@Irving
I'm curious, what voltage are you running that ATtiny at in the circuit in post #36?

5v. The TX6122 has an internal zener from VDD to GND. I could use a 78L05 there to reduce dissipation in the LED driver.

 

x300, even if only 1/3 were active at once, you'd be using close to a kW of electricity, 150W or more being lost as heat

I think I said I never envisioned ever having more than 3 or 4 active at one time. Does that help?

 

5v. The TX6122 has an internal zener from VDD to GND. I could use a 78L05 there to reduce dissipation in the LED driver.

I see, thanks.

 

WS2811 is a chip that allows you to daisy chain 300 (or more) together, all controllable separately from a single microcontroller. For each mug you would need s WS2811 chip, and a MOSFET, and probably a capacitor. They are all strung together with 3 wires. They are the same as the more common WS2812 used in strips, but don’t have the LEDs on chip, and therefore can control higher power LEDs.

I don’t think there is a simpler way to do this.

 

I think I said I never envisioned ever having more than 3 or 4 active at one time. Does that help?

OK, plan B

If we arrange Jugs logically (but not necessarily physically) in a grid, say, 8 x 8, with an LED driver per row and a controller per column would control 64 similar size jugs with the proviso that only 1 column could be on at a time, with 1 to 8 jugs in the same column following the same sequence. Extending that idea to 3 grids, one for each size of jug, would support 192 jugs with 1 to 8 jugs on in 1 column for each size - 3 to 24 jugs on in total. The drivers and controllers would no longer be in the jugs but in a separate 'control' box. It complicates the wiring a little but is more cost effective.

 

Why, when I offered a much simpler solution?

1 controller, 300 identical nodes containing only 3 or 4 parts each. Far less cabling.

 

WS2811 is a chip that allows you to daisy chain 300 (or more) together, all controllable separately from a single microcontroller. For each mug you would need s WS2811 chip, and a MOSFET, and probably a capacitor. They are all strung together with 3 wires. They are the same as the more common WS2812 used in strips, but don’t have the LEDs on chip, and therefore can control higher power LEDs.

I don’t think there is a simpler way to do this.

They can control external LED chain at 16.5mA nominal CC, but if you use them to control an external P-channel MOSFET (or an NPN BJT + N-channnel MOSFET) you lose the constant current feature and need a separate CC feed per channel.

So you could replace the microcontroller but the rest would be the same as my Plan A per jug, though it would work out slightly cheaper...

[edit] But they don't meet the TS requirements as ALL LEDs follow the same sequence, they are not individually addressable. Great for LED strips, not great for individual lighting.

 

But they don't meet the TS requirements as ALL LEDs follow the same sequence, they are not individually addressable. Great for LED strips, not great for individual lighting.

They are individually controllable, same as WS2812B. And yes, you need a MOSFET and resistor to control the LED. Like I said, 3 or 4 parts per node. Or, if you prefer, each chip could control 3 nodes. Then you needed 100 chips and capacitors, 300 MOSFETs and limiting resistors. You still get the controllable PWM per channel and only 1 micro and 1 signal line is needed for the entire setup. 100 chips cost $12 on Ebay, probably cheaper on Ali Express.

 

They are individually controllable, same as WS2812B. And yes, you need a MOSFET and resistor to control the LED. Like I said, 3 or 4 parts per node. Or, if you prefer, each chip could control 3 nodes. Then you needed 100 chips and capacitors, 300 MOSFETs and limiting resistors. You still get the controllable PWM per channel and only 1 micro and 1 signal line is needed for the entire setup. 100 chips cost $12 on Ebay, probably cheaper on Ali Express.

Ah I get you. Not individually addressable, but address implied by position in chain. Chain length determined by # of 24bit packets between 280uS 'breaks'. The bit time for 0 or 1 is 1 - 1.4uS * 24 = 24 - 34uS so 300 nodes is 10mS. OK, that works though that timing is based on LED strip with controlled impedances between nodes; I doubt that works for random length wires between jugs - would need to be tested carefully to see how low a speed you can go, maybe 100uS gap between nodes might work as can't increase bit timing within packet. Might need to split into multiple strings.

Dropper resistor for 1 LED/350mA = (24 - 3.2)/0.35 = 60ohm @ 7.35W (10W resistor!). 2 LED = 50ohm @ 6.13W! 5 LED = 22ohm @ 2.8W. These are not viable IMHO, you still need a CC buck per LED module, though that was the cheaper element. Overall would reduce per jug cost to around $2.50.

So worth pursuing subject to the reliability of the comms.

 

that timing is based on LED strip with controlled impedances between nodes; I doubt that works for random length wires between jugs - would need to be tested carefully to see how low a speed you can go, maybe 100uS gap between nodes might work as can't increase bit timing within packet.

I have used strips separated by 5’ of wiring with no special attention and it works fine.

 

As you can imagine, you guys have totally lost me…but that’s fine. I truly appreciate the discussion. This has been in my head for years and don’t see it coming to fruition anytime soon.
Can I still proceed in making the 1,2 and 5 bulb strips I’ve been making?
I’ve gotten pretty good at it but don’t want to sink anymore into it if I’m going the wrong way.

 

Those assemblies can work with the 3 or 4 designs proposed so far, but one wastes more energy than the others.

And since this project is still in the design state you can make sure they will work.

Your choices in the future may simply boil down to costs verses work. (more work=less cost, less work=more cost)

But you should at least commit yourself to actually finishing the project before continuing.

 

OK, plan B1..

Use the WS2811 approach, but use the 'grid' idea to provide n x CC buck at each 'shelf', each serving m jugs. As long as the controller keeps track of which LEDs are 'on' and avoids turning on more than 1 out of m per CC then this will be very cost effective. The cost for a CC buck is under a $ for the components, the bulk of the cost is the PCB, so packing say 8 or 16 onto a PCB per cabinet would make it much more cost effective. I'll work it up tomorrow...

The software will be fun...

 

Interesting idea.

 

But you should at least commit yourself to actually finishing the project before continuing.

Commitment and time I have plenty of…it’s the knowledge, skill and cash that are in short supply.

 

This not the best video but you can see what ws2812 leds kind of can do thats 300 leds full on there only 3 amps full white light is a little more 3.5 amps the lights are blinding and can feel a room with color. https://youtube.com/shorts/r6VVjjebhGQ?si=Waa29CiXXQuh6JzT

 

This not the best video but you can see what ws2812 leds kind of can do thats 300 leds full on there only 3 amps full white light is a little more 3.5 amps the lights are blinding and can feel a room with color. https://youtube.com/shorts/r6VVjjebhGQ?si=Waa29CiXXQuh6JzT

Neat, those LED strips are pretty impressive for ambient lighting in a room, or behind a TV, and that shows the capability of the WS281x devices (tbh I've not used them in anger before so I've learnt something from this exercise too). Having said that, the required light output of a LED inside a jug needs to be much higher in a much smaller space to get the TS' desired result.

I think we're close to a cost-effective and viable solution; as I said right up front, the devil is in the detail. Specifically how to drive a constant current through the 1, 2 or 5 LED with minimal losses but maximal flexibility.

 

Specifically how to drive a constant current through the 1, 2 or 5 LED with minimal losses but maximal flexibility.

I think you worry too much about the current control. A 5V supply and resistor should be fine. Each mug is going to filter a different amount of light through it anyway, so I think they will need to be individually calibrated via the PWM to make the effect consistent. The WS2811 makes that easy to do.

 

The ws2812 you can set brightness

 

The ws2812 you can set brightness

It can only output about 16mA, not enough to light these mugs. If an external transistor is used, you have to limit the current somehow. You can then control the brightness via PWM.