I had a protoype unit of one of my JangoFETs that I integrated a replaceable blade fuse into. Worked quite well and only made the unit a tad thicker while retaining the small footprint.

 

What about one of them self-resetting thermal fuses? Might not be compact enough for your design, but a self-resetting fuse is certainly convenient.

 

Yes, I've considered a small thermal fuse. Unfortunately they are only tripped by overheating components, and I am looking for something that trips when there is a short or something. But I can do both types, and just need to determine the normal operating temperature of my 'fets.

 

I'm referring to self-resetting PTC fuses, they're tripped by current, the temperature of the mosfet and such aren't a factor. They're called thermal fuses because they're designed to overheat at a certain current and this makes them expand slightly so they disconnect the contacts inside. When it cools down the contacts are re-connected again.

They are a standalone component and aren't necessarily installed on the mosfet unit itself.

 

They're called thermal fuses because they're designed to overheat at a certain current and this makes them expand slightly so they disconnect the contacts inside. When it cools down the contacts are re-connected again.

On a precursory search I'm having difficulty finding PTC fuses that actually meet AEG requirements, majority of the ones I can find 'trip' at 30A or less and take very long to do so. Littelfuse AHEF1000 holds at 10A, trips at 20A, and can handle 100A...but takes 15 seconds to 'turn off' at 50A draw. NIDEC PRCP-R1100-0-99 holds at 11A, trips at 22A, and can handle 100A...but takes 20 seconds to 'turn off' at 40A draw. Maybe I'm missing something but this doesn't seem very good for our application.

I would rather design current detection into your circuit, and have the microcontroller drop power if it measures anything past a certain threshold.

They were more focused on being completely modular without any added resistance of a fuse

You lose the resistance of the fuse and humbug of changing it but you also introduce three points of electrical resistance and mechanical failure. If it's working ok for you then that's great but I would only use this as a bench tester. A suggestion I would make for a new design would be to fix the heatsink to the PCB or something solid rather than having it float supported only by the MOSFET legs.

 

Awhile back I made two different types of quick change mosfet boards. They were more focused on being completely modular without any added resistance of a fuse. I made three flat ones for m4s buffers and then a shorter taller one for use under a ak dust cover. Down below is the ak one. Those screw terminals are strong and solid metal inside, the screws can tighten enough to squish the wires and mosfet legs without stripping. The mosfets are also installed cold, removing any risk of overheating them when soldering.

I can short out the motor, blow the mosfet out and swap it and be back in action in about a minute or two. I've so far killed 2 mosfet chips from minor oopies. Both fixed easily with a new chip. One was from contact bounce, other was tight shimming.

Yes burnting out the tvs diode or the resistors is possible but unlikely unless the input wires are wired wrong. A fuse would be more logical, but shorts aren't common, and tripping a polyfuse or burning out regular fuses when spamming semi can be irritating. Id think higher end mosfets track the amps to cutoff power to the mosfet before damage occurs.

 

PTC fuses are rather slow, that's why it's common to use several in parallell to get the trip time down significantly. Might not be ideal for your mosfet design though.

 

PTC fuses are rather slow, that's why it's common to use several in parallell to get the trip time down significantly.

That's not a good idea from a design standpoint, that's not what they're designed to be used for. That being said...

Taking the second graph to illustrate my point:

Red dots are, assuming the PTC fuses are identical (they're not), the current load each one sees. Vertical axis is time-to-trip in seconds, you will see that it varies greatly depending on what model fuse we implement. For a 20A current 'trip' using the 1.0/2.0/4.0/8.0A models we see a minimum of 5s and maximum of just over 10s. However, real-world results will be different due to manufacturing tolerance of components, and using fuses in parallel decrease reliability.

FYI on a related note, daisy chaining surge protectors is not recommended for the same reason - aside from being a potential fire hazard, it puts the protection circuits in parallel and effectively doubles the amount of power it needs to trip.

 

They seem to work well in my AEG's. Earlier I had a 3X10A PTC fuse with 3s trip time in my AK-104 with a 16 TPA SHS motor, 13:1 ratio gears and an M120 spring. The fuse would actually protect the motor from overheating if I used the gun heavily, with a short or motor stall (never had this happen), I imagine the fuses would trip very quickly.

It's also not a "daisy chain", it's an actual parallell connection, with common input and output, effectively tripling the trip current while keeping the individual trip time. Won't running them in parallell and wrapping them together actually reduce the trip time since they'll heat up more quickly?

 

It's also not a "daisy chain", it's an actual parallell connection,

My daisy chain comment was referring to AC mains surge protectors, running them in a chain does indeed put their protection circuits in parallel which is why they say not to do so in their instructions.

Won't running them in parallell and wrapping them together actually reduce the trip time since they'll heat up more quickly?

I just realized I forgot to mention in my previous post that I am analyzing the following in parallel:

1x8.0A, 20A each (green)
2x4.0A, 10A each (blue)
4x2.0A, 5A each (pink)
8x1.0A, 2.5A each (orange)

The red dots show the trip time of the PTC fuse array, in an ideal universe meaning all of them behave identically. Adding more of them divides the current however the time-to-trip at that current value is not the same and may be higher or lower depending on that specific trip current (20A) and using these specific components. In this specific ideal scenario, running 2x4.0A will trip ~2s faster than running 4x2.0A...in the real world it may turn out to be much different.

 

I see, I guess I have oversimplified things by just looking at the stated trip time and current in the specs filtering in the Digikey webshop. Looks like I got some homework to do.

 

I took a look on Digikey and yeah it is kind of misleading, the time to trip is always a graph and the given spec is for a specific amount of current (more current = faster trip time) at a specific temperature (thermal derating, higher temp = lower current handling). Not all datasheets on Digikey provide that information as going from the Littelfuse models that I was looking at yesterday on their own website, the Digikey-hosted datasheet appears abbreviated.

Example from one of the Eaton models: 20°C is where the datasheet provides accurate information, at 40°C it will have the same trip time etc. but at a 20% lower amperage.

 

I see, good thing the batteries I use will almost instantly trip the fuse anyway if there's a short.

 

A 15A fuse with the fastest available trip time is ideal then?

 

You need to know both the hold current and likely fault current of your application, along with the temperature that the PTC fuse will be at. An engineer will have to make a decision on which component is the best compromise for a specific application.

As a theoretical example I measured JG blue stall current at around 13A, so my hold current spec might be 15A or so. Fault would probably be anything above 50A, and I would need to figure out the operating temperature of wherever this MOSFET controller is going to be located in order to properly derate.