The BrightBoard is an ultra-bright LED panel for photography and videography. It is also a personal venture into the world of product design, and served as a series of important lessons in prototyping and engineering design. Let me list you the key features:
- 112 LEDs for a total flux of ~10k lumens (depending on input voltage)
- About 150 lumens/watt (rather efficient)
- Two color temperature LED arrays; 3000K and 6500K
- Color temperature and brightness can be controlled via Bluetooth
- Input voltage of 24V
The purpose of the BrightBoard is to be very bright; and to serve as a multi-purpose flood-light for studio and creative visual works. In the studio it can be used as part of a three-point lighting system for general video, as well as green-screening. Three-point lighting consists of a key, fill, and back light. Due to the BrightBoard’s adjustable brightness and color temperature, it could act as all three.
Another use is creative video works. Since the BrightBoard contains a programmable microprocessor, it can be used by filmmakers to set up timed lighting effects such as bright flashes and fades.
Any good electronics project starts with a solid schematic.
This schematic features an inexpensive Bluetooth SoC, E73-2G4M04S1B. Inexpensive, but a feature-full microprocessor which can be programmed to PWM the lights with a duty cycle value from a phone or other device.
The PWM frequency can be adjusted to several kHz to avoid flicker experienced by some cameras. Only a few hundred Hz is reasonably needed to make lights appear flicker-free to people and most cameras. The ability to PWM at several kHz is viewed as a huge bonus for potential customers.
The Bluetooth SoC can PWM the arrays through two N-channel MOSFETS. There are no FET drivers to save on unit cost. The FET chosen has a nominal gate charge, and I was confident that the SoC could drive the FETs directly. A pulldown exists on the SoC, but an external pulldown (or the pads for one) would have been a good idea. I have not tested FET thermal performance during PWMing, but I should upsize the FET package regardless (since the LEDs can get quite warm during normal operation).
Speaking of which, the LED array consists of 112 LEDs, split into two banks of 56. One LED array has a color temperature of 6500K, and the other is 3000K. A mix of these lights can produce a color temperature in between these two, which is a bonus to photographers or videographers who want to match the color temperature of existing nearby lights. Or, optionally, adjust the way skin tone is perceived on camera.
A 3.3V buck converter was used to provide the power to the micro. Voltage feedback can be used by the micro to guesstimate the power being delivered to the LEDs, and can be read back by the app. A barrel jack was chosen to provide compatibility with existing power supply units. Additionally, a snubber circuit was placed at the power supply entrance. The snubber values were identified using the “quick” method featured in this article. I’m a fan of finding the RC value experimentally, and the plan was to do it that way after the initial guess. The pads of the resistor and capacitor were oversized, just in case I needed to up-size something else later on.
The PCB size was chosen to be 8.25×11.5 inches, close to a US Letter sized piece of paper. This was mostly to provide space for the LEDs and heatsink space, but mostly to give the customer the impression that their LED array is very powerful.
The four “inner” mounting holes are intended to be mounted to a block of 3.5″x 3.5″x1/2″ thick hard plastic or aluminium. This block provides structural integrity to the light, as well as the 1/4-20 UNC mounting hole for a tripod.
The individual LEDs were provided with a large pad for heatsinking.
A prototype panel was assembled using a IR heater and hot air gun. No stencil was used; manual solder paste application only! Even on the tiny WSON 10 buck converter.
Problems and Lessons Learned
Problem 1. Buck converter blew up
The biggest problem I encountered was with the 3.3V buck converter. Everything worked fine on the bench with a lab power supply set to 24V (and even up to 26V!). As soon as I replaced the lab power supply with an inexpensive one meant for these kinds of applications, the buck converted immediately released magic smoke.
I suspected that the inexpensive power supply had a large voltage spike, and sure enough the oscilloscope confirmed by suspicions. The snubber was not enough to suppress the transient.
I wish I had an osciloscope capture, but I do not. Some sort of TVS diode or suppressor, or simply a higher-input voltage buck converter would have solved this issue.
Problem 2. Programming header was too small
Really simple problem, but the programming header was too big for my converters and adapters that I had. It made programming a pain.
Problem 3. The array is way too bright
Maybe this shouldn’t have been surprising, but this panel turned out to be way too bright. The panel has way too many LEDs. A 50% duty cycle is good for most applications. The next iteration needs to have less LEDs.
After using the panel for some general photography and video tests, I noticed the following things that could be improved upon:
- The length of the power supply I was using was way too short. The actual “brick” would hang in midair, even when the tripod wasn’t fully extended. The panel should really be battery-operated in order to prevent using a cord in the first place. If a cord is to be used, a longer one should be sourced.
- The panel is too bright. The number of LEDs should be reduced in order to save cost.
- The panel is also too big. The PCB itself is a large cost of the light, so the PCB should be reduced to save cost.
All-in-all, the idea of the BrightBoard is good, and the prototype is a very bright and useful panel. Lots of improvements can be made to the prototype to make it closer to a full featured product, starting with fixing Bluetooth functionality.
If you would like to see the project files you can find them on my GitHub.