How to select a fuse that blows before the mosfet does?

[Eight]

Hello.

I'm designing a custom lighting control circuit and I want to know how to protect its outputs.
Suppose you have a DC load rated 1A@32VDC (light bulb or a LED+resistor) that is driven by a low-side N-channel mosfet. The mosfet can drive the load via a PWM waveform or simply act as an on/off switch. This load is located off-board and connected by a wire. Now suppose something goes wrong and the load shorts out (designated by the SW1 switch on the attached image). I would like to protect the mosfet with a fuse, but unfortunately I'm having trouble understanding how fuses work (or rather how to select them for the given application).

How does one choose a fuse so that it melts before the mosfet does?
I did some reading on fuses and learned they are categorized (besides current and voltage) by their I²t rating, which supposedly represents the maximum energy before melting. Fuses apparently take longer to melt than some other semiconductors. So I'm wondering what's the criteria here? Can I get away by selecting a fuse only by its current rating so that the mosfet has a larger continuous drain current rating than the fuse or do I have to take the I²t value into account? If so, how does one compare a I²t rating of a fuse against a mosfet?

The PSU is rated 32VDC@10A (10A because it's powering some other things besides the DC load). The lowest absolute maximum rating of the circuit is 50V.

Thanks in advance.

 

[barry]

Th I2t rating tells you how long it will take the fuse to blow at a given current. For a slight overload, it will take a relatively long time to blow. I suspect for your application you’re more concerned with gross overload conditions. If you’re concerned about protecting the MOSFET, then just select a fuse with a lower rating than the MOSFET Idmax.

Just to throw some numbers out there, if you’ve got a maximum load of 1A, then use a 2A fuse and a 5A MOSFET.

 

[crutschow]

A saying I learned when I starting working in electronic design was that "The semiconductor is always faster then the fuse" meaning it's difficult to protect a semiconductor device from current overloads with a standard fuse.

But there are very fast fuses, called semiconductor or rectifier fuses, with relatively low I²t ratings which may be fast enough for your purposes.
It you look at the MOSFET's data sheet current-surge-rating versus time, you should be able to relate that to the fuse's I²t rating.

 

[std_match]

The light bulb load is also a problem, since the inrush current is much higher than the operating current.

 

[Eight]

Barry: This is the mosfet that I am currently using. It is rated for 2A continuous drain current according to the datasheet. I was going to use a 1.25A fast-blow fuse on it. How realistic is this scenario? You think I'd be better off by replacing it with one that has a higher current rating like this one? I'd like to avoid using a larger package than SOT-23 due to limited board space.

crutschow: Which parameter exactly am I supposed to be looking at in the mosfet datasheet? I find no references of surge rating vs time. There's thermal transient impedance vs pulse duration chart, Id vs Vds chart, there's rise/fall times in the characteristics table, but I can't find any current vs time references.

std_match: Thanks for the heads-up. I think it can possibly be solved by using soft-start PWM.

 

[std_match]

The relation between current/voltage and time is in the "Safe Operating Area" (SOA) diagram.

 

[barry]

Barry: This is the mosfet that I am currently using. It is rated for 2A continuous drain current according to the datasheet. I was going to use a 1.25A fast-blow fuse on it. How realistic is this scenario? You think I'd be better off by replacing it with one that has a higher current rating like this one? I'd like to avoid using a larger package than SOT-23 due to limited board space.

crutschow: Which parameter exactly am I supposed to be looking at in the mosfet datasheet? I find no references of surge rating vs time. There's thermal transient impedance vs pulse duration chart, Id vs Vds chart, there's rise/fall times in the characteristics table, but I can't find any current vs time references.

std_match: Thanks for the heads-up. I think it can possibly be solved by using soft-start PWM.

Right there in the data sheet is the maximum Pulsed drain Current parameter (8A/300uS).

What is your main concern? What are you trying to protect your MOSFET from? How fast does your fuse blow? How long will your fault condition last (>300uS)? How much current is drawn during a fault (>8A)? You need this information to make an intelligent decision. As a first-order guess I’d say sure, your 1.25 A fuse should work.

 

[Eight]

Main concern, what I am trying to protect mosfet from:
Overcurrent caused by a short on the off-board lighting element

How fast does your fuse blow:
I don't have a fuse yet, I am selecting one. I've been looking at this one:
hxxps://www.belfuse.com/product/part-details?partn=C1F-1.25
(please replace "hxxps" with "https", forum won't let me post links all of the sudden)

How long will the fault condition last:
Indefinitely. Or at least until something blows on the board. Hopefully, the fuse, not the mosfet.

How much current is drawn during a fault:
Normally I'd say as much as the PSU can supply (i.e. 10A or more), but I think the limiting factor will be the fuse and the mosfet Rds(ON) as it saturates.


Sorry, I'm still learning since this stuff is rather new to me. Regarding the pulsed drain current, I'm not sure I fully understand this parameter. It says the current is pulsed, not continuous. Does this mean I can still treat it as if continuous, and that the fuse should be chosen by its I²t to blow faster than 300us?

Like for this mosfet: 8A for 300us: I²t = (8A)² * 300us = 0.0192 A²s
And the fuse linked above is rated 0.023 A²s according to its datasheet.
Sooo... 0.0192 < 0.023... Uh oh.

 

[barry]

Main concern, what I am trying to protect mosfet from:
Overcurrent caused by a short on the off-board lighting element

How fast does your fuse blow:
I don't have a fuse yet, I am selecting one. I've been looking at this one:
hxxps://www.belfuse.com/product/part-details?partn=C1F-1.25
(please replace "hxxps" with "https", forum won't let me post links all of the sudden)

How long will the fault condition last:
Indefinitely. Or at least until something blows on the board. Hopefully, the fuse, not the mosfet.

How much current is drawn during a fault:
Normally I'd say as much as the PSU can supply (i.e. 10A or more), but I think the limiting factor will be the fuse and the mosfet Rds(ON) as it saturates.


Sorry, I'm still learning since this stuff is rather new to me. Regarding the pulsed drain current, I'm not sure I fully understand this parameter. It says the current is pulsed, not continuous. Does this mean I can still treat it as if continuous, and that the fuse should be chosen by its I²t to blow faster than 300us?

Like for this mosfet: 8A for 300us: I²t = (8A)² * 300us = 0.0192 A²s
And the fuse linked above is rated 0.023 A²s according to its datasheet.
Sooo... 0.0192 < 0.023... Uh oh.

Exactly.

As someone has already pointed out, semiconductors make better fuses than fuses do.

If this is really a concern, you may have to use some active current-limiting scheme.

 

[Eight]

Alright, thanks a lot for all the info and suggestions. I learned a lot about fuses and their operation in this thread. I think I'll replace my mosfet with the other one I mentioned. That one has a 17A current pulse rating for 300us, that's 0.0867 A²s, which is 4 times the amount and well within the fuse specs. I could also look for a faster fuse to gain some additional margin.

The short is unlikely to happen, but the fuse is there for just in case. A dead fuse is easier to replace than a charred mosfet. I decided to go with a fuse because it doesn't add much circuit complexity. There is another fuse present at the board power input, but it is of much higher current rating (10A). As I cannot expect it to blow, I decided to fuse the individual outputs with smaller fuses.

~8

 

[std_match]

You must look at the SOA (Safe Operating Area) diagram.
If you somehow can limit the current to 17A, with short circuit to 32V the maximum time is less than 100 us. Maybe 50 us.

 

[FvM]

Short comment.

It's effectively impossible to protect your small SOT-23 MOSFET with fuses. Electronic current limiting isn't much better, because the transistors have insufficient heat capacity to absorb power in linear operation. As already mentioned, you should also check if inrush current of incandescent lamps can be tolerated at all.

If you want short circuit proofness with small transistors, smart switches with internal protection are the only realistic option. Or fast fuses and massively overdimensioned transistors.

Electronic fuse (fast overcurrent switch-off with latch) may work but isn't very practical.

 

[Eight]

Aggghhhhh! $#!%#$%!%

FvM: Thanks, I think you have a point. I came to the same conclusion regarding fast fuses and oversized mosfets. I'm using SOT-23's, but even most DPAK's can't seem to handle currents beyond 10 amps at 32V for more than about 100us. Sigh. I'll check out the high-side switches and see, if I can find something rated for my voltages.

std_match: Out of curiosity I am wondering. Limiting the drain current of SQ2362ES to 10A (i.e. by PSU) would result in about 100us safe operating time, that's 0.01A²s. Likewise, limiting it to 17A at 50us is about 0.015A²s and that's at 32V like you suggested. Yes, those values are below the fuse rating, so, technically we have another "uh oh". But, hypothetically speaking, can we really assume that (in the event of a load short) the voltage drop across the mosfet is going to be at Vdd level (32V)? I mean, the fuse is in series with the mosfet and it has its own internal resistance. The mosfet RdsON vs Id chart shows about a constant resistance of 0.08 ohms. The (cold) resistance of the fuse is 0.21 ohms, so that's 32V over 0.29 ohms which would draw about 110 amps and gives us a maximum of 8.8V across the mosfet, not 32V, yes?

E.

 

[barry]

even most DPAK's can't seem to handle currents beyond 10 amps at 32V for more than about 100us.

That's totally not true. Where are you looking? A quick look on Digikey turned up 237 devices that exceed your requirements by a LOT. And that's just D-Paks.

Here's just one example. For US$0.87

 

[Externet]

Cannot achieve it. Protecting the mosfet would be totally against Murphy's law :
"Expensive electronic equipment always blows first to protect the fuse "

 

[Eight]

Yes, yes, thank you guys, I get it now.
I've read a number of forums/websites and pretty much all of them mention that a mosfet and a fuse is a no-no. Although, I wonder, if I can get away with it by using an overcurrent-protected mosfet (like A or B) in series with a fuse until it blows. Else, I'll have to settle for a high-side smart switch (I'm looking at A or B).

~8

 

[std_match]

Yes, yes, thank you guys, I get it now.
I've read a number of forums/websites and pretty much all of them mention that a mosfet and a fuse is a no-no. Although, I wonder, if I can get away with it by using an overcurrent-protected mosfet (like A or B) in series with a fuse until it blows. Else, I'll have to settle for a high-side smart switch (I'm looking at A or B).

~8

Why a fuse if you have a protected MOSFET? The MOSFET may reduce the current so fast that the fuse doesn't blow. Use a protected low/high side driver and skip the fuse.