Phil
You are right that a lot of work is going into higher channel counts, i am at 120ch and looking towards 512 from a single PIC next year, but, little or nothing is being done for individual local bulb control. Mind you copper is sort of cheap.
Some really good work is going on for RGB Pixel level control.
I like the idea and ultimately useful for more than just christmas lights, I'll have a few to run garden lights all year 🙂
Cheers
Aussiephil
Phil -
I like the idea. Curious if you have any more details on your idea. If going small we probably would like to use SOT-23 and other small SMT type devices. 10Fxxx have so little memory. it is a pity that 12F1822 is a future product. it might be small enough in a DFN and yet you get a lot more cpu power, memory, mssp, eusart in the package.
What protocol do you have in mind? something like 1-wire? or an in/out like renard?
what signal levels? just cmos or adding a level shifter?
what is the target bus dynamics? distance between nodes? max number of nodes? perhaps something like 128?
the output selection is probably the easiest - for low voltage DC maybe a small logic level FET? for AC a very small triac.
count me in for any help i can offer.
-Ed
j1sys wrote:Phil -I like the idea. Curious if you have any more details on your idea. If going small we probably would like to use SOT-23 and other small SMT type devices. 10Fxxx have so little memory. it is a pity that 12F1822 is a future product. it might be small enough in a DFN and yet you get a lot more cpu power, memory, mssp, eusart in the package.
Right now the project is similar in status to Renard. My focus has been on what I perceive as the hardest part, namely the firmware for baseline PIC. There are lots of different ways to incorporate the PIC into a controller, and that is more open right now.
The 12F1822 was pretty expensive the last time I looked at it. Besides, I've already written code for the 10F20x and tested it.
What protocol do you have in mind? something like 1-wire? or an in/out like renard?
The current protocol is in/out, more or less like renard. The character format is 38400, 8N2. The start of packet delimiter is 0x00, with data bytes ranging from 1 to 255 for dimmer levels 0/254 to 254/254 respectively.
what signal levels? just cmos or adding a level shifter?
what is the target bus dynamics? distance between nodes? max number of nodes? perhaps something like 128?
Again, there are a lot of different ways to do this. For my own project my current inclination is to go with just CMOS levels, subject to testing with actual lamps to see if there are any problems with noise. For my project I'm thinking of going up to a perhaps a few meters between nodes. The number of nodes would be limited to perhaps 128 by the communications baud rate (38400 / 11 / 20) with an update rate of 20 Hz or so.
There are some other related considerations - how a zero-crossing is created and (possibly) distributed, and how power is provided to the nodes.
the output selection is probably the easiest - for low voltage DC maybe a small logic level FET? for AC a very small triac.
Yes, or BJT or SCR. The current design caters to an AC application, so a small triac or SCR is the component of choice.
count me in for any help i can offer.-Ed
Despite what I've written above, I'm not completely fixated on this, and am curious to see other ideas. For all that I know there may be parts from Atmel or others that may be more suitable. And if these comments resonate with someone else to the point where they become obsessed (really obsessed) about some particular approach, I'm willing to switch over.
Phil -
Thanx for the info. i'll put the thinking cap on. i agree in/out is probably best, in essence signal regeneration. think maybe some sort of self clocking rather than classic serial might be better. i just wish they had a 10f400 with twice the flash.
12F1822? have you seen pricing advance notice. it is listed as a future project with no pricing on the web.
i would, in my design, probably assume DC LED only. that way we could provide ground, power, and a daisy chain i/o with no need for a zero-cross. also using FET or something to switch the power to the LED to remove the load from the processor line.
i agree a target of 128 or even 64 as a max in a string would be fine. we could then make a Renard to uRenard (micro renard) controller/driver.
-Ed
Phil, are you talking about something similar to Don's Chase-cicles with single SSR?
