Fuse vs no fuse

[ricko_uk]

Hi,
a client has a PCB made more than 9 years ago that they want to use on a new system.

The PCB does not have an on board fuse and it relies on the short circuit protection of the AC/DC power supply itself.
This is the external AC/DC power supply (24V 2A): https://www.xppower.com/portals/0/pdfs/SF_AFM30-60.pdf
The PCB has several regulators on it to generate various voltage 3V3, 5V, 12V from the 24V input. So if some failure occurs it can still happen on a subsystem without blowing any fuse at the input.
Also Some regulators have external switching mosfets with the top side connected directly to the 24Vin.

Obviously it is always preferred to have a fuse, but:
1) is it still ok to use that PCB without fuse IF that PSU specifically is used (i.e. rely only on it and its specs and short circuit protection)?
2) are there any regulations (Europe/UK/US) that for those input voltages/currents require a fuse at the input?
3) low voltage directive applies from 70V DC and 50V AC. What directives apply for 24V and below?

Thank you

[thm_w]

How do you guarantee that the given power supply would be used. If someone plugs in a 10A 48V plugpack, what is the end result?
Is the 24V used in any way? What is the maximum voltage of the regulators.

I would consider a fuse if you plan to provide a warranty for the product and want to avoid cases of catastrophic failure.


https://ec.europa.eu/info/business-economy-euro/product-safety-and-requirements/product-safety/consumer-product-safety_en

[ricko_uk]

Thank you thm_w,

The 24V does go to a second PCB but that is a newer board and is fully protected with its own fuse.

Yes if they agree to modify or redesign the PCB we will include it. But we are trying to figure out if it is ok as is when it comes to regulatory standards or if it is required by any regulatory standards.

Thank you

[Someone]

How do you guarantee that the given power supply would be used. If someone plugs in a 10A 48V plugpack, what is the end result?
Is the 24V used in any way? What is the maximum voltage of the regulators.

I would consider a fuse if you plan to provide a warranty for the product and want to avoid cases of catastrophic failure.


https://ec.europa.eu/info/business-economy-euro/product-safety-and-requirements/product-safety/consumer-product-safety_en

Even then an input fuse may not be sufficient. Modern safety standards are quite good at describing whats needed......    but first you need to figure out which standards apply.

[ricko_uk]

Thank you Someone,
where do I find the standard that might apply? The maximum voltage of the system is 24V (from which other lower voltages are derived) and maximum current 2A. All that is inside a plastic box. The power supply is external (link in the OP) and connected to the box via its own lead.

Thank you

[m98]

What's the application? Different standards apply to a household appliance than to a children's toy than to a medical device.

But in general, I don't think a fuse is strictly required by applicable standards in most cases. You just have to ensure the device can handle 1.5x the maximum fault current without bursting into flames. Sure, whether product liability insurance will like that might be something to consider.

[bdunham7]

Do the regulators not have current limiting?  Fuses aren't always the best answer for robust fault protection. 

[Someone]

But in general, I don't think a fuse is strictly required by applicable standards in most cases. You just have to ensure the device can handle 1.5x the maximum fault current without bursting into flames. Sure, whether product liability insurance will like that might be something to consider.

Yeah, no. Standards have moved on from such simple measures. But just determining what the fault current is varies depending on the specific standard.

[Someone]

where do I find the standard that might apply? The maximum voltage of the system is 24V (from which other lower voltages are derived) and maximum current 2A. All that is inside a plastic box. The power supply is external (link in the OP) and connected to the box via its own lead.

If you don't know where to start then its probably best to hire a consultant (they will rapidly identify the relevant standards and probably have them to hand) as this is a specialist area of practice you cant just learn in a day.

[ricko_uk]

Thank you bdunham7, m98 and Someone,

@bdunham7,
What alternatives are you thinking about?
Two of the regulators have external switching mosfets (i.e. the top one is connected straight to the 24Vin).

@m98,
the application is a desktop scanning and printing system

Thank you

[bdunham7]

What are better/alternatives to the fuse?

PPTC
Circuit breaker
Current limiting design (limits maximum current and all components can tolerate short circuit, like a lab PSU)
Foldback current limiting design (reduces current to a low value on overload--LM series)
Current overload reset design (disconnects on overload, needs power cycle to reset--ATX for example)
Impedance protection design (short circuit tolerant--Class 2 transformers, etc.)

[nali]

Do the regulators not have current limiting?  Fuses aren't always the best answer for robust fault protection.

That's a very good point which often gets overlooked. I've personally seen an installation where a (fortunately small) fire has been caused because the supply didn't have enough grunt to actally blow the downstream fuse

[T3sl4co1l]

UL508's limited energy circuit definition comes to mind (5A at up to 20V, 100VA up to 42.4V; that's VDC or Vpk and A RMS), though that's industrial and I'm not sure offhand which standards (and sections thereof) should apply to commercial/office/telecom equipment.  I don't recall reading anything about amperage in IEC 60950-1 but I may be forgetting something, and also 62368-1 is the newer version more or less so may have other changes.  Which coincidentally, the linked power supply has all the stamps from these (plus '601), so that would be good further reading if you need to be sure.

Agreed, good idea to plan for incorrect supplies; label appropriately (legal CYA), putting in a fuse is probably a good idea.

Note that SMT chip fuses -- fuse-fuses, not PTCs -- are cheap and plentiful.  This power level is kind of near the upper limit where PTCs are reasonable, so availability or cost might be poorer, and functionality worse (slower from the bigger thermal mass?), but they're also definitely doable.  Well, I haven't checked stock on 'em lately, but y'know... in ordinary times they'd be available, at least.

Probably plan for excess and reverse voltage, too -- maybe a crowbar diode or something.  For OV, might be good enough to just let a SMAJ size TVS cook off, fail shorted (potentially) and clear the fuse (or shut down the supply) that way?

What are better/alternatives to the fuse?

PPTC
Circuit breaker
Current limiting design (limits maximum current and all components can tolerate short circuit, like a lab PSU)
Foldback current limiting design (reduces current to a low value on overload--LM series)
Current overload reset design (disconnects on overload, needs power cycle to reset--ATX for example)
Impedance protection design (short circuit tolerant--Class 2 transformers, etc.)

You can't protect against absolutely everything, of course; they can always build a better idiot.  Say it gets cross-wired to 120, or even 240V mains, right?  So it's still a good idea to put in a conventional, reliable method, like a fuse, on top of these methods.  So you might combine a current limiter with an overvoltage crowbar, so that it's able to ride through excess voltage (whether fully functional during, or switched off by OVLO), but still confidently clears the fuse when pushed beyond absolute maximum.

And for that matter, mind the fuse ratings; typical SMT chips are like 50 or 100V and a rupture current of say 20-100A, that sort of thing.  Obviously not great if mains hit that, so you might want a larger (brick or cartridge) type in that case.  But it would be perfectly adequate for most low voltage applications, even against beefy-ass batteries if you have enough internal resistance on your end to guarantee the rupture current is not exceeded.  (Which is also a good way to use PTC fuses, have enough resistance to guarantee not exceeding Imax -- and you should get a reasonable number of cycles out of it.  Which, keep in mind, cycle life is like, dozens, they aren't terribly reliable parts.  Use them as you would any other fuse -- as a last resort!)

So that pretty comprehensively covers everything from the casual (none, or a fuse), to industrial or aerospace grade protection.

Tim

[SiliconWizard]

@m98,
the application is a desktop scanning and printing system

It's "interesting" that the board in question was designed without protection, if I got it right, and to be powered by a medical-grade power supply that contains the protection. Sounds more like an after-thought than a deliberate choice, especially since those "medical-grade" power supply adapters tend to be a bit expensive, so the extra cost is certainly more that what it would have taken to add appropriate protection on the board itself to begin with.

The problem with this setup as I see it, is that said power supply is an external, self-contained adapter, and with a standard plug, AND for a pretty general-purpose application here (so the user profile can't be controlled, unlike more specific devices, for instance). So it will be almost impossible to justify that this be safe enough - even if an added information was provided to the user, stating that the device shouldn't be used with any other power adapter, which should be mentioned anyway here - since the user could be using any other after-market adapter with the same plug and compatible characteristics, which are just everywhere as laptop adapters and similar... (and which a lot are just nasty crap!)

The bare minimum if you still had no choice but doing this, would be to change the power connector to something non-standard and a lot more "exotic", as a means of "foolproofing" it. Still a bit shaky, but better than the original.

[floobydust]

The rule in engineering is: if you touch a design - it's yours.
If you change a PC board, schematic etc. you are now responsible for the design (safety of the public) regardless of it's prior deficiencies. It's your baby now.

For this reason, do a safety assessment and add the fuse or two. Why be reluctant to do so?

[ricko_uk]

Thank you all as always!