Here’s an extended cut of the video display (which is displaying an RGB Test Sequence).  Note that the current maximum throughput is approximately 12-13 frames per second due to the RS-232 baud rate bottleneck, but I’m looking for ways to speed up the data transfer without requiring a faster oscillator.:

The Processing 1.1 code:
VideoDisplay (includes the .pde file and the video used)

As I noted in the previous post, there was an error on the first revision of the board and a pull-up resistor on RA5 (pin 4 in the schematic above) was necessary.  I added the MCLR resistor to the board along with a couple of other modifications:

• Removed the extraneous capacitors, we only need one.
• Added a breakout for the one remaining I/O pin.
• Added a small perfboard to the PCB with +5V and ground lines.
• Relocated the resistors to make them much easier to solder.  I may use a SIL resistor array in the future.

I kept the LED locations exactly the same because, hey, if I put all of the time and effort into the 4 boards that I previously ordered, then I might as well keep the same form factor.  My scheme for the short-term is to create a 4×4 array of the 4 RGB LED Controller boards, which will give me a 20cmx20cm 8×8 RGB LED display.

Here are some pictures of the boards:

Here is the updated schematic and board (note, you can open the BRD and SCH files in Eagle Layout Editor):

• PIC16F628 4 RGB LED DR1r3 Eagle Board

And finally, here is an updated firmware that improves the PWM performance:

• 16F628 Serial 4 LED PWM – 4-bit exp DR1r8.c
• 16F628 Serial 4 LED PWM – 4-bit exp DR1r8.hex

I will make sure to post when I have the full array put together (I currently only have 12 of the 16 boards I need for the 4×4 array).

Some insipration came from the BlinkM “smart LED” and the ShiftBrite RGB LED Module, but I was interested in using RS232 serial control.  Therefore, I chose one of my favorite simple-to-use microcontrollers, the PIC16F628.  The advantages include the built-in 4MHz oscillator, hardware USART, and ease of reprogramming.  A couple of features I had in mind during the design:

1. Multiple intensities for each color (using PWM)
2. Multiple individually controllable RGB LEDs
3. High-speed update rate
4. Daisy-chainable and addressable
5. Simple serial control

Schematic
Here is the full schematic for the driver. I chose to use a PIC16F628 as the microcontroller because it is cheap, has a internal oscillator (4 MHz) and an internal USART. NOTE: There is an error in this schematic and a pull-up resistor on RA5 (pin 4 in the schematic) is necessary.  See the bottom of the post for an updated schematic and board.

Board
I decided I would try getting a PCB printed for the first time, so I got boards created at BatchPCB.com for $5 each.  The total for 4 boards shipped was $32.36 (4 x $5 for the boards and $12.36 for shipping and handling).  They took a long time to arrive, but the quality was well worth the wait.  NOTE: There is an error on the first revision of the board and a pull-up resistor on RA5 (pin 4 in the schematic above) is necessary.  You can see how I compensated for the mistake in the second picture below (look on the back of the upper-left board).  This will be corrected in future revisions.  See the bottom of the post for an updated schematic and board.

Control Format
The current firmware has 8 commands (the 9th, self-test was removed to save space).  See the source code for the firmware for how the commands are implemented, but here is some example usage:

• Turn off all LEDs: 0h2000FF
• Make all LEDs full-intensity red: 0h2F00FF
• Make all LEDs full-intensity red: 0h1F00F00F00F00FF
• Make all LEDs at address 1 half-intensity green: 0h207001
• Make LED 2 at address 1 half-intensity green: 0h100007000000001
• Load start-up settings for all controllers: 0h40FF

Commands:

1. Update individual LEDs using 4-bit exponential update
   Byte 1, Nibble 1 = led1Red;   Byte 1, Nibble 2 = led1Green
   Byte 2, Nibble 1 = led1Blue;  Byte 2, Nibble 2 = led2Red
   Byte 3, Nibble 1 = led2Green; Byte 3, Nibble 2 = led2Blue
   Byte 4, Nibble 1 = led3Red;   Byte 4, Nibble 2 = led3Green
   Byte 5, Nibble 1 = led3Blue;  Byte 5, Nibble 2 = led4Red
   Byte 6, Nibble 1 = led4Green; Byte 6, Nibble 2 = led4Blue
   Byte 7 = theAddress
2. Update all LEDs using 4-bit exponential update
   Byte 0, Nibble 2 = led1Red = led2Red = led3Red = led4Red
   Byte 1, Nibble 1 = led1Green = led2Green = led3Green = led4Green
   Byte 1, Nibble 2 = led1Blue = led2Blue = led3Blue = led4Blue
   Byte 2 = theAddress
3. Save start-up settings
   Byte 1 = theAddress
4. Load start-up settings
   Byte 1 = theAddress
5. Get address
   Byte 1 = theAddress
6. Set address (one-time set)
   Byte 1 = theAddress to save
   Byte 2 = theAddress
7. Disabled in this release.
   Self test
   Byte 1 = theAddress
8. Update individual LEDs using 6-bit update
   Byte 1 = led1Red;    Byte 2 = led1Green
   Byte 3 = led1Blue;   Byte 4 = led2Red
   Byte 5 = led2Green;  Byte 6 = led2Blue
   Byte 7 = led3Red;    Byte 8 = led3Green
   Byte 9 = led3Blue;   Byte 10 = led4Red
   Byte 11 = led4Green; Byte 12 = led4Blue
   Byte 13 = theAddress
9. Update all LEDs using 6-bit update
   Byte 1 = led1Red = led2Red = led3Red = led4Red
   Byte 2 = led1Green = led2Green = led3Green = led4Green
   Byte 3 = led1Blue = led2Blue = led3Blue = led4Blue
   Byte 4 = theAddress

Source Code
The PIC16F628 needs to be programmed with the firmware below.  The Windows application is included as an example of how one could control a 2×2 array of the PIC16F628 4 RGB LED PWM Controller boards (4×4 LEDs).

• 16f628-serial-4-led-pwm-4-bit-exp-dr1r7.c (hex)
• 16 Led Sequencer DR1r2

The first couple of prototypes have worked well, but I am still working on refining the PCB, circuit, firmware, and software design.  Here are a few pictures of the boards:

Update (01/06/2008)
Here is an updated schematic and board with the corrected MCLR pullup:

The driver boards utilize 3 AMC7135 constant current ICs, which supply 350 ma each, wired in parallel. I picked the driver boards from DealExtreme, but judging from the forum comments on the page, I’m not going to expect these to have very tight tolerances. I could have picked up 1400 ma boards, but the increase in brightness is negligble for the 33% increase in power drawn. Once I pull together MOSFETs to drive them I should be good-to-go.

I also finally mounted the K2 emitters on aluminum breakout boards/heatsinks that I purchased from SparkFun, although admittedly the SparkFun boards are way too expensive at $2.95 each. DealExtreme sells similar boards for $1.91 for a 5 pack, but they appear to be too small for Luxeon K2s… I purchased the SparkFun breakout boards using a gift certificate that I won from Hacked Gadgets, so I’m happy (thanks Alan!).

I removed the 1 Ohm resistor and found that the LXK2-PR12-L00 drew about 0.7 Amps (due to the internal resistance of the batteries), which is the max current for this emitter binning. Powering the emitter without a current limiting resistor is a non-ideal solution because different batteries (i.e. NiMH, Alkaline, NiCd, etc.) have different voltages and internal resistances.

Another consideration is heatsinking the emitter. The flashlight had an aluminum ingot that acted as a heatsink for the LXHL-MW1D, so I used a dab of heatsink compound to secure the LXK2-PR12-L00 emitter to the ingot. As the emitter heated up, it drew more current, which in turn heated it up more. Therefore, I’ve made sure not to operate the flashlight without a current limiting resistor for more than 30 seconds with a 50% duty cycle.

The LXK2-PR12-L00 emitter’s radiometric power is ~200 mW and has a dominant wavelength of 455 nm (min = 440 nm, max = 460 nm). What does this mean? Well, the radiometric power is a measure of the electromagnetic radiation’s energy. Because the human eye is less sensitive to electromagnetic radiation greater than ~700 nm and less than ~400 nm, some blue LEDs (like the Royal Blue), IR LEDs, and UV LEDs will use radiometric power instead of lumens for a more accurate statement of energy output. While the human eye is less sensitive to the colors, the total energy output can still be dangerously high (as is the case with IR lasers and UVB and UVC light). The wavelength of blue light is normally between approximately 440-490 nm. The Royal Blue emitter’s output has a shorter wavelength than the average blue light (465 nm), so it’s output is stated in radiometric power instead of lumens. The interesting thing about having a shorter wavelength is that it can make certain colors and pigments fluoresce.

Playing around with my modified Luxeon flashlight, I found that some yellows and certain oranges and pinks fluoresced. Here are some before and after pictures:

Afghan – Before (with fluorescent ambient lighting):

Afghan – After (with modifed Luxeon K2 Royal Blue Flashlight):

Magazines – Before (with fluorescent ambient lighting):

Magazines – After (with modifed Luxeon K2 Royal Blue Flashlight):

The LED emitters pictured below are:

3 x LXK2-PD12-Q00 (Red)
3 x LXK2-PM14-U00 (Green)
3 x LXK2-PB14-N00 (Blue)
3 x LXK2-PR14-Q00 (Royal Blue)
2 x LXK2-PR12-L00 (Royal Blue)
2 x LXK2-PW14-U00 (White)

1 x LXHL-MW1D (White Luxeon Star)

Here are a couple of pictures comparing the different LED emitters: