Zeph wrote:A typical 5v pixel runs about 60ma each, or 3 A for 50. The most one would want to run without power injection is 60ish, or about 3.6A.
A typical 12v pixel runs about 30ma each, or 3A for 100, 3.7A for 128 (longest string Ray Wu sells, also RJ's maximum)
Ray's 45W waterproof 12V ps is 3.7A (45W).
The pixabulb is apparently higher current, but 3.7A should still handle about 70 @ 50ma.
The connectors I'm considering are rated for 3 or less than 4A.I think 3.7A or so would cover the most common use cases, but 3A would be fairly good. If you really want more than that, it's not unreasonable to use an external wiring harness and not go through the traces.
Ill try to see, 3.7-4a will need the traces to grow to 65-72 mil
For the power jack, for most of my pixels I have the 2.1mm jack on the power already. What about a hybrid?
With a little coercion I got the terminal to fit in the slots for the dc jack?
Zeph wrote:
It sounds like you are considering running at 5V. Is that mostly to be within ATMega328P spec's at 16 MHz, or to give a 5V swing to the outputs for driving pixels?Has that been tested? (Using a Zener to power the transceiver from 5v). My concern would be that it's not really regulated, so I hope the power would be steady enough not to interfere with the transceiver operation.
Real world diodes have a current to voltage curve rather than a cliff, so if the load varies there would be some voltage variation too; I'd be cautious about dropping that much in the diode.
Is there any chance of putting components on both sides?
Re 5v : its for you ;).
I have some 7805 and diodes so I was thinking of wiring one up on a breadboard and seeing how it goes. I'm not the best to tell if its ideal as I am without oscilloscope ;( (come on Craigslist .... Any day now ....) but I can see if it works. With there already having regulator on the circuit I was hoping the diode would do it. At $ .04 even if it needs a cap it would be cheaper than a second voltage regulator.
I was also thinking of 5v so that the expansion cards I'm planning have it as an option.
I know my 5v wall warts have heatsinked 7805 3.0a in them so I guess I could have omitted that all along?

dmcole wrote:I guess I'm still too dense ... why would you run the power for the pixels through the voltage regulator (assuming this as for why the x7805 would get hot)?I thought the MCU controlled the signal only and you'd send power down the line without passing it through the MCU?
Or am I way off base here?If all the voltage regulator is powering is the MCU and assorted other on-board bits, why would it get hot? (If it is powering pixels, I could see why it would get hot).
\dmc
PS: I realized while pondering these questions that we've never seen schematics of this project.
You may be right, I was worried about the heat coming from dropping 12v -> 5v. The pixels arent running through the voltage regulator.
schematic coming soon - almost done.
ok here is the first cut at what I am thinking.
Update: missing the series resistor on the Zener diode.
Per the Nordic spec the Rx mode of the radio is only using 12.3 ma when in receive mode. I adjusted for some room and used my closest value of 100 ohm.
I bypassed the regulator on a rf1 board and fed 12v into the header where the pixels would normally be and used DuPont cables to feed the nrf24l01 header back to the breadboard. There the series resistor and diode.
I got close but even though my diode is a 3.3 I was still seeing 3.6 on my dmm. That was too close to the max voltage so I didn't do a test 🙁
its late so ill redo the math tomorrow, AFTER We celebrate mom!
It would be useful if one could get by with using a simple Zener diode to power the 3.3V transceiver from the uC's 5v supply - small and cheap. I'll be interested in your breadboard testing results.
How were you thinking to check? Do some packet reliability testing and see if it's just as solid as with a 3.3v regulator?
I may be misunderstanding about how Zener's work, but I would have thought that if you put a 3.3V zener between the 5V supply and the transceiver, that would leave only 1.7V for the tranceiver.
---
Being able to use either a 2.1mm barrel jack OR screw terminals in the same holes? That's genius. Or at least very clever.
---
If the 5V regulator is supplying only the uC and the tranceiver, then how much current are we talking? I would have thought a modest amount for the '328P, some for the output pins driving the pixels, and some for the tranceiver. Would that overheat the regulator if dropping 7v? (or 8.7v if running at 3.3v?) How much power would the regulator need to dissipate? Hmm, should take into account that it's not rare for a 12v supply to run 5% or about half a volt high when under light load, so do the heat calcs with 12.5v input.
---
I am mixed about running the uC at 5V and 3.3v, actually. Jeenodes supposedly run well at 3.3v/16MHz, and some say the pixels are OK (at least at modest feed lengths, which is what the wireless options buys ya). The reason I want the extra pins however, is that I'd like to be able to run additional sensors or actuators from the node, like say a PIR sensor (this is also part of why I expect to run customized firmware and a more flexible over-the-air wireless protocol). Some sensors like 3.3v, some like 5v.
I see this as oh so useful for many things beyond lighting control. I'm thinking of using these or something similar to count metering pulses from the irrigation system, for example.
---
How much over 50mm is it?
I wonder if any space could be saved by exporting the reset button? I'm thinking that the only time you really need the reset is if you are programming it (you can power cycle if it's glitched, yes?). You need to add external hardware to program it anyway (eg: FTDI cable) What if you temporarily connected a reset button at the same time, with Dupont connections just like the FTDI interface? Or you might be able to just short two pins together, if you are flashing software which is not likely to need much debugging, like changing a #define to another established value. (I'm not sure this is a correct analysis, putting it out there to see if it makes sense to others).
---
I have to watch my words around you - it's like having a genie in the house. I brainstorm an idea (which could be half baked), and the next thing I know, it's already in the new revision
Some context...
I just received one of these from Ray:
Some of my (proposed) display involves 1/2" PVC pipe. I was thinking to hang wireless controllers on the pipes here and there to control pixels in their vicinity. The above power supply would strap onto an upright pipe (to keep it off the ground) without being terribly bulky, and could then feed the uC housed in another pipe segment also attached to the upright pipe.
Some of these will be right next to the walkway, and I was thinking to use sensors to detect people - PIR, ultrasonic, photo beams - the actual sensors may evolve over time with experience. There could be actuators as well, like servos. So it would be cool to run a few more wires out from the uC running the pixels, and have it also report sensor data wirelessly.
I have a mixture of pixels strings and some strips, and of 5V and 12V, 4 wire and 3 wire. I'd like to make wireless nodes which can be used for any of the FastSPI_LED2 supported pixels at 5V or 12V, with at most a reflash of the firmware and/or a cabling change - but even those might not be needed. One reason is so that I can have universal spares, able to replace any node without excessive work.

