PDA

View Full Version : Crazy Icicle light string idea



JFTiger
03-13-2009, 04:13 AM
I think it’s crazy, but thought it would make an interesting and unique display element.


Take an icicle light string and make each strand selectively sequence down separate of each other. Also could have an entire strand selectively turn on.


A rough gif of what I had in mind.

Penfold
03-13-2009, 09:19 AM
Hmmm...... that looks like something Galaxiaish. COOL!! Are you thinking of Charlieplexing the icecicles?

Warlock
03-13-2009, 10:25 AM
Something along the lines of the idea I was hoping to try if ledtriks could handle it for my mega tree...Great idea...I'd love to see it in action.
Joe..

P. Short
03-13-2009, 01:19 PM
Why do you think that it's crazy? Creating and building non-commercial decorations is largely what this forum is about. The challenge is coming up with neat, easy, or cheap ways to build your ideas, and (in some cases) finding easy ways to control them. And, for some people, none of these, they don't mind complicated, difficult, and expensive ways so long as the end result is cool or impressive.

JFTiger
03-13-2009, 03:10 PM
I got the idea from a Galaxia video. I've not done much research into charlieplexing so that is an unknown at this time. I've also not done much beyond a breadboard so I'm not sure if this will ever come to light through me. I may have to pass the torch to a more experienced developer if somebody is willing. I will definitely be posting a lot of requests for help if I do start working on this one.

In a nutshell that is what I was fighting myself over. I’ve thought it over for weeks before finally throwing caution to the wind saying. “It might not be worth it, but at least the Idea is now in the hands of the forum.”

My original thoughts are to take a Renard standard controller, preferably one with dc or low voltage ac to power or control the Icicle controller. The Icicle controller will then sequence the lights. The problems I foresee are how to control the speed of the sequence and turn them all on at once. Would dimming work if we used a resister, capacitor network? Also would dimming be possible without using a pic? A move advanced idea would use a led tricks type configuration so we can use the Icicle’s as a marquee. Like the Beijing Olympics and in the Galaxia video.

scorpia
03-13-2009, 05:41 PM
it could be done very simply by connecting a renard pic at the top of each run with its 8 outputs driving 8 led's down each icicle.

like this

--pic-----pic----pic---pic---etc
| | | |
led led led led
| | | |
led led led led
| | | |
etc


basic asci drawing were not my best. but hopefully you get the idea. (damm stupid forum deleting extra spaces. drawig didnt work)

could be done with the dip pics or even better using the little smd versions.

probably not the best way to do it. but it would be easy enough.

P. Short
03-13-2009, 06:38 PM
it could be done very simply by connecting a renard pic at the top of each run with its 8 outputs driving 8 led's down each icicle.

like this

--pic-----pic----pic---pic---etc
| | | |
led led led led
| | | |
led led led led
| | | |
etc


basic asci drawing were not my best. but hopefully you get the idea. (damm stupid forum deleting extra spaces. drawig didnt work)

could be done with the dip pics or even better using the little smd versions.

probably not the best way to do it. but it would be easy enough.

Like this? I still had to hack it a bit, and I don't know if it will look OK on all browsers...


--pic-----pic----pic----pic---etc
| | | |
led led led led
| | | |
led led led led
| | | |

holtm
03-13-2009, 07:00 PM
I've had the same thoughts. The channel count could go through the roof, but adding a serial board is easy. I've thought maybe some canned routines would work and then each strand of 8 could be one channel. It would interpet certian dimming levels as chase down, chase up, fast or slow fade. With some mods to Vixen it would be easy to sequence and fit into the scheme.
I thought if I could solder the leds on a flexible strip of plastic and stick the string into a semi opaque plastic tube I might get a continuous effect.
I think that 8 leds on all together might be too much power for the PIC but haven't checked. Power could come from the same Cat5 as the serial, maybe as 12V regulated to 5v at each string, or if loss isn't a big deal 5V.

P. Short
03-13-2009, 07:30 PM
I've had the same thoughts. The channel count could go through the roof, but adding a serial board is easy. I've thought maybe some canned routines would work and then each strand of 8 could be one channel. It would interpet certian dimming levels as chase down, chase up, fast or slow fade. With some mods to Vixen it would be easy to sequence and fit into the scheme.
I thought if I could solder the leds on a flexible strip of plastic and stick the string into a semi opaque plastic tube I might get a continuous effect.
I think that 8 leds on all together might be too much power for the PIC but haven't checked. Power could come from the same Cat5 as the serial, maybe as 12V regulated to 5v at each string, or if loss isn't a big deal 5V.

Each IO pin on that PIC is good for 25 mA, and the power/ground pins are good for 250/300 mA respectively.

What would really drive the channel count through the roof would be using RGB LEDs.

scorpia
03-13-2009, 07:53 PM
yep phil, thats what i had in mind.

not the best way to do it but it can be done by simply wireing up led's to the current pics and code that is available now.

very simple to do,

as for RGB, now that is something i have been thinking about for a while. i need some help with some code to multiplex rgb led's off a single pic. could go from 8 single color leds to 8 rgb led's by simply runnign 2 more wires and a bit of pic code.

any takers :)

Peter

ErnieHorning
03-13-2009, 11:23 PM
Each IO pin on that PIC is good for 25 mA...Each IO pin on that PIC is good for +/-25 mA, so each output is good for at least two channels.

awhaley
03-14-2009, 12:49 AM
It looks like an SMD only chip... but the Philips PCA9532 is a 16 channel i2c chip with PWM all generated onboard... Or of course the TLC5940 could be used. You would then use one chip for every 2-4 strands of icicles. I think this is a great idea and could lead to some REALLY cool effects.

For a while I was thinking about a pixel thing of this sort using the 16F54 PIC... but the lack of onboard oscillator, interrupts and UART convinced me that the support components and programming difficulty would more than offset the cheap price of the chip. The 16F722 could drive 16 LEDs very easily... Renard code exists for it... it's fast enough to handle DMX if you wanted more pixels/more standard protocol....

And Peter, I've ALSO been pondering cheaper RGB pixels for quite some time! We should trade ideas at some point!

As far as control of all of this goes... I think the idea of a combination of direct control of every pixel if you want it, or a 'head unit' at the top of the chain that takes signals and then runs preprogrammed sequences on the pixels is a cool concept. I'd want access to every pixel most times...but there would be times when one channel to select and effect, a second channel to select the speed, and a third to set the color palate for the effect would be very cool. If we were doing individual pixel control of a lot of pixels, I think the next thing Vixen would need would be a pixel mapping system... LEDTRIKS uses a sort of one, so adapting to a more flexible pixel mapping system may not be all that hard!

Keep thinking on it and feel free to ask ANY questions of the group you'd like. This is a fine place to learn from step 1!

Art

scorpia
03-14-2009, 02:10 AM
i have been thinking about a pwm driver chip as well but is it worth replacing a $1.50 pic with a $2.70 pwm chip that still needs a pic to control it. do the advantages outweigh the disavantages?

i think the 16f722 chip could drive 16 led's quite easy and be able to do alot more.

just a thought.

ps how do you have negitive ma, surely its just a ma rating weather the pin is sinking or sourcing the power.

P. Short
03-14-2009, 05:53 AM
The other idea that I've been musing over is to use a Charlie-plexed scheme with a PIC16F688. In this setup each PIC16F688 would have new firmware (perhaps using the Renard protocol) that would allow it to drive 56 LEDs. Using the Charlie-plex scheme each LED would have a 1:8 duty cycle, with up to 7 LEDs on at any given time (requiring buffer transistors for whichever LED terminals are 'common'). I'm thinking that each LED would have 32 dimmer levels, would be on/off for 26 uS (or so) at a time, and would have a 150 Hz overall refresh rate. These parameters would give good, reliable operation. It might be possible to increase the number of dimming levels if the on/off time is decreased and/or the overall refresh rate is decreased, but I think that these are becoming risky.

Under these circumstances I don't think that it would be hard to come up with firmware for this, although I'm busy on other things at the moment.

scorpia
03-14-2009, 07:57 AM
is it really worth the issues of charlie plexing. i mean how are you going to physically wire up 56 led's to use in a string with them charlie plexed. i would think you would be better off cutting the numbers down a bit and making the wireing less complex. sure its a few more pics but in the end i think it would be alot easier to do.

holtm
03-14-2009, 11:01 AM
Each IO pin on that PIC is good for 25 mA, and the power/ground pins are good for 250/300 mA respectively.


The 16F688 Spec sheet I have shows max current sunk or sourced by PortA and PortC combined at 90mA. The LEDs from the LEDTricks are 20mA. 20x8 > 90, am I missing something? Maybe duty cycle? Sorry if this is a dumb question.

P. Short
03-14-2009, 11:05 AM
The 16F688 Spec sheet I have shows max current sunk or sourced by PortA and PortC combined at 90mA. The LEDs from the LEDTricks are 20mA. 20x8 > 90, am I missing something? Maybe duty cycle? Sorry if this is a dumb question.

The datasheet that I have (41203B) shows 200 mA. This was downloaded from the link provided by Mouser, I wonder what gives.

P. Short
03-14-2009, 11:09 AM
The 41203D that I just downloaded from Microchip does indeed show 90 mA...I guess you should never trust Mouser to keep their links up to date. That certainly puts a kink in these plans.

djulien
03-14-2009, 01:23 PM
i mean how are you going to physically wire up 56 led's to use in a string with them charlie plexed.

How about something like this (use cat5 for the groups of parallel wires):

don

awhaley
03-14-2009, 02:39 PM
Don, that looks like a really logical layout for a multiplexing rig... Charlieplexing is a little more complicated. Your diagram would require 15 pins to individually control all those leds, and the 16f688 Phil mentioned only has 12 available IOs. Your scheme WOULD work with a larger PIC, like the 16F722 we mentioned earlier... and this might be a better idea than trying to squeeze too much out of the 16f688 anyway, especially now that we've found out it's rated for less total current than we thought.

For charlieplexing, some of the diodes are wired back to back, and some of the common will serve as positive at some times and negative at other times, and be set to Hi-Z other times to allow other diodes to work. It's more complicated on the software and hardware side... but greatly reduces the number of pins required! :)

Art

djulien
03-14-2009, 03:39 PM
Charlieplexing is a little more complicated. Your diagram would require 15 pins to individually control all those leds, and the 16f688 Phil mentioned only has 12 available IOs.

The diagram I posted is for charplexing - it puts an LED between each pair of pins. It requires 8 I/O pins, and 8x7 connections gives 56 LEDs (only 1 on at a time, though).


For charlieplexing, some of the diodes are wired back to back, and some of the common will serve as positive at some times and negative at other times,

Yes, they are wired as back-to-back pairs, but you have to follow the wires to see that. For example, if you follow pin 8 from the connector, in the first cluster it goes thru an anode to pin 1, and then in the last cluster (upper right) is the other side of that pair - pin 8 goes thru an anode and back to pin 1. They are wired as a square matrix with a corner-to-corner diagonal missing.

don

P. Short
03-14-2009, 03:41 PM
awhaley - as far as I can tell the drawing that djulien posted depicts an alternate way of drawing a charlie-plexed setup. You need to trace things through, but I think that you will find that there are only eight inputs needed to drive the 56 LEDs (on the lower right), and for every LED there is another one that is in parallel with it pointing in the opposite direction.

P. Short
03-14-2009, 03:43 PM
Djulien - if the controller had sufficient current drive it would be possible turn on up to seven LEDs at a time (it could be all seven LEDs in any one of those eight blocks).

djulien
03-14-2009, 04:29 PM
Djulien - if the controller had sufficient current drive it would be possible turn on up to seven LEDs at a time (it could be all seven LEDs in any one of those eight blocks).

Good point! I was thinking in terms of one of the Renard PICs (90 mA) and no drivers, but if drivers or a 200 mA PIC is used, that would allow a whole group to be on at a time and ease up the addressing timing a lot.

don

djulien
03-14-2009, 04:49 PM
depicts an alternate way of drawing a charlie-plexed setup. You need to trace things through

Here is another view of the same diagram, but it's a little easier to see the Charlieplexed pairs (the LEDs that are "upside down" are the complementary LEDs to the upright ones). This diagram was based on a post from Phil a while back about redrawing Charlieplexing to look more like row/column multiplexing.

don

scorpia
03-14-2009, 05:33 PM
How about something like this (use cat5 for the groups of parallel wires):

don


NICE. i stand corrected.

ps if current is an issue why not just add a couple of drive chips. uln200? or jsut add transistors as originally mentioned.

ErnieHorning
03-14-2009, 06:22 PM
The 41203D that I just downloaded from Microchip does indeed show 90 mA...I guess you should never trust Mouser to keep their links up to date. That certainly puts a kink in these plans.An even bigger change was that the “Maximum current out of VSS pin” went from 300mA to 95mA and the “Maximum current into VDD pin” went from 250mA to 95mA.

P. Short
03-14-2009, 07:43 PM
NICE. i stand corrected.

ps if current is an issue why not just add a couple of drive chips. uln200? or jsut add transistors as originally mentioned.

The ULN parts would not work, I think, but the discrete transistor method is fine.

The part that I would select is the PIC16F627A, which (in the I/P flavor) costs about $0.10 more. This part still seems to be spec'd for 200 mA (I'm wondering if the 90 mA spec for the PIC16F688 is a mistake, as that is the value for the 8-pin PICs). And 2N4401 transistors are cheap....

ErnieHorning
03-14-2009, 10:26 PM
I'm wondering if the 90 mA spec for the PIC16F688 is a mistake…It was changed in rev DS41203C back in 2006.

awhaley
03-15-2009, 01:56 AM
Sorry! I took a quick glance at the diagram and thought I saw bussed inputs that would require independent control of the commons! Sorry for posting without more careful study!

I guess that's evidence of your good work too! Charlieplexing diagrams usually look so complicated! The fact that you make it look simple enough that the ignorant (me) confused it with simple multiplexing is impressive. :)

Art

djulien
03-15-2009, 03:21 AM
I guess that's evidence of your good work too! Charlieplexing diagrams usually look so complicated! The fact that you make it look simple enough that the ignorant (me) confused it with simple multiplexing is impressive. :)

Actually, it was Phil's post in another thread (but I can't find it now) that redrew the charlieplexing as a row/column matrix.

don

awhaley
03-15-2009, 05:42 AM
If you make it not look like a mess to wire, it starts to look legit. ;) I still think the name 'charlieplexing' makes it sound too much like something your uncle charlie cooked up on a napkin... but... ;)

I'm torn on using the charlieplexing for this partiular application though... It cuts down on the number of ICs... but greatly increases the number of wires running from each one of them and I'm not sure that bundles of cat-5 coming off and then wires splitting out here and there makes for a pretty icicle curtain device... at least when compared to one IC every two or three icicles with fewer wires coming out. But it's hard to beat the cost and efficiency of getting 56 individually controlled pixels off on 16f688... so, like I said. Definitely torn.

Art

P. Short
03-15-2009, 11:03 AM
As I see it, the wiring complexity is more or less the same in both cases. For charlie-plexing in the eight-output situation the wiring would be more or less as depicted in djulien's drawing above. It would be eight wires (like perhaps a CAT5 cable) leading off from the controller. Then, for each side-strand, there is a tap taking a small number of wires down that side-strand. The number of wires that are tapped off depends on how many lights there are on the sub-strand, I would think it could be as few as three wires if there are three or six lights in the sub-strand.

JFTiger
03-19-2009, 04:00 AM
I think I opened a can of worms with this one ;) But if it does come to fruit, it will be worth the effort. I envision using as many as ~200 strings on my house so if I can be of any help, I’m in.

I also have little idea how charlie-plexing works, so I’m confused on how this will work out in the end. I’ll google charlie-plexing and learn more when I have more time.

I personally like the idea of varying sizes of light strings. I’d prefer it to be random or user selectable as to the length of each string would be though.

Ronp
03-19-2009, 09:24 AM
This idea is doable most of the work is in the wiring I have something similar the bench now using the original ledtriks panel wiring.
My prototype has 8 leds per drop using 9 wires …8 anode wires daisy chain to the next 8 leds the full length of the 48 icicles string the 9th wire is the cathode witch runs the full length of the string from each icicle to the controller on the end .
If I ever complete this, the finished string would be 48 ft long using 6 inch spacing between the icicles or 2 strings at 24 foot each. Syncing them to music isn’t possible with the controller so I put this on the back burner with hopes a new controller would come along.
Also my setup only used one color possible to use bi-color to get 2 but would cut the string length in half

Ronp
03-20-2009, 05:26 PM
To give you the idea of what im doing here is a icicle picture 96 in a row spaced 6 inches
Also only 7 led to make 1 cat 5 work

ErnieHorning
03-20-2009, 11:31 PM
I like the idea Ron! Try one with the LED’s facing towards the inside instead and see if the tube will light up any. I know the light mostly goes straight out the front, but not 100%. Maybe try varying the depth of the LED’s also. This is similar to an idea that I have.

Ronp
03-21-2009, 05:39 PM
The leds I have are 20 only degree Im hoping from 20 ft upand 25 away they will look good (My display is only veiwed from the street)

this picture is just another idea im thinking about.this should keep them straight and help
with storing .

Ronp
04-03-2009, 11:23 PM
I havent had muck time to play but its getting there...maybe buy sunday ill do a test run
i just have to conect the 8 cathodes and fire it up.

P. Short
04-04-2009, 12:40 AM
It looks very interesting.

I'm noodling around with the idea of using RGB LEDs with a small ($0.50) PIC on each light to do something similar (each LED would be independently dimmable). Obviously this would be somewhat more costly than what you're doing. I haven't given it much thought yet, but it might be more economical to use either a 7x6 or 9x8 charlie-plexed scheme with a small PIC for every few risers.

Did you look into using some sort of transparent or translucent tube (or channel) in place of the PVC tubing?

vziukas
12-20-2009, 02:39 PM
Just want to bring same topic back ...
What if to use LEDTRIKS - TRICK-C board for that ?
By default matrix is 48x16, so its possible to have 48 hanging icicles with 16 leds each, or wire to 96 icicles with 8 leds each.
Just currently no way to control from vixen nice way as separate channels ...
But will be quite challenging project to have another plugin for LEDTRIKS to address each pixel of 768 led as separate chanel for vixen ...

I have snow on my ledtriks. Watch at about 1:10 min on
http://vimeo.com/8026792

What you guis think ?

P. Short
12-20-2009, 07:22 PM
The plugin would be trivial for a number of people here...

It would be a wiring nightmare, I think...

BillAd
12-20-2009, 07:45 PM
A local hotel has a Christmas tree decorated with about about a dozen 'snowfall light tubes'. These are strips of leds that simply cascade from top to bottom to look like falling snow. Each strand is independent such that the cascade is not synchronized with any of the others. The look was really nice and seems like it might be one application of the design discussed in this thread, although the control possible by Vixen could be quite an improvement.

A Christmas store in the area sells the snowfall light tubes on their internet site:

http://shop.christmascentral.com/Items/snowfall%20sf1000mm%205%20tubes?&caSKU=snowfall%20sf1000mm%205%20tubes&caTitle=Snowfall%20-%20Set%20of%205%20Single-Sided%2040%22%20LED%20Christmas%20Light%20Tubes

David_AVD
12-20-2009, 07:50 PM
One chip I'm partial to is the UCN5821 / MIC5821. It uses a S/R input and has 8 outputs. That would get the micro I/O pin count down and still have plenty of current capability. Of course this is for non-multiplexed wiring. One good thing is the micro doesn't have to chew up CPU cycles just refreshing a matrix. There's plenty of other LED driver chips out there from Allegro, etc. Just a thought in case the wiring of a matrix gets out of hand.

holtm
12-20-2009, 08:41 PM
I've been looking at Microchip for a simple solution or a least a chip than can source/sink enough current. They have a 16F1826 chip on the horizon with 32MHz internal oscillator and in the I/P version can sink 200mA from all ports or 25mA by any I/O pin. Could it be possible to use a pcb header for each icicle with just 1 chip and then 12 LEDs hanging below? I'm thinking single color LED for cost and simplicity.

rsriverrat
12-21-2009, 02:28 AM
all this talk is kind of ironic, i was just talking with someone about using a triks to make a mega tree, most people use 16 strands of lights around their tree, so if one were to make 16 rows whicj would be the uprights with 48 leds spaced 3" apart you could get a 12' tall mega tree, use 4 different color leds and 4 triks and you could have a full 4 color mega tree. just put it all in some kind of clear plastic tube, maybe the kind that protects flourescent lights, that was gonna be my project for 2010 among a few others.

Ratt

budude
12-21-2009, 04:27 AM
Keep in mind that LEDTriks does not support dimming - only on/off so if that's OK for your Mega-Tree then it would be possible. It would also be very tough to sync up the LEDTriks to your sequence...

Entropy
12-21-2009, 12:03 PM
It looks very interesting.

I'm noodling around with the idea of using RGB LEDs with a small ($0.50) PIC on each light to do something similar (each LED would be independently dimmable). Obviously this would be somewhat more costly than what you're doing. I haven't given it much thought yet, but it might be more economical to use either a 7x6 or 9x8 charlie-plexed scheme with a small PIC for every few risers.

Did you look into using some sort of transparent or translucent tube (or channel) in place of the PVC tubing?

I've basically been toying with the same idea, using an ATTiny25 instead. They're unfortunately more expensive ($1.10/chip or so) than the PICs, however they're faster and have a hardware USI (kind of a midway between pure bit-banging I2C and a fullblown I2C slave in hardware). The $0.50 PICs seemed slower and didn't have any hardware I/O facilities. Plus I have no experience with them. An experienced PIC programmer probably could pull it off though. More on this after Christmas - I have 10 ATTiny85s on my Christmas gift list, and 50 10mm RGB LEDs on order from Evil Mad Scientist Labs - hopefully they arrive before I drive home to New Jersey!

Individual RGB PWM per node provides the easiest wiring. See mrpackethead's tree as an example - that wouldn't be possible with any multichannel driver solution.

For something like this, an NXP PCA9635 per "icicle" might do the trick. Someone commented "why have it when you still need the microcontroller to drive it" - You would only need one uC to drive multiple PCA9635s. So something like a 9635 for each 5-light RGB "icicle" (16 channels -> 5 RGB triplets with one spare) with a single uC controlling all of the 9635s in use.

If you don't mind the wiring challenge from hell, the Rainbowduino can control up to 192 individual LEDs in a charlieplexed matrix fashion (64 each of red, green, blue is the standard configuration, specifically it's designed for an 8x8 RGB matrix) with 4 bits/color dimming according to the documentation.

Edit: I was somewhat wrong about the Rainbowduino - it's a standard matrix driver, not charlieplexing.

Ronp
12-21-2009, 12:28 PM
The wiring is a nightmare.I think the problem I ran into was when you seperate the ledtriks pannel adding long wire between the leds the controll signal began to breakdown.
That and I did burn a trace on the ledtriks pcb during testing and had a few chips fail over all.
So I put this on the back burner for a wile.

Ronp
12-21-2009, 12:49 PM
all this talk is kind of ironic, i was just talking with someone about using a triks to make a mega tree, most people use 16 strands of lights around their tree, so if one were to make 16 rows whicj would be the uprights with 48 leds spaced 3" apart you could get a 12' tall mega tree, use 4 different color leds and 4 triks and you could have a full 4 color mega tree. just put it all in some kind of clear plastic tube, maybe the kind that protects flourescent lights, that was gonna be my project for 2010 among a few others.

Ratt
I did this test with a mini tree but the original plan was a mega http://www.youtube.com/watch?v=IgOeeTMg9NU&feature=related
Good luck wiring it lol

Photovor
12-21-2009, 10:25 PM
I've been down this road part-way, though only using red leds on a bread board, but I have a schematic and code for I believe 12 charlieplexed LEDs with PWM (fade out as it drops down) built for a 12F PIC. I wanted to attempt it this year, but never got the LEDs to make it work. I'd be more than happy to collaborate on the plan.

dmoore
12-21-2009, 11:21 PM
A local hotel has a Christmas tree decorated with about about a dozen 'snowfall light tubes'. These are strips of leds that simply cascade from top to bottom to look like falling snow. Each strand is independent such that the cascade is not synchronized with any of the others. The look was really nice and seems like it might be one application of the design discussed in this thread, although the control possible by Vixen could be quite an improvement.

A Christmas store in the area sells the snowfall light tubes on their internet site:

http://shop.christmascentral.com/Items/snowfall%20sf1000mm%205%20tubes?&caSKU=snowfall%20sf1000mm%205%20tubes&caTitle=Snowfall%20-%20Set%20of%205%20Single-Sided%2040%22%20LED%20Christmas%20Light%20Tubes

Here are some nice snowfall lights:

http://www.christmas-leds.com/category.aspx?categoryID=313

smartalec
12-22-2009, 12:04 AM
the icle lights i got was in banks of 3 drops, 1drop 5lights, 2drop 4lights, 3drop 3lights,
an there was 10 channels on shorter sets an 12 channels on the longer sets.
i ended up tieing 4 sets to a 7channel chaser in 2004, running as my extended fence, total length was about 30-40meters i think off hand.

djulien
12-22-2009, 11:37 PM
I've been down this road part-way, though only using red leds on a bread board, but I have a schematic and code for I believe 12 charlieplexed LEDs with PWM (fade out as it drops down) built for a 12F PIC. I wanted to attempt it this year, but never got the LEDs to make it work. I'd be more than happy to collaborate on the plan.

I am somewhat interested, although I'm not quite sure what my final direction will be. I'm wanting to do some extensive triks-like subsystems, but the physical characteristics of my applications are somewhat different than what the normal swetup would be. But if you already have some of the pieces of the puzzle figured out, that would be a head start!

don