Powering NeoPixels from 3.3V - Wrong but everyone does it?

[sebmadgwick]

NeoPixels are prolific, used in countless hobby, industrial, and commercials products.  They have become synonymous with "RGB LED".  There a several variants but the datasheets all specify an operating voltage of 5V, more specifically, 3.5V to 5.5V.

Nevertheless, I have seen many designs power NeoPixels from 3.3V.  I do not do this myself and will always provide a 5V supply, using a charge pump or similar if necessary.  It just seems unacceptable to me to end up in a situation where a product I have designed starts misbehaving because I ignored clear engineering data.

I am also not satisfied assuming that empirical data overrides datasheet specifications.  Just because the first ten thousand units work, it does not mean that the next ten thousand will.  Manufacturers change components characteristics knowingly or unknowingly and only the datasheet specification can be relied upon.

So my question is, can NeoPixel 3.3V operation be verified analytically, such that a design engineer can take an informed risk?

For example, perhaps it is known that the logic is 3.3V compatible CMOS but the LED Vf is 3.5V which would mean reliable operation at 3.3V but with compromised brightness / colour mixing.

[aliarifat794]

If the NeoPixel uses 3.3V compatible CMOS logic, it can generally accept logic signals at 3.3V without issues. This means that the control signals from a microcontroller operating at 3.3V should be compatible with the NeoPixel.

[langwadt]

If the NeoPixel uses 3.3V compatible CMOS logic, it can generally accept logic signals at 3.3V without issues. This means that the control signals from a microcontroller operating at 3.3V should be compatible with the NeoPixel.

datasheet for them usually says Vdd min 3.5V, probably because if the Vf of the LEDs, and Vh_min 0.7*Vdd

so 3.3V violates the datasheet for Vdd, and 3.3V logic isn't enough for 5V Vdd

(Neopixel is an Adafruit brand name, the LEDs are WS2812 or similar)

[sebmadgwick]

Vh_min 0.7*Vdd... and 3.3V logic isn't enough for 5V Vdd

It sounds like you've misinterpreted how logic thresholds are specified.  If Vh_min is specified as a function of VDD then VDD refers to the actual voltage on the VDD pins, not the recommended typical value for VDD.  For example, if VDD was 4V then Vh_min would be 0.7*4V.

[langwadt]

Vh_min 0.7*Vdd... and 3.3V logic isn't enough for 5V Vdd

It sounds like you've misinterpreted how logic thresholds are specified.  If Vh_min is specified as a function of VDD then VDD refers to the actual voltage on the VDD pins, not the recommended typical value for VDD.  For example, if VDD was 4V then Vh_min would be 0.7*4V.

I said "for a 5V Vdd", and 5V is the most likely supply voltage

[thm_w]

Ignore the first response above.
We are not talking about logic levels, OP is talking about supply voltage specifically.

WS2812 just has a recommended range of 4.5 to 5.5V, no specified value I see. Blue Vf = 3.2 to 3.4V
WS2812B datasheet mentions 3.5 to 5.3V, but then later gives a recommended value of 4.5 to 5.5V. Blue/Green Vf = 3.0 to 3.4V
SK6812 mentions 4.5V, 5V, and 5.2V without giving a specific minimum. Blue/Green Vf is stated as 3.0 to 3.3V.

So no, you can't reliably have any of these working at 3.3V Vdd because the Vf of the LED could also be 3.3V or higher (as langwadt mentioned). You might get a bit of light but not the specified amount.

So my question is, can NeoPixel 3.3V operation be verified analytically, such that a design engineer can take an informed risk?

For example, perhaps it is known that the logic is 3.3V compatible CMOS but the LED Vf is 3.5V which would mean reliable operation at 3.3V but with compromised brightness / colour mixing.

If you don't care about brightness meeting spec then use SK6812 or something with a low Vf as possible.
But then whats the point of having a design without consistent brightness?

[paulca]

They do, usually, run on 3.5V or even 3.3V.  When the chips are running on 3.3/3.5V  they "usually" respond absolutely fine.

What isn't fine is running the LEDs on 5V and sending 3.3V logic levels.  This usually results in corrupted pixels or flashing on chips which default to white on error.

When I say "Usually" it's a YMMV thing.  I have had strips that just wouldn't work at 3.3V and others that didn't care.

[kevin.gibbs]

The problem is most likely the controller's power supply. If it is powered from 3.3V, its voltage may not be enough to control the neopixel when LEDs are powered from 5V.

 Indeed, a low power supply is likely to affect the intensity of blue and green colors. I believe that you can make any variant for yourself as long as you are satisfied with the result. But when you offer someone to replicate the design (or sell it), you should stick to the specifications or honestly warn about shortcomings.

[Siwastaja]

Remember that most 3.3V rated microcontrollers (etc) are rated to max recommended operating voltage of 3.6V. Sometimes this is all you need to interface (or like in this case, share power) with some "weird" chips, without needing multiple supplies or level shifting.

If you do use 3.6V, you might want to carefully look at regulator tolerances and maybe pay a bit premium for a 1% part instead of the absolute cheapest 3% accuracy regulator, though.