm.
The total current drawn from the LED array is about 3 A. To reduce the power loss and heat effect due to the resistance of the PCB leads, a large cross-section is required. With 4 mm lead width and 0.07 mm PCB copper layer height. The cross-section of the lead is equivalent to that of an AWG 23 wire. The voltage difference across the power lead is less than 0.04 V at 3 A, which is around 0.3% of the voltage across the penta-LED series. The power of the LED is provided by a computer power supply with the 12 V ATX output. This minimizes the Ohmic loss of the circuit and, more importantly, provides the same voltage to each LED to guarantee a uniform light distribution.
Two metal-oxide-semiconductor field-effect transistors (MOSFETs) are used as the switch to control the current. The two MOSFETs are connected in parallel to share the current burden. The nanosecond-level switch time of the MOSFETs is essential for PWM switching. Their relatively small profile enables them to be integrated into the PCB, making them the ideal choice over mechanical relays, despite the latter having a typically larger current rating.
The sensors circuit use the fixed-value resistors (5 kΩ for thermistor and 10 kΩ for phototransistor) to form voltage-divider circuits. The values are chosen to create output voltage ranges with magnitudes around multiple volts. This maximizes the resolution and avoids overshooting the 10 V measuring range of the DAQ digitizer.
On the PCB, the control lead connecting the gate of the MOSFETs is placed through the gap between the source and drain terminals. In this way, the entire circuit can be printed in one layer. This enables using aluminum as the backbone of the PCB, which conducts the heat to the waterblock effectively.
Wiring the Electronics and Water-Cooling Loop
The MOSFETs’ base pin is connected to the analog output port (AO0) of the DAQ controlled by the computer. The DAQ generates a 250 Hz square wave, with a 10Vpp amplitude and 100% offset, which is equivalent to a 0 V – 10 V pulse. With the 5 kHz sampling frequency, each wave period consists of 20 sample cycles. Therefore, the duty cycle of the square wave can change in steps of 5%, providing 20 different irradiance levels.
The sensors are connected in series with their corresponding resistors (5kΩ for thermistor, and 10kΩ for phototransistor). The outputs are connected to the analog input ports (AI0 for thermistor and AI1 for phototransistor). With the PWM control mentioned above and the common ground, the four wires forms cable 1 connecting the PCB and the DAQ.
Most of the components (the LED array, the sensors, the robot arm stepper motor drivers, and the water pump) use 12 V DC provided by the ATX power supply. The ribbon cable consists of AWG 14 wires, which conducts large current with minimal heat effect. The LED array PCB is connected to the dual + 12V output (conventionally used to power the CPUs) with the Molex connecter which brings ease when switching the board. The sensors are powered by the same 12 V power as the LEDs. This is achieved by the PCB circuit connection. The water-cooling pump and the stepper motor drivers are powered using the 12V and GND pins of the peripheral output with custom AWG 22 wires. An illustration of the wiring is shown in Fig. 5.
The stepper motors of the robot arms are connected to their corresponding driver modules with a proprietary 4-wire cable, through which both driving signal and power are transmitted. The driver module is controlled by the DAQ via low-voltage (5V) digital signals. A 4-wire cable is used to connect the PUL+ (to p0.1 on DAQ) for motion-stop control pulses, the DIR+ (to p0.2 on DAQ) to control the direction of the motion, and the grounds (PUL-, DIR-, to GND on DAQ). The ENA+/- ports are directly connected to the 12V power and the corresponding ground ports, as the motors are always enabled. The stepper motor-driver module for the other direction is connected in the same way (using the DAQ p0.3 and p0.4 ports respectively).
Optics Assembly with 3D-printed mounts
The combo of the LED array and the water block forms the source part of the optics. A 3D printed mount is used to hold the PCB and the water blocks and provide a flat bottom to mount the module onto the optical breadboard. Two more pieces of 3D-printed bracket are used to fix the position on the optical breadboard with ¼ inch screws. Another 3D-printed clip is used to align the PCB to the center of the water block to keep the optical alignment (Fig. 6). Therefore, the PCB can be easily changed if a change in wavelength is required.
The 2-inch Fresnel lens is held 1 inch above the LEDs by four cage system rods. The distance 1 inch is determined experimentally to maximize the light intensity at the sample plane while not compromising the uniformity. The rods are attached to the bottom of the platform through the through-hole flat screws.
Human epithelial ovarian cancer cell line NIH:OVCAR-3 (OVCAR3) was purchased from American Type Culture Collection (ATCC, HTB-161™) and cultured in T75 flasks (Fisherbrand™, FB012937) in a humidified incubator at 5% CO2 and 37°C. OVCAR3 cells were maintained in RPMI 1640 Medium (Gibco™, 61-870-127) with 20% heat-inactivated FBS (R&D Systems, S11150H), 1% penicillin/streptomycin (Fisher BioReagents, BP295950) and 0.01 mg/ml bovine insulin (Sigma Aldrich, I0516). Cells were passaged at 80–90% confluency; TrypLE™ Express Enzyme (ThermoFisher Scientific, 12604021) was used for lifting the cells. Cells between the 5th and 30th passages were used for the experiments.
Before plating cells for PDT experiments, cells were suspended at 20,000 cells/ml in the complete growth medium. 100 µl of cell suspension was added to 90 wells of a black-walled flat-bottom 96-well plate (PerkinElmer, 6055300) and 100 µl complete growth medium with no cells was added to the remaining 6 wells as the Media Only control. On day 4, old media was removed by inverting the plate. Then the drug is added to different treatment groups with media as shown in Fig. 2b. Each plate had 16 groups, including 12 treatment groups, three control groups (drug with no light (DNL), light with no drug (LND), and no light with no drug (NDNL)), and one media only (MO) group where no cells present) and all the groups had six replicates. The photosensitizer is added with 100 µL fresh media to all 12 treatment groups and DNL group, while the fresh media without the photosensitizer is added to the LND, NDNL, and MO groups. The well plate is then incubated for 1.5 hours before performing PDT.
Performing PDT and Viability Measurements
In the effectiveness comparison experiments (PWM LED vs continuous LED, and PWM LED vs laser), the PWM controlled, LED-based therapy is performed by the setup mentioned above. The continuous LED therapy is performed by replacing the computer power supply with a variable DC supply and keeping the PWM duty cycle being 1. The laser therapy is performed by replacing the LED array with a fiber-coupled diode laser (ML6500, Modulight). The output of the fiber is placed on the focal point of the Fresnel Lens. Due to the power limitation, the PWM LED vs laser experiment is performed at 50 mW/cm2. The dose applied to each series are 2, 4, 6, 8, 10, and 12 J/cm2. The LND control is treated with the PWM controlled LED with a dose equal to the maximum dose applied to a treatment group (12 J/cm2) and power equal to the treatment groups (100 mW/cm2 for PWM vs continuous, 50 mW/cm2 for LED vs laser). A video demonstration of a routine use of this setup is available at https://youtu.be/OJX8PF69tTI.
In the effect of power experiments, the optics setup is the same as the PWM vs continuous test mentioned above, except the robot arm is not being used. The light power applied to each series is 45, 90, 135, 180, 225, 270 mW/cm2, and the light dose is 8 J/cm2 for all groups. The light power is measured by the power meter before performing treatment on each group. The LND control group is treated by light dose 8 J/cm2 at irradiance 270 mW/cm2.
The well plate is incubated for 72 hours after the illumination. Then the cell culture viability was measured using CellTiter-Glo® Luminescent Cell Viability Assay (Promega, G7570). The plate and its contents were equilibrated at room temperature for 30 minutes. CellTiter-Glo reagent was formulated by mixing CellTiter-Glo® Buffer and CellTiter-Glo® Substrate equilibrated at room temperature. 100 µl reagent was added to each well and mixed the contents on an orbital shaker (DragonLab, SK-O180-E) for 2 minutes. The plate was then incubated at room temperature for 10 minutes and the unfiltered luminescence was recorded by using a plate reader (BioTek, Synergy LX Multi-Mode Reader) at 1 second integration time. Plate reader data was further analyzed by using GraphPad Prism 9. NDNL control group was used to define 100% viability and MO group was used to define 0% viability. EC50 values were determined using a nonlinear fit (Inhibitor vs. Normalized response – Variable slope) on GraphPad Prism 9.