Thin fuses. Wanted resettable SMD but they might get too hot

I'm working on my first real PCB and wondering what fuse I ought to use because I want my project to be as thin as I can reasonably make it. I was going to use THT resettable fuses bent over but I think they will be thicker than necessary. Each MOSFET shouldn't switch more than 2 A and the total circuit not more than ~5 A. I think I can replace the smaller fuses with SMD ones without issue but for the main fuse I was thinking about using two 3 A fuses in parallel. The main problem I think is the heat that will come from the dual diode which at 5 A will be 6.25-8.75 W (250-350 milliohm on the diode) 1.25-1.8 W (250-350 mV on the diode) depending how hot the diode gets and the heat will go straight through the tracks into the fuses. I will use thermal vias on the diode's pad and the enclosure will be made of aluminium core PCB with a corresponding exposed copper area to get the heat off the the diode but might it still melt the fuses? I know there's usually some amps between I(hold) and I(trip). I was also thinking of using a glass fuse or an SMD chip fuse? Hopefully I won't be blowing fuses too frequently.

The thickest part will be the terminals but I will have cut-outs in the top part of the enclosure anyway because they need screwdriver access so it would be nice to keep the rest no higher than the top of the terminals including the thickness of the top part of the enclosure.

You can see the schematic/PCB in better detail IF YOU SCROLL TO THE BOTTOM here (the top two have a bug and are out-of-date):

The fuse areas on the PCB right now look a mess because there's multiple components on the schematic fighting to be chosen.

Edit: Incidentally, there is a capacitor next to the 3.3 V regulator (external to the Arduino - I didn't want to use the Arduino's because I want the voltage more stable for the ADC). The capacitor is across the regulator Vin and GND. Should I put a small resistor in series with it to stop inrush current? I just watched this video that says ceramic capacitors can kill electronics without a resistor in the situation that I think my regulator and capacitor are in: SDG #088 Ceramic Capacitors will Blow Up your PCB - YouTube

What is they might get too hot; what temperature is that. As long as you do not exceed the manufacturers specifications the fuses should be fine.

I'm not sure how hot they'll get but like I say, that diode will produce a considerable amount of heat. The enclosure won't have good ventilation either. And when the fuses get hot, they blow at lower currents.

You are very cautious with lots of current limiting and pull down resistors some of which, in reality, could be omitted. I suggest also including the schematic directly in this thread to get other comments. You may be able to further rationalize the design.

The decision to have a separate 3.3v regulator is good because, depending on the Nano, the 3.3v is available only when powered via a USB cable.

Iā€™d probably use glass fuses. These can be used with low profile mounting clips. The Nano is anyway, depending on which headers are populated, will have a higher profile since you appear to be concerned about the dimensions.

If you need reverse polarity protection, reverse a diode across a (glass) fuse. Certainly try to avoid designing protection circuits which generate a lot of heat during normal operation.

Maybe also just ensure that access to the usb socket on the Nano is not blocked by other components, especially if it is soldered in place.

1 Like

diodes are not rated in milliohms, take a look at the current - voltage curves. The forward drop should be in the 0.5v area...

Getting even 2 watts off a board isn't trivial. I suggest you look into using a MosFet for reverse voltage protection.

I've never seen two fuses in parallel. I doubt they will "share" the current.

1 Like

Thanks for your reply. I've added the schematic to the original post.

I currently plan to use the onboard 3.3 V for the voltage divider for the thermistors so I'll need to make sure it's working when powered via Vin. I'm not sure whether I will use a Nano or a Nano Every. Maybe I can find a footprint that would allow the Nano Every to be soldered down with it's castellated pads.

If I use a glass fuse, I guess I will use those ones with axial leads.

The dual diode was intended so I could potentially have my mains-based power supply and a battery bank plugged in at the same time.

Unfortunately, I've further compacted the layout of the board to shave of a few millimeters since I last posted and the USB port would be blocked but maybe I can breakout an ISP header somewhere.

Oh that's right - it's 0.25-0.35 V so only 1.25-1.8 Watts.

I read on StackExchange that polyfuses should self-heat thereby tend towards equalising their current but that would require them to be more separated from each other and this other source of heat.

Looking into MOSFET reverse polarity protection now. Would it be possible to have two inputs at the same time without one backfeeding into the other? I wanted to be able to have a battery and mains-connected PSU at the same time.

Have you breadboarded a prototype to confirm operation?

What is the project and why so compact?

Thanks.. Tom... :smiley: :+1: :coffee: :australia:

By the look of your very, very thin traces, you already have fuses.

I have not breadboarded anything yet except the switches! I've got the code working with the switches just how I want it. When I figure out the design, I'll have a BOM and try to get the parts I need to breadboard the other aspects.

The project is electrically-heated clothing and this will be a four-zone thermostat for that. I want to be able to put the thing on clothing (velcro), underneath additional clothing. I will, of course, take care not to position it over any heating elements (carbon fiber tape). Unfortunately it looks far too late for this winter but maybe I'll be comfortable the one after that!

I've increased the width of the medium traces from 1.1 mm to 1.5 mm. They're intended to carry 2 A max. I might be able to increase the width of the others a bit if I can decide on a fuse. Worst case, I can upgrade to 2 oz copper.

If you are concerned about dimensions, use an SMD version of the ATmega328P instead of a Nano. Break out an FTDI and ICSP header. If you use the internal oscillator, you need only 2 or 3 additional components.

Here are sample fuse clips: but smaller ones are available.

I prefer electrolyte capacitors, their efficiency is higher! In fact, they write in various articles that ceramic is better in terms of their technical indicators! But I have my own point of view! Difference between electrolytic capacitor and ceramic capacitor!!! - YouTube

I might do that but I'm afraid of introducing more points of failure. This is already the biggest Arduino project I've done. It would be pretty amazing though; my first PCB and first embedded MCU not using a development board. Would the ATmega4809 be much more difficult than the 328P? I haven't gotten around to using the SSD1306 OLED library yet but I've seen people report struggling with the amount of memory it uses and figured that the Every/4809 would let me side-step struggling with that.

Then you already lost me when you're using an Arduino Nano as a component. Just stick a 328P on there with the necessary oscillator circuitry and if you so desire a Serial <> USB interface. SPI is convenient as well.

So leave it out. As you say, if you want a reverse polarity protection, consider a different approach. Why burn several Watts just...well, 'just'? It makes (too) little sense IMO.

Neither am I convinced of the need to individually fuse each heating element.

Have you actually thought of the failure modes/scenarios you're protecting against? It sounds more like you've been micro-managing ("oh, a heating element may short so I need to protect that bit") instead of looking at the bigger picture (what is the actual cascade of failures a single problem will trigger, how do you stop that, and which ones are the valuable components to save vs. the sacrificial ones you accept will go to heaven?)

In short - I think you're creating more problems than you're solving at the moment.

PS: how does your OLED physically connect to the PCB? I see its pads, but the OLED itself seems to act mostly like a lever that will rip the traces off your PCB the first time you pick up the PCB and turn it upside down :wink:

The heating elements will be carbon fiber strips machine sewn onto the outside of a fleece top. I want to keep it all accessible for repairs. It's conceivable that it could be powered while the clothing not being worn or otherwise some partial short circuit that would be within the tolerance of the main fuse but become dangerously hot and start a fire. Looking at datasheets, it seems that fuses already tend to allow quite a lot of extra current for a considerable time before they blow, except in the case of resettable fuses that are already overheating. It's not just about saving the components but preventing the possibility of burns or a fire.

The OLED will have to be physically attached by some kind of double-sided tape or sticky pad. I think that's how they're attached to the development boards you can buy.

Ok, that sounds sketchy for sure. How about a more active current limiting circuit then, instead of a fuse?

Fine for me; I just wanted to alert you to the physical part of your project. It's easy to get carried away in the electrical domain and then end up with something that physically just won't work.

I do intend to have the Arduino monitoring the current via the current sensors and switching off the MOSFETs if it deviates from expectation and I also intend to use the watchdog timer in case it hangs but I can't only trust my badly written and tested code and there's a possibility that a MOSFET will fail (maybe ESD from the clothing?) and the Arduino will be unable to turn it back off. In that case, the buzzer will urge me to intervene at my earliest convenience. Someone suggested "crowbaring" the supply in that case. I'm afraid to find out how my battery pack responds to being crowbared however. Eventually, I might make up my own battery pack with a quality BMS but in the meantime, I don't know how my power bank will respond. It's meant to be good for 5 A at 12 V. The cells inside it are good for 32 A between them (4P 3S Samsung 35E Li-ion).

Yeah, I saw that. Firstly, I'm not too concerned about the MOSFET if you properly protect it with a a TVS diode. Secondly, why rely on the Arduino and your code to do the protection? Your current sensor has an analog out, you can feed this into a comparator (opamp) / flip-flop which you can use to trigger a MOSFET to break the circuit. You can do this within a millisecond or so, regardless of the operational status of the Arduino, and without loss of the Arduino's ability to monitor current as well.
BTW, I wonder why you're using a current sensor that has such poor documentation; I searched, but could not find a complete datasheet that wasn't in Chinese.

Yeah...might work, but I'm not a fan of it to be honest. I can see the use of a crowbar circuit, but here I'd solve it differently.

The 5 A version of the ACS712 is out of stock but there are still these Chinese copies. The Chinese copy apparently gives a higher mV/A as well. I used Google Translate to interpret the datasheet and it's pretty similar to ACS712 but is missing the high frequency filter which is a feature I don't understand anyway.

I will look into the possiblity of using a comparator with the current sensor instead of a fuse. Last time I suggested using another MOSFET though, someone said using two MOSFETs in series isn't trivial.

But it's the feature that will save the day in this case...also, it is the feature that makes the ACS712 a reliable tool instead of a cheap toy. In this case you'd be better of with an ACS712 10A version than a Chinese copy rated for 5A, regardless of the apparent higher resolution of the Chinese one, which in reality will not materialize...

So use one. You only need one switching element after all. It's fairly easy to work out a circuit that will act as an AND port at the gate of that MOSFET!

I understand now, thanks. I think I have enough information to Google my way to updating the design. Hopefully I'll back again in a few weeks, having made some progress (I can only work on this in spare time). Thanks for your help (and everyone else that's posted).