As a cautious analog-centric engineer, I have always liked traditional thermal-based fuses and been somewhat skeptical of electronic fuses (also written as e-fuses or eFuses) due to their active components. After all, when it comes to reliability, simple is almost always better, and there is nothing functionally simpler than a thermal fuse. I don’t mean to denigrate them technically, as they actually incorporate advanced materials science and technology; it’s their functionality that is simple.
Face it: a standard fuse does one thing, does it well, and does it in a fully defined and immutable way within a given set of specifications. You can’t mess them up because they don’t give you any hooks or handles with which to grab them (Figure 1). If you specify the appropriate fuse parameters – current rating, time lag, material, physical size – you really can’t go wrong. I know, everyone has their story of when a fuse didn’t do what it was supposed to do, but those are still the rare outliers; an interesting quirk of English: while a “fuse” opens on overcurrent, “to fuse” means to join two materials.
 |
||
| Figure 1. | These are three commonly-used schematic symbols for the thermal fuse; both the schematic and functionality of this two-terminal passive component are simple, which is major virtue. |
|
However, the thermal fuse has several drawbacks, beginning with the time it takes to react. Depending on the overcurrent value compared to the threshold, it can take anywhere from tens of milliseconds to tens of seconds to react and open the circuit. In today’s lower-voltage designs, the overcurrent is often a modest value, so the fuse reaction time may be too slow to protect sensitive circuitry. Also, the standard fuse must be physically replaced after it opens, which is a disadvantage in many (but not all) applications.
Conceptually, an electronic fuse is a simple circuit that provides an alternative approach to current limiting and cut-off with unique advantages, as it measures current but is not dependent on the thermal heating and the subsequent open circuit of an in-line element. It is built from several analog components: a precision current-sense resistor, an amplifier with accurate scaling resistors for capturing and “gaining-up” the voltage across the resistor, a comparator circuit to “switch” at the preset value, and a MOSFET to allow/break the path of current flow in the line being monitored (Figure 2).
 |
||
| Figure 2. | The basic block diagram of a typical e-fuse shows its apparent simplicity and ease of connection. |
|
The circuit function is fairly straightforward. Typically, the resistor value is chosen so the voltage drop across it will be between 50 to 100 mV at maximum current. The e-fuse is connected between the power rail (or supply source) and the load to be protected. The current to be monitored passes through the resistor, and the resultant voltage across this resistor is sensed and scaled by the current-sense amplifier (CSA).
 |
||
| Figure 3. | An e-fuse with additional functions is programmed via simple external passive components; other e-fuses incorporate addition features and functions. |
|
While it is possible to build an e-fuse from individual components, most users instead opt for a complete IC-based e-fuse, which incorporates the needed circuitry including the FET (Figure 3); some even have an internal sense resistor. Other IC-based e-fuses also include additional functions and features such as user-programmable undervoltage lockout, overvoltage clamp, and auto retry, as well as a start-up time that can be set by means of external components. That last feature is useful to keep in-rush current under control during startup and hot-swap operations, and so they are used extensively in that application.
My initial sense of e-fuses was that they have their role in lower current/voltage circuits, that’s as far as they can go. I wasn’t confident they would be a good fit for higher-range applications that need Underwriters Laboratories (UL) and International Electrotechnical Commission (IEC) certification, where a thermal fuse is a fully accepted means of protection.
That’s why I was interested in a just-published note [1] from Texas Instruments. This note discussed how e-fuses fit into the UL/IEC certification process, and why they are suitable under the right conditions. These certification requirements are so complicated, with so many clauses, mandates, exceptions, and rules that any knowledgeable advice is very welcome, and any component that allows you to skip a step in the process is a good thing.
The brief note explained the ins and outs of e-fuses with respect to these regulations, and it also reminded me of a point that it’s easy to overlook: fuses protect against overcurrent situations and the risk they pose to systems and people. They are not for high-voltage protection, even though we may unconsciously associate “danger” with AC-line voltage of 120/240 V and the large amounts of current AC lines can deliver. The TI note lists approved e-fuses that go up to 4.5 to 60 V at 6 A – clearly not line voltage, but still a substantial amount of current and double-digit voltage. Protection against high-voltage events is not a task for a fuse; instead, that is the role of MOVs, spark gaps, and other components.
The reality is that e-fuses have distinct differences compared to thermal fuses, and these are often beneficial: accurate current limiting, much-faster reaction time, and the ability to self-repair and restore the connection when the fault clears (depending on the e-fuse model chosen). In many design situations it makes sense to give serious consideration to an e-fuse in place of the classic thermal device.
Still, an e-fuse is much more than a simple, extremely reliable, limited-function thermal fuse; it’s an active device, albeit a simple one. Perhaps some designs need an e-fuse plus a thermal fuse for maximum confidence, as a sort of “belt and suspenders” approach.
But would that indicate that the designer was cautious and prudent, or that the design lacked confidence in the failure analysis and subsequent protection scheme? What’s your view on the use of e-fuses in higher-current applications?
