Possible to DIY a Low-Current "NeoPixel"?

Thundercat

Well-known member
Background:

Hi all, been researching hours to try to find a way to incorporate a bunch of NeoPixels into a Teensy 4.1 project. Basically, I want to have a rotary encoder with 16 LEDs around it light up as the encoder turns, like you see on many professional mixing desks. There will be 8 encoders, so that's 128 LEDs! And each one to have RGB capability. So that's a lot of power.

Obvs this is way too much on a Teensy 4.1 plus I have other devices that will require power too, and I don't want to add an external PSU. I want it all to work over USB-C. Haha, maybe impossible.

I had considered using these tiny NeoPixels, but again, too much power: https://www.aliexpress.com/item/32958032852.html

I'm not convinced that trying to use them with a low brightness would work to keep power consumption into a useable range for the project. I have a 2A regulator, which I could bump up, but USB spec only allows for 5A when following the rules.

Desire:

I found some ultra low-current LEDs that are tiny, and look perfect for the use-case. They consume 2mA or less, and they are RGB: https://www.digikey.co.uk/en/products/detail/sunled/XZCMEDGCBD110W/10449796

The question is, is there a way to use these and have the same functionality as a "true" neopixel, with the individually addressable and dimmable LEDs etc?

I assume we would need a driver IC either per LED, but after a lot of googling I can't seem to find anything that makes sense. Would each LED need its own dedicated driver IC like the NeoPixels?

If so, what might work? All the googling returns is how to use existing NeoPixels, and searching for "DIY NeoPixels" results in how to USE existing NeoPixels instead of how to MAKE them.

Also, I don't want to use a bunch of pins on the Teensy 4.1 because I'm already using a ton.

Thanks for any hints or tips.

Mike
 
There are some types of addressable LEDs also in different sizes. Adressing all 128 LEDs via one pin is handy and very quick.

Why not use, e.g., 10% of power (or even less)? That is roughly the same as the mentioned 2mA per LED. WS2812 and similiar in all it's available versions should be more than bright enough for this application today.
I would get some of those WS2812 LED rings and play around with it to get a feeling and measure the real current. In my experience the current is lower than often expected.
 
There are some types of addressable LEDs also in different sizes. Adressing all 128 LEDs via one pin is handy and very quick.

Why not use, e.g., 10% of power (or even less)? That is roughly the same as the mentioned 2mA per LED. WS2812 and similiar in all it's available versions should be more than bright enough for this application today.
I would get some of those WS2812 LED rings and play around with it to get a feeling and measure the real current. In my experience the current is lower than often expected.
Thank you @TomChiron. That's a great idea. I guess I was concerned they would cut out completely - this happens on my other RGB LEDs if PWM duty cycle goes too low.

Are you suggesting then that I do just that - use, say, 20 instead of 255 brightness? I'm not sure I'd have a useable range of brightness if it works at all. I'd want the user to be able to bump the brightness up or down and this makes for some very noticeable steps in the graduation of brightness.

I'll order some of the standard ones and see how it goes. I'd still need to make my own configuration, and I'd use the smaller SK6812 1515 with it.

I'm still curious if it's possible to make a DIY'ed NeoPixel using the lower current LEDs, as it would allow for a larger range of brightness graduations while using PWM, instead of the stepped, noticeable result when you start at a super low value from total possible of 0 - 255. I'd have the whole range. But I suppose only testing with existing hardware will tell the tale best.

Thank you again,

Mike
 
The neopixel chips define the output current, so the only way to get less bright ones is to find less efficient LEDs, not easy these days, and I'm not sure you can get the neopixel chips separately from an RGB LED package.
 
The neopixel chips define the output current, so the only way to get less bright ones is to find less efficient LEDs, not easy these days, and I'm not sure you can get the neopixel chips separately from an RGB LED package.
Thank you. That was my suspicion - and hence the question, are there any IC chips that can drive LEDs in a similar fashion...Finding low current LEDs isn't hard, but as you say the neopixel ones are married to their IC.

Thanks Mark.
 
A few more thoughts:

You can get the WS2811 as single chips. These one are on the 12V LED strips, where one chip drives 3 RGB LEDs.
I don't know if or how it would be possible to use them with low current LEDs.

There are smaller versions than the very common 5050 WS2812(b) LEDs, e.g.:
They seem to draw less power. Final judgement would have to be made by trying out and measure.

With a bunch of multiplexers (CD74HC4067 or so) there would be probably something possible. But that is less flexible than being able to just add another 100 LEDs without having to change anything of the rest of the circuit.
 
A few more thoughts:

You can get the WS2811 as single chips. These one are on the 12V LED strips, where one chip drives 3 RGB LEDs.
I don't know if or how it would be possible to use them with low current LEDs.

There are smaller versions than the very common 5050 WS2812(b) LEDs, e.g.:
They seem to draw less power. Final judgement would have to be made by trying out and measure.

With a bunch of multiplexers (CD74HC4067 or so) there would be probably something possible. But that is less flexible than being able to just add another 100 LEDs without having to change anything of the rest of the circuit.
Excellent thank you 🙏

I have also found SK6805 RGB chips - with built in LEDs and just 2mm x 2mm.

They claim to use 5mA constant current, although I can’t seem to find out if it’s 5mA total, or 5mA x 3 LEDs.

Either way that would be a win.

Worst case, 128 LEDs x 15mA = 1.92A.

Likely I’d run them very low, perhaps 20% or less, so suddenly this seems doable.

20% x 1920mA = 384mA, and we’re in business.

And 1/3 this if it’s really only 5mA each.

I’ll post back as I develop this further.

Thank you all!

Mike
 
Back
Top