A chance to spend some time working on #Yapolamp means I can incorporate the conclusion that the driver circuit used in #TritiLED and #Yapolamp could really increase safety by adding a fuse. This surfaced in the comments when @Ted Yapo kindly pointed out the possibility of the MOSFET staying open, either by its own failure or the microcontroller's.
I am familiar with fuses and their role in AC domestic systems but the only fuses I have seen in low voltage DC systems are the PTC resettable fuses that are ubiquitous in 3D printer controller boards, such as RAMPS. The convenience of a self resetting fuse is very attractive in a torch that I want to enclose for good ingress protection, so I looked into the difference between single use and resettable fuses.
How to choose between single use and resettable fuses
Littelfuse has written an accessible design primer (also see the PDF version at the side which includes the graphs) to help you choose the right type of fuse for your application. Fuses don't have neglible resistance, in the order of an Ohm or few, so I will need to see what impact this will have on the performance of the Yapolamp circuit. I was also interested to learn that PTC fuses still allow current to pass when they are "tripped". They typically increase their resistance by around 5-10 times their "untripped" resistance. This initially put me off selecting one as I thought that in a "MOSFET-stuck-on" failure mode, I wanted to be sure that nothing overheats and so I thought a traditional fuse had to be the answer.
But the fact that so many computer and mobile device manufacturers demand resettable fuses because of fuse replacement service costs and user inconvenience, and they use the same lithium battery technology, made me think harder. So, what exactly are we frightened of happening?
If the MOSFET fails ON, we end up with a current continuously running through the inductor. The properties of the inductor will restrain the rate of current increase but after a very short while, only the equivalent series resistance of the inductor and RdsOn of the MOSFET (around an Ohm) will limit the current. Using Ohm's Law, we get I = V/R = 4V / 1 Ohm = 4A. So what happens at 4A?
First line of defence - the battery protection module
The first thing to remember is that the DW01 protection chip on the TP4056 charging module has overcurrent protection. Using the DW01 datasheet (and not checking with the actual boards and a controlled current power supply - for now) if the equivalent RdsOn of the paired MOSFET on the TP4056 module is 25mOhm, the overcurrent protection will kick in at 3A. So, we expect that the module will protect the Yapolamp if its MOSFET fails ON.
But until I get around to testing these boards for real world performance, let's indulge some "belt and braces" additional protection, given that this is designed for children...
As a reminder, here's the Yapolamp basic circuit, with a resistor acting in place of the fuse.
In the failure scenario, current will pass from the battery, through the inductor, the fuse and the MOSFET. The LED won't see this current as it's only ever driven by the inductor, when the MOSFET is OFF. Assuming that the battery is adequately protected by the DW01 at 3A, let's look at the MOSFET and the inductor (we'll come on to the fuse in a bit). The MOSFET is going to be around 50mOhm RdsOn and at 3A, the power dissipated is I^2 * R = 9 * 0.05 = 0.45W. This is within range of the maximum power dissipation of many SOT23 logic level MOSFETs but we're close to the limits of some I'm considering. Also the 3A max is above some I've looked at, such as the NTR4170NT1G. The inductors aren't this resilient though - looking at the Coilcraft MSS1210 series inductors, for the ranges of 1mH to 10mH that we're interested in for Yapolamp, the DC current at which the inductor will lose 10% of its inductance when current is removed ranges between around 1A to 0.3A respectively. Clearly the inductor is a key part of the circuit and we'd like to protect it from damage that would affect performance.
Sizing the resettable fuse
According to LT Spice, the currents that are generated in the MOSFET or the LED aren't affected by the addition of a 2 Ohm fuse in series between the inductor and the MOSFET (I compared a 0.1 Ohm resistor value to a 2 Ohm resistor value as a proxy for this). We're looking for a trip current around 300mA or below and many PTC fuses are availabel from Littelfuse, Bourns or Multicomp in this range. Remember that the holding current isn't the same as the trip current and be aware that the trip time is not that quick for currents in the same order of magnitude as the trip current.
For holding currents in the order of 100mA, a 2 Ohm untripped resistance is achievable and the trip current is 250mA. The tripped resistance of around 15 Ohms is typical and this would result in a current limit of I = V / R = 4 V / 15 Ohms = 260 mA. This needs checking against the power dissipation capacity of the PTC fuse - in this case 0.26 ^ 2 * 15 = 1.17W, which is in the region of the typical power dissipation of around a Watt for fuses with these ratings. Do check whether this is a typical or maximum value on the datasheet of the PTC fuse you are looking at.