EEVblog 1377 - The Amazing UNPREDICTABILITY of Fuses

[EEVblog]

How long does it take for your 400mA multimeter fuse to blow at 600mA?
Grab a chair and watch!
The amazing unpredictability of fusing current ratings at low overloads.

[Tek_TDS220]

Despite their shortcomings, fuses have saved me from stupidity (mostly mine) many times.

[SilverSolder]

The simpler a component is...  the more complex its behaviour! 

[edpalmer42]

I came across this post & video on non-destructive fuse testing.  I used this method to confirm that some 10A SIBA fuses from China were actually fuses rather than just a big piece of wire.

http://www.kerrywong.com/2020/04/12/non-destructive-testing-of-fuses/

[kingsolmn]

I would love to see more videos about this topic! I think it'd a fun project to design and create some kind of automated test jig to batch test fuses. Anyone want to help me? I certainly know I'm not the smartest member of the forum so I will definitely need some help along the way  . But I need a new project and invite anyone along for the ride! 

 - Steve

[Ian.M]

I came across this post & video on non-destructive fuse testing.  I used this method to confirm that some 10A SIBA fuses from China were actually fuses rather than just a big piece of wire.

http://www.kerrywong.com/2020/04/12/non-destructive-testing-of-fuses/

That page is a bit dubious:

For most of the fuse material, the melting point is relatively low, so it is reasonable to assume that the melting temperature is at between 250 and 750 degrees

Most fusewire used in plain wire fuses is tinned copper.  Copper melts at 1085 deg C.  Although the melting point of tin is only 232 deg C, once it melts it will diffuse into the underlying copper to form a bronze alloy.   To lower the melting point of the resulting alloy below 800 deg C it will require a composition exceeding 10% tin, which is unlikely for all except the thinnest fusewires.

[edpalmer42]

Most fusewire used in plain wire fuses is tinned copper.

What's the source of this info?

Various documents[1] agree that even a 40ga copper wire has a fusing current of about 1A. so smaller fuses likely aren't copper wire.  I suspect that the composition of a fuse wire is more like a 'secret recipe' to get the exact characteristics that the manufacturer wants.

[1]
https://www-d0.fnal.gov/hardware/cal/lvps_info/engineering/wirefusing.pdf
https://www.powerstream.com/wire-fusing-currents.htm
http://www.nessengr.com/technical-data/fuse-equations/

[Ian.M]

Yes, once you go below about 1A, its difficult to reliably mount and connect to a thin enough plain tinned copper wire, so low current fuses must use other wire compositions, and IMHO, its probable there are special purpose higher current fuses (other than thermal fuses) where the required I²t characteristics favor the use of alloy or non-copper wire, but  I would expect that plain tinned copper wire is still dominant, if only because of its well understood fusing characteristics, acceptable mechanical strength at elevated temperatures, and that the relatively high melting point decreases the effect of ambient temperature on its fusing current.

Sorry I don't have sources handy - I last looked into this in detail many years ago, and even in this internet age, reputable fuse manufacturers aren't keen to publicly disclose their 'secret sauce'.

[SeanB]

I have taken apart 200A fuses, where the fuse element is multiple silver foils, swaged into a carrier frame made from copper, that then was swaged onto the outer ceramic case after filling with the powdered silica sand. Lots of low current fuses in parallel, so that eventually enough will melt on overload to fail the remainder fast. This one failed after around 3 decades of running at close to nominal current, with a few months of running at 2 times current for hours at a time as new loads were added to the grid segment it fed. They only changed the one out of the three, the other two, despite being made in the 1950's, were still fine.

[rs20]

Dave, at 24:06, I suspect the curve on the SIBA graph isn't "resetting". The curve to the left of 4 is the lower bound (perhaps a guaranteed lower bound on trip time at the given current) and the curves to the right of 4 are the upper bounds (perhaps guaranteed upper bounds on trip time).

[firewalker]

Are you using a constant current source? In most cases the fuse will be in a constant voltage circuit. The positive thermal coefficient would increase the fuse time.

Alexander.

[Refrigerator]

I remember seeing some fuses that had two springs inside that were soldered in the middle, so that when they got hot the solder would melt and the springs would release creating an open circuit.
I wonder how would those hold up in such test.

[dreamcat4]

So here is what i dont undesrtand: for lower current applications we cannot cheaply make reliable fuses for less than about 1 amp. Even the pricier ones are not consistent, right? Yet also in these modern times, the cost of a lot of low power electronics has come down a lot. So surely as an alternative an efuse is going to make more sense below a specific number of milliamp threshold. Assuming that an efuse can be made cheaply enough.

And this also comes to another comment somebody made about li-ion cells. Where you have to actively limit the current with a mosfet in some similar way. Because fuses simply cant do that job accurately enough. Or consistently enough.

But I have more questions about fuses:

* Well what about thermal fuses? Are they also inconsistent / problematic in similar ways to regular fuses?

* How about some type of an online tool (or even just a big flow chart)... to calculate / help us determine which type of fuse is best for a specific application? Because quite frankly it's still hella confusing to me. Dave only looked at 2 specific types here for his MMs, all within a certain specific mA range. However the overall landscape is vast with many different options. It really makes my head hurt.

[PartialDischarge]

There are 2 reasons why low current rated fuses behave in unpredictable ways:

1) Temperature, as mentioned in one of the slides of the video

2) Material thickness uniformity. A 10um copper wire in diameter (yes 10 micrometers) can easily carry 100mA even for a 5cm wire length. So this means that a small variability in the thickness in the wire used (usually not copper but other materials) impacts the fusing current more than higher rated fuses.

[PartialDischarge]

So surely as an alternative an efuse is going to make more sense below a specific number of milliamp threshold. Assuming that an efuse can be made cheaply enough.

The real value of a fuse, and something people usually forget, is not just breaking the path of a particular current, something that a mosfet can do. Is that they offer a large current breaking capacity something that semiconductors cannot offer (comparatively speaking)

[exe]

This explains some of my observations. I've seen recommendations to put fuses with double the expected maximum current. I didn't know that, so I always put fuses just big-enough for the worst power consumption case. Most of the time it worked well, now I see why.

Apart from tolerances, their rating applies to the maximum allowed ambient temperature.

[SilverSolder]

Are you using a constant current source? In most cases the fuse will be in a constant voltage circuit. The positive thermal coefficient would increase the fuse time.

Alexander.

The fuse is in series with the load, and has a fraction of the load's resistance -  so the experience of a fuse is of a constant current (determined by the load, not the fuse).

[Kleinstein]

It is interesting that some of the longer lasting fuses got really hot before they blew. Some of the faster failing ones did not gat that hot at all.
For slower blowing fuses it should be the temperature to limit the current, so these may be more predictable than the fast blowing fuses.

The fuses with sand inside may have the problem of the sand moving on transport and stressing the filament in a variable way.

[SilverSolder]

Could be interesting to compare with the glass fuses that just have a thin filament suspended in air...   they have a chance of being more predictable?

[Huluvu]

Fuses and Fuse Holders occupied quite some remarkable amount of time in my work life.
Hours and Days of testing, thermal imaging, cross sectioning and endless discussions with vendors....

I found the same or very similar behaviour as seen at Daves Video but it is even worse if you consider the full picture.
Fuse Holders can make things even worse (especially with closed fuse holders)
Fuses can burn down your PCB and should really carefully design in if not avoidable.
Fuses are not easy to test during the vendors manufacturing process and you really have to rely on the process capability.
Fuses can age over time if you use them above the nominal ratings.
+ many things more 
 

[dreamcat4]

this ^^   

[agtrbt]

I missed the ceramic or thermosetting bakelite fuse holders so much, nowadays most fuse holders are just softy plastic rubbish.

On another occasion, i had a 1 amp slow blow fuse in a EI transformer, lasted 200ish inrush cycles before it eventually blown.

Now I have replaced it with 1.6A slow blow that is still alive today, and I put another 3A fast blow fuse to arrest rapid mishaps

[SilverSolder]

I guess if you really want an accurate fuse, you have to think in terms of some electronic assist (crowbar)?

[Refrigerator]

I guess if you really want an accurate fuse, you have to think in terms of some electronic assist (crowbar)?

I think the point of fuses is to keep them simple, if you start talking about adding electronics to fuses then why not just go all out and implement overcurrent protection in your electronics (if possible of course).
But i do think that the venerable fuse can still be improved, and the spring loaded fuse i mentioned earlier does give me some ideas that i want to try out and experiment with actually.

[Kleinstein]

For slow reaction there are relatively accurate motor protectors. They are resetable and some are adjustable. However they are a bit slow for the normal shunts.