Multiple identical fuses in series, does one blow first, or ...?

[incf]

Hi,

I have a system where there are multiple power sources and multiple loads on a bus. Each has (what is effectively) a mini automotive 5A blade fuse on their connection to the bus.

In the event of a short circuit fault, one or more power sources and/or loads will source/sink about ~100 amps (sometimes more, sometimes less). The short circuit fault current will pass through several identical fuses.

Questions, in the system described above during a short circut:

• In common practice will one fuse blow first 99% of the time?
• Can they all blow simultaneously?
• Can can one blow and the rest be damaged/fatigued due to heat stress and fail/blow at some unknown value that is below their rated current at a later date?
• Or something else?

***The fuses really do have to be identical, because multiple identical units are installed in parallel

If anyone could recommend any literature, or share their experience, I would be grateful.

[wasedadoc]

If I understand correctly you have multiple PSUs with their outputs in parallel. A VERY bad idea.

[Monkeh]

In short, no, there's no predictable behaviour with identical fuses in series, or even with different fuses which are too close in rating when subjected to a substantial fault current. The art of making sure you know which device will open is called discrimination.

[incf]

If I understand correctly you have multiple PSUs with their outputs in parallel. A VERY bad idea.

This is a battery backup system for a power bus. It's just batteries in parallel (with diode on each to prevent backflow) for sufficient run time.

[wasedadoc]

If I understand correctly you have multiple PSUs with their outputs in parallel. A VERY bad idea.

This is a battery backup system for a power bus. It just batteries in parallel (with diode on each to prevent backflow) for sufficient run time.

That additional info does not explain why you said the fuses are in series.

Still a bad idea as equal current sharing requires the batteries to have the same voltage and the diodes to have matched voltage drops.

Suggest you provide a schematic that shows all the relevant details

[PCB.Wiz]

I have a system where there are multiple power sources and multiple loads on a bus. Each has (what is effectively) a mini automotive 5A blade fuse on their connection to the bus.

In the event of a short circuit fault, one or more power sources and/or loads will source/sink about ~100 amps (sometimes more, sometimes less). The short circuit fault current will pass through several identical fuses.

Questions, in the system described above during a short circut:

• In common practice will one fuse blow first 99% of the time?
• Can they all blow simultaneously?
• Can can one blow and the rest be damaged/fatigued due to heat stress and fail/blow at some unknown value that is below their rated current at a later date?
• Or something else?

Given fuses are your only line of defense, All of the above. 

Fuses are not identical, so yes, one will blow first, that then increases the load on the slower ones, and the next one will blow, in domino fashion.
To a user that will appear as a blow simultaneously event, but to a physicist each will have a ns time stamp.

Unless the massive overload is removed very fast, you are unlikely to have any fuses survive, but yes, fuses can suffer thermal fatigue.

[incf]

If I understand correctly you have multiple PSUs with their outputs in parallel. A VERY bad idea.

This is a battery backup system for a power bus. It just batteries in parallel (with diode on each to prevent backflow) for sufficient run time.

That additional info does not explain why you said the fuses are in series.

Still a bad idea as equal current sharing requires the batteries to have the same voltage and the diodes to have matched voltage drops.

Suggest you provide a schematic that shows all the relevant details

Series connections of fuses occur due to there being multiple sources and multiple loads. Sources may only emit 5A, loads may only consume 5A, and no segment of the bus/wiring may pass more than 5A. Some wiring/bus connections have dedicated fuses. Therefore, when a short occurs, it often might involve multiple identical fuses.

For example if a short occurred at a load, the fuses involved would be:

• one or more batteries each with a fuse
• one or more bus/wiring links each a with fuse
• one load with fuse

There is no requirement for equal current sharing.

I am primarily concerned with the long term (mis)behavior of the fuses under and after fault conditions. I fear a blowing and replacing a single fuse may render the portions or even the whole bus unreliable due to fatiguing/damaging of multiple series fuses.

[Gyro]

It is probably worth mentioning that automotive blade fuses are about the worst characterized fuses that there are (short of fuse-wire rewireable ones). They are housed entirely in plastic and are likely to melt if operated for an extended time beyond their maximum rating. The element is also exposed to the air, so prone to high temperature oxidation. [Edit: Their interrupt current capacity is also low. actually the larger blade types can be up to 1kA at 58V DC, surprising].

[incf]

It is probably worth mentioning that automotive blade fuses are about the worst characterized fuses that there are (short of fuse-wire rewireable ones). They are housed entirely in plastic and are likely to melt if operated for an extended time at or beyond their maximum rating. The element is also exposed to the air, so prone to high temperature oxidation. Their interrupt current capacity is also low.

Size, packaging and availability drove us away from traditional glass fuses. We are okay with the "current vs time to blow" specifications that are typical of these fuses.

These fuses will live well below their rated current their entire life. Although temperature and environmental contaminates are certainly not guaranteed.

I do wonder about the interrupt capacity, I will need to check that. Although, given the number of connectors, and length of wiring, I have a hard time imagining the system providing anywhere near enough to hit the interrupt limit.

I feel as though the track record of mini blade fuses in automobiles appears "pretty good" *** , they've been around for a few decades - maybe I am mistaken?

*** As long as the fuses are protected from corrosion, and are not counterfeit.

[Gyro]

As my edit above, interrupt rating is actually surprisingly good - well it surprised me anyway. Of course, there will always be cheapies around.

[calzap]

If there is a sudden overload, it can be hard to predict which protective device in series will open first even if they have different amperage ratings.  I once installed a 240 VAC wall switch incorrectly.  It was fed in from a 30 A breaker in a panel in the same building.  That panel had a 100 A main breaker which was fed from a 100 A breaker in a panel in another building 100 m away.  When I flipped the switch to on, it was the 100 A breaker in the other building that tripped.

Mike

[garrettm]

So, just some thoughts. Diodes are very lossy for this application. What you want is an "ORing" circuit using a MOSFET or "hybrid relay" (MOSFET in parallel with a normal relay for very, very low conduction losses). You can use either DCCTs (DC current transducer) or current shunts to detect current magnitude and direction as well as triggering a protection mechanism for sustained surge currents. DCCTs from LEM can be had for fairly cheap on eBay. There may be commercial battery ORing products readily available, but I don't know of any.

Fuses are not really the best approach here (usually too slow and unpredictable), and can be a significant source of resistive losses in the system if you have too many of them.

https://www.eetimes.com/fundamentals-of-power-system-oring/

https://e2e.ti.com/support/power-management-group/power-management/f/power-management-forum/972417/lm66100-best-topology-for-oring-battery-and-usb-power-sources-with-rpp

https://sound-au.com/articles/hybrid-relays.htm

Hybrid relays are a pretty neat idea: they reduce EMI and conduction losses as well as reduce the size of mechanical relays/contactors and the solidstate switch. The solidstate component is only used during turn on and turn off events to minimize arcing and bounce of the contacts. This also allows for safely switching off DC voltages that would otherwise produce a continuous arc (from metal vapor), as DC has no zero voltage crossing like AC to naturally extinguish the arc.

[incf]

So, just some thoughts. Diodes are very lossy for this application. What you want is an "ORing" circuit using a MOSFET or "hybrid relay" (MOSFET in parallel with a normal relay for very, very low conduction losses). You can use either DCCTs (DC current transducer) or current shunts to detect current magnitude and direction as well as triggering a protection mechanism for sustained surge currents. DCCTs from LEM can be had for fairly cheap on eBay. There may be commercial battery ORing products readily available, but I don't know of any.

Fuses are not really the best approach here (usually too slow and unpredictable), and can be a significant source of resistive losses in the system if you have too many of them.

https://www.eetimes.com/fundamentals-of-power-system-oring/

https://e2e.ti.com/support/power-management-group/power-management/f/power-management-forum/972417/lm66100-best-topology-for-oring-battery-and-usb-power-sources-with-rpp

https://sound-au.com/articles/hybrid-relays.htm

Hybrid relays are a pretty neat idea: they reduce EMI and conduction losses as well as reduce the size of mechanical relays/contactors and the solidstate switch. The solidstate component is only used during turn on and turn off events to minimize arcing and bounce of the contacts. This also allows for safely switching off DC voltages that would otherwise produce a continuous arc (from metal vapor), as DC has no zero voltage crossing like AC to naturally extinguish the arc.

I'm currently opting for simplicity despite there being some power dissipation. We've got enough surface area that I doubt a few watts of heat will matter.

We have a relay switch. But we opted not to perform ORing.To do it properly (to be "as reliable and fool-proof as a plain old diode") would have required either a microcontroller, or a ton of discrete logic. Prone to "chattering" and backfeeding when there are many battery units in parallel.

ST's field effect rectifier diodes (FERDs) are pretty good. Vf=325mV typical @ 4A.
https://www.st.com/en/diodes-and-rectifiers/ferd60m45c.html

If it wasn't for FERDs we probably would have been forced to do relay-ORing.

I've done designs with MOSFETs, etc. as relay replacers in aircraft and it's just really hard to make them as "foolproof" and durable as a diode + relay ( + fuse + current limiting resistor + etc.). The on-resistances of HV FETs are lousy. They are fairly fragile even when surrounded by protection devices. They can oscillate. etc.

---

I'm okay with the fuses we chose. We simply design around their published ratings so that in the final system tolerates the worst case faults indefinitely.

The losses from the fuses seem negligible at a few amps. (<30mOhms in the socket, or so it seems) We lose a lot more elsewhere.

[thm_w]

One will blow and the rest will degrade, whether that degradation is serious enough to care about, not sure you'll have to research and decide for yourself.

Obvious solution sort of alluded to above is to have one fast blowing element, like a resettable breaker, that you know will react quicker than the slow blow(?) fuses.

https://www.sciencedirect.com/science/article/pii/S0142061519318423

[incf]

Maybe I'm not thinking creatively enough, but I am having trouble imagining where a "weak link" would fit in.

Suppose a fault could occur anywhere in the system shown below (items are physically far apart and have wiring that must be protected):

edited to remove doodle and close thread

If the fault can occur anywhere, where does one place a single weak link?

*Admittedly, I have not drawn the system correctly (actual arrangement is more complex and has fewer "obviously unnecessary" fuses), but I think it illustrates the concept of redundant series fuses existing along wiring/bus links

[SteveThackery]

Haven't your original questions been answered? If all fuses are identical, it's probable that one will blow and the rest will be stressed.

[incf]

Haven't your original questions been answered? If all fuses are identical, it's probable that one will blow and the rest will be stressed.

Correct, the primary question has been addressed, leaving only one or two secondary questions. The thread is now in the "clearing up loose threads" phase if its life cycle.

[coromonadalix]

Maybe I'm not thinking creatively enough, but I am having trouble imagining where a "weak link" would fit in.

Suppose a fault could occur anywhere in the system shown below (items are physically far apart and have wiring that must be protected):
(Attachment Link)

If the fault can occur anywhere, where does one place a single weak link?

*Admittedly, I have not drawn the system correctly (actual arrangement is more complex and has fewer "obviously unnecessary" fuses), but I think it illustrates the concept of redundant series fuses existing along wiring/bus links

this schematic  just to show examples  contains multiples weak links

the supplies  put together with a fuse between them v1 v2 or v3 v4 ....        unless you have exactly adjusted the same way  or provide load share,  for sure  there is possible trouble there

as other wrote    even identical fuses currents specs, they are not perfectly made, there is some slight differences  ... the weakest of the one in serial will burn and the others will be stressed out for sure

@incf      learn about power distribution ...

[floobydust]

The schematic is dangerous, there are fused nodes with 5A, 10A, 20A available. You can't mix serial and parallel feeds and not have wiring burn up.
Power feeds are typically a tree-structure, feeder and branches in a star architecture.

[incf]

Okay, thank you everyone for your input about fuses.