Color Changing Christmas Light Project

[DynamoBen]

John is a contributor on the PicBasic forum. He created a DMX controlled RGB light string. Thought you guys might be interested.

http://www.response-box.com/rgblights/index.shtml

[A Marchini]

John is a contributor on the PicBasic forum. He created a DMX controlled RGB light string. Thought you guys might be interested.

http://www.response-box.com/rgblights/index.shtml

Cool, I remember seeing that project, I believe , when he first started. I remember that he only had an appartment balcony at the time. It is good to see that he continued on with the project and got a house to decorate with it.

T.

[JEC]

Hi Folks,

Tony sent me an email mentioning this site and the link to my project, so I thought I'd take a look.

I've attached a screenshot of my current design, which may be a couple revs newer than what's posted on the website.

It's exclusively SMT with the exception of the LEDs. Final board size is 2.30 x 1.85 inches. Six LEDs of each color - makes for an exceedingly bright fixture. Each pixel has an RS-485 receiver included.

Also, I found a decently inexpensive source of CAT-5 patch cables. So each pixel has two RJ-45 connectors for daisy-chaining.

I've actually ordered a trial batch of 30 of these from my overseas manufacturer. I'm expecting delivery later this week if all goes well. They're doing the PCB, parts and assembly.

If anyone is interesting in some of these boards, just let me know. Assuming the trial batch works properly, I'll be placing a larger order within the next few weeks.

John

[DynamoBen]

Wow, welcome...this is like seeing a coworker outside of work. ;)

Yeah, I saw your link while searching for an old DMX post on the picbasic forums and thought the folks here might enjoy it.

Have you been here before?

[JEC]

First time on this new site. Looks like it's growing bit by bit!

[stempile]

Any estimates on what the per board cost might be?

ms

[JEC]

Price will definitely come in sub-$11 per pixel.

I'm hoping to get it even several dollars lower, but at this point I haven't seen the freight and/or customs bills yet.

The PCB is fully assembled and populated per the schematic on the project website link above. You'd be responsible for providing a clean DMX signal and a stable source of +12V DC. You also should terminate the last pixel in the chain with a 100 ohm, 1/2W resistor.

Each pixel draws a maximum of just over 180 mA, so you can size the power supply accordingly.

My most current firmware was last touched about 6 weeks ago. At this point, pixel addresses are hard-coded into the pixel source code.

Set an address, burn a pixel, increment an address, burn a pixel.

Repeat ad absurdum.

I'm working on a firmware version which will allow field-programming.

[A Marchini]

Price will definitely come in sub-$11 per pixel.

I'm hoping to get it even several dollars lower, but at this point I haven't seen the freight and/or customs bills yet.

The PCB is fully assembled and populated per the schematic on the project website link above. You'd be responsible for providing a clean DMX signal and a stable source of +12V DC. You also should terminate the last pixel in the chain with a 100 ohm, 1/2W resistor.

Each pixel draws a maximum of just over 180 mA, so you can size the power supply accordingly.

My most current firmware was last touched about 6 weeks ago. At this point, pixel addresses are hard-coded into the pixel source code.

Set an address, burn a pixel, increment an address, burn a pixel.

Repeat ad absurdum.

I'm working on a firmware version which will allow field-programming.

If you are interested, we have a Co-Op(erative buy) section in this group list. If you offer them, to the public, perhaps a quantity buy in the several hundreds could be made, dropping the unit price.
This is , of course, if you were interested in being the point man for your own device, which is no easy task.

Tony

P.S. perhaps maybe just a survey to figure out numbers even.

[A Marchini]

The PCB is fully assembled and populated per the schematic on the project website link above.

Wait a minute, do you mean you a purchasing them assembled? or you are assembling them??
T.

[JEC]

Yes - the shop overseas has my BOM (bill of materials) as well as a full set of gerber files. They source all the components per my specs.

The company fabricates the PCBs and then folks in the building next door populate the circuit boards and do all the soldering. My batch of 30 is likely too small for machine loading, but the larger run will be more automated.

For what it's worth, there's a local shop I use for short-run assembly of other projects.

I asked them to quote this one for me, and their labor-only price was more than the overseas folks wanted for the entire job, turnkey.

Turn time seems to be about two weeks from my 'go.' The second batch will likely be faster, since the films for the circuit boards are already on file.

---- Edit -----

I could probably ask the factory to run a batch (tens to hundreds) of circuit boards and send them over sans components. If people were interested in assembling by hand, that may be more economical. It just depends on the price of your time. *grin*

I also have a known-good PCB layout which features dual 5-pin .1" headers instead of RJ-45 jacks. All of the parts on this board are thru-hole, which may be easier on the fingers. The parts on the SMT board are SOIC, 0805 resistor and SOT-23 transistor.

These bare boards could be in addition to a largish batch of populated PCBs.

I'll keep an eye on this thread for a week or two and see what, if any, interest exists.

John

[DynamoBen]

Folks with other DMX devices need to keep in mind that each of these color changers count against your 32 device limit. An opto isolator/repeater will need to be added to your system to split the DMX line. Something to consider... ;)

BTW this limit is a bit higher depending upon the transceiver used in the receiving device. Some of the newer transceivers say the limit is 128 devices.

[A Marchini]

I also have a known-good PCB layout which features dual 5-pin .1" headers instead of RJ-45 jacks. All of the parts on this board are thru-hole, which may be easier on the fingers. The parts on the SMT board are SOIC, 0805 resistor and SOT-23 transistor.


The thru-hole may be better for kit stuff. Out of curiousity does your current firmware care if the packet is limited to less than 512 channels? I was thinking of a supplimental system of routing, though I suppose the said router could just pad out to 512 empty channels if it is necessary, which probably is a no.

tony

[JEC]

The firmware is happy with 3 channels, 512 channels, or any number in between.

All that's required is that there are *enough* channels present to support the pixel's start address plus 2.

If your console outputs 192 channels and your address is 191, Unpredictable Things (c) may happen. Most likely, the blue channel will just never be picked up properly.

--------------------------

Heh - after typing this, I realized that I've used a word-sized variable to store the DMX address. So one *could* design a DMX driver which includes many, many more than 512 channels in a single packet. Up to ~65K, actually.

Like multiple universes in one cable.

Right now the firmware can run circles around a full-bore, 512 channel, 44 Hz datastream. The PWM subroutine is fairly tightly written and is fast enough that the main DMX program doesn't mind being interrupted. DMX byte length is 44 uS, which equals 88 machine cycles on the PIC. The dimming routine only uses maybe a third of that. So there's lots of room to play.

Running more than 512 channels could definitely drop the system refresh rate.

----------------------------

Other known issues related to the firmware. None of them are fatal, but I haven't had time to work on them:

----------------------------

1. If data applied and the dimming works properly, and then data is lost/reapplied, the software likely won't recover from the error. It may be necessary to power-cycle the pixel.

Solution could be to set up TMR2 to roll over every hundred milliseconds or so. If the DMX receive routine doesn't tickle the timer from time to time, the timer interrupt could kick the processor and make the DMX routine look for a new start code. The PWM routine would continue running with the last known good data received.

2. Firmware doesn't currently discern between zero and non-zero start codes. This is a dead simple fix, (and the code is even commented where the test should take place) but I've not been back to fiddle with it. If a rig of these were ever deployed in the wild, this change would be necessary.

3. Dimming levels, though matched 1-to-1 to the DMX channel values, are decidedly non-linear to the eyeball. There's code space enough that a simple look-up table could be implemented, mapping DMX values to linear steps of PWM intensity. Unfortunately, I don't have any light measuring equipment to do the calibration.

--------------------------

My only other concern is that the +12v supply may going to drop too much over lengthy cable runs, even with three of the CAT5 conductors carrying the power. Since that's the voltage which drives the LEDs, there could be some brightness variation.

The solution, of course, is to add some sort of buck/boost converter to each pixel, guaranteeing a solid +12 and +5 output, regardless of the input supply.

Pretty much kills the budget though.

JEC

[cmurray]

Any pictures of the completed units?
I seen you mention an acrylic dome. What do you use for a base?
Where do you get them and what is the additional unit cost for the entire enclosure?
Pictures please of the entire unit if you do not mind sharing.

My guess is that if the unit count gets into the hundreds the fully assembled surface mount will be cheaper than a thru hole PCB and your own parts.

The thing I caught was that the PIC is soldered onto the board BUT each PIC has to be programmed with a unique address.
I am sure that someone can point me to a SMT chip clip for programming the PICs.

Joel

[DynamoBen]

3. Dimming levels, though matched 1-to-1 to the DMX channel values, are decidedly non-linear to the eyeball. There's code space enough that a simple look-up table could be implemented, mapping DMX values to linear steps of PWM intensity. Unfortunately, I don't have any light measuring equipment to do the calibration.


I have an S-Curve table that I'm putting in my AC dimmer; you are welcome to use it in your code. That will fix your linearity issue.

[JEC]

The thing I caught was that the PIC is soldered onto the board BUT each PIC has to be programmed with a unique address.
I am sure that someone can point me to a SMT chip clip for programming the PICs.

Joel

The best solution I've come up thus far is twofold:

First - finish the firmware so that addresses can be flashed in the field. The easiest way to do this would be for the pixel to repond to, say, a start code of '10' by accepting the next two data bytes as the new start address.

So a packet containing

10 0 6

Would set the pixel address to DMX 6.

A packet containing

10 255 250

would set the address somewhere in the high 500 range.

The PIC has lots of internal EEPROM, so storing a few bytes isn't difficult.

-------------

The second step would be to build a little programmer (probably another 16F688 and a DIP switch or LCD screen) which can send out these specially formed DMX packets.


FInally, the pixel PCBs include an ICSP header, so a SMT clip isn't necessary. That five pin header on the top of the board contains +12, gnd, Vpp, clock and data for programming.

-------------------

I'll work on some photos this evening.

John

[ErnieHorning]

The solution, of course, is to add some sort of buck/boost converter to each pixel, guaranteeing a solid +12 and +5 output, regardless of the input supply.

To increase the voltage will mean that you draw more current on the Cat5 and less current locally and adding to the problem; you can’t defy ohms law. The solution would be to increase the input voltage and regulate it down locally.

[DynamoBen]

If you are using the DMXPro you could set all the start addresses via RDM. You would need to make one mod to your current hardware design. The Pic would need to control Rx/Tx on the 75176. As far as software the standard is pretty easy.

A more simple option could be dipswitches. I know how much people LOVE dipswitches. ;)

[JEC]

If you are using the DMXPro you could set all the start addresses via RDM. You would need to make one mod to your current hardware design. The Pic would need to control Rx/Tx on the 75176. As far as software the standard is pretty easy.

A more simple option could be dipswitches. I know how much people LOVE dipswitches. ;)


Ugh - the DIP switch would be larger than the pixel itself.

RDM makes more sense, but I've never used it. The PIC still has several free I/O pins, so it wouldn't be difficult to control the RS-485 receiver.

Can anyone send or post more information on how RDM is used?

[RavingLunatic]

John,

I like what you have done. The effects that you create are AWESOME! :shock:

I'm curious about the topic that hasn't been mentioned yet - controlling software/program.

On your website you mentioned that you would run it via
a larger PIC or one of Parallax's new Propeller chips .

Which did you end up using? Do you have multiple routines or a single one that repeats over & over?


{Thought Bubble} Maybe KC could create a VIXEN UI for these for next year. :wink: Probably would be a huge task to track the color and dimming for each pixel for every time interval.

RL

[DynamoBen]

Can anyone send or post more information on how RDM is used?

I can talk about how it works and how is works in general but like most things ESTA related its $40 for a copy of the standard.

RDM is a bidirectional (half-duplex) communication between a device and a "server." The communication between the server and device is the same as DMX and doesn’t require any changes to the wiring but will require a small change in how the 75176 is controlled (see above). The server in this case would be a PC and is the “master” of communications each RDM device is a slave (only “speaks” when “spoken to”).

The main difference between an RDM packet and a DMX packet is the Start Code and the data contained in the packet. RDM uses a very specific start code and packet format. The ESTA document covers all this along with timing; for example how long a receiving device should remain idle prior to taking control of the line and transmitting back to the server.

Beyond timing and format the ESTA document lists all the different message types that can be sent back and forth, and gives a list of the required messages for all devices. What you include beyond the minimum messaging depends on the application or in your case how much room you have. ;)

Its not too bad and for your application may be just the ticket.

[DynamoBen]

Maybe KC could create a VIXEN UI for these for next year. :wink: Probably would be a huge task to track the color and dimming for each pixel for every time interval.

I guess patching and multi-channel control (aka personalities) are closer than I had hoped. ;)

[JEC]

John,

I like what you have done. The effects that you create are AWESOME! :shock:

I'm curious about the topic that hasn't been mentioned yet - controlling software/program.



Controllers - Rev 1:

In 2005, through a somewhat circuitous process, I was actually driving my 32 pixels from a video files in real time. I used a suite of programs on my Mac called Max/MSP + Jitter to tie everything together.

Basically, I opened a video file (640 x 480), then averaged neighboring pixels together to create a smaller 32 x 24 pixel video file. Then, I just chose one row of pixels. I extracted the red, green and blue components from the video scan line into a trio of 32-element arrays.

Then, I fed those arrays into a third-party patch to the Max program. This patch communicated with my Enttec DMX USB Pro dongle and, by inference, the outside world. Unfortunately, this patch isn't available for the less expensive 'open' DMX-Interface.

The end result was that I could use arbitrary video files. Some I created myself, others were screen captures of iTunes visual displays.

-------------------

There are a myriad of video / graphic to DMX pixel hardware and software programs out there. Entry fee starts at about $1K and goes way up from there.
Most of these packages let you define an array of pixels, determine the DMX patch for them, and then apply still and video files on top of this array. The software then extracts the proper RGB data and sends it out the door.

Bigger lighting rigs use ARTNet, as a single DMX universe will only control 170 pixels or so.

The programs have built-in effects generators with many parameters which can be adjusted. Instead of programming a rainbow chase one pixel and one frame at a time - a process excruciatingly tedious at best -, the 'rainbow chase' effect can be applied to a group of pixels. The software handles the behind-the-scenes rendering.

-----------------

Here's an idea I've been ruminating on for a while, but haven't sat down to implement in code. It should fit handily on a $5 PIC.

The theory is this...

A (mostly) linear series of pixels can best be described as three mathematical arrays - one each red, green and blue.

So define three arrays in your processor's memory, each as long as your display. Make it 120 columns, just for fun.

Here's how to do a simple chase. The intensity of a comet trail might look like this:

[FF FF FE A0 70 30 20 10 5 3 2 1]

Now, take that intensity map and colorize it.

For a cool color scheme, 0xFF might be pure blue and 0x00 might be pure green, smoothly fading between extremes.

So intensities of [0 FF] would colorize thusly:

[0 FF] Blue
[0 0] Red
[FF 0] Green

Then, step through your output matrix one pixel at a time and add colors appropraitely.

To chase your new comet tail, just shift the output matrix values once to the left or right every so often.

A nearly endless variety of color maps and patterns could be stored in memory. Trig functions (or their lookup table equivalents) could be overlapped to make flickering / shimmering effects.

Strobes are easy (actually, I tried this one. I sent out one DMX frame of red, then a frame of green, then a frame of blue, repeating over and over. To the eyeball it mixed to white - until you moved your head!)

I'd love to try this on a Propeller, but don't have any experience with their tools just yet. Having one cog for pattern generation and timing, and another for DMX transmission is attractive.

But for now I'm using a '4550 @ 40 MHz.

If the only tool you have is a hammer...

Enough rambling for now.

John

[DynamoBen]

I think effects for such a string would be a snap in software (so says the non programmer); in the end its just mathematics. If the software knows the physical layout of the string and then DMX values for the layout then control should be easy.

See my attachment for an idea of software control. The slider on the right makes entire string change color. The circular graph associates colors left/right/up/down based on the pointers location. The drop down determines the pattern those colors are represented in.

[JEC]

Actually that might be fun to try...pipe a function generators into a pic and convert the values into DMX. Instant light show. ;)



Here in town, there's a musician who puts on a 2000 seat, two night, more-than-sold-out theatre show every Christmas. It's mostly solo piano with occasional guests and backing musicians.

I've been lucky enough to help with lights and video for the past few years.

He normally plays a 9' grand piano, which occasionally has comes with a MIDI brain installed. If properly connected, the grand can trigger other MIDI gear...

So I keep meaning to build 88 x 1 foot diameter acrylic spheres, and hang them in a row from a batten above the stage.

1 per piano key.

Then, I'd use a suitable combination of hardware and software to capture the MIDI note and velocity data from the grand piano.

Munge those numbers with some nifty effect generators and voila - the audience can see the music as it's played.

Live. A bit like the iTunes engine.

Or the pixels can just run ambient patterns and washes as part of the overall lighting design

Or some combination of both.

You heard it here first, folks!

:)

[DynamoBen]

BTW JEC, I would be interested in a string for use as front boarder on a stage. ;) I would probably use frost globes normally used on ceiling fans.

[JEC]

I think effects for such a string would be a snap in software (so says the non programmer); in the end its just mathematics. If the software knows the physical layout of the string and then DMX values for the layout then control should be easy.

See my attachment for an idea of software control. The slider on the right makes entire string change color. The circular graph associates colors left/right/up/down based on the pointers location. The drop down determines the pattern those colors are represented in.


For just under $1K, the Color Kinetics iPlayer 3 and associated software is really, really hard to beat. The editor works in Windows and OS X. Patching and fixture grouping is very intuitive. The show is timeline based and all effects are drag and drop. Once you're happy, the show is rendered and then downloaded to a stand-alone playback machine. Their playback machine supports 2 universes of pixels.

[JEC]

BTW JEC, I would be interested in a string for use as front boarder on a stage. ;) I would probably use frost globes normally used on ceiling fans.

Best price I've found is for some translucent 6" globes at just over $4. They were buried deep inside an online lighting retailer's website.

I tested one from Home Depot on the bench and was very happy with the diffusion.

I don't recall the online company's name, but I found them without too much hassle.

[DynamoBen]

For just under $1K, the Color Kinetics iPlayer 3 and associated software is really, really hard to beat. The editor works in Windows and OS X. Patching and fixture grouping is very intuitive. The show is timeline based and all effects are drag and drop. Once you're happy, the show is rendered and then downloaded to a stand-alone playback machine. Their playback machine supports 2 universes of pixels.

No doubt, but that’s what you get with a for profit lighting company.

I think this string could really be a cool thing. The only issue at this point is the per pixel cost. It would be awesome to get that under double digits.

[JEC]

I think this string could really be a cool thing. The only issue at this point is the per pixel cost. It would be awesome to get that under double digits.

I'm confident of sub $10 per pixel and cautiously optimistic that a run in the range of several hundred boards will be even several dollars less.

[JEC]

Well, this morning the overseas factory sent me photos of the sample batch, which they planned to ship me.

It turns out that they used the wrong LEDs (I specified flat-top LEDs, which have a wide, smooth dispersion pattern).

For some reason, they just used regular high-intenstity 5 mm LEDs instead.

So they're going to replace them with the proper type and then ship.

I've attached a photo of the finished product.

Aside from the horrid fisheye lens used to take the photo, they PCBs look very nice. The two silver jacks on the left side are RJ-45s for DMX and power.

John


************* EDIT***************

I may ask the factory to install the LEDs on one side of the PCB and the supporting components on the other. This would allow the LEDs to shine forth unimpeded.

Once the samples are here I'll know if it's really an issue.

[DynamoBen]

Something else that might help is an inexpensive fresnel lens above the leds. This should eliminate any "hot spots."

Even a cheap diffusion gel might get the job done.

[wjohn]

JEC,

VERY NICE WORK!!

[DynamoBen]

Its like your project but bigger... ;)

http://www.makezine.com/blog/archive/2007/09/10000_color_changing_ligh.html

[ErnieHorning]

Here's pictures. Scroll sideways to see them all.

http://www.house42.net/htm/lif006.html

[TerryD]

Here in town, there's a musician who puts on a 2000 seat, two night, more-than-sold-out theatre show every Christmas. It's mostly solo piano with occasional guests and backing musicians.

I've been lucky enough to help with lights and video for the past few years.

He normally plays a 9' grand piano, which occasionally has comes with a MIDI brain installed. If properly connected, the grand can trigger other MIDI gear...

So I keep meaning to build 88 x 1 foot diameter acrylic spheres, and hang them in a row from a batten above the stage.

1 per piano key.

Then, I'd use a suitable combination of hardware and software to capture the MIDI note and velocity data from the grand piano.

Munge those numbers with some nifty effect generators and voila - the audience can see the music as it's played.

Live. A bit like the iTunes engine.

Or the pixels can just run ambient patterns and washes as part of the overall lighting design

Or some combination of both.

You heard it here first, folks!

:)

If you're interested in MIDI code let me know. I use PIC based MIDI contollers for my light show.

Terry

[JEC]

I spoke with the overseas factory late last night.

They've received the new flat-top LEDs I specified and are now re-working the boards. They say they'll be finished in three days - I interpret this to mean they're shipping to me on Thursday of this week. The package will be shipped FedEx international and should arrive in 2-3 days.

Once they arrive and are tested, I'll place the larger production order.

Lead time on my 150 pixels (plus whatever interest exists here in the forums) is said to be 20-30 days.

So the larger order arrives toward the end of October, best case.

Still working on final pricing for the larger batch.

At this point, my preference is to only order assembled and tested SMT models. As I mentioned above, there is a programming header on the circuit board, so the firmware can be changed as required.

John

[A Marchini]

At this point, my preference is to only order assembled and tested SMT models. As I mentioned above, there is a programming header on the circuit board, so the firmware can be changed as required.

John

I was going to mention something but I see that it is already discussed. I noticed that it wasn't easy to do on the through hole version but how easy is it to free up pins 2-3 on U2 (can you cut them with a knife)? The idea, which I guess has been brought up is to jumper pin U2-4 to U1-6 and U2-3 to U1-7. Allowing bi-directional comunication with the pixel.

Tony

P.S. It appears from zooming in on the SMT board that the trace goes to pin 2 before 3 . So assuming pin 3 on the 485 driver is still the driver enable line, it should be possible to free it up using a knife.
On the thru-hole design it went to pin 3 before 2 so it made freeing this pin up more difficult.

[JEC]

How about I just edit the schematic and board layout, making this a permanent change? If the layout they're working on now works properly, adding these two traces won't change any existing functionality.

Homebrew RDM-able RGB pixels. Hooray!

EDIT: A new schematic is attached as a PDF file.

[stempile]

Will the height of the rj-45 connections cause shadows in the light output? They appear to be taller then the LEDs and may block.

ms

[JEC]

Will the height of the rj-45 connections cause shadows in the light output? They appear to be taller then the LEDs and may block.

ms

Possibly. For my application, which has the pixels enclosed in translucent containers, I don't think it will be a problem. There should be enough light bouncing around in there to make it work.

I'll know for sure in a few days. The test boards I made and have on the bench didn't include the jacks.

I first planned solder CAT5 cable directly to the PCB, but disovered that the solid conductors break after being flexed a few times. Our roofline is high enough that once everything is installed, I'm not going back up there. :)

We could also ask the factory to mount either the LEDs or the jacks on the reverse side of the PCB.

[stempile]

A reverse makes sense, might impact the PCB design though. Though there might be enough light the shadow still might be enough to cause the color on the side of the jacks to be effected...

ms

[A Marchini]

How about I just edit the schematic and board layout, making this a permanent change? If the layout they're working on now works properly, adding these two traces won't change any existing functionality.

Homebrew RDM-able RGB pixels. Hooray!

EDIT: A new schematic is attached as a PDF file.

Won't this effect your ship date??
One thing, you may want the /RE pin to to be selectable (solder blob perhaps) between the PIC output and ground. Some protocols, it is handy to "hear" what you are saying. Not a show stopper or anything.

Hey, and while you are at it, add through hole connectors to the pull down and high side, if you don't populate the diodes, you could use this to drive an AC FET control. I was thinking of it as a stepping stone to dim individual C9 bulbs.

Or not... :)

[A Marchini]

A reverse makes sense, might impact the PCB design though. Though there might be enough light the shadow still might be enough to cause the color on the side of the jacks to be effected...

ms

Perhaps you could use the RJ45s that are perpendicular to the board on the back, then the RJ-45s could "come from the top" with the LEDs facing downward. Then a diffuser could make it radiate out a little.

Now all we need is the extensible DMX type signal, something to break the pixels into rows and you got LEDTriks x10.

T.

[JEC]

Won't this effect your ship date??

Not terribly much. I'm not going to change the trial batch, but we can tweak the larger run without problems.


Hey, and while you are at it, add through hole connectors to the pull down and high side, if you don't populate the diodes, you could use this to drive an AC FET control. I was thinking of it as a stepping stone to dim individual C9 bulbs.


Look here:

http://www.tmb.com/products/festoon/

It exists already!

TMB is a giant A/V rental house with shops across the country. They're very friendly... if you're an A/V staging company. Last I heard, they're happy to cross-rent but they don't sell to the general public.

I'm not sure the price, but I'm sure it's not inexpensive. The 5W dimmer is built in the 5W lamp base and hundreds of thousands can be connected together. Standard two-wire zip cord, too! The data is somehow overlaid on the power line without needing a separate conductor.

They also claim a 20 Hz refresh rate for dimming. I'm very jealous.

[A Marchini]

Won't this effect your ship date??

Not terribly much. I'm not going to change the trial batch, but we can tweak the larger run without problems.

Look here:

http://www.tmb.com/products/festoon/

It exists already!



I didn't actually expect that such a thing didn't exist, though it is close to the thing I was thinking, only I was going to use a separate comunication wire. They could be using those devices that superimpose comuncations channels on AC , could even be proprietary, that kind of technology has existed for a long time. They may have a custom asic to get all the functionality in one chip.

I was looking to make something that the common person could make in their own time.
Realistically, all of this stuff exists out there in one form or another. We are not doing anything new, just rehashing other ideas for public consumption.

They changed the voltage on the lamps to, so they are controlling the bus power which makes the remote control easier. Also don't need big fets like the ones I intend to use to PWM AC.

Those are neet though.
Tony

[JEC]

I didn't actually expect that such a thing didn't exist, though it is close to the thing I was thinking, only I was going to use a separate comunication wire. They could be using those devices that superimpose comuncations channels on AC , could even be proprietary, that kind of technology has existed for a long time. They may have a custom asic to get all the functionality in one chip.

I was looking to make something that the common person could make in their own time.
Realistically, all of this stuff exists out there in one form or another. We are not doing anything new, just rehashing other ideas for public consumption.

They changed the voltage on the lamps to, so they are controlling the bus power which makes the remote control easier. Also don't need big fets like the ones I intend to use to PWM AC.

Those are neet though.
Tony

It looks like they're using a +48V DC supply for the lights. Maybe the control signal is just modulated AC piggybacked on the DC. One could extract it with a diode and an op-amp or two.

[P. Short]

Out of curiosity, what is the light output of the board (various colors)?

--

Phil

[A Marchini]

Out of curiosity, what is the light output of the board (various colors)?

--

Phil

Nope, the trick on these is individual lamp control.
Using tri color LEDs and a bus of cheap pics (maybe not even pics, perhaps those chinese controllers you see in toys from time to time) you could do something in the realm of mini lights with mulicolor control.

I see the control of C7 /C9 lights question come up from time to time,and even though it is a costly thing to do, I think it may be worthwhile for a design/solution to exist. That way is someone wanted to take it upon themselves, and had the means... there would be a solution for them.

The idea had been around for a while but Johns pixel devices sort of fit the bill, just need the led section removed and replaced with a single channel SSR. Then packaged into a small board (possibly potted) and vampire tapped into a C7 /C9 light string.

Tony

[JEC]

Out of curiosity, what is the light output of the board (various colors)?

--

Phil

If the question is directed at me, I have the following specs from the LED manufacturer:

Blue

Luminous Intensity Typ Iv (mcd) : 4000(Typical) ~ 5000(Max)
Viewing Angle : 85 ~ 100 Degree
Max Power Dissipation : 80 mw
Max Continuous Forward Current : 30 mA

Red

Luminous Intensity Typ Iv (mcd) : 4000(Typical) ~ 5000(Max)
Viewing Angle : 85 ~ 100 Degree
Max Power Dissipation : 80 mw
Max Continuous Forward Current : 30 mA
Max Peak Forward Current : 75 mA

Green:

Luminous Intensity Typ Iv (mcd) : 4000(Typical) ~ 5000(Max)
Viewing Angle : 85 ~ 100 Degree
Max Power Dissipation : 80 mw
Max Continuous Forward Current : 30 mA
Max Peak Forward Current : 75 mA


---------------------

I'm not very good at calculating light intensity, but there are six LEDs of each color.

That makes me think that output, per color, is somewhere in the 24,000 - 30,000 mcd range, and over a fairly wide viewing angle.

It doesn't count if you're super bright but only over a degree or two.

The LEDs should be running at 20 mA peak, so there's probably even more brightness that can be squeezed out.

John

[P. Short]

John,

Thanks. Those are pretty good numbers for a viewing angle of 85-100 degrees, they come out to something over 6 lumens (typ) each using this candela-lumen calculator (http://led.linear1.org/lumen.wiz). Of course, you didn't give the minimum intensity, so I don't know what the minimum light output its.

That makes your pixel lights a pretty good deal. The only LEDs that I can find in the mouser catalog that have similar output cost around $3.55 each in quantities of 500, and take a lot more current than 20 mA each (or appear to be multiple LEDs in series). An example of this type of part is p/n 697-SSP-LX6144D6UC. Typically Ultra-bright LEDs with output intensity of around 4000 mcd appear to have viewing angles of 30 degrees, which translates to a total output of under 1 lumen. So, even ignoring the cost aspect, you seem to have found a family of very efficient LEDs.

--

Phil

[JEC]

Of couse, the posted specs could be totally bogus and the manufacturer is up in the night. :)

Unfortunately, I don't have any way to measure or verify light output...

I can say that when viewed directly and 100% intensity, the pixels leave a decent imprint on my retinas.

I first found the LEDs from this eBay seller and used them in a set of prototype pixels. Retail auction price for 50 each red, green and blue cost was about $20.

My overseas manufacturer has since tracked the source and a much better price.

Here's a link:

http://cgi.ebay.com/20x-Green-5mm-Wide-angle-Flat-Top-LED-Free-Resistors_W0QQitemZ300158400601QQihZ020QQcategoryZ 66949QQssPageNameZWDVWQQrdZ1QQcmdZViewItem

I specifically chose wide-angle LEDs because of their uniform light output across a large field. I'd experimented with other high-intensity LEDs in the past and wasn't pleased by the circular 'hotspots' they projected.

John

[P. Short]

Thanks.

--

Phil

[P. Short]

I ordered some of them so see what they look like. I suspect that if you find them acceptable, so will I. At this point I've got the firmware for the PIC10F200 (prototyped on a PIC12F629) more or less working, so the next step is to determine what type and how many LEDs that to put on each controller, and then do a larger scale test.

In the end, though, it may turn out to be cheaper for me to just order Pixels from JEC.

--

Phil

[JetMech]

In the end, though, it may turn out to be cheaper for me to just order Pixels from JEC.

--

Phil

Yeah, but what fun would that be? After all the site is "Do It Yourself Christmas"
I get more joy when I say "I did that" if I don't think about how much it cost me to do "that". :D

John

[P. Short]

John,

While I wait for my LEDs to arrive, what is the Vf of the various colors?

--

Phil

[JEC]

John,

While I wait for my LEDs to arrive, what is the Vf of the various colors?

--

Phil

Green 3.2-3.6 ===> 9.6-10.8 for chain of 3
Red 2.6 - 3.0 ===> 7.8 - 9 for chain of 3
Blue 3.2 - 3.6 ===> 9.6-10.8 for chain of 3

[P. Short]

Hmm...

I ordered these parts from that same guy,

http://cgi.ebay.com/ws/eBayISAPI.dll?ViewItem&item=300051632308

(Red)

http://cgi.ebay.com/ws/eBayISAPI.dll?ViewItem&item=300059268722

(Green)

and

http://cgi.ebay.com/ws/eBayISAPI.dll?ViewItem&item=300093865429
(Blue),

which I thought were the same parts that John was using. I'm not so sure that I got the right p/n, as the Red and Blue ones are far dimmer than I was led to expect. The on-axis peak intensity was far less than some LEDs that I purchased from Mouser (putting 25 mA through each). Has anyone seen these LEDs, and what are your impressions?

--

Phil

[JEC]

Those appear to be the same parts. Maybe you got a different bin than I did?

As I was breadboarding my strings of three, I accidentally shorted two legs of the LEDs together, effectively reducing the string length by one and dramatically increasing the forward current.

Though it was only for a second, I noticed that the LEDs were permanently less bright than an (untouched) triple string of the same color nearby.

John

[P. Short]

It's getting to be a bit off-topic, but I think that I'd like to (temporarily) exchange boards just so that I can see how bright your LEDs are. It may simply be that my expectations are unreasonably high.

Here is a picture of my first board. The upper three LEDs are some Kingbright LEDs that I installed just for comparison (high intensity, but fairly narrow angle). The bottom nine LEDs are the ones from that eBay guy.

--

Phil

[JEC]

It's getting to be a bit off-topic, but I think that I'd like to (temporarily) exchange boards just so that I can see how bright your LEDs are. It may simply be that my expectations are unreasonably high.

Here is a picture of my first board. The upper three LEDs are some Kingbright LEDs that I installed just for comparison (high intensity, but fairly narrow angle). The bottom nine LEDs are the ones from that eBay guy.

--

Phil

PM me with your address and I'll drop one in the mail.

I've also got half a tube of 10F200 DIP chips if you want a few.

[P. Short]

Regarding the LEDs from the Ebay link provided by John...

Good news and bad news.

The good news is that the brightness appears to comparable with the factory LEDs that John has been using.

The bad news is that I've had quite a few failures...almost 50% of the green LEDs (both flat-top and 'straw hat'). They start out OK, but they burn out one after another over a few hours to a day or two. I'm pretty sure that I'm not doing anything wrong...I'm driving them with a constant 25 mA current source (so that one LED burning out will not put extra stress on the other ones).

--

Phil