UKH

The sort of geeky question I'm curious about that somebody on UKC will know the answer to....

https://datasheets.raspberrypi.com/pico/pico-datasheet.pdf

4.5. Powering Pico on page 19.

Why a Schotkky diode (in red on figure 15) to supply Vsys  rather than an ordinary diode?

Does the Schotkky give better protection against static electricity, over/reverse voltage or something else in some way? 

Is it Schottky has reduced forward voltage so smaller loss of supply voltage?

Background - got myself a pico W to control RGBW neoxpixels. Works nicely although you need to take care not to exceed maximum current output of laptop.

However it works fine headless when I use an Ikea USB charger rated at 3.4A overall and 2.4A per individual output with all 4 channels (RGBW) of thirty neopixels at maximum (255) for extra bright white.

Not measured the current but I think it is 30*4*0.02 = 2.4amps or nominally 12W.

The plan is to feed USB power to neopixels and take feed from neopixels to power pico on Vsys. I do not want to feed power into or damage laptop if I forget to unplug Ikea charger. Hence I think I should feed power in to Vsys rather than Vbus. In my case voltage drop across an ordinary diode should not matter as USB 5V supply is nowhere near pico minimum of 2.3V.

Overall the pico W is a great little beastie. Got the W (wifi) version so it's networked and running a web server (Python, not Apache etc) that listens for http requests to control the lights.

I'm not much good at semiconductor theory.  But the main advantage of Schottky diodes is the switching speed which gives higher efficiency.

Usually these kind of dev boards cheap out on the regulator so they have a massive dropout voltage and your need every volt you can get. However this one seems to use a dcdc converter so not sure that's the case here.

Also check your datasheet for your specific LEDs. Pretty sure WS2812b use 60mA at full bore. 

And watch out for taking power from the far end of a long strip of them. Voltage will drop significantly. Test while they're on.

Low forwards voltage drop and high reverse impedance at low voltages mean it effectively separates two 5V supplies, protecting them from each other and giving precedence to the one applied to VSys (“supply oring”).  Faster switch to conduction than a normal diode if VSys removed as well.

Much love for the Pi Pico and it’s documentation vs Arduino land.

ESP32 all the way

I looked at the circuit and it seems to be DC. So switching not an issue.  I suppose, as you said, it's just about voltage drop.  Indeed they suggest a mosfet solution to  "overcome the efficiency and voltage-drop issues with the diode-only solution."

I like the PIOs on the Pico, almost as good as having a little FPGA in there…

Thanks all, never really know what Shottky diodes were until today. Never realised forwards voltage could be so low.

I don't think I need faster switching for DC power as I plug stuff in/out and I don't think I need lower forwards voltage so I'm hoping an ordinary small signal diode will be just as good for me.

The familiar Raspberry branding, quality and documentation appealed as never used a microcontroller before.

Switching time isn’t really important, it’s the low forwards drop.

You might be misreading the docs; that shows the circuit on the Pi Pico board, it’s not an application note.  All included in the £4 PCB.  

Yeah. Low Vf so we don't lose too much of the VBus 5V supply.

Fast recovery/switchover if the external supply goes, so avoiding brownouts.

No, I realise all that circuitry (except the red stuff) is on the board.

My only concern is not to fry my laptop if I have Ikea USB PSU plugged in when I plug pico into laptop. Hence feed Ikea power into Vsys rather than Vbus so laptop protected by diode on pico board.

I suppose the extra (red) diode protects the cheap Ikea PSU more than the laptop.

A couple of times I put all 4 channels RGBW on full power when plugged into laptop. Mouse & keyboard stopped working which was scary as I thought I'd gubbed new laptop. Thankfully unplugging & plugging in again fixed it.

Quite amazing what you can get these days with Pi & Pico or SPI sensor - looks like a transistor but it's a microcontroller with a communication protocol. 

The Schottkys prevent unplugged power supplies pulling the VSys rail down. As wintertree says, they provide a 'power OR' function, with isolation of unpowered supplies.

I bought a usb isolator jobby and connect through a powered usb hub for this sort of thing. They're cheap. Far too easy to toast a motherboard. They're not cheap.

Scaring myself by taking too much juice out of a computer - I remember I first time I did that in 1985.

I'm a slow learner.

What Wintertree said but I doubt a junction diode would be a problem.

Jk

https://shop.pimoroni.com/products/flexible-rgbw-led-strip-neopixel-ws2812-...

So far I've not found a datasheet for RGBW rather than RGB so I've guessed the W of RGBW also 20mA, hence 80mA rather than 60mA but it's just an assumption in the absence of better info.

Brightness doesn't seem to be any different at the "far" end (1m, 30 pixels).

Actually measuring the voltage supplied to the pico and neopixels when headless and powered by an Ikea USB supply...

LEDs off:
5.1V and LEDs consume 0mA

LEDs varying colour and FAR from full power (eg averaging 20 out of 255)
5.1V and LEDs consume 16-100mA, mostly about 30mA

LEDs - 255 for all 4 of RGBW
4.6V and LEDs consume 1200mA which is half what I expected, it should be 1800mA for RGB only and I really want that extra brightness.

Estimate resistive losses in ordinary USB charging cable 0.5V. I can try other cables and I've ordered a bare USB plug so I can solder together something with thicker wires to supply LEDs and take feed from pico from that. 

Measuring Ikea USB charger - 5.2V no load, 5.1 V when LEDs on full power (1200mA) so 0.1V lost within charger and 0.5V in USB charger cable.

4.6V not a problem for pico, I'll need to check if 4.6V too little for neopixels and see if currents per pixels higher when resistive losses reduced for for 5, 10 & 15 pixels of RGBW all 255  - circuit debugging in software!

And measure voltage at far end for resistive losses within strip.

When I reduce the pixel count from 30 to 10, 5 & 3 for RGBW (all 255) the current per RGBW pixel remains about 40mA.

Laptop USB output reduced from 4.9V (no pixels) to 4.2V (10 pixels RGBW on full) but current per pixel remains consistently 40mA for 3, 5, 10 pixels on full RGBW.

Looks like despite voltage fade the LED constant current drivers are working but at ~10mA per LED (40mA for RGBW on a pixel) and I expected 20mA per LED (80ma per RGBW on a pixel or at least 60mA for RGB).

Looks like adding more volts by reducing series resistance unlikely to increase current or light output.

Time to buy a diode variety pack on ebay as laptop not happy driving pixels plus mouse & keyboard.

If you’re experienced at tapping in to cables and cutting the red wire without disturbing the yellow wire, you can convert a USB cable in to a data only cable.  Or you can use one of these [1] which handily switches the power line but not the data lines.

[1] - https://www.amazon.co.uk/Sabrent-4-Port-Individual-Switches-HB-UM43/dp/B00J...

I've seen that movie. Never cut the red wire!

It's still all too easy to put too many volts onto one of the data lines, which is something I've very nearly done too many times. 

Oh ok, over 30 pixels or ~1m you probably won't notice the droop. On the scale of all the way round the kitchen cupboards you definitely do. Easy enough to feed extra power in further along though.

And if you do ever switch uc allegiance to something 3.3v, https://hackaday.com/2017/01/20/cheating-at-5v-ws2812-control-to-use-a-3-3v... is a useful trick.

Curiously pico is 3.3V and this is the "circuit" I use without level shifting.

https://learn.adafruit.com/assets/98790