WS2811 question

prof

Apprentice elf
Generous elf
Joined
May 26, 2010
Messages
89
Location
Bundanoon
Data sheets that i have seen for the 2811 list the two operating frequencies for the chip, then just say the LEDs are driven by constant current.
I presume this is a PWM signal - if so what is that actual PWM frequency? (Do not have a scope to check it myself).
Thanks.
 

David_AVD

Grandpa Elf
Community project designer
Generous elf
Joined
Jun 12, 2010
Messages
4,404
Location
Victoria Point (Brisbane)
As far as I know the 400KHz / 800KHz speed refers to the data input / output, not the LED PWM rate.

I don't think I've ever looked at the actual LED output PWM rate.
 

AAH

I love blinky lights :)
Community project designer
Joined
Dec 27, 2010
Messages
3,802
Location
Eaglehawk
The PWM frequency is about 2kHz.
They are constant current via a current limiting circuit within the IC. The logic turns on the output transistor until the current reaches 18.5mA and holds it there for the duration of the ON time loaded in (0-255). The IC will typically have a couple of volts between GND and the R,G and B channels when the transistor is turned on. As the voltage along the string drops there will be less and less voltage between the channel and GND and eventually the chip won't be able to regulate the current to 18.5mA and at that point the colours will tend to lose faithfulness and brightness.
 

prof

Apprentice elf
Generous elf
Joined
May 26, 2010
Messages
89
Location
Bundanoon
Thanks, for the answer. Thought it might have been something like that.
Reason for asking is i am thinking about power supply and cable sizing.
Maybe i am over thinking it.
Running the controller at 30% brighness on the port settings for example will reduce the average power consumption (and brightness of the LEDs), but because the pixels are PWM they will still be drawing the peak current - though at a reduced duty cycle.
Should we be designing the power delivery (psu/cables etc) and pixel string lengths in our display around the peak power demand, then throttling back the brightness. Rather than cramming in as many pixels as we can get away with because we have throttled back.
The transients on the supply lines in the pixel strings due to all the pwm switching must be absolutely crazy - considering the pixels themselves only have the bare minimum of smoothing capacitors fitted to them.
Light bulb moment - could the very nature of the PWM load be a contributing factor to the pixel failure issues being seen?
The pixels are actually being stressed at a higher level than we realise because we are only thinking in terms of the "average" power draw and not the peak levels.
 
  • Like
Reactions: AAH

TerryK

Retired Elf
Joined
Feb 9, 2020
Messages
395
Location
West Central Ohio
I would not be too concerned about pixel stress. If the pixel manufacturers design the pixels correctly they should run at a constant 100% white providing the pixel does not overheat. And I think few of us drive pixels that hard.
Yes, I agree the hash on the supply lines is quite obnoxious but it is not just the pixels. The power supply contributes as well and I would not be surprised if there is some being generated in the controller too. Distributed capacitance and resistance in the cables will attenuate some of this as will the already mentioned smoothing capacitors in the pixels (and power supply and controller). Good design practices in the ICs should enable the IC to work properly in spite of the noise unless it become to excessive.

Regarding the supply and cable design, the pixel IC datasheets are rather vague but my impression is they have a free running internal oscillator to clock through the RGB pixel data and clock out the PWM outputs. As such the pixels will be operating asynchronous and from the power supply's perspective see an averaged current from the pixels.

So, for your rather intriguing "Should we be designing the power delivery (psu/cables etc) and pixel string lengths in our display around the peak power demand, then throttling back the brightness. Rather than cramming in as many pixels as we can get away with because we have throttled back." question, I aim for the latter, although I target 70/80% rather than 30% so I can strobe the neighborhood if I want.
 
  • Informative
Reactions: i13

David_AVD

Grandpa Elf
Community project designer
Generous elf
Joined
Jun 12, 2010
Messages
4,404
Location
Victoria Point (Brisbane)
The PWM of each pixel is free running so the current spike won't be syncronous anyway.

You'd have to put a 'scope across a current shunt to see what the actual current spikes were as a whole.
 
  • Informative
Reactions: i13

i13

Senior elf
Joined
Jul 5, 2013
Messages
994
There is a different possible problem with using brightness to reduce current draw. This technique relies on the WS2811 data to limit the current draw. If this data becomes corrupt and the pixels flicker, you're no longer able to limit the current draw.

I design for 100% brightness but I also run the pixels at 100% brightness. I've seen the reduced brightness technique working in practice though. There isn't really a perfect answer.
 
Top