Methods of Circuit Protection

Overvoltage, overcurrent, and heat are the three most likely events that can destroy our expensive silicon-based components or reduce our product’s life expectancy. The effects are often quite instant, but our product might survive several months of chronic overstress before giving up the ghost in some cases. Without adequate protection, our circuit can be vulnerable to damage, so what should we do? Or do we need to do anything?

For instance, if the added cost of a circuit protector is too high, we may choose to live with the consequences. Either way, as a designer, we should take some action and document our decisions in the form of a technical record so that in the future, our choices can be seen and justified. Ideally, we start by listing the threats, those things that might be expected to happen. Here, it’s worthwhile being as thorough as possible, from considering poor handling practices by the customer (including ESD effects) to the effects of indirect lightning surges producing potentials measured in kilovolts at our PCB. All credible threats need to be covered if the complete picture is to be considered.

Threats

• MIL-STD-883 ESD (electrostatic discharge) event
• IEC 61000-4-2 ESD event
• IEC 61000-4-5 indirect lightning surges (AC power lines and signal/data lines)
• AC power surges - how good is the stability of our local supply?
• Customer misuse (a tricky area for sure)
• Over-temperature including unexpected and prolonged climatic conditions

Measures that Don’t Typically Protect PCB Components

It’s worth mentioning those techniques used in our industry that are not particularly useful for our PCB protection components. I’ve heard engineers say that a fuse provides overcurrent circuit protection components to a MOSFET or a diode, but this is naive because fuses can take a significant time to blow with moderate overcurrent and don’t blow quickly enough to protect against fast-rising transients.

Fuses (and Circuit Breakers)

We see them in many places, and we certainly see them mounted on printed circuit board protection, but what do they bring to the party, or what do they protect? The main fuses prevent infrastructure damage; they will avoid wiring fires inside walls, and they will prevent consequential burning in our product should a component (such as an AC transformer or SMPS) fail a short circuit. They are an acceptable fire insurance policy, but they cannot be relied upon for PCB protection components, other than in a few select situations.

Here’s one unique example. Electronic circuits used in hazardous gas or dust environments are usually protected by “Zener barriers,” so named as they employ Zener diodes. The barrier will restrict the voltage, current, and power that feeds into an “intrinsically safe” circuit within the gas or dust zone. However, to protect the Zener diodes from an upstream overpower situation (should a fault have occurred), the barriers rely on a fuse to open-circuit before the Zener diodes are compromised. However, in the main, a fuse cannot be relied upon to protect an electronic component from overcurrent surges. For instance, most integrated circuits (ICs) or transistors will fail due to overcurrent long before the fuse adequately disconnects the supply. The thermal lag of the fuse is just too long compared with most semiconductors. A fuse is several orders of magnitude too slow to protect our silicon from permanent damage.

Circuit breakers also cannot be relied upon for protecting our PCB components. Like fuses, they take too long to react, and the damage can be done in tiny fractions of a millisecond or even less.

RCD, RCCB, GFCI or ELCB

These devices measure (or infer) the presence of an undesired ground/earth current and are used to prevent the risk of an electric shock to humans and livestock that may touch the device. They do not necessarily offer any circuit protector to our PCB or its components.

Measures that Do Protect PCB Components

These devices and techniques are considered suitable for protecting our PCB’s electronic components from overvoltage and overcurrent protection circuit events.

Zener Diodes and TVS Diodes

Zener diodes have been used as a circuit protective measure for many years. However, for handling high surge pulse energies (such as from an ESD event or an indirect-lightning event), TVS diodes are the preferred solution because of their larger PN junction surface area. This makes them capable of safely conducting much higher levels of current. But this comes with a drawback; TVS diodes usually have a much higher terminal capacitance compared to a Zener diode and will, in some applications, cause significant degradation of high-frequency signals that wouldn’t be seen with a Zener diode. Fortunately, there are many to choose from, and there are some types available that are designed for high-frequency circuit protection components but at the cost of a lower surge handling capability.

The main feature of a Zener diode is that it can regulate voltage more accurately than a TVS, and that’s why they are used in voltage regulator circuits. This is precisely the application that Zener diodes are designed for, and it is this type of application that dictates how they are made and how they are tested before being shipped. Hence, Zener diodes are not surge tested before shipping. TVS diodes, on the other hand, are generally chosen to be non-conducting (held in reserve in our circuit), whereas Zener diodes are presumed to be conducting and regulating. For this reason, TVS diodes are surge tested before being shipped.

Both can be effective in overvoltage applications, but a TVS (for the same package size as a Zener diode) can handle much higher peak surge energies. In other words, the TVS diode is specifically designed for ESD and surge events. And finally, a TVS can be generally relied upon to fail a short circuit protection circuit; thus, in death, it still protects our valuable circuit (even if a fuse blows) whereas, the Zener diode cannot be relied upon to fail in this way; it might fail open-circuit, and thus, our valuable circuit is no-longer protected against the next “event.”

Metal Oxide Varistors (MOVs)

Like TVS diodes, they are used to restrict the extreme voltages caused by high energy surges. However, the MOV has a particular usage area that a TVS diode cannot currently fulfill, AC power lines. TVS diodes can be found with stand-off voltages up to about 500 volts peak. However, their low current handling capacity (less than 1 amp at this voltage level) makes them unsuitable for providing indirect lightning surge protection devices on AC power lines.

Dramatically diode, they can respond quickly, typically within a few nanoseconds for most MOVs, which is more than adequate for AC line surges that have rise times measured in several microseconds. However, one of the main disadvantages is that a MOV will degrade a little each time it handles a surge. The more powerful the surge it receives, the more it will lessen until eventually it will suffer a thermal runaway (due to an increased leakage resistance) and short circuit protection circuit the AC supply or starts to burn. That is why, when using a MOV in a power AC application, a fuse must also be used in circuit protection.

Crowbar Protection - Gas Discharge Tubes (GDTs)

GDTs are helpful in some applications because they can handle massive surge currents. They achieve this because once they have initially arced, they “crowbar” the voltage down to a low level. Therefore, they turn from a device with a high stand-off voltage to a device that clamps at a much lower voltage. Watt for watt means they can handle higher current surges. The main downside is that unless there is a mechanism for extinguishing the low-voltage arc (such as the AC voltage passing cyclically through zero volts), they will remain permanently on. This is why they are not favored for DC voltage systems because there is no natural extinguishing mechanism. They are used in large numbers on telecom lines because the current of a regular phone line is insufficient for arc-sustaining once the surge has been dealt with.

Another disadvantage is that they are relatively slow compared to MOVs and TVS diodes. For that reason, they are sometimes used in conjunction with a MOV; the MOV handles the initial fast-edged surge, and after a microsecond or so, the GDT takes over to safely manage the bulk of the surge energy. GDTs can also degrade with repeated usage and fail short-circuit; hence, they must be fused when applied to AC power applications.

Other Types of Crowbar Protection

A standard and straightforward method is the use of a thyristor (or Triac); once a surge voltage is “detected,” the thyristor activates and clamps the voltage supply to a low level (in a very similar way to a GDT). Usually, a fuse is used that open-circuits the line. Still, there are more sophisticated designs available than use current limiting circuits (and timers) so that the circuit returns to “normal operation” once the surge has “gone away.” This latter type is known as a self-resetting (or active) crowbar circuit. If we have an expensive and vulnerable electronic circuit fed from a relatively inexpensive local voltage regulator (that could fail to short-circuit), a crowbar can provide reliable secondary circuit protection. However, they don’t continuously operate quickly - they can take up to several microseconds before the protection is adequate. It is normal to choose to incorporate a more conventional device such as a TVS diode (or even a bulk capacitor that can suffice in many cases) to cover this short initial period.

There are also TVS thyristors (available from the usual vendors) with trade names such as Trisil, Sidactor, and Thyzorb. These are mainly found in high-speed data applications protecting ethernet devices from differential surges. They have low capacitance and are rated to withstand moderate pulse energy levels. However, because they are a type of crowbar device, they can outperform (size for size) small TVS diodes for mild surge currents. They are usually bidirectional devices, meaning they operate identically on positive and negative polarity surges. In this respect, they are pretty similar to a DIAC (a diode for alternating current used in some thyristor crowbar circuits).

Ground Isolation Techniques

All disruptive surge events are modeled as voltage sources referenced to ground/earth, and if the PCB is also grounded/earthed, then the surge energies that need to be handled to prevent failure of the circuit can be high. So, if we can galvanically isolate our circuit’s interface with the outside world, then we Dramatically reduce the surge energy intensities that need to be managed.

Think of ethernet magnetics and what they bring to the party. In some cases, they provide galvanic isolation between a PCB and its external wiring up to around 6 kV. It follows that the prospective surge energy is significantly reduced; there is no conduction path for the surge current to flow to earth. Yes, the transformer has to be rated for the surge voltage. Most ethernet products use a transformer for data integrity reasons; adding a few additional insulation layers isn’t a showstopper.

However, there are some things we must get right to achieve significant surge withstand capabilities. We need to maintain clearance distances on our PCB so that kV level surges do not “cross” over the isolation. If necessary (in extreme cases), incorporate PCB cut-outs but, whatever we do, we should do it in a balanced way (not just for data integrity but for surges also).

Balancing techniques must be applied to the isolated input/output signal traces. For example, an isolated RS485 interface IC might still be vulnerable if its line IO terminals are not balanced. The surge protection level we are trying to meet might be 6 kV (just as an example), but if we don’t impedance balance our input circuitry and traces (to ground) with care, a usual common-mode surge of 6 kV can produce a localized differential voltage transient of tens of volts and destroy our IC.

Using common-mode capacitors to the ground can introduce significant problems; these capacitors’ values won’t be precisely the same and will naturally introduce an imbalance that results in a damaging differential surge and defeats all our good intentions. If we have an isolated input circuit, we should do our best to keep it this way for all possible situations.

Opto-isolation is also another proven and well-trodden technique available.

Let’s Not Forget the Obvious Things

Capacitors

We use them all the time, but they are excellent for surge protection, especially from ESD events. Consider that a particular “human body” (HB) ESD event is equivalent to a capacitor of 100 pF charged to (say) 2,000 volts, which is then discharged onto our valuable PCB via a 1.5 kΩ resistor. Forget about the resistor and ask yourself what value capacitor can be added to my sensitive input to reduce the voltage down to (say) 20 volts. The “charge” on the 100 pF capacitor is capacitance multiplied by voltage, which equals 200 nC. Now apply this charge to a 10 nF capacitor, and we find that the voltage becomes 20 volts. In other words, capacitors are excellent at dealing with ESD events. So, if our circuit can tolerate a 10 nF capacitor on its input, do yourself a favor and pick this obvious and inexpensive overvoltage protection circuit scheme. Simulation tools are our friends in this situation.

Diodes are tremendous at diverting surge events to our power rails. This is a widespread method for protecting a sensitive input from shifting energy from a motor being driven by an H-bridge. Though, to be effective, we have to provide local decoupling capacitors so that the energy transfer doesn’t pass down long inductive tracks and cause more significant problems as a consequence. In other words, something has to receive energy without generating an excessive voltage, and decoupling capacitors are the solution. However, this is not rocket science and is easy to implement.

When a coil is deactivated, relay coils need flyback diodes to “catch” the coil’s stored inductive energy. Virtually any diode is suitable because it only handles the coil current for a few tens of milliseconds. However, if the relay (magnetic device or solenoid) needs to turn off quickly, we can place a resistor in series with the diode to burn off the stored energy more quickly. If the coil current is 20 mA and we have a 1 kΩ resistor in series with the flyback diode, the peak voltage seen at the coil driver transistor will be Vcc (say 12 volts) plus 20 volts across the resistor. You’ll need to choose a transistor that can withstand this voltage with some comfort; maybe consider at least 45 volts in this example.

Resistors

I’ve seen some designers trying to shoe-horn a TVS diode into their circuit to protect a vulnerable input from an ESD event, but despite their calculations, they just couldn’t make the numbers work. “I just can’t find a TVS diode with low enough voltage to protect my CMOS input,” I hear them saying, and of course, this won’t quickly work without unnecessarily affecting the CMOS high signal level. So, we use a series current limiting resistor after the TVS diode. The series resistor limits the current into the CMOS input to maybe 1 mA (or whatever value is stated in the datasheet), and we choose a TVS diode that clamps at (say) no more than 10 volts. This means that if the CMOS supply is 3.3 volts, it will drop 6.7 volts across the resistor (at 1 mA) and protect the input. That works out as a 6.8 kΩ resistor, which will be inexpensive.

Basic RC or LC filters

We use these a lot to shape signals, but they can also remove a lot of the higher-frequency energy from a surge event. Low-pass RC filters will naturally do this without too much trouble. The resulting peak output voltage (and surge current protection circuit) can be significantly smaller, making the choice of a TVS diode easier. The series resistor will still be subject to significant power, but it’s easy to simulate. The surge waveforms can be approximated by straight lines, and virtually all commonly available simulation tools offer the ability to measure the power and energy consumed by any component in the protection circuit.

Low-pass LC filters need a little more care, but using a simulator tool pays dividends again. The high-frequency attenuation of a low-pass LC filter is twice as good as an RC filter with the same cut-off frequency. However, the excess energy isn’t burned off in a resistor. Sometimes, the resulting peak output voltage waveform is just as devastating to an underpowered TVS diode as the original surge. So, use a simulator and be prepared to use a resistor in parallel with the inductor (to reduce the Q-factor) or a resistor in series with the inductor for the same reason.

Many data transmission systems have no spectral content below a few MHz, and using a high-pass filter (or a band-pass filter) can also prove to be an effective surge energy reduction technique.

Filter techniques (as mentioned above) apply to individual IO lines when the circuit is not isolated from the ground, i.e., the filter output capacitor shunts some residual energy to the ground/earth. Trying to use filters of this type with isolated IO circuits can cause differential surge voltages to appear at vulnerable IC pins and, quite often, can make the problem worse. Refer to the section on ground isolation techniques for an explanation as to why this is.

Common-mode chokes are also subject to the same small print. They can be great for noise reduction, but they can’t work effectively without grounding/earthing components such as capacitors. The capacitors can then create unwanted differential surges, as previously mentioned.

Overcurrent Protection

When we think of this, we might consider the consequences of an output signal shorted out. A typical example is in an audio amplifier; many amplifiers use extra transistors to detect output overcurrent protection circuits and restrict the levels driven to the loudspeaker, but how would we implement circuit protection if our circuit has many such output signals? We are in the realm of protecting against customer misuse, but there’s not always a clear solution.

For analog outputs such as op-amps (operational amplifiers), we can add a series resistor (within the negative feedback loop) that limits the output current protection circuit. However, we might consider choosing devices that can survive if the outputs are shorted for digital circuits that need to drive differential lines at a relatively low impedance (circa 50 Ω). We might also need to choose devices (and employ techniques) that can withstand indirect lightning surges.

For data and signaling lines, IEC 61000-4-5 (surge immunity tests) specifies that instead of a 2 Ω signal source (as used for AC power lines), data line surges are applied via a 42 Ω resistor; thus, a 2 kV surge can generate a prospective short-circuit current of 47.6 amps (as opposed to 1,000 amps for AC power lines). Although this current still sounds high to the uninitiated, many TVS diodes of small size and low capacitance can handle such levels.

At the risk of sounding a bit repetitive, simulation tools can help out a lot.

Thermal Protection

Electrolytic capacitors will degrade more rapidly than any other component of our power supply that gets too hot over a sustained time. Any power module (such as an SMPS) will likely be the hottest part of our protection circuit, so it’s worth considering what we can do to alleviate thermal effects.

We could choose to disconnect our power supply from the incoming power. This would need a particular circuit and possibly a relay or solid-state relay (SSR), or we could simply use a thermal fuse. A thermal fuse can be chosen that breaks if the local temperature rises above a certain level for a certain length of time. Once broken, some thermal fuse types would need replacing, but others operate using the bimetallic strip principle and will reset once sufficiently cooled. The self-resetting type is usually intended for low-voltage DC power applications.

AC power-compatible non-resettable fuses can be obtained relatively easily and inexpensively. However, we have to weigh up what might be more critical to our customer; should our product protect itself from a sustained high over-temperature and be serviceable for decades, or should our product battle on despite the high temperature to keep working but eventually fail a year or two down the line? A tricky choice.

Have more questions? Call an expert at Altium and discover how we can help you with your next PCB design.