Custom LED Flash Design

This page is the current central source of research and design documentation for the Modular LED Flash Unit project.

Background

The BenchBot Image Acquisition System and likely PlantMap3D applications, have the need for a relatively low cost (for PM3D at least) and waterproof (for both) camera flash system.

We have not been able to find a suitable COTS device. Existing industrial flashes are both too expensive $1500+ per unit, and not bright enough (RHI200-DO lighting for logistics - Smart Vision Lights).

The existing consumer grade flash we use (Godox Witstro Ring Flash AR400 AR400 B&H Photo Video (bhphotovideo.com) Is typically bright enough for Ag Image Repo applications, although in certain circumstances, such as very harsh daylight shadows, there could be a benefit to having a more intense flash. Although it is more than bright enough for PM3D applications, which have less intense image quality requirements. However, this type of flash is not suitable for long term usage in our application. First, there is no ingress protection. Second, it is hard to integrate both mechanically and electrically into a robotic system. And third, while a reasonable cost ($450) for BB/IAS application, it would not be affordable for PM3D MultiOak / cam where 4-8 units would be needed for each system.

Important Links

Design Files on GitHub

BOM on PartsBox

To Summarize:

Existing Consumer Flash Units

Awkward to mechanically integrate
Awkward to electrically integrate
No meaningful ingress protection
low durability / reliability
Still too costly for the multi camera systems

Existing Industrial Flash Units

Expensive ($1500+) per unit
Not bright enough for BB / IAS applications
Very few options / suppliers available
- Limits install flexibility

For these reasons we are exploring the design of an in-house flash unit to fulfill our needs.

Direction

The flash system will likely be LED for the following reasons:

LEDs are more physically durable than Xenon flash tubes.
A well implemented LED flash should last much longer than Xenon tubes.
LED flash systems require much lower voltages (50-100V vs. 300-500V) which helps size, cost, complexity, and safety.
Building a flash from multiple LED modules provides more flexibility than Xenon tubes which come in a limited number of sizes and shapes.
An LED flash can be controlled in a much more precise way than Xenon tubes, where if implemented both brightness and duration can be precisely controlled.

Initial research has found that there is a significant lack of information / standard designs with LED camera flashes. So, the design phase will be proceeded by a more significant research / prototyping stage than our typical hardware projects.

For the development of this flash system, we will need to characterize the output of LEDs in many different overdrive conditions. Typically, in electrical / electronics design we can utilize the components data sheet for this type of information; However, LED manufactures typically do not publish data beyond 200% output levels. But research and commercial products show that it is possible to significantly overdrive an LED chip without damaged for short periods dramatically increasing the light output.

Some References Related to LED Overdriving:

Testing Setup

For testing we need to be able to measure the output and duration of light pulses generated by our prototype and COTS units (for comparison). This is actually a very difficult measurement to make. Commonly available luminance meters have a very slow acquisition time and are thus only suitable for capturing continuous light output.

There are some photography “flash meters” available, but they are only able to measure total energy in a given period of time (thus negating any peaks or other time series behavior). And they do not report units in a very helpful way when trying to measure lux directly.

After considerable research, including investigating building our own meter, we purchased this device: ILT5000 Research Radiometer/ Optometer / Pico-Ammeter (internationallight.com)

We are in the process of building a 2' x 2' x 2' light box for testing of these flashes in a relatively controlled manner.

Preliminary Specifications:

Enclosure rated to handle outdoor environments for extended periods of time unprotected (weatherproof & IP 65 or better). What Does IP65 Mean? | OnLogic
Accepts common input voltages present in our systems (12, 24, or 48 VDC).
Design or resulting units are configurable for various applications (such as PM3D vs. AgIR).
An individual unit suitable for PM3D application will cost <$200 in quantity to build.
Incorporated proper heatsinking for good longevity of LEDs and continuous operation.
Form factor accommodates integration into out hardware systems.

Prototype Design

Here various design decisions and calculations will be documented through the protype design phase.

The first step is to choose what LED(s) will be used. There is a large variety of form factors and suppliers to look at, both individual properties of the products themselves need to be considered, as well as their form factor. Common form factors include chip LEDs that get individually soldered and large COB (Chip On Board) modules. What makes one a better choice over the others is related to several factors:

Light output vs. Area.
Ease of integration into a thermal management solution.
Optics availability and cost.
Labor needed to assemble units (more smaller LEDs may take longer to build unless they can be done on an SMT machine).
Cost in dollars per lumen output (brand / model dependent).
Efficiency (brand / model dependent).

After analyzing all of these factors and the current availability of modules the SPHWHAHDNM271ZT3DB Samsung Semiconductor, Inc. | Optoelectronics | DigiKey was chosen for preliminary testing.