Yes, the internal oscillator is the reason I selected the A6281 over the A6280. Also, the 3mm x 3mm QFN is small enough to fit inside the LED leads and enables a compact design (the limiting factor is really the headers). And when you're ordering a few thousand of something, DIP versus QFN isn't a factor except in PCB real estate, because you're not worried about ease of hand-soldering.
The flicker while dimming I *think* is a result of the PWM oscillators out of sync, although the datasheet claims that a latch event won't actually reset the oscillator. Regardless, I found that pulsing the Enable line during a Latch pulse would smooth the dimming fade immensely. That removes the need for a continuous ~1MHz PWM pulse train. But there are some situations you want a synchronized pulse train anyway, like video display. The A6281 actually lets you select the external clock and do that too, just like the A6280.
Nothing about the ShiftBrite is really a secret, it's just a breakout board for the A6281 with an included LED and preset current control resistor. I certainly have no problem with people deciding to build their own through-hole version if they have a cheap source of PCBs and some quiet winter months to solder a few thousand pads. The ShiftBrite is priced where it is because of the convenience, I still get people buying them in 100+ quantities because they need the function right now, and don't have time to design their own board, wait 2+ weeks for the PCBs, source all the parts, and spend a week soldering and testing. These are people who will calculate the hourly cost of their time when considering their options.
For hobbyists (like me, actually) who really enjoy the challenge and sense of accomplishment brewing devices like this up from scratch, the extra hours can be counted as leisure time and you might actually come out ahead building your own.
Anyway, feel free to use me as a resource for running ShiftBrites or DIYBrites(?); cabling ideas and sources, coding, interfacing to various controllers, etc. I've been primarily using the Arduino as a controller, it could easily serve the purpose of your outboard shift register and PWM clock generator. It has an embedded USB to serial converter (on the official boards) so it's a ready made interface to any PC.
Thanks for your interest and discussion, I had not actually considered the holiday lighting market when designing these. If you have suggestions for an improved device tailored to this type of application, it's very possible for me to create a customized version.
Garrett, Thanks for the follow up! Even after getting some samples to play with, the DIP version is still extremely challenging to work with considering the clock that you need to create. Even though I've started this "DIY" version, the pricing scheme to buy the ShiftBrites themselves is still very attractive, considering the overhead that you don't have to worry about, the easy to use breakout board with headers, etc... Now if I could only hand-solder up the QFN packages 🙂
In other development updates.... I'm starting to use the CCP module along with spi_write to create the appropriate clock and data sequences. I don't yet have a working example, but I hope to see something soon. I need to get my head wrapped around PIC interrupts.
By chance are you still working on this project? I have this idea to cover my house in about 50 frosted plastic Snowflakes http://www.partycheap.com/Sparkle_Snowflake_11_in_p/20370-11.htm (or something similar) and light each of them up using Shiftbrites. The idea being that I could control each one separately to create chases and fades across my house in different colors using Vixen. My control PC is located in my workshed, about 30 yards from the house.
Given the above info. Do you think it's possible using your idea to control the Shiftbrites via the parallel port over about 100 yards of cable? And if not, what do you think it would take get Vixen to do what I'm trying to do?
I'm aware of the JEC Pixels, but the complexity (to me at least) of DMX is not something I'd like to try and take on. It seems very simple to use the parallel port, but I worry about the reliability of the signal over that great a distance.
Your thoughts, and work on this project are greatly appreciated!!
crperryjr wrote:By chance are you still working on this project?
Not sure what is happening with that project, but you might want to check out the project that JEC and I have been working on at http://response-box.com/rgb/2009/06/micro-pixels-first-test/ . The new 'tripix' pixel are three rgb leds and an A6281 on a board, complete with connectors and the like. These things are really bright. We are just completing a controller unit at the moment, which will allow DMX control of them. If you wanted to do a differnet kind of control you could as well no problems there. The construction of these is 'fiddly', so we have had them factory made. They will be avaiable for sale ready made very shortly.
After having too many problems trying to work with the DIP version of the chip, I took a break from this project. In my attempts to make the allegro chip work, I found it much easier to just program a 12f PIC to use PWM and do the same exact thing, with much more flexibility.
In the same note, running a parallel port signal over a hundred yards or so isn't going to be your best friend either.
