sporadic wrote:I was curious how responsive the circuit you referenced would be, so I punched it into LTSpice and put a typical 2811 signal through it (1000ns high, 250ns low). Rise time was way too slow w/ the 10k's, but looks decent with a 220.
MOSFETs, almost the perfect device if it wasn't for that parasitic gate capacitance...
Using 220 ohm puts the drive current over spec for the ESP, which can only drive 12mA per IO pin, the best we could do is a 330 ohm resistor. It's a shame there are no buffers like the one you used in a through hole package, that would be the most ideal solution.
madsci1016 wrote:MOSFETs, almost the perfect device if it wasn't for that parasitic gate capacitance...Using 220 ohm puts the drive current over spec for the ESP, which can only drive 12mA per IO pin, the best we could do is a 330 ohm resistor. It's a shame there are no buffers like the one you used in a through hole package, that would be the most ideal solution.
Ah, good catch. I had planned on sticking with the MC74VHC1GT125 buffer anyways.
I'm playing catch-up on the ESP8266 stuff. Grabbed a nodeMCU clone w/ the ESP-12 module for prototyping which should be here the next day or so. Thinking about laying out a board similar to my PixelStick, but using the ESP-01 if these things can perform as well as my current Xmega + nRF24L01 setup. What toolchain setups are you all using? Anyone using the Espressif SDK or pretty much sticking with the Arduino and LUA options?
sporadic wrote:...Thinking about laying out a board similar to my PixelStick, but using the ESP-01...
Something like this:

ESPixelStick by smerrick, on Flickr
The others can probably offer more recent experience. This project fell to my back burner back in January. But at that time the native toolchain was a royal PITA to get set up. It involved a half dozen different packages and compiling it yourself after modding the downloaded source.
--Jon Chuchla--
Sent from my iPhone using Tapatalk
jchuchla wrote: the native toolchain was a royal PITA to get set up. It involved a half dozen different packages and compiling it yourself after modding the downloaded source.Sent from my iPhone using Tapatalk
O it is still is a Royal PITA, especially if you aren't tooled/familiar with *unix toolchain / development environments. My background is in AVRs before Arduino was a thing, with a recent bout of NXP ARMs so I'm used to the super integrated and easy to get going combination IDE/toolchains where you go from installed to compiling in 5 minutes flat. I spent a weekend trying to get Expressif going with Cygwin and gave up. Five minutes later I had the Arduino IDE flashing the ESP with no issues.
The Arduino core for ESP was fairly buggy at first, but has made improvements in leaps and bounds. Besides the 2 different pixel versions I'm working on and the Wifi-Renard bridge, I'v also got 2 non blinky projects going with it so far and have been nothing but impressed.
I'm not a coder but I can somewhat follow along with the Arduino environment. Seems like everytime I get something going I find something else I would like to add to the code. I am working on a configuration page for it and now I need to learn some html. If someone likes the eclipse environment there is an alternative for ESP. Its called Sming. From what I gather and if I'm saying this right the API is like arduino. Seems like it is little more advanced with the accomplishments thats been made from what I see. It's also supposed to be native firmware for the ESP. Thought I would throw that one out there if anyone might be interested in checking that one out.
Update on my testing - I setup a circuit with a level shifter and everything worked great. For comparison I went back to the old way without a shifter and it still worked great. Eventually I found that one of the pins on the data line was not making a good connection and was causing the issue with the fades (and all other data actually). So now I don't know if I should add a shifter or not.
I use the E6804 controller and it has a jumper to raise the logic level to the pixels from 3.3v to 4v. I have always left that jumper in the 3.3v mode. Is there a spec for the WS2812 that specifies the data line voltage?
Ceer123 wrote:Update on my testing - I setup a circuit with a level shifter and everything worked great. For comparison I went back to the old way without a shifter and it still worked great. Eventually I found that one of the pins on the data line was not making a good connection and was causing the issue with the fades (and all other data actually). So now I don't know if I should add a shifter or not.I use the E6804 controller and it has a jumper to raise the logic level to the pixels from 3.3v to 4v. I have always left that jumper in the 3.3v mode. Is there a spec for the WS2812 that specifies the data line voltage?
I am a bit fuzzy but i seam to recall it dropping out at 3.1-3.2 ish
Ceer123 wrote:Update on my testing - I setup a circuit with a level shifter and everything worked great. For comparison I went back to the old way without a shifter and it still worked great. Eventually I found that one of the pins on the data line was not making a good connection and was causing the issue with the fades (and all other data actually). So now I don't know if I should add a shifter or not.I use the E6804 controller and it has a jumper to raise the logic level to the pixels from 3.3v to 4v. I have always left that jumper in the 3.3v mode. Is there a spec for the WS2812 that specifies the data line voltage?
The WS2812 sheet I have says 0.7Vdd minimum for a logic high, which would be 3.5V given a 5V Vdd. Lots of people get by running 3.3V data lines and don't have issues with it. It's up to you, but given how flaky the signalling can be driving these things (800kHz single ended transmission line can get noisy quick - especially on the cables we typically use), I'd recommend driving them to spec if possible.
what about the original ones and clones? 2811 i know had issues every time i tried. I may have had signal inegraty issues though my scope is not fast enofe for pixels.
