Well that is a much simpler idea than mine, I like it!

Twice?!? That's awesome to hear! Congratz! It's great to hear that the ESP can be used for some long term solid builds as well. I had already lost faith in them, but now am reconsidering thanks to you :)

Rusty, my best advice will be to ensure they have a stable power supply. Initially I was super frustrated whilst using PC USB only to drive these little wifi wonders, but I soon learnt that by adding a beefy capacitor, they are able to run very stable - even when facing variable temps in the attic... Try it, I am sure it will restore your faith with these chips :D

Nois, thanks for the tip! Is beefy in the range of 500-1000uF? Yeah, I must admit I have been using a not-so-stable power supply. Will give it a shot. Thank you! :)

I use 2200uF and have no issues. I vaguely recall Andreas Spiess had a youtube video on this topic many moons ago where he found that wifi traffic caused the spikes, and he measured with an osiliscope that a 1000uF was just enough to smooth the power requirements. I had 2200uF caps and if it is worth doing, it is worth overdoing :D Because I run mqtt, traffic is constantly running and I wanted it to run reliably - proof is in the pudding.

You can really only reliably use 6 GPIO pins on a ESP8266 based NodeMCU board for output - 4,5,12,13,14,15. On boot various boards will have unreliable output levels, as some go high, some low.

Exactly. The LED in the isolator and the indicator LED will almost certainly drop at least 3v. Infrared could be in the isolator so it could be as low as 1.5v. Visible LEDs are normally >=1.7v. So the two has a drop of 3.2v or better. If the power into the LEDs is 5v, then the MCU has to pull it down to at least 1.8v before the relay triggers. an MCU at 2.25v or greater probably has a Vio-high of 1.8 or greater and thus a high level turns the relay off since it can't get the voltage dropped low enough for the isolator to power on. The Vio-low on the MCU is usually around

Thanks for the insight :-)

By the way, it is also sending some garbage on the serial port before it is properly booted.

David Watts Almost all bipolar transistors are light-sensitive, btw. You can actually shave the casing off of a transistor, and make a crude phototransistor. ;) As long as you use specific pins on the ESP to hook up to the serial pins to the shift register, it should be fine on power-up / reset. But, as Mats said, some ESP pins can do weird things during reset, and obviously certain pins get driven High or Low during programming, or even when the USB UART chip connects to the PC. Definitely not ideal to power the relays from the ESP board. Even if the +5V rail has enough juice to power the relays... Although that board does seem to have a protection diode on each relay (to circulate the back-EMF spikes from the coils), they're not always ideal, and the voltage spikes can get back into the +5V supply rail on the MCU board, and sometimes upset them. It's probably fine in projects like this, to be fair, it's just something to watch out for if your MCU starts acting up when the relays switch. (the back-EMF spikes mainly occur when the coil is switched off, and the magnetic field collapses.)

David Watts btw, if you just saw a sub from me, I was already subbed, but UKposts still has an issue where I can't click the bell icon unless I unsub and re-sub, for some reason? It's quite bad, as lots of channels probably aren't sending out all notifications. It does give an error message when the bell is clicked, but many people might not know about the workaround.

For the record, this is not garbage ;) github.com/esp8266/Arduino/issues/3047

Undefined Lastname - Good point! Experimentation needed, to see if the GPIO pins only toggle during power-on, or if they also toggle during a hardware or software watchdog reset. Alternatively, a quick search suggests GPIO’s 4 & 5 are stable (high impedance) during boot-up. So another option (if you can't just use these for up to 2 relays), would be to connect the tri-state buffer’s OE to one of these stable GPIO pins, with your code then controlling when the relay outputs are actually enabled. For this approach you’d need an OE pull-down resistor, or, use one of the more common active-low OE tri-state buffers with a pull-up resistor. :)

Undefined Lastname - What you are describing for debounce is essentially a digital circuit low-pass filter. This could also work, but only if: (a) The unwanted power-on / reset pulses are acceptably short. (b) The latency (delay) this introduces into the relay switching is acceptable for the application. (c) The added minimum cycle time (relay on or off time period) this introduces is also acceptable for the application. If you were to use the tri-state buffer approach, then there’s no real added relay switching latency or extended minimum cycle time, during normal operation.

Undefined Lastname - LOL. Yes, the simplest approach would just be to use a better behaving MCU. :) The appeal of the ESP devices is clearly low cost performance for applications requiring wireless connectivity. But as you’re pointing out, we should be selecting an appropriate MCU choice to suit the application. I’d agree an ESP8266 is very likely not the right choice if implementing a non-wireless relay controlling application.

Undefined Lastname - Totally agree with you on your analysis, especially in relation to newbies and Arduino with ESP. I was attracted to the ESP32 initially as being best fit for a BLE project I was working on (I think it still is). But prior to my wireless needs, I was most recently using PIC, coded in MPLab’s XC C compilers. Also agree, having used microcontrollers going way back to Motorola MC68HC11’s, you would generally expect that digital GPIO pins were initialised as inputs. I don’t have much experience with ESP8266, other than some Wemos Mini’s I picked up for playing around with. :) The ESP does have merits for the true hardware hackers who delight in hardware hacking and discovery (like Neil Kolban) I do also have that discovery urge, but when you just want to design an end project, you don’t want to be figuring out how the tools actually work! So, as always, with so many microcontrollers to choose from these days (and languages), you should always be considering what's the best fit for purpose. :)

"tons of bootstrapping pins that eat up a few of the already limited pins" There's 11 GPIO pins that are FULLY usable on an ESP-12 if set right with pullup/down resistors (GPIOs 0-16 but less the 6 that access the flash.) Even on the ESP-01, which has 8 pins exposed, you have 4 GPIO's available (you may have to be careful about peripherals that may try to pull them up or down at startup.) That's still plenty for a simple remote light switch, LED controller, etc. The only thing the ESP-8266s lack is a decent number of ADC pins. "the typical way is pull resistors to define the default state" That's the typical way to set an INPUT pin, not an output one. Pullups/downs can't do anything if the pin is set as output at bootup. True, it might be best to have the pins all default to input at boot but if they are set as output, then resistors mean nothing.

He must be joking or trolling us I guess. connecting trigger of a relay to RX and then to TX too!

Or an RC delay loop for the relay power. Then you don't need the OR-gate. Just one transistor (or MOSFET) to switch on the relays (with a delay).

The solution to prevent it could be not to hook up relays to pins that are connected to the flash chip (or just to...I dunno, maybe RTFM?. The SD0-SD3 are NOT intended for GPIO use.)

My answer is a generic GPIO not limited to SD0-3!

It won't help if it's in a situation where manually flipping a switch isn't an option. If there is a brief power out and then the NodeMCU reboots, it will possibly activate some of the relays. In my application even they activate for a fraction of a second it will be a problem. I'm only using the relay as a momentary switch but not to switch mains.

Why is that? this board also have 8 bit data out you can even use it to control 255 relays

They didn't last long :-)

@@DavidWatts they seldom do

there is a chrome extension "equalizer" for these.

As a lefty it is because otherwise I would not be able to see what I have just written. It would be on the other side of my hand, righties don't have that problem.

Thanks David, that actually makes a lot of sense. Being right handed myself(obviously) and as you say, not a problem for me, I had never even considered the problem existed! My son is a lefty but he's not writing yet, at least I'll understand now if he has problems as he learns. Cheers!

SimoWill75 also..... When writing with ink/ballpen you avoid smudging the ink with your hand in case it wouldn't be fully dry yet

Ed19601 I had considered that, but then I tried writing right to left and it didn't happen, my right hand stayed clear of the fresh ink. David's reply makes perfect sense though.

SimoWill75 it does make perfect sense, but the ink argument was told to me by left writers in my youth........ But mind you, I am from a time that we learned to write with liquid I know you had to dip a pen in and only when I was about 8 the first ball-pe s were introduced...... And the were not as good as now

I wouldn't be using these little Songle "10A" relays to switch more than 1-2A anyway. I tend to use 16A or 30A relays to reliably switch higher currents (even 10A).

i think optocupler prevent you from using transistors and diodes to drive the relay coil . and also provide total isolation from the coil also. You only are driving a led from your esp. not a transistor and a coil that can generate a HV spike

Thanks for the time you took to write a comment, I thought I might take the time to reply even if your comment is not aimed at me. Thanks for your comment about my mathematics ability, perhaps you would be better off with the numberphille channel if that is your bag. The root of the video is based of another person's observations and not mine, in fact I think I said I did not understand why people had experienced what was described in the email feature, I couldn't exactly replicate it myself. When I spoke about driving the relays from the ESP I was talking about bridging those power pins which as I am sure you are aware exist to separate the controller from the power to the relays. I am sure the designer had their reasons. At the start I mentioned that the pins I used did not cause in odd behaviour in the relays so I explored which pins might do it and offered a solution in the form of a shift register without (mentioned in the video multiple times) knowing the reason for the 'mystery behaviour' I don't know if all your comments are like this but this but the community I exist in is all about sharing experiences, knowledge and helping other people. This channel is an electronics vlog, not a tutorial channel or featuring someone with any advanced training. I also don't make a living from it, I just enjoy sharing my experiences. While I dislike your approach to the comment I am sure it comes from a place of experience so if you care to input, please do it respectfully.

I still stand behind WHAT I said, even if I may not have been the most politically correct about HOW I said it. Yes, it was more on the snarky side than it should have been. But the points still remain. And I just see so much junk on UKposts with people trying to explain something without really even understanding the issue (and actually, your video was FAR from the worse I've seen in that regard) that sometimes it really gets to me. But that being said....... You never explained how you got the 2.8v number. Either it's going to be close to 3.3v (or whatever the high logic-level is for whatever's connected to it) or it's close to ground. It's probably not going to be 2.8 with an ESP8266 unless you're powering it from under 3v (although the datasheet does say the high level out can be as low as 0.8xVDDIO (which should be 3.3v on both of those boards and thus a high output COULD be as low as 2.64v.) If the resistor connects to 5v and the visible LED is connected to something else, then either the something else is pulled down to (close to) 0v and the LEDs conduct/light up or it's set to something above around 1-2v, in which case the LEDs can't conduct (not enough voltage drop available) and the control pin will either be 5v (if the control pin isn't being driven at all but is more or less floating) or whatever the controlling circuit manages to drive the control pin to. On the relay board, the designers had their reasons for adding that jumper BECAUSE the MCU or other control source may be at too high or too low of a voltage to properly operate the relay board. Or maybe the MCU board simply can't handle the load of the relay board, you're worried about high voltage if the relay shorts out and thus want to isolate the power. etc. So you have flexibility. You CAN power the board from the same source as the MCU, as long as there's not too much other peripheral load or you can power only part of it (the visible LED and the LED half of the optoisolator.) But my point about powering the relay board from the ESP board was simply that, under normal use, one relay should be fine. Four at the same time MIGHT be pushing it, though. And it was the "exploring other pins" that was part of the problem. If you have a device that seems to be flakey in how it works, you need to FIRST know what the pins are for. A search on the ESP8266 shows what IO are usable all the time (1, 3, 4, 5, 12, 13, 14 & 16), which are used at bootup but can still be used afterwards as long as you do it right (0, 2 & 15) and which ones are NOT (easily or at all) usable (6, 7, 8, 9, 10 & 11.) Then you'd find out what pins are what on each board. On the Lolin one (the long, skinny one) you'd find all the pins labeled D* are the readily usable GPIOs (but again, keep in mind the ones used at boot.) So you'd then try those one at a time and say "Hmmm...none of those caused an issue. Maybe they connected the relay to one of the reserved pins and THAT's what caused the issue. Let me try those. AHA! These ones labeled SD* seem to cause issues because they HAVE to go high during parts of the bootup process. That's probably what happened is that they simply were using the ESP wrong. But......JUST in case they really did use one of the proper pins and it really is a quirk of the ESP, here's how a buffer chip may help." I.e be methodical about it instead of "well, I don't know what this pin is for and haven't researched its use but I'll just plug something to it and see how fast the magic smoke appears." And one last note on the video. The resistor feeding the LEDs was probably a 1kohm, not a 100ohm. If it'd been a 100ohm, then driving an LED with a drop of 2.2v (whether it's one or two that add up to that), you'd be pouring 28ma into the ESP, which is only rated for 12ma max on the GPIO pins. With it being a 1k, it'd only be using 2.8ma, which would be fine for pretty much any controlling chip. But basically, the the two LEDs probably have a total voltage drop around the 3.5-4.5v area (depending on the color and, of course, we have no idea what color the one in the isolator is without a datasheet.) With that drop and the 100ohm, the current would be 5-15ma and could still fry some boards. So, as mentioned, my guess is that it was a 1k, not a 100 so that it'd be compatible with pretty much anything thrown at it.

Thank you for the detailed response, it was very interesting to read.

Dead on Mike Anderson. Accurate and brilliant reply. Fact I'm the only one who liked and thanked you for it shows just how bad it is.