Total Newb: LED Display

[Q__Hybrid]

Hey there kids,

here's the deal: I have absolutely no idea what I'm doing. I can marginally read a circuit, as in I can understand interconnections and can recognize several symbols, but I probably won't know why anything is placed where it is. Pretty bleak, eh?

My idea is not very original in the fact that its architecture is a grayscale LED display. I need someone to walk me through the entire process so I don't get lost or have a terrible accident. (Please oh please god, somebody agree to commit to this!)

Concept: "Illuminated Human"
This is quite ambitious and stands as a true challenge. If it helps persuade someone with serious skill to sign-on, it'll probably be very quick and simple to design and build. Heads up: it requires multiple ICs.

This concept utilizes an LED matrix of an indeterminate size (one diode per square 0.5"). In its final form, the LED matrix is applied to or embedded within a garment of clothing. I have chosen to make a coat as the prototype's fashionable debut. The simplest explanation is as follows:

Information is read from a compact storage medium (possibly CompactFlash, etc) and stored in memory where it can be accessed in its entirity (without lag) by a microcontroller. The micro's job is to access memory and use it to display a grayscale image within the LED matrix. The micro and memory must be able to display full grayscale animation at 30fps for at least 30 seconds, preferrably 60 seconds. I plan on using the scanline method to display each frame. A touch-sensitive switch in the form of embroidered wire thread upon the cuff of one of the sleeves will turn the system on and off.

The major hurdle in this project is that all of the components must be engineered to take up the least amount of space possible, weigh the least amount of weight possible, and use energy as efficiently and/or sparsely as possible. Memory will be particularly intensive because of how much image data will need to be stored for each "pixel."
Most of the system design is oriented around the architecture of the microcontroller, memory, timing of animation, things like that. The actual display structure itself will probably be really simple and easy, albeit monotanous.

I don't have any plans of copyrighting this technology because it's too cool of an idea and too easy to build. But I do plan on selling custom shwag with this design and similar luminescent concepts (EL wire, strips.) So, I suppose being able to capitalize off of the final design is an added incentive. I know I'm gonna be "testing" the original every weekend. :twisted:

Always good to grab some attention...

Anybody want to help me build this really nifty thing?
+Q__

 

[bonxer]

Unless you have a friend / co-worker / unguarded storage room that is going to supply you with the LEDs, that's gonna be hella expensive. I don't know how you would do greyscale lighting, but single "bulb" red-green-blue combination LEDs are easy to come across from https://futurlec.com/LEDGen.shtml. They're 50 cents each. Spaced every half inch, it's going to be $2 per square inch. If you'd rather just do white, they're 70 - 80 cents each, depending on size, so you might as well do color. As for scanlines, making an LED display is different than a TV display. With a TV, the cathode ray can ping each pixel once right after the other, top to bottom, then start again. With an LED display, the minimum duty cycle is 10%. So for however long you visite them with a voltage pulse, you can only have it off 9 times that long before you have to hit it again, otherwise it's just not going to be visible. So if you have only 10 lines of LEDs, no problem, hit each line sequentially. If you have 20 lines, you're going to have to do them in parallel. 40 lines? Going to have to drive 4 lines at the same time. And that just accounts for the horizontal "scan lines". You're also going to have to load the data patterns into the columns of LEDs, so that when the signal is ready to turn on a specific row, all of the patterns are ready to go on every column. It may be best to use latches so you can run a single data bus to all of the sections, and 3-to-8 decoders (or 4-to-16) for addressing the latches. As for testing it every weekend, is this to see how well it attracts or repels females?

How committed are you to this idea?

 

[Q__Hybrid]

Bonx:

LEDs? That's simple. I'll get them from All Electronics at as low as $0.04/ea. I wish I could use less bulky bulbs, but I can't pass up that price.

Displaying the image is actually really simple. If I used something like the TLC5940RHB driver chip, I would be able to illuminate 16 diodes at specific values of brightness. By using multiple TLC driver chips to control the brightness of each diode in a column and then strobing through each row of the image, it really isn't that hard to draw each frame.

The data that defines the value for a given diode is stored in memory where it is then supplied to the appropriate shift register/channel of its respective driver chip. This information is stored so that as the image is strobed, the appropriate line of image data is "called" to the display. In other words, in programming jargon, the enumerations of the scanline's index and the scanline's image data are the same.

If there would still be an issue with the duty cycle, then the system could be divided into parcels of memory equal to the maximum allowable size. In this configuration the operating loop is the same, but the strobing is quicker.

A black and white palette would make things very easy, but grayscale is so much better. Nothing else will do.

I am extremely commited to this concept.

Ok, maybe not every weekend. Repel? Who wouldn't dig a jacket with full-motion animation all over it? To me, that's badass. The only thing cooler than that would be a jumpsuit and matching helmet (like that guy in Daft Punk with the custom cast acrylic thing.)

+Q__

 

[Q__Hybrid]

Moderator:

My wireless connection is really really terrible. One of those drill-click just to make anything transfer scenarios. So, uh, delete these duplicates please. Thanks.

+Q__

 

[Q__Hybrid]

Bonx:

LEDs? That's simple. I'll get them from All Electronics at as low as $0.04/ea. I wish I could use less bulky bulbs, but I can't pass up that price.

Displaying the image is actually really simple. If I used something like the TLC5940RHB driver chip, I would be able to illuminate 16 diodes at specific values of brightness. By using multiple TLC driver chips to control the brightness of each diode in a column and then strobing through each row of the image, it really isn't that hard to draw each frame.

The data that defines the value for a given diode is stored in memory where it is then supplied to the appropriate shift register/channel of its respective driver chip. This information is stored so that as the image is strobed, the appropriate line of image data is "called" to the display. In other words, in programming jargon, the enumerations of the scanline's index and the scanline's image data are the same.

If there would still be an issue with the duty cycle, then the system could be divided into parcels of memory equal to the maximum allowable size. In this configuration the operating loop is the same, but the strobing is quicker.

A black and white palette would make things very easy, but grayscale is so much better. Nothing else will do.

I am extremely commited to this concept.

Ok, maybe not every weekend. Repel? Who wouldn't dig a jacket with full-motion animation all over it? To me, that's badass. The only thing cooler than that would be a jumpsuit and matching helmet (like that guy in Daft Punk with the custom cast acrylic thing.)

+Q__

 

[Q__Hybrid]

Bonx:

LEDs? That's simple. I'll get them from All Electronics at as low as $0.04/ea. I wish I could use less bulky bulbs, but I can't pass up that price.

Displaying the image is actually really simple. If I used something like the TLC5940RHB driver chip, I would be able to illuminate 16 diodes at specific values of brightness. By using multiple TLC driver chips to control the brightness of each diode in a column and then strobing through each row of the image, it really isn't that hard to draw each frame.

The data that defines the value for a given diode is stored in memory where it is then supplied to the appropriate shift register/channel of its respective driver chip. This information is stored so that as the image is strobed, the appropriate line of image data is "called" to the display. In other words, in programming jargon, the enumerations of the scanline's index and the scanline's image data are the same.

If there would still be an issue with the duty cycle, then the system could be divided into parcels of memory equal to the maximum allowable size. In this configuration the operating loop is the same, but the strobing is quicker.

A black and white palette would make things very easy, but grayscale is so much better. Nothing else will do.

I am extremely commited to this concept.

Ok, maybe not every weekend. Repel? Who wouldn't dig a jacket with full-motion animation all over it? To me, that's badass. The only thing cooler than that would be a jumpsuit and matching helmet (like that guy in Daft Punk with the custom cast acrylic thing.)

+Q__

 

[bonxer]

Alright. Sounds fun. What do you plan to use to store the image data and drive your TLC's ? Have you ever used a microcontroller before? A friend and I are currently working on (read: putting off for months ) an implementation of Conway's Game of Life on a 32x32 LED grid, using latches, controlled by a single 40-pin PIC microcontroller. We've got all of the LED matrices mounted, resistor networks, latches, and header pins soldered, so the display unit is finished. We just need to make a few dozen serial cables and a board with lots of headers to plug into, and all the hardware is done. I've written a little chunk of code to visit each of the 64 individual cells in a single 8x8 matrix. It just needs to be expanded to a larger scale.

Have you started on anything yet? Acquiring LEDs? Wiring stuff up? Where are you in the process?

 

[checkmate]

Well, for an absolute eletronic newbie, I do not know how much development/learning time you have, but all those up there could take about a year without help. You'll need microcontroller, CF interfacing, port expanders (A LOT OF THEM). With that kind of density, I would foresee that majority of the weight would come from the wires and batteries.

Each LED consumes 20mA, each square inch consumes 80mA. A single sleeve could easily eat up ampere-level currents. Make a dress out of it and you have a walking generator :lol:

 

[audioguru]

You are too late, they are already being sold.
I saw the jackets and sweaters on TV a few months ago. They are very, very expensive but selling like crazy in rich cities. They look cool displaying full-motion colour videos and marquee signs, "I am sexy" or something! They have their own small smarts module and you download the info to it from a Palm Pilot or RIM-berry. Of course they can also have a wireless feed to show TV or computer video presentations like a monitor screen.

Theirs has very many tiny surface-mount LEDs not just a few discrete ones like yours. By the time you make a crude one, theirs will be improved, truely mass-produced and much less expensive than yours.
Sorry.

 

[jrz126]

I have some experience with using a semi-large quantity of LED's in a project. (62 to be exact). **broken link removed**
I also have a couple of movies:
**broken link removed**
**broken link removed** I'd be willing to lend you some support, although speaking from personal experience, it%27s rather hard and time consuming.' I'd be willing to lend you some support, although speaking from personal experience, it's rather hard and time consuming

 

[Q__Hybrid]

audioguru: @#$$%!!

Damn. Somebody beat me to it? I knew that would happen.

It's important to know if the entire surface of the exterior of the clothing was populated by LEDs. If not, then I still have the upper hand with my concept. Kinda. :x

In regard to the ampere level power usage and stuff, that's not true. All that would be required is enough voltage for the maximum number of diodes within the longest row in the matrix structure. Because we're strobing through the rows one at a time at a very high speed, this is the maximum amount that the system could possibly ever need or use. Still, you're right about how heavy the battery would have to be. I imagine it would be a suitcase-sized thing, etc.

In terms of how far along I am, it's nothing more than a "wow, that would rock!" concept. Now that someone else has already made a topnotch design of the same idea with intentions of mass producing, I'm not as enthusiastic about it. I won't give up on it because I don't have verifiable proof, I just think I want to work on another project in the mean time.

On the subject, it would be awesome to cover the outside of something like a '98-'99 Mitsubishi Eclipse with tri-color SMT diodes...And of course the engine would have to be overloaded to make sure it could win any race...

I'm going to post about my other project, message will probably be titled something like "multi-purpose controller" or something. So, go there because I'm putting this idea on hold.

+Q__

 

[audioguru]

The back of the jackets and sweaters had a large, flexible high-definition colour display. I don't remember seeing any on the front or arms. The controller with its built-in battery was very small. I don't think they said how long (or short) the display battery would last.
Last year i saw a clown with his large hat covered with blinking, colour LEDs. Sometimes I look like a clown and I frequently carry around my blinking LEDs. :lol:

 

[Rescue1]

I could be reading this wrong,but I used to work EMS at a sign show.They have companies that use fiber-optic strands embedded into just about anything,then a series of leds do the job of lighting it.This would be a better way to go as the strands are easier to place and run,and the electrical benifits speak for themselves..... :shock:

 

[audioguru]

Ouch! Glass in a garment?
I have a lamp with a fiber optic "head of hair". It doesn't look very good anymore, bits of glass fibers cover the table under it.
When any glass thingy gets old, the glass becomes very brittle and breaks easily with movement. Plastic isn't much better and is frequently worse after a while.

 

[Halogrunt1234]

about the jacket LED concept already been sold, Just go to a Blue Man Group concert. they had a full 1960's style dress completely covered by LEDs that had a couple of special effects playing on it. It looked extremely heavy and wasnt in any way appealing in fashion, but it was still an interesting concept. the idea is in no way new, and i have no experience whatsoever about the kind microcontrollers and programming that would be required with this almost humanly impossible feat, but i still like the concept.

while we are on the concept of LED matrices, I was just wondering... cant you put an LED matrix in place of an LCD screen such as one on a Ti-83 or any other black and white LCD pixel display? this would most certainly require a massive ammount of LEDs - one per pixel - even for a meager 94 x 62 screen (resolution of a Ti-83), and I wouldn't know how to even start thinking about connecting it to the gadget being altered, because all of the LCD's that I have come across were connected with this strange foam on some transparent terminals on the screen itself. Just wondering

 

[skraiduolis]

would you have any additional info on these garments?

I have started exploring the idea of a wearable display last September... after spending a week at Burning Man... 'nuff said.

I now have a pretty good idea what happens right after the serial control signal gets to the TLC5940RHB IC... this thread has some examples... Cool stuff.

Who could recommend the following specs!
A controller/micropocessor to supply the neccessary serial signal which is derived from a media file, residing on either removable media or transported over IR, wifi or bluetooth.

or,

programming a cascade/ladder type script that would be used by the controller to drive the lights and which would be easily updatable through an SD/USB interface on the processor.

The actual color information would be generated in this case through RGB value tracings from something like Macromedia Flash or Maya.

The plan:
a matrix of at least 64X48 RGB (3072) RGB LEDs distributed over a piece of clothing.
controlled by 576 TLC5940RHB's (3072 X 3 for RGB, one 16 channel IC for 5 LEDs) to achieve 24 bit color depth.

The pricing for this stuff is in the $Thousands of dollars (576 x ~$3.20 for an IC in quantity). Not even mentioning the cost of rgb LEDs, and these would have to be in 2-3mm range.

There is a supplier for conductive metalic thread that could be used for LED lead sewing.


Thanks for any and all infor, in the meantime, I am gethering samples from manufacturersa to start testing the driving and installing these into something wearable...
Good forum!,


skraiduolis





I saw the jackets and sweaters on TV a few months ago. They are very, very expensive but selling like crazy in rich cities. They look cool displaying full-motion colour videos and marquee signs, "I am sexy" or something! They have their own small smarts module and you download the info to it from a Palm Pilot or RIM-berry. Of course they can also have a wireless feed to show TV or computer video presentations like a monitor screen.

Theirs has very many tiny surface-mount LEDs not just a few discrete ones like yours. By the time you make a crude one, theirs will be improved, truely mass-produced and much less expensive than yours.
Sorry.[/quote]

 

[badrilal prajapati]

i want to design fixed celluler terminal by using GMS module please give me suitable guidelines and how to access AT command