Electronic Fuse Circuits For DC and AC

In the article Electronic Fuses - A Collection Of Useful Ideas, I've examined a number of different topologies.  Most are 'instant' acting, so there's no delay between an over-current condition and operation, although there are options to provide 'slow-blow' capability with some circuits.  In contrast, the circuit described here is designed for use with an Allegro ACS712 (or ACS758xCB where 'x' is maximum current) Hall-effect sensor, and there are only limited choices.  As Devo was once heard to tell us in song, "Freedom of choice is what you got, freedom from choice is what you want".  While this is true enough (especially now), presenting a number of circuits often confuses the beginner, so I'll keep this simple.

One thing that's lacking in many of the circuits you may find on-line is latching.  This is important, as the last thing anyone wants is for the fuse to disconnect and re-connect repeatedly if there's a fault.  'Real' fuses become open-circuit when blown, and cannot be reset other than by replacement.  Circuit breakers trip, and require a manual reset.  An electronic fuse must have similar capability, but it will usually be disconnection of an external (auxiliary) power source that forces a reset.  Another critical factor is that an electronic fuse should never allow the protected circuit to receive power if the auxiliary supply is not available, or is turned off.

The switching mechanism has to be able to interrupt the fault current, and this isn't always easy to achieve.  Relays may not be a good choice, because if the fault is serious, many relays cannot interrupt the fault current, especially if it's DC above 30V.  Once you exceed the 30V threshold, the ability of a relay to interrupt a current of several (or many) amps is not good, as there is every chance that an arc will be formed that literally melts the contact assembly.  Using semiconductors always includes some risk, because when they fail, it's almost always short-circuit.  Therefore, I recommend that a standard fuse is always included, so that some protection remains even if the switching device fails.

Part of the process is also minimising the voltage-drop caused by the electronic fuse.  Some published circuits will drop 1-2V, which may be a significant part of the supply voltage.  This problem gets worse as the supply voltage is reduced, because the burden voltage remains constant.  Such circuits also add impedance (or just resistance) into the supply line, so some powered products may misbehave due to the series resistance added.  This is avoided in the designs shown, which use a Hall-effect current sensor (DC) or a current transformer (AC).

The circuits all include a standard fuse, which will be rated at something greater than the protection limit set, but with a value that will definitely fail should a major fault occur and the e-fuse doesn't operate for any reason.  The idea is that under all normal conditions, the e-fuse will open at the designated current, protecting the load and the supply.  Because no electronic circuit is 100% reliable under all foreseeable conditions, not including a conventional fuse could lead to dire consequences.  Most people will agree that anything 'dire' is best avoided.

The recommended ACS712 is a Hall-effect sensor, available in three current ratings - 5A, 20A and 30A.  They are rated for over-current of up to 5 times the rated current, but the PCB would need to be very robust to handle that with the 30A version (150A).  The output voltage is centred on +2.5V, and the details are shown below ...

Where higher current is required, the ACS758 series can be used (these aren't shown in the drawings).  With current ratings from 50A to 200A, these are heavy-duty ICs with a substantial lead-frame for the current sensor.  While it's unlikely that these would be included in most DIY projects, it's worth pointing out that they exist.  They can handle an over-current of up to 1,200A at 25°C.

These ICs come in two versions, bidirectional (AC) and unidirectional (DC).  Both have an offset - Vcc/2 for the AC types and 600mV for DC.  If you wanted to use these, the values of R3 and R4 would need to be changed to suit.

I'll concentrate on the 5A version of the ACS712, but substitution is easy, and only requires a threshold change to set the current.  They are bidirectional devices, so can monitor AC or DC.  For use with AC, I insist on monitoring both polarities, because a fault can occur where only one polarity is affected.  It may be unusual, but it can (and does) happen.  Doing this is a nuisance with a device having a 2.5V DC offset, because the offset means that full-wave rectification is difficult.  For AC applications, it's easier to use a capacitor to remove the offset, then full-wave rectify the result.

Of course, there's an ideal device for AC - a current transformer.  These have virtually no resistance (only that of the wire passing through the centre), and are available for currents from 5A to 500A (and more).  Needless to say, they do not work with DC, where the ACS712 can be used with either AC or DC.  However, the ACS712 devices are SMD (surface-mount devices) and are fiddly to use.  Consider buying a module intended for use with an Arduino, as these have the IC pre-mounted and include a terminal block for connecting the supply and load.  Note that the polarity is important!

With the 5A version, at zero current the output will be at 2.5V.  If we need a trip current of 4A, the output is 180mV/ A, so it will rise to 3.22V at our maximum current.  If the current exceeds 4A, the e-fuse should trip.  A sensible design will lock out after it's tripped, otherwise there will be an endless cycle of over-current, trip, short delay, over-current, etc.  This is not helpful, and will almost certainly cause more damage if there's a 'real' fault (as opposed to a one-off condition).

The ACS712 uses a Hall effect sensor, which is completely isolated from the current-carrying pins.  The claimed resistance of the internal conductor is 1.2mΩ, so losses are very low provided the PCB has wide traces (and/ or thicker than 'normal' copper).  Because the device has isolation between the sensing and control circuitry, an electronic fuse using the ACS712 can be used with any desired power voltages, and there is no requirement for any common link between the main supply and the fuse electronics.

However, it should be noted that if a relay is used as shown below, the monitored supply voltage must be limited to around 30V to prevent arcing.  I suggest that anyone contemplating higher voltages refer to the Relays articles (especially Part II which deals with contacts).  Higher voltages can be used by operating two sets of contacts in series.  Alternatively, you can use the Project 198 MOSFET relay.  This can be made with MOSFETs selected to suit your supply voltage and current.

A basic circuit is shown below, and it was found (then lost again) on the interwebs.  At first look there's nothing wrong with it, and it is set up to latch with an over-current (via D1), which wasn't included in the original.  Without latching, an electronic fuse is a waste of time, and it can cause additional damage due to constant cycling if there's a fault.  Once tripped, it can be reset by using the switch or by disconnecting the 5V supply.  However, it has a serious flaw that makes it completely unsuitable for use in any real circuit.  It's not immediately apparent unless you already know what to look for.

Figure 1 - General Idea Of An Electronic Fuse (Do NOT Use)

So, where is this mysterious flaw?  Note that the relay uses the normally closed contacts to connect the load.  What happens if the 5V supply is missing for some reason?  Nothing, and that's the problem.  If there's no 5V supply, there is no protection!  This makes it unusable for anything more serious than a quick experiment.  To be safe, the relay must remain activated all the time, and it's deactivated by an over-current fault (the load is also disconnected if the 5V DC supply is missing).  While this does add a few small complexities, without the change the circuit is best described as dangerous, as it gives a completely false sense of security.  Normally I would not show a design that should never be used, but in this case it's important to highlight the problem so readers can use the same criteria to critique other designs that may be found.

Any protective circuit that's so easily disabled is not worth the components used to build it.  Simply by omitting the 5V supply (whether by design or by accident), the circuit will happily pass current to the 'protected' circuit.  With no protection at all, a fault will cause damage, with the real possibility of fire because you're relying on an inactive circuit to protect against excess current.  Naturally, it can do no such thing.  Because it also lacks a 'final backup' in the form of a wire fuse, it's not just a false sense of security, but dangerous.  The following shows how it should be wired.

Figure 2 - Fail-Safe Version Of The Electronic Fuse

This shows how it must be wired to provide safety.  The comparator is reversed, so its output is normally low, thus keeping the relay activated in the absence of a fault.  It goes high when a fault current occurs.  R6 (560Ω) is included to ensure that the voltage can reach close to 5V when the comparator turns off.  The output voltage must exceed the maximum output from U1 (3.4V at maximum current).  When a fault causes the relay to release, the zener diode that allows the peak collector voltage on Q1 to fall to around -17V.  This ensures that the relay can release as quickly as possible.  Without the zener, most relays will not release faster than around 10ms after the coil current is interrupted.  By including the zener diode, this is reduced to less than 3ms (typical).

During normal operation, the output of the comparator (U2) will be low, and Q1 is turned on thus energising the relay.  Power is delivered to the load via the N/O relay contacts.  Should the current exceed the threshold at U2 Pin 2, the comparator output goes high, turning on the 'Trip' LED and turning off the relay.  The comparator's output is also fed back to the non-inverting input (Pin 3), which latches the fault condition and prevents the relay from re-engaging after the current falls to zero.  Pressing the 'Reset' button disconnects the positive feedback and restores power to the load.

The circuit is latching, and requires a manual reset if the e-fuse trips.  Should the fault still be present, it will trip again instantly, and will cycle on and off while the 'Reset' button is depressed.  This should not be maintained for any length of time or the relay contacts may be damaged, along with possible further damage to the protected circuitry.  The relay uses a 5V coil, so the current will be greater than you'd expect with a higher voltage version.  Q1 can handle a relay current of up to 100mA easily.

VR1 is used to set the trip current.  The voltage at the comparator's inverting input is the reference.  Provided the output from the current sensor is less than the reference voltage, the circuit remains inactive.  Even the smallest 'transgression' above the set point will trigger the comparator, and D1 provides positive feedback to ensure that the circuit is latched.  The current setting pot is a compromise, because of the different sensitivity of the sensors.  The maximum increase with the 5A sensor is 900mV (above 2.5V, so 3.4V), for the 20A sensor it's 2V (100mV × 20) and for the 30A version it's 1.98V.  With the values shown, that means that VR1 cannot be advanced to maximum for the 5A device, and the 20A unit will only get to 17.3A with VR1 at maximum (26A for the 30A version).  This can be altered by changing the values of R3 and R4 if desired.

The two LEDs need to be high-brightness types, as only a small current is available.  The 'power on' LED isn't critical, but it's easier to make them the same.  The LED current is only about 300µA, but that's more than enough for high-brightness types to give an acceptable indication.  R5 should not be reduced, as that may cause the circuit to fail to latch if it's set for close to maximum current.  Make sure that the Hall-effect sensor is nowhere near any stray magnetic fields (e.g. high-current wiring, mains transformers, etc.), as you may get erratic results if stray fields are present.

If you need a slow-blow fuse, increase the value of C3.  The time constant is set by R2 and C3, and as shown it's 100µs.  For a fuse that can withstand an overload for up to 100ms, simply increase the value of C3 to 10µF.  As an example, if C3 is 100nF, the circuit will tolerate a current of 16A for 850µs, or 22A for 600µs.  These times are extended to 8.5ms (at 16A) or 6ms (22A) respectively if C3 is 1µF.  You will need to perform experiments to determine the optimum value of C3 for your application.

As many readers will know, I like current transformers (CTs).  They can handle very high current, and if properly terminated are not damaged by fault currents that could (literally) melt the power wiring.  Because their output is a current, it has very high impedance, and rectifying the output current only needs four ordinary diodes.  It stands to reason that this is the detector of choice for an AC electronic fuse.  For full compatibility with the DC fuse, I've retained the 5V supply, but it can be increased to 12V with a few resistor changes.

By using a current transformer, rectification is easy, and the circuit still only needs one comparator and a few other parts.  The relay is unchanged, but with AC it can handle higher current without arcing.  Naturally, the relay manufacturer's ratings should not be exceeded.  A 1:1,000 current transformer provides an output of 1mA/ A.  The CT's secondary is effectively open-circuit until the diodes conduct, so their forward voltage has little effect on the rectified output until the AC current is less than ~100mA or so.

Figure 3 - Fail-Safe AC Electronic Fuse

The reference voltage at U1 Pin 3 is 320mV - enough to ensure reliable operation of the comparator.  The output of U1 is normally high, turning on Q1 and energising the relay.  Power is delivered to the protected circuit via the N/O contacts as long as Q1 is turned on.  If the current exceeds the preset value, the comparator's output goes low, turning off Q1 and disconnecting the load.  The collector of Q1 goes high (5V), and this is fed back to U1 Pin 2 via the 'Reset' switch and D5 (1N4148).  This ensures that even if the current falls below the threshold (which it must as the load is disconnected), the output of U1 remains low and the relay cannot re-energise.

With the values given, the lowest current that can be set is around 850mA peak, or 600mA RMS.  The maximum is effectively unlimited if VR1 is set to zero ohms, but that would not be sensible.  Note that the circuit is peak sensing, so it will trip the instant the preset peak current is detected.  Slow-blow operation can be achieved by increasing the value of C2, but I don't recommend anything greater than 1µF.  This will operate with 700mA RMS in about 28ms.  As with the DC version, you will need to experiment and run careful tests to determine the optimum values for your application.

VR1 acts as the burden resistor for the current transformer.  At low current settings, the value is much higher than the recommended value (100Ω), but distortion is unlikely because the current is low.  Even if the CT does distort (due to core saturation), it doesn't matter because its output is still predictable.  If you wanted to detect very low currents, VR1 can be increased to 2k, allowing detection down to about 100mA.  While there's probably no requirement to be able to trip at such a low current, it demonstrates the flexibility of the circuit.  It can be used with low-voltage AC or mains, provided the relay is rated for mains switching.  The CT provides complete isolation of the sensing circuit.

The circuits shown are certainly not the only way to build a reliable DC or AC electronic fuse.  However, they have the advantage that the sensing and switching are isolated from the control circuit, both are fully adjustable, and very flexible.  'Slow-blow' operation is easy to configure, but when an e-fuse is used, most of the time you're after a 'hard' limit.  If sluggish operation is acceptable, it's easier and cheaper to use a standard glass fuse - they aren't accurate, and they are definitely not fast (even 'fast' types), but they do provide protection from serious fault currents, and are used in their millions in all manner of systems.  These range from DIY and hobby projects, through to automotive and industrial applications.  They remain one of the most common 'protection' systems in use.

You will note that I've shown the circuitry for both DC and AC fuses.  With AC, additional circuitry and some reorganisation is necessary to provide full-wave rectification and provide effective detection.  While you can use dual comparators (with one for each polarity), this makes the circuit harder to follow and construct.  If the DC fuse is used as shown with AC, only one polarity is protected (only the positive half-cycles), which may allow excess current if the fault is asymmetrical.  Such faults are common during the inrush period for transformers, and can also be created if one diode in a bridge rectifier fails.  AC protection should always be full-wave.

The circuits have been pared back to the minimum, avoiding any unnecessary complications.  Performance can be improved in any number of ways at the expense of more parts, but in most cases there's no need to make the circuits any better than they are.  Because they are intended to protect equipment, it's important that circuitry should be reliable, so skimping on the relays (for example) is not advised.  It's very important that a 'fail-safe' circuit is used, so most predictable issues (loss of 5V DC supply in particular) are overcome.  Obviously, no electronic circuit can be relied upon to be 100% effective forever, so using a 'traditional' fuse to protect against major faults is highly recommended.

• Electronic Fuses (ESP Website)
• ACS712 (Allegro MicroSystems Datasheet)
• ACS758LCB-050U-PFF-T (Allegro MicroSystems Datasheet)

Main Index Projects Index