Suitable TVS diode for low-voltage IC

[Dannyx]

Good day folks. I'm working on a small project to control some light fixtures - sounds boring, I know, but I just couldn't find the right product to do the job off-the-shelf, so I figured it's cheaper (and more satisfying) to create something myself which precisely suits my requirements. I'm only going to briefly touch up on the operation of the thing and focus on the power supply design aspects instead.

TL;DR: which are considered some of the "best" practices for protecting a low voltage IC from transients and other electrical disturbances caused by nearby relays and power supplies ?

Details: I based my project on an Arduino Micro. I use it to switch 2 relays in a certain sequence whenever a button is pushed (an I/O pin is pulled low). The project called for some time-based actions as well and I didn't want to bother using an RTC for the Arduino itself because figuring out a way to SET it after the project's been packed up and tucked away in a corner was going to be......tough at best and annoying at worst, since I didn't account for a display to actually allow you to see what you're setting - that would be another can of worms I wasn't going to deal with - so I ripped the clock module out of a programmable timer switch that's got everything built-in (you can see it in the picture): the programmable clock itself, a display and an open-collector output, so interfacing with it was just a matter of connecting it to one of the I/O pins and writing my code accordingly, so that was the programming taken care of.

This is where the project became slightly more complex than I had originally planned: the clock's backup battery was part of the circuit I pulled it from and my project will get switched on/off from the light switch, along with the lights I'll power it FROM, so the clock would reset each time. I did a bit of brain-storming and came up with a solution: I used an 18650 cell as my backup battery to keep just the clock module alive when the main supply and Arduino are off. In order to charge it when the light switch happens to be on, I use it in conjunction with a "UPS module" I found for cheap on Ali. When power is on, 5v and "SYS GND" come in from a power adapter and charges the cell an powers the Arduino. When power is removed, the Arduino goes off, but the 18650 cell continues to output voltage via the "UPS" until it reaches the cut-off threshold (I think it's 2.8v). The board switches the negative leg of the load, so that's why I felt the need to clarify the adapter GND (SYS GND) can be separated from the battery GND there.

There was still a catch though: the clock module runs on 3.3v and the battery board does not regulate the output - it spits out whatever the cell voltage happens to be, which can be as high as 4.2v when fully charged. The clock does not like that and acts up if I hit it with more than 4v - I tried it with my bench supply - it locks up or displays random characters. Fortunately, I quickly took care of this issue as well: I dropped an LDO between the battery board and the clock, so no matter what the battery's doing, the clock always gets 2.8v. The LDO is a DI62062.8. I had to go lower in output voltage to prevent the battery voltage from going below the lowest operating voltage of the LDO. Thankfully the clock is happy to go all the way down to a 2.2v input by my testing, so 2.8v is plenty.

I put the circuit together and it all works, as you can see. It's not the prettiest product out there - sure, but since it's for my own personal use, looks don't matter. Durability does, so I quickly realized it's lacking in fuses and ESD protection. It's going to be located next to a pretty beefy PSU and two relays (visible in the upper-right). I was going to make the relays part of the board, but since they'll be switching 230v at considerably high current, I thought it'd be better to have them as far away from the electronics as possible. It's probably not enough, but still better. Then there's the matter of ESD and such: I figured I should add some TVS diodes to protect both the clock and the Arduino. Not knowing exactly what I need, I looked up through the selection of TVS diodes an electronics website has to offer, but the options for low voltage applications is pretty slim: only a couple of results going all the way down to 3.3v or thereabouts and even then, the clamping voltage is still way up there - 10V and up, which I'm not sure the clock module would tolerate, even if it's a very brief pulse.

Come to think of it, the I/O pin connecting the clock to the Arduino could use some protection too.....maybe.

Just in case, I'm planning on adding F1, F2 and F3, which weren't in the original "design". I was going to make F3 100mA or something small like that, but it's probably superfluous anyway, because unless TVS2 fails dead-short, the fuse would not open in time in case of a spike.

This is where more experienced circuit designers can come in and provide some insight on how to protect input pins and other low-voltage applications like these. Don't be too critical of my circuit - it's fit for purpose

[jwet]

I saw you got no replies, maybe this can start a conversation.

Don't get carried away with fuses.  Ask yourself what the real function is.  The main input fuse is good, "consumer" 18650's have internal protection, if you're using surplus batteries from old packs, etc, add a fuse.  Having a TVS on the 5v input can't hurt. (note, the series diode is likely a problem, LiIons charge to 4.2v and you will have something likely 4.3 on your charger after the diode.)  You might put the diode downstream of the fuse across the line.  The line from the Nano to the clock could have a 1k series R to limit latch up currents. 

The big deal is which you don't show is how you drive and protect the relays.  I don't know what your load is but you might consider solid state relays, these usually have an opto coupler input so be isolated from your circuit.  Relays are isolated as well really, but there is a lot of capacitive coupling and inductive kicks that come back from relays.  The capacitive stuff is couple from the load.  The inductive stuff is mainly the coil drive- use a diode across the coil and RC snubber across the contacts.  All should be appropriately rated.  Fuse the HV path.

The most important thing you could do would be to nail everything thing down and put this in an enclosure.  The biggest danger is contact with something or a wire coming loose.

I've said enough here to get a few people to correct me no doubt.  Good luck- be careful.

[Dannyx]

Thanks.

I added that diode to prevent reverse-polarity - just to stop the average idiot, including myself, from connecting the DC supply backwards. It's true that I didn't take into account the voltage drop at that time because the battery part wasn't part of the project then....or did, I ? With a supply voltage of precisely 5v from my bench supply, the voltage after the diode is still a decent 4.8v, with both relays active and the cell charging. Turns out the diode is not 1N4007 like I originally though, which would drop almost a full volt - I took a closer look and although I can't see the whole package, since it's laid down on the board, I can clearly see "N5" on there, so it could be that I thought of this after all and chose one with a low Vf. Can't remember for sure what it is - must've had it in my junk bin, which is how this whole project came to be, outside the Arduino

The relays are pre-assembled like this and used those jumper wires to connect to the headers, to keep them as far away from the micro as possible. Here's an actual picture of one up close. It's got the usual diode in parallel with the coil already on the board, so that part's taken care of. It's got 5v, GND and a control signal going to it, so it's no doubt got an NPN on there which pulls one leg to GND when the control signal is "high" - haven't gone through the trouble of reverse engineering it. This was great for my needs because it saved me the trouble of assembling this myself. One relay switches a 24vDC SMPS which powers some LED strips, the other switches a 230v LED lamp, which is also a SMPS at the end of the day, except much smaller.

Although I didn't show it in my crude schematic, the I/O pins for the button and the output of the clock are actually debounced like this, except the resistors are under the Micro, so you can't see them.

Yes, it will have an enclosure, with a cutout to be able to access the timer.