I am doing some work on a matrix that involves some custom electronics. My circuit is connected to the output side of a 12V seed pixel string. During testing today I noticed a weird problem with the regulated +5V voltage rail of my circuit. I measured the voltage with a DMM at about 7V. That voltage immediately concerned me and I began investigating the issue. The +5V rail is regulated from the 12V supply used to power the string of seed pixels. I measured the voltage of the 12V supply to be about 12.4V. What was peculiar was that the 5V rail went to 7V when the lights were turned on. When the lights are turned off the rail voltage drops back to 5V. After quite a bit of investigative steps I measured the data waveform coming out of the last pixel of the seed pixels string using an oscilloscope. The string is 36 pixels of 12V 4 wire pixels. The data waveform coming out of the last seed pixel of the string has a peak voltage about 7.5V. That surprised me as I was expecting it to be 5V. This explains the issue with the voltage of the power rail. When the lights are on, the data line is acting like a charge pump and increasing the rail voltage because of the input protection diodes on a buffer connected to the data line.
What I do not understand is why the data line of the pixel output is more than 5V. Has anyone else looked closely at the voltage of the data out of a 12V seed pixel? Can anyone else confirm this peculiar observation?
My pixels Intest with came from Ray and are thought to be "data return" pixels. I have identified the +V, DATA and GND wires. The fourth wire is unknown. It should be the data return signal so I think that implies it should be common for all pixels from end to end but it measures about 43k ohms end to end (36 pixels). During my tests this line was left unconnected.
The spec for WS28xx pixels allows them to operate at up to 7V. That means the data line can go up that high as well. It also means the resistors dropping 12v to 5v actually only need to drop to 7v on no load and are allowed to drop to 4v on full load.
MartinMueller2003 wrote:
The spec for WS28xx pixels allows them to operate at up to 7V. That means the data line can go up that high as well. It also means the resistors dropping 12v to 5v actually only need to drop to 7v on no load and are allowed to drop to 4v on full load.
Thanks for taking the time to provide feedback.
Referring to the ws2811 v2.2 data sheet the maximum rating for Vi (logic level input) is VCC+0.7V and the Max rating for VCC is 5.7V
Referring to the ws2812B-2020 v1.5 data sheet, the maximum rating for VDD is 5.3V and the logic level (DIN) max Vi = VDD+0.7V
Referring to the ws2814 v1.1 data sheet, the maximum rating for VDD is 5.3V and the logic level (DIN) max Vi = VDD+0.7V
Referring to the ws2815 v2.0 data sheet, the maximum rating for VDD is 13.5V and the logic level (DIN) max Vi = 5.7V
Referring to the ws2816B-2020 v1.0 data sheet, the maximum rating for VDD is 5.5V and the logic level (DIN) max Vi = 5.7V
Referring to the ws2818 v1.5 data sheet, the maximum rating for VDD is 5.5V and the logic level (DIN) max Vi = VDD+0.7V
Unfortunately, there are no Vo specifications for the DOUT pin. However, the logic level Vi specification must imply the maximum level for the logic Vo otherwise a ws28xx chip would destroy its downstream neighbour. Thus Vo(max) must be less than or equal to 5.7V. This has been my understanding.
What appears different with the seed's I have measured is the Vo(max) is much larger than the logic level VDD (5V typical) you would expect. I suspect (I will measure this today) that Vo of the unloaded DOUT signal when disconnected from my circuit may be as high as 12V. The circuit I have the DOUT signal connected to has an TPD4E05U06 ESD protection device that clamps the voltage. The clamp breakdown voltage is 6.5V (min) to 8.5V (max) so I suspect this device is clamping the voltage to about 7V on the data signal. This ESD device appears to have done its job and protected the buffer from failure; though, the buffer diodes are conducting the higher voltage onto the 5V rail and charging the output capacitor of the regulator in a "charge pump" manner. Of course this means the buffer chip is experiencing an over voltage on its VDD power input. So, the buffers are being stressed beyond their design.
Such a large Vo for the data line has other implications as well, notably for EMC emissions. The higher voltage will radiate a lot more noise.
The measurement of the DOUT signal with no load is about 8.6V. That appears to confirm my hypothesis the ESD device is clamping the data signal.
I would be very surprised to see the Do drift that high.
