Fire hazard question

[Derek Potter]

I wonder if anyone can help me?

I am trying to find out what the current "best practice" would be
apropos fire hazard protection in a particular situation as follows:

The unit is a (small) fire alarm, it has a 12V external alarm line.
The curreent level there is normally 120mA but because it is external,
there is a significant risk of someone interfering with the wiring and
shorting it out. The 12V is essentially unlimited current but the
circuit is protected against gross faults by a 1A fuse. The external
alarm driver is a TO92l MOSFET, Ron ~2 ohms, which will obviously rise
as it heats up, so I can envisage it taking a couple of amps for
several seconds before the MOSFET literally burns out. Whilst most
small semiconductor burn-outs take place harmlessly in the air above
the PCB, it is obviously a fire hazard, albeit a remote one. To
provide some protection my colleagues have suggested a semiconductor
current limit.

My question is what we should be aiming to do - it is common enough
for people to talk of "single component failure" i.e. good practice
requires that a hazard will not be caused by any single component
failure. Unfortunately I believe there are strings attached: IIRC,
generic European and American standards specify that potentially
hazardous faults should be monitored or discoverable through routine
testing. In addition I believe they only count if they are spontaneous
internal component failures, not external conditions caused bu human
interference or bad installation.

The problem is thus that the current limit could fail spontaneously -
a single component failure - and this not be detected. So we are back
to square one, with the system vulnerable if someone causes an
external short. This causes an internal burn-out which may be a slight
fire hazard. All in all, the frequency of such fires occuring is
probably incalculably small, but I think the system could escape the
"single component failure" criterion if this is applied strictly.

This query has appeared in alt.electronics. Please note that I am not
asking for circuit suggestions as I can design circuits in my sleep
(and frequently do). I have had a string of circuit suggestions
ranging from relying on a PCB trace as a fuse, through to using a PTC
thermistor as well as the current limit, all of which are unacceptable
for various reasons. Clearly such additional protection would be "belt
and braces" but, in any case, such specific circuit suggestions only
reflect one individual's ingenuity, they do not elucidate the safety
design *critera* which, in the worst possible scenario could be
invoked in court under the heading of "due care"... I just want to
know what regulations, recognised best practice, and general standards
(US and Europe) have to say about this kind of situation: where
multiple faults are needed to cause a hazard but the faults may not
meet the criteria of being

1 spontaneous
2 internal
3 monitored

TIA - remember, no circuit suggestions, thank you.

 

[[email protected]]

The unit is a (small) fire alarm, it has a 12V external alarm line.

Consult the applicable UL standards. There's no simple answer to your
question, however:

The problem is thus that the current limit could fail spontaneously -
a single component failure - and this not be detected. So we are back
to square one, with the system vulnerable if someone causes an
external short.

This is considered a multiple-point failure. Of itself, the failure of
your protection component is not hazardous.

UL is very specific about which conditions need to be
monitored/supervised, and as far as I can recall from detailed study of
these requirements six months ago, this is not one of those conditions
(at least, it is not one of the conditions in UL864 commercial fire;
I'm not as intimately familiar with the residential requirements but in
general they are looser).

Also note that UL has very specific requirements about the flammability
of the enclosure (if nonmetallic) which would appear to obviate the
need for this heart-rending on your part.

 

[colin]

I wonder if anyone can help me?

I am trying to find out what the current "best practice" would be
apropos fire hazard protection in a particular situation as follows:

The unit is a (small) fire alarm, it has a 12V external alarm line.
The curreent level there is normally 120mA but because it is external,
there is a significant risk of someone interfering with the wiring and
shorting it out. The 12V is essentially unlimited current but the
circuit is protected against gross faults by a 1A fuse. The external
alarm driver is a TO92l MOSFET, Ron ~2 ohms, which will obviously rise
as it heats up, so I can envisage it taking a couple of amps for
several seconds before the MOSFET literally burns out. Whilst most
small semiconductor burn-outs take place harmlessly in the air above
the PCB, it is obviously a fire hazard, albeit a remote one. To
provide some protection my colleagues have suggested a semiconductor
current limit.

My question is what we should be aiming to do - it is common enough
for people to talk of "single component failure" i.e. good practice
requires that a hazard will not be caused by any single component
failure. Unfortunately I believe there are strings attached: IIRC,
generic European and American standards specify that potentially
hazardous faults should be monitored or discoverable through routine
testing. In addition I believe they only count if they are spontaneous
internal component failures, not external conditions caused bu human
interference or bad installation.

The problem is thus that the current limit could fail spontaneously -
a single component failure - and this not be detected. So we are back
to square one, with the system vulnerable if someone causes an
external short. This causes an internal burn-out which may be a slight
fire hazard. All in all, the frequency of such fires occuring is
probably incalculably small, but I think the system could escape the
"single component failure" criterion if this is applied strictly.

This query has appeared in alt.electronics. Please note that I am not
asking for circuit suggestions as I can design circuits in my sleep
(and frequently do). I have had a string of circuit suggestions
ranging from relying on a PCB trace as a fuse, through to using a PTC
thermistor as well as the current limit, all of which are unacceptable
for various reasons. Clearly such additional protection would be "belt
and braces" but, in any case, such specific circuit suggestions only
reflect one individual's ingenuity, they do not elucidate the safety
design *critera* which, in the worst possible scenario could be
invoked in court under the heading of "due care"... I just want to
know what regulations, recognised best practice, and general standards
(US and Europe) have to say about this kind of situation: where
multiple faults are needed to cause a hazard but the faults may not
meet the criteria of being

1 spontaneous
2 internal
3 monitored

TIA - remember, no circuit suggestions, thank you.

Some regulations can be quite restrictive and at the same time a bit
ambigous. to get round a similar problem of a single point failure going
unnoticed and subsequent failures/short circuit cuasing something to fry and
emit smoke flames etc, the solution was to make the component non
flamable/contain the flame/smoke. a simple resistor designed to fail open
circuit without emiting any smoke was used aka fusable resistor. some of
these just had a plastic sleave covering an ordinary resistor.

Colin =^.^=

 

[Rich Grise]

I wonder if anyone can help me?

I am trying to find out what the current "best practice" would be
apropos fire hazard protection in a particular situation as follows:

The unit is a (small) fire alarm, it has a 12V external alarm line.
The curreent level there is normally 120mA but because it is external,
there is a significant risk of someone interfering with the wiring and
shorting it out. The 12V is essentially unlimited current but the
circuit is protected against gross faults by a 1A fuse. .....
TIA - remember, no circuit suggestions, thank you.

Well, I don't know if you'll disqualify this as a "circuit suggestion",
but why not run a constant-current driver instead of constant-voltage.
It's inherently short-proof, but you don't say what kind of logic
you're using in the overall system (i.e., a contact closure to 12V?
An open a la broken foil?) How you're sensing that "external alarm
line" makes a lot of difference.

Or, just current-limit the 12V supply, and report when the current
limit is exceeded.

Good Luck!
Rich

 

[Derek Potter]

Well, I don't know if you'll disqualify this as a "circuit suggestion",
but why not run a constant-current driver instead of constant-voltage.
It's inherently short-proof, but you don't say what kind of logic
you're using in the overall system (i.e., a contact closure to 12V?
An open a la broken foil?) How you're sensing that "external alarm
line" makes a lot of difference.

Or, just current-limit the 12V supply, and report when the current
limit is exceeded.

Good Luck!

Sorry, the external alarm line is an output alarm like a sounder/lamp
etc, not an input. It is not monitored. the relevant BS to this
particular application does not require it.

A constant current driver is, in effect, what my current limiter
becomes under overload. It's failures in the constant current driver
that I'm worried about. As I said, it wouldn't cause an immediate
hazard but it would leave the system vulnerable and as it's not
monitored this could mean it can't be counted as protection.

Obviously testing the current limit would require a periodic check
whether automatic or manual.

With your current limited supply, you can't report when the current is
exceeded as it won't be until the external fault occurs. By that time
a failure of the current limit could have occurred and not been
noticed.