Very high channel count controller - Requirements gathering

[Aussiephil]

Hello everyone.

There has been considerable discussion in the chat room about the need, desire for very high channel count controls.

Now by very high channel count we are talking about 256 channels and above for the exclusive function of LED control at an individual level.

this thread is about gathering your requirements or ideas, it is not about any actual product at this time.

Certianly i would see this being taken on by members of this forum as RJ certianly has plenty on his plate at this time.

There are considerable hurdles such as power supplies to be overcome but it is my belief that this is achievable.

I will kick off the requirements list and we can see what comes out. I will be happy to gather together the requirements and see if there is any common ground

1. DC supply for the board and LEDS, seperate if required
2. DMX enabled
3. 256 channels
4. 512 channels (maybe)
5. outputs via 0.100 header pins
6. current limited outputs at 20ma max
7. SMT Design to keep board size small


Phil

[rrowan]

Hi Phil


Thanks for posting the idea of a VHCCC.

This would help people that need a lot of low power but lots of channels. I don't think there is at this moment another board like it. I know 3 or 4 people in CHAT that have ask for something like this, of course not many people on DLA go to the chat room so there might be more out there.

Your requirement list looks like a good starting point.

When the old brain comes up with anything extra I could need I'll post it.

Cheers

Rick R.

[KeithTarpley]

Greetings,

Knew I'd respond, didn't ya?

1.  I'd prefer the power supply separate.  Assuming a 5 or 12 volt standard, that allows all the various computer and other supplies to be used.  And power supplies do fail.  Much easier to make a connector that can be used with whatever stable source people have.  That also keeps the ac away from the board, solving and preventing related issues.  Reduces board size for what would be an unneeded part, in my belief.

     For me that also allows any power supply noise and size issues to be moved a short distance away from the board.  Allows models to be more likely to have controllers placed inside, nearer the leds. 

2.  DMX seems to be a good way to go.  We already have that part both hardware and software wise dealt with.

3 and 4.  Channel numbers.  There's a point of diminishing return.  2048, I would think.  :-)

     Working with speed limitations of pics, etc., 512 makes one universe go round.  256 acceptable, as that may be what many people would see as enuff.  Half a universe is better than none.

5.  Header pin outputs.  I can work with that.  Seems the simplest solution.  Can fit in small areas.  Easy connectors to find and make.  Kind to small animals...

6.  20ma seems the proper current.  Works with known leds.  Can always be used as a starting point for a rare input to other things.

7.  SMT should be doable.  As far as the parts, we really aren't using anything really exotic here, are we?

     I'll put my support behind this, as I'm looking already at ways to do this.  Let me know what can be done from this end.  And official statement of thanks for looking at the possibility, where ever it leds....

Keith

[christmas_in_VA]

hello, sorry for being so late! I would like to input my ideas. I am very interested in making a cool screen that can control every LED individually with 255 levels of dimming. I am going to try and use modules to run a screen. As in 64 LED's that make up a panel and I connect those with cable to make a universe.

Ben

[MoeJoe]

Has anymore work ever gone into this idea as I'd love to play with a board like this as it really does meet with my requirements in every possible way...

[knguyen916]

Has anymore work ever gone into this idea as I'd love to play with a board like this as it really does meet with my requirements in every possible way...

Hey MoeJoe, as stated in Aussiephil's message it's only ideas and nothing more. Hope that helpsanswer your question

this thread is about gathering your requirements or ideas, it is not about any actual product at this time.

[n1ist]

  My DMX Status display is close to what Ben is looking for.  I haven't tried displaying images on it, but it is a 16x16 matrix, with 256 levels per LED.  See
You are not allowed to view links. Register or Login for the project.

  I am not sure whether a scanned matrix like this is best for the VHCCC.  It may make sense to handle it with a bunch of shift registers, possibly 8 or 16 chains clocked out at once.

  What's the plan for the 20mA current limit?  If we go with individual resistors  or resistor packs, the value would be dependent on the LED color and number in series per channel.
Using just TLC5940s without multiplexing would be too expensive ($80 for 512 channels).
A dual FET constant current circuit per channel cuts that in half.

Sure sounds like a fun project.
/mike

[awhaley]

I think a scanned matrix could be made to work for the project...  The only problem with this might be that, to my understanding, a scanned matrix has to overdrive the LEDs, since they're working at a limited duty cycle?  Does this make it more or less difficult to make the controller fairly universal, as in, able to drive most LEDs without modification?

Art

[Murl]

Thanks for posting the links.  I hadn't seen that before.

You might also want to check out the LedTriks section on the You are not allowed to view links. Register or Login forums.  It's one of RJ's earlier designs for a 16x48 LED sign that you can sequence using Vixen.  The original design was to drive the sign directly from a PC parallel port, but one of the other members (TimW, I believe) designed a board called the Tricks-C to off-load most of the cpu load.  Together, the LedTriks and Tricks-C make a pretty good setup.  I built one using 1" pegboard to poke the LEDs through, which gives me a 4' wide sign that's pretty easy to animate.  Just be prepared to do a lot of soldering - 768 LEDs take a while.   

RJ is also working on an updated version called the LedTriks II during his free time.   There's a section devoted to it on this site, and it looks pretty promising.

-Murl

[RPM]

  My DMX Status display is close to what Ben is looking for.  I haven't tried displaying images on it, but it is a 16x16 matrix, with 256 levels per LED.  See
You are not allowed to view links. Register or Login for the project.

Interesting... did you ever get more boards made up? It would be nice to have a visual display of the DMX universe at the computer and it looks like two of these would do it

  Robert

[Aussiephil]

Well, good to see some people have put forward more ideas.

Looks like i may have a LED controller starting with 32 channels and scalable to at least 128 if not 256, though 256 will force me to learn assembler and change some bits.
Projected cost at this time is around $1.50 a channel including 2A per channel outputs.

I had planned to start at 48ch but my circuit board software ran out of pins allowed.

Phil

[softshelpro]

Bump Bump.  So what the status on this cool toy?

[Aussiephil]

Status:

This was largely an ideas thread, though out of it last year came a 48ch DC low voltage DMX controller. This year will see a 120ch DC Low voltage LED controller (100mA/ch) aimed at RGB control for my Megatree project. I need 864 channels for the tree!
Next year we will modularise the 120ch design so it is scalable to a full 512ch per PIC Controller board. Sadly for all my US based friends i have no plans to support AC based dimming, only DC.

DMX has given us the ability to have 512 channels with just one dongle. RJ and others have given us all cost effective hardware so it becomes easy to have lots of channels. RGB just triples any channel count you may dream of.
In some ways the only thing holding back a serious explosion in channel counts is the current software we use becomes (imho) increasingly harder to use for sequencing as counts go up and  the sequencing time just expands to fill all spare time.
Personally i still believe that 1000 channel count setups will become normal.
Cheers
Phil

[Kwajtony]

In some ways the only thing holding back a serious explosion in channel counts is the current software we use becomes (imho) increasingly harder to use for sequencing as counts go up and  the sequencing time just expands to fill all spare time.

Here Here!

[Aussiephil]

In some ways the only thing holding back a serious explosion in channel counts is the current software we use becomes (imho) increasingly harder to use for sequencing as counts go up and  the sequencing time just expands to fill all spare time.
Personally i still believe that 1000 channel count setups will become normal.

It's funny that you should mention the same thing - I stated this during my presentation at our DIY workshop last weekend, that I felt in 5 years you would be purchasing RGB strings for a bit more than standard strings today.  Out of that comes very high channel counts and out of that comes complications in sequencing.  I think you are right that 1,000 channel count displays will not be that uncommon in a few years - mainly due to RGB.

I think what will come will be software that allows you to define elements (arches, trees, borders, etc) and bulb color and the software will setup the initial sequencing.

Interesting that it is alreeady being spoken about, thanks for that update

You know i would take that one step further, I would say that in 5 years we will have indivually addressable pixels for what we pay for decent LED lights now.

The price of 30LED/metre Flexible RGB string last i looked was heading under $10US a metre in limited quantity direct from the factory this indicates that 10LED/metre (4" spacing) should could easily get under $4 a metre or $40 per hundred leds now if someone wanted to buy in bulk and that is probably quite pesimistic pricing and cheaper than buying seperate colour string (in Aus at least)

Speaking of software, i have been thinking hard on writing a basic utility to handle the pattern generation for my megatree this year.

The tree will be 72 RGB strings each broken up into 8 sections vertically, each string will be paired for 36x8x3 = 864 channels, original plan was to not pair them and run 1728 channels! Maybe next year. The cost was the main factor in halving the count.

With the above complexity whilst Vixen can run the sequence i'm not sure i can use it well enough to make the seqence

Cheers
Phil