Fuses for Supplementary Protection:
Safety Versus Reliability
A case study by
John McBain, PSE Society
Jim Krepelka, Reliability Society
18 February 2014
Disclaimer Statements
• The data in this presentation are intended
to convey concepts. Therefore, data charts
may have “synthetic data”; that is,
qualitatively correct but not quite real.
…
• No functioning analytical devices were
harmed in the making of this presentation.*
*verified by the SPCI (Society for Prevention of Cruelty to Instruments)
What is the purpose of a fuse?
Product Safety
Reliability
As supplementary
protection the fuse
must open in order to
remove power during
a short circuit or
overload condition.
As a component in a
high reliability system
the fuse must be very
stable over time and
not have unintended
failure modes.
These purposes are not incompatible,
but they require different evaluation.
Fuses work well -- mostly
• Fuses work very well for most applications
where the reactive load is minimal (p.f.~1).
• In those cases fuses tend to see current that is
either steady-state or short-circuit.
• Therefore those applications are not the focus
study of this talk.
• Too easy
Up for a challenge?
• Power up effects for large capacitive or
inductive loads can play havoc with fuse
behavior.
– This can affect both reliability and safety.
– Especially true for electromechanical devices
(extended in-rush time).
– For example, where there is a strong
relationship between line frequency
and rotational speed
• Not so easy
Story time ... A Case Study
• Once upon a time there was an R&D project
that needed lots of vacuum pumping.
• The project team chose a vacuum pump using
a large universal AC motor.
• But they were under severe space constraints
when it came to the system’s supplementary
AC protection devices.
• They opted to go with board mounted fuses.
Specifying a fuse by “rule of thumb”
• This product needed about 4 amps in steady
state for the vacuum pump.
• So 6 amp fuses went into the first prototypes.
• But the fuses failed after a few system starts.
• So 8 amp fuses were substituted.
• But the fuses failed after a couple dozen system starts.
• So 10 amp fuses were substituted.
• But the fuses failed after about 50 system starts.
Well what happened next??
• Well ... Jim’s telephone started ringing.
• “Hey Mr. Reliability – We did everything right.
Want to come over and tell us why these lousy
fuses keep failing?”
• Time to measure in-rush.
In-rush, who needs it anyway?
• Inrush current, input surge current or switch-on
surge refers to the maximum, instantaneous
input current drawn by an electrical device when
first turned on.
• Here are in-rush plots from devices with minimal
reactive load. Note the short durations.
In-rush for a universal AC motor has a
different decay characteristic!
• It may take many cycles to reach a steady state
condition. (When do you stop calling it in-rush?)
• This extended in-rush has to do both with charging the
inductive load as well as with having the motor reach
its nominal rotational velocity.
Fuse Curve vs. In-rush Distributions
Several trials under different specified conditions.
Example of good reliability – distributions all below line.
In-Rush RMS Versus Time
Desirable EUT RMS Time-Current Curve
No Fuse Degradation
100
Will Open
10
May Open
Time (sec)
1
0.1
0.01
Will Not Open
0.001
0.0001
0
10
20
30
Current (amps)
Min Time, 10 A TD, SPT 5 X 20
Max Time, 10 A TD, SPT 5 X 20
RMS Current for EUT
40
50
60
What is the practical effect?
• Each time the RMS current goes into the “May
Open” range, the fuse filament degrades.
• When the filament degrades, the practical effect
is to lower the amperage rating of the fuse
slightly, shifting the fuse curve towards the origin.
• With the fuse curve closer to the origin, future inrush events will cause even faster degradation of
the filament – positive feedback.
• Eventually the degraded fuse will open during a
normal start.
• The higher the fuse rating, the longer the fuse will
last (other factors being equal)
In-Rush RMS Versus Time
Undesirable EUT RMS Time-Current Curve
Fuse Degradation
100
Will Open
10
May Open
Time (sec)
1
0.1
0.01
Will Not Open
0.001
0.0001
0
10
20
30
Current (amps)
Min Time, 10 A TD, SPT 5 X 20
Max Time, 10 A TD, SPT 5 X 20
RMS Current for EUT
40
50
60
Leaving no chance for failure
• To comply with the “Reliability” in-rush
requirements for the fuse, the specified fuse was
about three times steady-state current.
• To comply with the “Safety” requirements for the
fuse, the specified fuse opened within a few
seconds during single-fault locked-rotor tests.
• If an intermediate high current could occur under
overload conditions, then the certified motor’s
internal thermal interlock should activate.
• The construction and test results met internal
specifications and external safety standards.
And they all lived happily ever after ...
• Then came a report from the field that charring
had occurred on fuses and fuseholders.
• The system failed safely ...
for clarity let me just say
SAFELY…
• Still, the product needed a new fuse board and
the customer was irritated by the smell of burnt
plastic.
• Now John’s phone starts ringing….
• “Hey Mr. Safety, How come this happened? Isn’t
the fuse supposed to provide overcurrent
protection?”
First hypothesis … circuit inductance
• Some information from the field
– Incident in Japan (nominal 200Vac from mains)
– Step-up transformer was used with product
– Start up was after maintenance
• Initial speculation
– High temperatures usually are from high current, and
lower line voltage would have higher current.
– If maintenance created a potential short (falling
debris?), high inductance from the transformer could
help start and sustain it.
– Maybe the impedance of the entire ac input circuit
prevented the wall breaker from tripping?
Nice try, but …
• Reasons why the facts did not align:
– Maintenance was not in the ac input area.
– Charring occurred around the fuseholder, not at
places that easily could be shorted by debris, such as
connectors.
– If a high current was maintained for a while, why
didn’t the fuse open quickly and stop the incident?
• Maybe it is just a random failure of a defective
part?
Not so random?
• Then came a second report of a similar failure.
• Both failures were on older products.
• Perhaps the vacuum pump is wearing out and
drawing increased current??
– Not enough to activate the motor’s thermal
interlock but …
– Just enough to cause the fuse and fuse holder to
overheat and char without opening the fuse?
Second hypothesis … current heating
• Let’s check out how hot the fuse can get with
continuous current
• Experimental setup:
– Use low-voltage DC current source
– Monitor temperatures on fuse and fuseholder
– Try different currents near & above the fuse rating
Start some current runs
Fuse Failure Curves
300
Fuse body temperature, Degrees C
250
200
20 Amps F4 (C)
19 Amps F4 (C)
150
18 Amps F4 (C)
17 Amps F4 (C)
16 Amps F4 (C)
100
50
0
0
200
400
600
800
Time in seconds
1000
1200
1400
1600
Getting hot but not producing the
damage of the field events
Less current = longer fuse survival
• Survival time increased from hours to days.
• In all cases, there was no sign of charring.
• Serious degradation effects were:
– Annealing of spring contacts in the fuse holder
– Solder leaking from the fuse body
• But there just did not appear to be sufficient
energy available to cause charring or auto
ignition of the available materials.
It must be something else …
• Using high current by itself was not enough
to reproduce the failure.
• Must be something more than just I2R.
• Perhaps one of the thermal degradation
modes was activating a new mechanism – a
mechanism which would have enough
energy to char and reach auto ignition
temperatures.
Another field report … sigh
• This just in … Field report of a fuseholder charring 5
seconds after fuse replacement!
• Clearly, this event has really high power dissipation.
• Could this be an electric arc event??
• Let’s see exactly how the fuseholder works.
Third hypothesis … electric arc
• The fuse cap has 2 wiper arms that press
against the contact ring of the fuseholder.
• If both wiper arms lose contact – and a small
gap opens in the circuit – an arc could occur.
• Obviously, we can create a gap by manually
replacing a fuse with the power on … but who
would ignore user safety instructions?
Fitting all the pieces together …
• Motor starts to fail and draws unusually high current.
– Not enough to trip the motor interlock
– Enough to over heat the fuse and fuse holder.
• The fuse cap wiper arms anneal and lose their spring
force.
• Eventually both arms lose contact with the metal
contact ring and an arc is initiated by the back EMF of
the high inductance electric motor.
• Within a short time, the arc chars the plastic, vaporizes
the wiper arms and eventually terminates current flow.
Can we make a fault happen?
• Three “gap scenarios” to test:
1. The thermal degradation hypothesis – bend fuse
cap wiper arms so they are just barely not
touching the contact strip of the fuse holder.
2. The human error hypothesis -- fuse cap is not
fully locked into the fuse holder.
3. The foreseeable misuse hypothesis -- fuse cap is
hot plugged into the fuse holder.
Testing … 1
1. The thermal degradation hypothesis – bend fuse cap
wiper arms so they are just barely touching the
contact ring of the fuse holder.
(Videos of several trials removed.)
#1
#2
#3
#4
#5
Results
• No failure. Either the contact touched or it did
not. No arcing or charring.
• But we will come back to this scenario later.
Testing … 2
2. The human error hypothesis -- fuse cap not fully
locked into the fuse holder.
#1
fuse not turned
#2
fuse barely turned
Results
• No failure. Either the contact touched or it did
not. No arcing or charring.
Testing … 3
3. The foreseeable misuse hypothesis -- fuse cap hot
plugged into the fuse holder (motor switch is on).
An arc is possible!
Results
One wiper arm gone – but
the motor continued to run!
Results (continued)
• We have a winner!
• It appears that if you plug the fuse quickly
enough, without any hesitation, you can start
the motor and not arc or char the assembly.
• However, if the fuse contact makes and breaks
after the big inductive vacuum pump motor has
started then a substantial arc initiates.
• AC line by itself does not appear to be sufficient
to strike the arc. However the inductive back
EMF from the motor plus the line voltage has
no problem producing an arc and charring.
Results (continued)
Another hot plug test
example - both wiper
arms are gone and the
motor stops.
Let’s compare to the field return
• The damage location is the same.
• The damage to the wiper arms is the same.
• The damage to the fuseholder body and the pcb
is similar, but less severe in the experiment.
• Note that the experiment used new fuseholders
that never had experienced thermal degradation,
repeated on/off cycles or previous fuse
replacements.
Let’s talk more about scenario 1
• The experiment to get thermal degradation
(scenario 1) did not produce an arc or
charring.
• But our test was not done “real world”.
• The pump needs to be running at the instant
the wiper arm loses contact inside the
fuseholder.
• Then it will look exactly like the hot plug
scenario 3.
Isn’t it Ironic …
• We put the fuse in to protect against
overcurrent & thermal events.
• But it was the thermal event.
DAMN!
Lessons learned
• Ratings OK, construction OK, standard testing OK –
but that may not be enough.
• Interactions between devices and the resulting
wearout characteristics must be considered.
• Incorrect user behavior must be considered, in spite
of warnings about what to do (or NOT to do).
• Fuses and fuseholders can sometimes be the source
of thermal events - instead of stopping them!
• Keep looking until you find the REAL reason for
failure.
Our recommendation
• Don’t fall into the trap of component myopia.
• A fuse does not exist as an ideal component
suspended in a state of grace.
• It is a physical and mechanical device and forms
a system with the fuse holder.
• Depending on your product, a fuse may not be
the best answer – no matter how convenient.
Supplementary Information
• “Rules of thumb” as described in these documents are
often very good guidance.
• In-rush curves with extreme time, current, and shape
characteristics may or may not fit the rule.
www.littelfuse.com/.../Fuseology.pdf
http://powerelectronics.com/ .../DesignFeature.pdf
www.schurter.com/.../file/Guide_to_Fuse_Selection.pdf