ESD Protection

Electrostatic discharge can be a mysterious phenomenon. In this article, Stuart tries to shed some light on how it works, and how to deal with ESD in your circuit designs.

WHAT ESD IS

There are multiple kinds of electrical discharges, such as lightning, arcing from motor brushes, spark plugs in a gasoline engine, and numerous other things. Many types of electrical discharge can cause upset or malfunction in electronic circuits, as described in my last article (“Combating Electrical Noise: Keeping Your Circuits Quiet,” Circuit Cellar 403, February 2024) [1]. But here I want to look specifically at ESD and how you can protect circuits from ESD damage. ESD is a specific type of electromagnetic discharge.

The electrostatic part of ESD occurs when an insulator accumulates a static high-voltage charge. The discharge part of ESD happens when the voltage is high enough to break down the insulating properties of air, and arc over to the nearest conductive path. The breakdown voltage of air is about 30kV/cm, depending on humidity and altitude, so to draw an arc 1” long requires about 75kV. In lightning, the high potential voltage accumulates on the clouds and arcs to the earth. Given the altitude of the clouds, it takes millions of volts to trigger a lightning strike.

WHAT CAUSES ESD

Some years ago, I was on a business trip, and the hotel room had synthetic blankets. I didn’t want to use them, so I peeled the blanket off the sheet, and the separation of the two materials created a line of blue arcs where the two fabrics separated. It was quite impressive in a dark room.

Except for those of you who live in a humid climate, nearly everyone is familiar with walking across a carpet and touching a doorknob or electric switch and getting a momentary shock. You may have seen demonstrations of rubbing a balloon on your hair and using it to bend a thin stream of water from a faucet. Many years ago, my father had some sand pumped into a storage tank situated well above the ground. They pushed the sand through PVC pipe. By the time the sand reached the end of the pipe, there was quite a lightning show between the pipe and the tank supports. I didn’t see it, but arcs several inches long were occurring, and it was apparently quite disconcerting to the guy who was sent into the building to watch for problems.

All of these are examples of triboelectric static, caused by a charge transfer between materials with opposite triboelectric characteristics. Electrons are absorbed by one material, and since the material is an insulator, the electrical charge has nowhere to go. Some materials are more prone to give up electrons than others, and this is illustrated in the triboelectric series; a quick Internet search will find multiple different triboelectric tables, listing different materials from most positive to most negative. Quartz, which is the primary material in most sand, is at about the middle of the “positive” spectrum on the table. PVC pipe is near the bottom of the “negative” end of the table. That explains the dramatic lightning show in the sand-pumping incident. Large quantities of quartz sand moving through a PVC pipe can build up a lot of charge.

ESD CHARACTERISTICS

Figure 1 shows the waveform of a typical ESD event. The rise time of the initial pulse is about a nanosecond, representing a high-frequency signal, about a gigahertz. The current, which can be 8, 10, or 15 amps, drops off as the energy is expended. Not every ESD waveform looks exactly like this, and if the ESD entry point is the end of a cable (such as a USB cable) the impedance and capacitance of the cable may change the waveform by the time it reaches the other end.

You can see why you would prefer not to have multiple amps of high-frequency energy flowing through your 3.3V integrated circuits.

Esd In Embedded Systems

Many embedded systems, including consumer electronics, will experience ESD. The primary cause of ESD is a user building up a charge (say by walking across a carpet or rolling a desk chair around) and then touching the product. I’ve generated a mildly painful shock by sliding across a car seat and then touching the door frame. Points of entry, where the energy discharges, can be the case, a keypad, a connector on the end of a cable, or the battery contact when the user changes the batteries. In general, the ESD energy, like water flowing downhill, will tend to flow to the lowest electrical potential, which will usually be one of the power supply voltages. Usually the ESD energy will find its way to the system ground, although it’s going to get there through the ICs, power supply filtering capacitors, or whatever path is the lowest impedance. Note that the lowest impedance path for ESD is not necessarily the same as the lowest resistance path for DC.

ESD PROTECTION METHODS

Figure 2 shows a transient voltage suppressor (TVS) diode on a signal line. TVS diodes are effectively two fast-response zeners connected cathode-to-cathode in a single package. TVS diodes are available in large wattage and voltage ratings, but they have the drawback of having relatively high capacitance, on the order of 100pF. This is a characteristic of the junction size needed to handle high power, but it’s problematic for high-frequency signals. When an ESD event occurs, the TVS diode will clamp the voltage at the signal line to a safe value.

TVS diodes are typically bidirectional, and function as a Zener for either positive or negative voltages. Unidirectional TVS diodes are also available; they clamp the signal line at the breakdown voltage in one direction and at a fraction of a volt in the other direction.

TVS diodes aren’t magic, and they need a good ground to be effective. In most PCB applications, this means a ground plane must be used, and it must have a low-impedance connection back to the power ground. If the ground impedance is high enough that the ESD current can lift the PCB ground voltage away from the power supply or from other components in the system (remember, we’re talking about amps of current), then the ESD-generated voltage (or at least part of it) will still be applied to the parts in the system. Grounding is essential when using TVS diodes. That means not just DC grounding but a low-impedance path for the high-frequency ESD voltage as well.

In a system with a chassis, like rack-mounted equipment, ESD will preferably be referenced to the chassis, not the PCB ground plane. Ideally, you would reference the TVS to the chassis using a dedicated ground plane. But that may be impractical, so referencing to PCB ground and then having a low-impedance connection to chassis ground may be the best you can do.

ESD DIODES

TVS diodes are available in high-power versions, but to protect the electronics, the clamping voltage must be within the maximum voltage of the device pins. A 36V TVS may clamp the signal line at approximately 36V, but if the IC pins have a maximum rating of 5V, the damage will already be done by the time the TVS diode turns on.

ESD diodes specifically for protecting electronic signals are available. These are smaller TVS diodes with lower capacitance and correspondingly lower power ratings. I randomly selected the ESD341 from the TI family of parts—this clamps at about 6V and has less than 1pF of capacitance. It is in an 0201 SMT package. Depending on the part you are protecting, even the 6V clamping voltage may be too high; you have to select a part that will protect your specific circuit based on the circuit inputs. The datasheet will show the breakdown voltage, holding voltage, and peak pulse power.

POWER

When an ESD pulse is applied to the diode, it will dissipate some power. 4A into the ESD341 will be about 24W in an 0201 package. It’s a brief surge of power, but you will want to pick a part that won’t fail due to excessive power dissipation.

SIGNAL COUPLING

When an ESD event occurs and is clamped by a TVS, there is a significant current flowing into the circuit. It’s a brief surge of current, but if there is an adjacent trace on the PCB, the ESD signal may couple into it. The ESD pulse is high-frequency, and the tendency to couple into an adjacent trace is a function of the frequency, current, and capacitance between the traces. Higher frequency couples better, higher capacitance couples better, and higher current couples better. So, where you have ESD-protected signals, keep unprotected signals as far from the protected trace as practical. Ideally, have the unprotected signal run at a right angle to the protected trace, and even better, put them on opposite sides of a ground plane. All of those things won’t always be possible, but they minimize the coupling between the protected and unprotected signals.

FILTERING

For low-speed signals, you can often filter out ESD with a low-pass filter (Figure 3). The ESD pulse is comprised of high-frequency components. So a low-pass filter can be effective. This has the advantage that it doesn’t require matching a TVS to the maximum IC voltage, and it limits spurious effects in the circuit. In Figure 3, a signal is filtered with a 1KΩ resistor and a 100nF capacitor. This has a roll-off of about 1600Hz, so frequencies above that will be attenuated. This is well below the hundreds of MHz that characterizes the ESD pulse. But is it adequate?

Let’s say that an ESD pulse of 2500V and 60ns duration occurs. To keep it simple, we’ll say that the voltage is constant throughout the pulse, which is close enough for a first approximation. The time constant of the resistor and capacitor is R x C, which in this case is 100µS. So how much will the voltage on C1 rise in the duration of the pulse? The voltage will rise to about 66% of the ESD voltage, or about 1650V, in 100µS. But the pulse is only 60ns long, so the voltage will rise to about 60ns/100µS or 0.0006 x the ESD voltage, about 1.5V. So, this filter would filter out the damaging ESD unless the part it is driving has a low maximum input voltage threshold.

I made some simplifying assumptions here, but this is a reasonable outcome. Note, though, that ESD damage could occur if the ESD pulse is negative and the IC being protected can’t handle inputs below ground. In that case, a negative pulse would build up negative voltage on C1 until something in the IC starts conducting, and then all the current would go through the IC. So you have to look at both positive and negative limits.

All the previous rules apply to this circuit; the current through R1 is passing through C1, so C1 must have a low-impedance connection to ground. If R1 has high capacitance, then this would have to be included in the calculation as a capacitance in parallel with R1. But few resistors have enough capacitance, even at 1GHz, to affect the result when C1 is 100nF. C1 needs to be a capacitor with low impedance leads and high Q, probably a SMT ceramic, due to the high frequencies involved. And this approach will significantly attenuate high frequencies; it is unsuitable for high-speed signals, as it will filter out the desired signal as well as the ESD voltage.

Littelfuse has a series of EMI/ESD protection parts that incorporate both an ESD protection diode and high-frequency filtering into one part (Figure 4). An example is the SP6001 series. Combining the TVS diode with the filtering allows a much lower capacitance on the signal line; the TVS diode limits the voltage to the filter so the roll-off frequency can be much higher since it only has to filter the clamped voltage, as opposed to the unclamped voltage in my example. This allows the use of much smaller resistors and capacitors for the filter portion. The Nexperia PESD3V3S4 contains four 3.3V ESD protection diodes that can withstand up to 20A and 30kV of ESD energy. For parts like this, the datasheet will describe the voltage/current/power limits and capacitance of the parts.

You can purchase connectors with TVS diodes built in, and with TVS diodes and an L/C filter as well. But again, the effectiveness of these connectors is only as good as the discharge path you provide to ground. That current has to go somewhere.

INTERNAL PROTECTION

Many electronic devices, ICs in particular, have internal protection diodes. These usually consist of a diode to V_CC and another diode to ground (Figure 5), limiting the ESD excursion to one diode drop above V_CC and one diode drop below ground. The diodes are small, so their power-handling capacity is limited. And they have another problem: if you drive a diode-protected input with a signal that forward-biases either diode, you may destroy the device if the current is high enough. This is particularly problematic if you are driving an input while the device is not powered on; any voltage that appears at the pins will potentially turn on the diodes and may even attempt to power the device up through the diodes. So you have to be sure not to overdrive the protection diodes in normal operation. The same consideration applies to external protection diodes, but since they are larger, they are less likely to be destroyed this way.

In addition to the importance of a low-impedance ground, you will also need a low-impedance connection between ground and V_CC. Current from a positive ESD event will get passed into the V_CC supply via the internal diodes. If there is insufficient coupling between V_CC and ground, the entire V_CC voltage rail can be raised high enough to exceed the maximum rating and damage the parts. Generally, you will want some low-impedance ceramic capacitors between V_CC and ground, and ideally, you’ll have a V_CC plane on the PCB layer adjacent to the ground plane, with a thin dielectric layer between them. The close proximity of the two planes will provide a significant distributed capacitance between them that will help in coupling the high-frequency ESD energy. An additional ground layer on the other side of the V_CC plane is even better, but that may require the board to be manufactured with an extra layer pair and could be a cost issue.

ANTISTATIC BRUSHES

If your equipment has a moving conveyor belt or any rotating parts that are known to create static electricity, and you can’t use conductive parts for some reason, you may be able to stop the ESD at the source by using antistatic brushes. These are brushes made of metal or other conducting material that are placed near the moving part (like the belt on a conveyor belt system) to absorb the ionization as the part moves. That doesn’t prevent external ESD caused by touching a cable or something, but it can eliminate internally generated ESD from an unintentional Van De Graaff generator. The brush must be grounded. An antistatic brush works by induction (the same principle that causes bits of paper to be attracted to a charged balloon or comb) so it won’t completely remove the charge. But it can significantly reduce it.

ESD UPSET

Any time an ESD event occurs, and even if you have the circuits protected, it is likely to produce a spurious input to your system. An ESD event on an audio cable that you pick up and plug into an amplifier will probably just produce a click in the speaker. But an ESD event when pressing a key on a keypad or an event that toggles an interrupt line is likely to produce an unexpected transition in your MPU. You will want to handle those by recognizing a spurious key transition, by having error-correcting or error-retry capability in the communications interfaces, and by a full system reset in extreme cases. You may want to have a keep-alive timer that does a reset if it isn’t periodically reset by the software. That way, if the software just “goes into the weeds,” it will recover. You have to think through what happens to any ongoing processes and how such a reset may affect other MCUs in the larger system. Maybe the communications interfaces need to have a message that tells the other MCUs, “I’ve had a probable ESD event and I did a reset.”

ESD TESTING

ESD simulators are used to simulate electrostatic discharges. They don’t use triboelectric generation but instead use internal voltage generators to produce the high voltages needed to produce an arc. You can use those to “zap” various parts of the circuit or system to see how susceptible everything is to ESD, and where the susceptible areas are. Some equipment certifications require ESD testing, and testing in your facility before sending the system out to a compliance testing laboratory can save a lot of time and expense. When you do this testing, you may be surprised at where ESD effects occur and what parts of the circuit are affected. ESD doesn’t always behave the way you’d like and the path it finds to ground may not be the path you expected it to take. I’ve seen an ESD event at the corner of a large chassis, far from the electronics, produce problems in the circuit. It’s all a matter of what path the ESD takes to close the circuit.

CONCLUSION

ESD is a real thing, and with the decreasing size and operating voltage of modern ICs, susceptibility to ESD is more of a problem. It’s important to be sure that your design can survive whatever ESD events it may encounter in normal operation. Sometimes ESD looks like black magic. I hope this article has provided some insight into this often mysterious problem. 

RESOURCES
Littelfuse | www.littelfuse.com
Nexperia | www.nexperia.com
STMicroelectronics | www.st.com
Texas Instruments | www.ti.com

REFERENCES
[1] Stuart Ball, “Combating Electrical Noise: Keeping Your Circuits Quiet.” Circuit Cellar 403, February 2024.

PUBLISHED IN CIRCUIT CELLAR MAGAZINE • APRIL 2024 #405 – Get a PDF of the issue

Sponsor this Article