Barnabybear
Thanks for the summary. It seems really cool and you must have alot of time and effort into this. Excited to see what comes next. I am an extreme novice here on this but with some learning, maybe I can offer some help. (not likely but I can try).
P. Short wrote:They're probably ignoring any possible issue.Over 6 feet there might not be too much of an issue, since the propagation time of the signal down the cable is about the same value as the signal rise/fall time, around 6 ns each. And the capacitance of the cable is perhaps 70 pf, so you might not even need a high-power driver for it (although ground-bounce might start to be a problem if there are a lot of simultaneous transitions). Things become much worse over longer distances, though, and even the facile solution (single-ended, high-power driver) might not be adequate as the capacitance rises and the propagation delays over the cable mount up. A lot depends on how sensitive the receivers are to ringing and/or smeared-out transitions on their inputs.
Hi, Phil.
At the moment there are 3 different approaches to using the SPI bus to drive more than one set of pixels. My personal angle is to use shift registers to drive all the pixel strings in parallel. I appreciate that this should be ok for 4 wire pixels, but becomes a little trickier for 3 wire (timing issues).
I am playing with the 74HC595 at the moment, but haven’t tried driving it above 1MHz (I estimate needing 16MHz for 8 strings and 32MHz for 16 strings of 3 wire pixels (2811s and thas if they will accept slightly adjusted timing). My thoughts are along the lines of MPH’s (but probably for very different reasons). That I would use a 74HC540 as an output stage. Quite simply – this is the same IC that is used in the E681 and in theory should offer the same operating performance (which has been accepted and work rounds formulated). However if MPH’s combined power supply / differential decoder becomes popular, it maybe necessary to offer the option of an ST485 (or something similar) as an output stage.
Null pixels are cheap and SMT (MPH’s board) isn’t yet something that people are happy messing with, but who knows what this year will bring. I can’t see me moving on to output and board considerations until the end of Februarry.
I don't think the PI will make the best pixel driver (but a usefull atribute). Its strengths are in data manipulation as outlined in a previous post.
Thanks for the insight.
Phil.
I just read this thread an my head is spinning. I got an Arduino LAST christmas that has come out of the box, once. For that reason I hesitated to jump on the Rasberry Pi bandwagon, but it sounds like I'm gonna get sucked in at some point this year.
Barnabybear, yes, thanks for the summary. I have to read it a few more times and hopefully more will sink in. I think a PI is somewhere in my near future!
I think it is always good to look at the wiring. I prefer to wire the control and power separately. Using simple AC cords to distribute the power is the best for me. Then I use the waterproof power supplies at the point of use. Also, I like the idea of running one control line to each group of lights. I am not much for daisy chaining.
If the run of the wire is short there does not have to be a buffer. However, if you want to be safe you can buffer all of the signals with 422 drivers. They are reasonably cheap and according to the spec you can drive 12 meters at 10MHz. My preference for driving control signals is either this or RS-232. You can easily do 50 feet with no issues.
My bridge for both E1.31 and SPI, that I am building is all surface mount, but I was going to leave off the buffers, so different drivers could be used. I figure the parts are about $30 with an additional $15-20 for the circuit card. It seems silly to add a distribution board to the PI that cost more than a PI. Is there a pain threshold for costs?

