Project with many speakers, sensors and leds

Hello all!

I'm hoping you can help give me some guidance and advice on this project. I have worked with all of these technologies on a smaller scale, but I am embarking on a project that will use many instances of what I mentioned in my title. So I feel there are some issues, warnings and technologies I should be aware of!

I am planning to use 500 - 1000 small 6-8ohm speakers, pir sensors, and leds for an installation. To clarify, that is 500-1000 instances of each of those elements. I don't have specs because I am not quite at that stage. But to give you an idea the leds would be RGB maybe around 12v. About 4 of these groupings would be placed every meter in a grid around some walls.

I would ideally like a centralized system so I can easily program this project. what kinds of technologies should I be looking for to interface all these groups of speaker/sensor/led, to make the wiring and programming simpler? I will have a crew to help me, but if there are some technologies that would make this easier/more feasible I would appreciate your recommendation!

Thank you :)

To clarify, that is 500-1000 instances of each of those elements.

Then I think you have moved outside the realm of micro controllers for controlling all of it. What audio do you plan on having the speakers produce.

This sounds like a project a friend of mine did called “sonic pixels” although only 36 units were involved and it is complex and expensive.

Thank you, and a cool project! The audio will just be tones, can be whatever is the easiest waveform to produce. Every element is very rudimentary, it is the scale and behavior that would make the piece impressive. The budget is quite large, so expense is not a huge problem. I am open to a decentralized system of many microcontrollers or simple circuits, but I would love to have control and easily program it if feasible!

Any chance of getting a point in the right direction if I am to be leaving microcontrollers? I was planning to use my computer softwares to control the behaviors and maybe use DMX universes/Drivers to control some elements. But I am really unsure of how to interface all of those PIR sensors ://


Can you please tell us your electronics, programming, Arduino, hardware experience?

Thanks..Tom.. :)

One DMX network will control 512 nodes, so you can get up to your required number with two networks. There is commercial software you can get you can run on a computer or laptop to handle this number of nodes.

All you have to do is to design and build the nodes. :-*

You will have a big problem distributing the power to these nodes as the total current you will need for this many will exceed what you can put through a flexible cable. This was a big problem with the sonic pixel project and that had only 36 nodes. I think the trick is to distribute the power at a high voltage and regulate down at each node with a switching regulator to trickle charge a battery.

I would recommend you try something like a 20 node system before embarking on the bigger project. I think you might have underestimated the cost of all this.

Hello, thanks for the responses!

I have been working with programming/electronics/arduino the last 6 years. Doing lot's of smaller scale projects with various led lights, motors, sound, and hobby sensors. The largest motor project I did was interfacing 50 servos to replicate earthquakes (Through arduino). The largest LED light project I've done is with 40 12v LEDs (using LED DMX driver).

My budget is $70,000 - $100,000 do you think that is not big enough? I will only make enough groups within the confines of the budget (that's why I have a wide rage of 500-1000 groupings). Unfortunately the organizers of this are rushing everything... I dunno if I will have enough opportunity/time to experiment with smaller setups, unless they are really small (The location of the actual project is in another country!)

Unfortunately the organizers of this are rushing everything.

Been there a few times and let me tell you it never turns out well.

Scaling up any electronics project is never a case of just doing more. You get all sorts of problems you never thought of, like the power problem I mentioned before.

do you think that is not big enough?

Touch and go, while $100 per unit sounds likely do you have to factor in installation in an other country and travel in this budget?

At $100 to $200 per node you don't need to worry about the individual nodes. My worry would be how you are going to establish bidirectional communication with 1000 nodes. You need some way to address each node but having to manually assign an address to each would be a major pain. It would be cool to have connectors on all four sides and wiring the nodes in a mesh. Each node would wake up and probe the four neighbors to see where it was in the mesh. Some "master node" would have a jumper set and the mesh members could then work out where they were relative to the master. Like: "The node on my right says he's three rows up and four columns to the left so I must be three rows up and five columns to the left".

So I guess each board needs four bidirectional communication interfaces. Events have to be sent toward the master so they can get to whatever device is taking the inputs (PIR). And the commands for Tone and RGB have to get to the target nodes. I don't think a Master/Slave interface like SPI or I2C would work by itself. Maybe SPI where the Slave has an extra 'interrupt' line to signal that it wants to send. Then you have the problem of some interfaces (like UP and LEFT) being Master and the others (DOWN and RIGHT) being Slave. The ATmega328p only has one hardware SPI interface so it can't be both Master and Slave. It can be Master to two Slaves so that might help.

Once the hardware is sorted out it's just a small matter of programming. :)

Grumpy_Mike: Touch and go, while $100 per unit sounds likely do you have to factor in installation in an other country and travel in this budget?

I do have to factor in travel, but it is a tiny dent in the budget. The installation is in Shenzhen, I'd imagine it will be cheaper to do this project there than here in the US.

Thanks so much for your advice all. I may have to scale the project back/reconsider some things. Unfortunately this project is rushed and I don't have time/means to prototype effectively, although I plan to bring an electrical engineer on board at least to help figure out power and issues around that.

The installation is in Shenzhen, I’d imagine it will be cheaper to do this project there than here in the US.

In that case… :slight_smile: Put three processors on each board:

  • Master has the peripherals, acts as SPI Master connecting UP and LEFT, and acts as an I2C master connecting to Slave1 and Slave 2.
  • Slave 1 acts as an SPI slave and connects DOWN.
  • Slave 2 acts as an SPI slave and connects RIGHT.

The slave processors don’t need a full ATmega328p. They could use any ATTiny that has both I2C and SPI slave capability.