next topic »

[on:]

Hi, thanks for checking this. I plan to build a large installation that will essentially be a huge low-res display of roughly 200x100 pixels built from piranha rgb leds. The idea is to put each piranha in a small white plastic lightbox to make the display square rather than seeing the leds directly.

I've been programming for a while but I'm relatively new to Arduino and want to find out the best way of going about this, if this is possible and the kind of hardware I would need.

Is there something that already exists on the market like this as tiles, or is this something I would need to get manufactured to order? It may be there's an alternative to piranha that will achieve the same effect.

If anyone could give me some pointers I'd be very grateful!

Cheers,
Alistair

[Reply #1 on:]

20,000 RGBs is a bit much for arduino.  It would take quite a few arduino mega boards to address 60,000 color levels at usually a byte each.  Maybe a multiple FEZ or Netduino solution would work but be in for a heavy duty project.  Also be prepared to slowly work your way up and deal with the limits you will hit along the way.   This is kind of a high end / high budget project for an arduino solution - particularly if you are new.  You may be better off with a pro solution if you have a deadline.

[Reply #2 on:]

I'd look for commercial advertising screens for this, rather than building your own. It's probably cheaper and easier to just buy a high-res large TV anyway.

[Reply #3 on:]

Thanks for your reply Marklar, I was imagining that Arduino may not be the correct hardware for something on this scale. To be honest, I don't mind what the hardware is (could be a pc or mac) as long as it achieves the effect I'm after.

So, my revised questions:

Do you think you might be able to suggest a pro solution? I'm planning on getting someone in to work on the project with me with the relevant expertise.

Also, does anyone know what would be the maximum number of piranhas I could power? would 1728 be possible? (72x24 res).

What would be the best platform for this if not Arduino?


Hi singul4r1ty, I'm looking to cover a wall 4m x 3m so an actual screen this size or near it would cost a fortune I was kind of hoping I may be able to build something v low res within £5k


Thanks for any help in advance!
Alistair

[Reply #4 on:]

~2000 leds x 60mA = 120 A plus too much wiring.
I think,  projection TV what you need.
http://www.pcmag.com/encyclopedia_term/0,2542,t=front-projection+TV&i=43520,00.asp

[Reply #5 on:]

I'm looking to cover a wall 4m x 3m so an actual screen this size or near it would cost a fortune  I was kind of hoping I may be able to build something v low res within £5k.

Do you want to "watch" this like a TV or is it something you want for some type of artistic display? 

If the idea was to use leds to create a TV the projector idea is hands down the best idea.  You should be able to get a nice new projection system for that amount and the results will be something you can watch.  Do not buy a used one unless you can test it and know you a new bulb and the used projector for much less than a new one.  You can do lots of cool stuff with a projection TV but they have limited "hours" they can run before you have to pay in again. You need space on the ceiling and between the projector and the screen for projection so there may be logistical considerations.

If you are looking for some type of artistic display and you are willing to scale down to 1800 LEDs then you can fit that into the same budget.  To run 1800 LEDs you need three 200 watt power supplies (1800*.3 watts = 540 watts).  Note that the 540 watts would be only when on full white which would be "crazy bright", normally you will be pulling 1/2 or less running patterns.  You can also run say just the top line on very low output to pull the same power as a 15 watt night light and provides a great effect.  There are brightness, capability and long life advantages of using an LED solution.  But if you plan on attempting to create a watchable screen then the LED solution is the wrong direction.

Any details you can provide about the usage of the end product will help in us providing you with advice and direction.

[Reply #6 on:]

Thanks again Marklar, my idea is to create a more artistic installation rather than a screen to watch videos on. I'm planning on creating the content especially for this display as opposed to trying to play video clips on it. The display would run as a kind of ambient animation and I may connect a kinect to it to cast shadows as people walk past. There is also the possibility of making custom apps for if to allow the display to be controlled from a web browser. I plan to place translucent white plastic over each LED to define the 'pixel' shapes.

I think your LED suggestion sounds best, with this I would aim for a 72x24 resolution display so 1728 LEDs in total.

Do you think piranha LEDs are the best sort of LEDs to use for this and do you know if they are available pre-soldered in tiles? Soldering 1800 LEDs may take me a while If you could give a brief overview of the best approach that would be great.

Any guidance you can give is much appreciated, cheers,
Alistair

[Reply #7 on:]

You can see a demo of my living room installation in the context of a hacked stereo as a controller here.
http://arduino.cc/forum/index.php/topic,67382.0.html
Note: In the related video you will see the lights being shown during the daytime with the blinds slightly open.  At night they are way brighter than they look in this video. 

Soldering 1800 LEDs may take me a while  If you could give a brief overview of the best approach that would be great.

In the above video you can see the lights are in a tiny little channel, this is because they are pre-soldered with 4 inch wires between each LED.  Each LED also contains the chip needed to control the colors and naturally the related resistors.  They daisy chain with 4 wires - data/clock/vcc/ground.  This is hands down the easiest way.  I have implemented 595 and TLC solutions - the wiring / soldering was crazy.  Once I found the soldered solution I never looked back. if you really want to learn wiring and implement a low level solution for the learning experience then you may want to start with some traditional matrix approaches.  If your interested in having an end result with the least work possible and you have around a $2USD per LED budget then I would suggest the soldered / chip included approach.

It sounds like you want to program it yourself.  What you will find is controlling 1800 LEDs takes a ton of memory and you will hit some speed limitations.  Even using controllers you have to track the color of each one at a minimum.  That is 5400 bytes.  You will need a mega to address that many LEDs and quite some time to get it all worked out.  I would suggest if you are programming this yourself that you start with a smaller sampling of lights and work your way up to the 1800 level.  1800 LEDs would be mind bendingly bright in a LR setting. 
 

[Reply #8 on:]

You can get some ws2801 based chips at sparkfun.  This is the chip I use and love - it is easily the best chip out there for this form factor.

I buy factory direct in bulk but did test this strip and it works well.
http://www.sparkfun.com/products/10312
You could get 32 of them and have a 32/32 matrix for $1500 USD for the LEDs (plus the cost of the power and controller(s) ).

Sparkfun does not sell prewired versions of the 2801 based leds but they have an individual breakout.  Waaaay over priced but here is the link so you can see them - http://www.sparkfun.com/products/10504.

Also - the LPD based chips have 32,000 colors instead of the 16 million provided by the ws2801 chip.  That said, you can find some nice stuff out there with this chip.  A guy posted a link to his stuff and that is where I learned of such an animal.  I have ordered leds and spectrum analyzer chips form this site.
http://www.bliptronics.com/item.aspx?ItemID=98

You can get the spectrum shield based on the above guys design at sparkfun.
http://www.sparkfun.com/products/10306
I have purchased 20 of these from sparkfun as well - and love them.

Hope those links help.

[Reply #9 on:]

Hiya, this is incredibly useful stuff thankyou.

I think the pre-wired LEDs you mention: http://www.bliptronics.com/item.aspx?ItemID=98 or the LED strips http://www.sparkfun.com/products/10312 are definitely going to be the neatest solution, your video is v impressive!

Also found this link that has lots of detailed description of this hardware and code libraries that may be useful to others: http://www.ladyada.net/products/rgbledpixel/

The area I'm going to cover is 3mx2m so I'm thinking of having 6 sets of 32 strips which is 192 strips in total. This is going to be very bright for a small area though I am not sure if there is an alternative or if I can just send less current to the LEDs.

If you could advise what hardware would be required to drive this many I'd be really grateful. As far as I can gather I would need to get:

- pre-wired RGB LED strips or pre-wired 'naked' bulbs containing chips (thinking of getting these from UK supplier, I'm in London: http://www.coolcomponents.co.uk/catalog/product_info.php?products_id=706 )

- 5v power supply with sufficient current to power the number of LEDs 

- Arduino mega (would one mega be able to handle this many LEDs?)


Are there any additional connectors/hardware I would need? I plan to control the display from live data so won't have any physical controllers connected initially.

So, my next task is to order some strips and start doing some prototyping.

Thanks,
Alistair

[Reply #10 on:]

The area I'm going to cover is 3mx2m so I'm thinking of having 6 sets of 32 strips which is 192 strips in total. This is going to be very bright for a small area though I am not sure if there is an alternative or if I can just send less current to the LEDs.

6144 LEDs is pretty heavy duty.  Do you need 1 every 1.x inches?  For a 6.5ft x 9.75ft wall - every 2 inches brings your count down to roughly 2281, every 3 inches (a standard for a large matrix) would be only 1014 leds.  These numbers are much easier to manage from a power, memory and update speed perspective than 6144.

If your going to invest in this many LEDs - go with the WS2801 solution on the LPD ones - they are just much better from an end result and ease of programming perspective.

You need some good 5v DC power supplies to run this and make sure if your AC is are 220 / 240 you specify.  To run full power - white you need 1800 watts to run 6144 LEDs but only 685 to run 2281 and 304 to run 1014.

Yes this will be bright but you can decrease the brightness as you may have seen in my video.  This is one by the chip, not the input current.  This is great because it allows you to set the entire scene to say 20% power and then have a full or more power "icon" run around in there and the results are pretty wicked and even shows up cool in video form (rare for LED lights - usually videos don't do justice to the lights). 

I plan to control the display from live data so won't have any physical controllers connected initially.

Not sure I follow this.  When you say live data - do you mean computed on the fly at the controller level or live data from a computer steam or other?

Are you a programmer and if so - PC and/or Micro-controllers?