next topic »

[on:]

Hello Folks,

As you can tell by my topic, this is an ambitious project.  I've read the FAQs, topics, and such, and I'm sure this topic has come up before, so I'll try and make it unique.  Apologies in advance if I'm asking questions that can be found in these forums.

I'm creating a 196x32 RGB Led matrix.  Before getting down to buying the components, I would like to propose the architecture and receive feedback:

Here's the scope:  196x32 matrix = 6272 led's.

I've decided to use PWM to control the color of each LED.  I would use cascading shift registers to drive each led individually.  This would essentially require 4 outputs from the arduino:  3 for the modulation of the red, green and blue, and 1 for the data being sent to the shift register.  

I'm concerned with driving each LED individually. that it would not be able to achieve the refresh rate required for smooth transitions.  Each LED would be on for a very small amount of time, and on top of that, it would be modulated.  Since only one LED is on at a time, the desired color would be affected as the intensity of each color (RGB) would be dropping to 0 and back up each time the LED is selected/deselected.  

Am I making any sense or am I completely in left field? I've read the article on Charlieplexing, but I'm going to read it again.

Thanks in advance,

Eugene.

[Reply #1 on:]

a very ambitious project

since you can always buy hardware after the fact, might i suggest not quite investing in the led's right away until you figure out all the hardware?

you may also need to look at using multiple microcontrollers and segmenting the display in order to get reasonable refresh rates my guess is that chaining this many drivers together will not yield satisfactory results. you will probably want segments controlled by microcontrollers that at as frame buffers, and microcontrollers dedicated to feeding display information to those frame buffers. you may also find out that the arduino isn't the best platform for this project. you also need to look at total power usage for the led's which will be quite high.

what do you want to do with the display? that might provide some more opportunity for feedback. do you also have a budget for this project? i would expect it to run easily into the thousands of dollars range.

it's doable though it will just take alot of planning, calculations and cold hard cash do you live near any universities? speaking with seasoned faculty on the subject will answer many questions as well.

[Reply #2 on:]

I've consulted an engineer and he suggested I create led modules... he suggested 16x16 .. each module would require its own voltage regulator, microcontroller and drivers.. you could then plug the modules into each other to create the size of matrix desired..  a master microcontroller would be responsible for sending the data to all the modules, as well as syncing, etc...

i never even considered power requirements for when the matrix is entirely illuminated (duh)..

so yeah, im into thousands of bucks..

what I dont get is:  these matrix's already exist.. the DOT furniture store has a 24x168 RGB display (although i think each led can have 6 colors, not 16.2 million).. do you think these displays cost thousands???

thanks,

-E

a very ambitious project

since you can always buy hardware after the fact, might i suggest not quite investing in the led's right away until you figure out all the hardware?

you may also need to look at using multiple microcontrollers and segmenting the display in order to get reasonable refresh rates my guess is that chaining this many drivers together will not yield satisfactory results. you will probably want segments controlled by microcontrollers that at as frame buffers, and microcontrollers dedicated to feeding display information to those frame buffers. you may also find out that the arduino isn't the best platform for this project. you also need to look at total power usage for the led's which will be quite high.

what do you want to do with the display? that might provide some more opportunity for feedback. do you also have a budget for this project? i would expect it to run easily into the thousands of dollars range.

it's doable though it will just take alot of planning, calculations and cold hard cash do you live near any universities? speaking with seasoned faculty on the subject will answer many questions as well.

[Reply #3 on:]

Here's a place that does them: http://www.selectricsigns.com/  you can find out how much they cost from them.

There is a datasheet there - input power requirments were (IIRC) 12V @ 3A. Not huge, but not insignificant either.

As for how to wire it up - modules may be a the right way. One thing I was thinking is that these kind of displays have been around for a long time, going back to the 8080 or Z80 era. It would not have been usual in those days to have many processors. It would probably haev been done with parallel loaded shift registers.

It might give a clue as to how to structure a modern design.

Also, I seem to remember there being a magazine project some years ago, too. (Probably Elector, but maybe the UK magazine Practical Electronic.

Its a good project - certainly gets the brain going.

Just found this kit: http://www.hobbytron.com/vk5600g.html

Runs off a single 18 pin 18C54 PIC - no sign of anyother ICs, so must be multiplexing the whole lot. I doubt this is how it is done on the commercial units.

Mike

[Reply #4 on:]

Thank you sir.  Yeah. single color led matrix's, I can get my head around that...  but the rgb matrix where I want to control the intensity of each color separately for each led (to give me millions of colors / led), is giving me a headache... I may just try a small 8x8 rgb matrix, 6 colors per led for now..  maybe this is the way to go... anyway, thanks very much for the info...

Here's a place that does them: http://www.selectricsigns.com/  you can find out how much they cost from them.

There is a datasheet there - input power requirments were (IIRC) 12V @ 3A. Not huge, but not insignificant either.

As for how to wire it up - modules may be a the right way. One thing I was thinking is that these kind of displays have been around for a long time, going back to the 8080 or Z80 era. It would not have been usual in those days to have many processors. It would probably haev been done with parallel loaded shift registers.

It might give a clue as to how to structure a modern design.

Also, I seem to remember there being a magazine project some years ago, too. (Probably Elector, but maybe the UK magazine Practical Electronic.

Its a good project - certainly gets the brain going.

Just found this kit: http://www.hobbytron.com/vk5600g.html

Runs off a single 18 pin 18C54 PIC - no sign of anyother ICs, so must be multiplexing the whole lot. I doubt this is how it is done on the commercial units.

Mike

[Reply #5 on:]

something very similar you may look into are gambling machines.

often they can have decent sized arrays of smd rgb led's
here is something i came across
http://tinyurl.com/2x7u7y

as for do these things really cost thousands, i think the answer is yes .. easily.
commercial electronics in general are very pricey especially things that aren't mass produced. you would be amazed how much things can cost.

in the above gambling machine example, this is an array that is much smaller than what you are proposing and the seller states that the panel itself cost thousands of dollars originally.

this is interesting:
http://www.ledtronics.com/ds/rgb1004/