Grounding is the most basic characteristic of all types of electrical equipment. In Compliance magazine and other publications there are a large number of high-quality articles on specific topics, mainly about printed circuit board (PCB) level grounding. This article focuses on an inaccessible path, the grounding at the system level, that is, the grounding of the equipment actually used in the factory.
There are several key aspects to grounding, including safety, ESD, EMI, and signal integrity. Although this and other magazines have published detailed articles on one or more of these topics, this article combines them to help equipment users and tool manufacturers understand what is important and how to achieve optimal ground performance. This article does not deal with PCB grounding (there are many excellent articles on this topic) and double insulated portable tools without grounding.
Safety always comes first. Too many experts in electrostatic discharge (ESD) and electromagnetic interference (EMI) have not received professional training in electrical safety. This article is far from a comprehensive safety guide, nor does it cover all important safety points. The entire purpose of this section is to make factory ESD and EMI experts and tool designers pay attention to electrical safety, otherwise they may not know that grounding is a safety item. I strongly recommend that those dealing with such issues take electrical safety courses, make friends with licensed electricians in the factory, or join the factory safety committee. In this article, we will only touch the surface and touch the basics.
So why is grounding a safety element? For example, let us consider a typical industrial device, such as an integrated circuit (IC) processor or a surface mount technology (SMT) pick and place machine (or any other tool you are familiar with). Each of these tools draws power from AC power, which means that usually anywhere from 100VAC to 440VAC will enter the equipment. If the live wire in such a machine or tool becomes loose for any reason, it may touch and energize metal parts accessible to the operator (ie, supply power to it). Now this metal part, such as the housing, is under high pressure. As long as the operator touches such a part, it is easy to get an electric shock.
This is where grounding can solve the problem. If all metal parts accessible to the operator are properly grounded, the live loose wires that touch the part will effectively ground any live voltage, and the resulting excessive current will trigger the circuit breaker to cut off the power to the tool. For all of these to work, the following conditions must be met:
All conductors accessible to the operator must be grounded1; and
How should the conductivity of the ground path trigger the circuit breaker? There are several different standards and guidelines on this topic, but the basic answer is that the ground path should be at least as conductive as the live or neutral path. If your power cord uses AWG12 (or 2 mm diameter) power cord, you cannot use a ground wire thinner than this. The ubiquitous AWG18 green wire will not work.
Must all ground wires in the tool be as thick as the power cord entering the tool? unnecessary. When grounding for non-safety purposes (such as ESD/EMI) and where there are no voltage conductors, you can choose the grounding wire according to other standards (see further content in this article).
Usually, the ground wire and the neutral wire are reversed in the facility wiring or the internal wiring of the equipment itself. This causes the return current to flow through the ground instead of the neutral wire, which can cause many functional problems in addition to safety issues. The ubiquitous three-LED socket checker cannot detect this. The easiest way to check it is to use a simple AC current clamp to measure the AC current entering the ground wire of the device (make sure the ground wire is correctly identified). If the ground current exceeds 0.1 A during the operation of the equipment, it should be investigated. Even if the wiring is correct, this does not explain the excessive leakage current in the device.
After safety, the second most common use of equipment grounding is to solve ESD problems, and more specifically, to provide a grounded discharge path for conductors and static dissipative materials. If the static charges accumulated on the electrical "floating" conductors and dissipative materials are not released to ground potential, they may carry unwanted voltages and cause problems with ESD sensitive devices.
How can we effectively ground these objects? Standards such as ANSI/ESD S6.1 [1] and the "comprehensive" standard ANSI/ESD S20.20 [2] provide good recommendations. Here, we will add some useful narratives.
What makes me wonder is that engineers and technicians who deal with grounding problems do not ask the most important and logical question of grounding, which is what is the ground voltage? Not a resistance, because resistance is just a means to reduce the voltage of the grounding component. The whole purpose of ESD grounding is to create an equipotential environment.
Currently, neither the ESD Association nor the IEC has issued any consistent standards, standard practices, or technical reports to cover this topic and provide any details about verification. However, this is the most important issue of equipment safety in the process. The only document that solves this problem is the SEMI standard SEMI E.176 [3], which I will introduce later in this article.
How do we make sure what really needs to be grounded? There are implicit and explicit methods of providing ground connections. The implicit method consists of mechanically fastening the conductive parts of the tool to the ground frame so that there is no obvious ground wire, but the electrical connection by mechanical fastening is still present and sufficient. The problem with such implicit connections is that they are uncontrolled. Depending on the construction of the tool, any component in the electrical connection chain can be changed in the next revision of the tool or during repair or service to the extent that electrical connection is no longer ensured. In any modification, maintenance or repair process, metal washers can be replaced with nylon washers, otherwise the original bare metal parts may be anodized and so on.
There are two ways to prevent such problems. One method is to add (and strictly follow) requirements for adequate ground connections in tool specifications, maintenance procedures, and verification documents. Another method is to use a clear, separate grounding method. Either of these two methods is feasible, and the choice is up to the device user, because the manufacturer may not understand the importance of proper grounding to ESD.
An example of explicit grounding is shown in Figure 1. I will return to this diagram later in this article.
Figure 1: "Explicit" grounding in the IC processor
Various ESD-related standards, such as ANSI 6.1, ANSI/ESD S20.20, ESD S10.1 [4], IEC 61340 [5] and some other documents, as well as proprietary plant-wide documents, provide grounding guidance . This section is just an attempt to clarify some details.
For "explicit" grounding and grounding of floating metal parts, these documents specify (or recommend) a ground resistance path of less than 1 ohm. Although it is easy to achieve this goal with fixed equipment, it may be very difficult to achieve and not feasible for some moving parts.
If the part moves a little bit (or even just a few centimeters, which is common in many tools), a flexible steel cable (very similar to a bicycle brake cable, see Figure 2) is usually used to complete the grounding. The material, the bending radius, and the number of bending cycles of this type of cable need to be carefully selected to avoid the cable from breaking during use. Obviously, the conductor of steel is not as good as copper, but it is much more durable. Also, because the cable lines are short, resistivity is not a real problem.
Figure 2: Use flexible steel cables to ground moving parts
Longer movements require longer ultra-flexible cables protected by flexible conduits, as shown in Figure 3. The internal structure of this flexible cable cannot support a sufficiently thick wire. Therefore, many super-flexible cables contain an extra layer of PTFE or similar material around each wire to reduce the coefficient of friction and allow the wires to slide against each other when they are bent.
Figure 3: Flexible cable on the robotic arm
As shown in Figure 2, this happens with any super-flexible cable with or without external wiring harness. The result is that this type of wire has a higher resistivity, and considering all interconnections, it is almost impossible to achieve the 1 ohm requirement for the entire connection along the chain. The requirement for the total resistance of the flexible ground connection usually varies between 2 and 10 ohms, depending on the factory, although I have also seen a requirement of 20 ohms. Will this increase of more than 1 ohm significantly change the ESD environment in the process? In practice, this is unlikely, but the cause of the problem is the loss of the ground connection.
The reliability problem with using dedicated conductors for explicit grounding is that the failure of the grounding connection may not be immediately obvious. After all, this grounding or lack of grounding does not change the basic function of the tool and may be overlooked for some time. Unfortunately, I have witnessed a large number of cases where the "explicit" ground wire is disconnected due to tool maintenance, but not reconnected, but the wires are completely removed or their ends "dangling", making the tool look a bit Like a hedgehog. These problems usually arise when there is a need to solve "bursting" ESD or EMI problems.
One of the ways to solve the problem of land loss is ground monitoring. There are many ground monitoring devices on the market. This monitor is independently connected to the grounding point and the reference ground, and sends an alarm when the ground connection fails.
The wristband and/or wristband cord contains a 1 MOhm resistor and ground wire for a simple reason, namely to prevent electric shock. If an operator wearing a wrist strap accidentally touches the grounding conductor, the current passing through the operator should not exceed 0.5 mA (ANSI/ESD S1.1 Appendix B [7]), which is consistent with several broader safety standards. At 250V RMS, this is the highest RMS AC voltage among ordinary power outlets, and the minimum resistance should not be less than 500 kOhms (regardless of the resistance of the operator's body). A 1M resistor can meet this requirement, including dual wrist straps, and two parallel resistors between the operator's body and the ground. Unless verified, try to avoid using low-cost wrist straps and cords.
Should the same 1 MOhm resistor be used to ground other objects (such as metal objects or dissipative materials)? The common reason for using 1 MOhm resistors in this type of application is "slowing down the discharge rate". Will it really "slow down" the discharge?
Let us consider an electrically floating metal object that needs to be grounded. The object will have a capacitance that depends on its size (among other things). Assuming that the object is at ground potential, whether the object is grounded through a sub-ohm resistor, grounded through a 1 M resistor, or remains electrically floating, will the discharge characteristics be very different?
Figure 4 shows a highly simplified equivalent electrical schematic of this connection (parasitic inductance and capacitance have been omitted for clarity). The device (IC) has a certain capacitance C1 and is charged to the voltage V1, which may be the result of being lifted from the tray. The arm of the IC processor is about to place this device on the shuttle (the metal tray used to move the IC in the processor). When the IC is in contact with the shuttle, the voltage is almost immediately equalized.
For practice purposes, we will assume that the shuttle is implicitly grounded through the resistor Rg instead of mechanically. Finally, any charge left on the space shuttle will be dissipated to the ground through Rg. But the problem we are trying to solve is the role of Rg in the characteristics of the discharge itself.
The resistance Rc of the contact between the IC and the shuttle is negligible and may be only a few milliohms. If we set Rg to 1 MOhm, most of the action will occur between the IC and the shuttle, because Rg is too large to participate in voltage equalization during a short nanosecond discharge period. If we take this situation to the extreme and assume that Rg has infinite resistance, will this "slow down" the discharge? Of course not, because the discharge waveform is only defined by the capacitance and contact resistance Rc of the metal part. ESD practitioners know that touching the floating board of the CPM is easy to produce discharge, just like touching a fully insulated metal door handle. The only function of Rg is to finally dissipate any small amount of charge shared by the IC and the shuttle ground and bring the shuttle voltage to ground potential.
The same applies to static dissipative pads. Inserting a 1 MOhm resistor into the ground connection will not change the rise time or amplitude of the discharge. On the contrary, it will only slow down the dissipation of charge to the ground. In the case of static dissipative materials, this may put these materials under voltage during a fast-paced process. Although existing practices allow the use of 1 MOhm resistors in ground circuits with dissipative materials, in reality it will be counterproductive.
We finally arrived at the most interesting part of grounding, the high-frequency voltage of grounding, or EMI. In this case, the term may not meet the needs of purists, but since it is widely used in the industry, we will also use it.
Every electrical device generates some kind of parasitic signals, such as unplanned or unwanted signals. Automation equipment contains a large number of high-frequency voltage and current signal sources [8], the strongest of which is generated by pulse-driven motors (servo, stepper and VFD-variable frequency drives) [9] and switch-mode power supplies, including those in LED lighting Those ones. These high-frequency signals "leak" to the ground through parasitic capacitances, causing very undesirable voltages between different grounded parts of the device. This is never good news, but it is especially bad news for sensitive equipment and test and measurement.
Why do we focus on high frequency voltage and not any other voltage? Simply put, traditional grounding methods can handle DC and low-frequency AC voltages well. Considering their low ground path resistance (see previous discussion), they will sink any leakage AC and static DC voltages that happen to appear on the metal and static dissipative parts of the device to the ground. Due to the parasitic inductance and capacitance of the conductor and their mutual influence, only high-frequency voltage signals are left. Although DC and low frequency ground resistance paths may be very low, this is not the case for high frequency signals, and we will analyze them in detail.
A simple straight line that is great for ESD and safety grounding is actually an inductor. Although calculating this inductance may be a bit complicated, there are many useful Java-based inductance calculators on the Internet, which are more practical than manual calculations [10].
As a reference point, a 1 m long 1mm diameter wire (AWG18) has an inductance of 1.5 µH. At 1MHz, this will present an impedance of 9.42 ohms. This only applies to straight lines, the typical service loop for ground wires only increases impedance. There are also calculators [11]. For example, winding five turns of the same wire in a 6-inch (15 cm) diameter coil will produce an inductance of 6.1 μH, and an impedance of 38 ohms at 1 MHz. The resistance of the same wire at DC is only 0.06 ohms.
At high frequencies, the current is "pushed out" by the magnetic field generated by the current, the so-called skin effect. The higher the frequency, the thinner the conductive layer. At 1 MHz, the outer conductive layer is only 66µm thick. The skin effect does not increase the resistance like pure inductance (1m AWG18 wire constitutes 0.09 ohm, if there is no skin effect, it is 0.021 ohm), but they will all add up. Stranded wires are helpful because the larger the surface of the wire, the lower the resistance. But often the number of strands of wire found in the manufacturing environment is too few to function.
Two wires running in the same conduit affect each other through capacitive and inductive coupling. In Figure 2, there are drive signals between the wires in the flexible channel, which are connected to the servo motor on the robotic arm, and there is a wire that grounds the robotic arm itself, all of which are very close. A typical robotic arm of an automated equipment has three servo motors, one for each degree of freedom. This is equivalent to nine wires carrying a pulse voltage, usually with a peak voltage of 200V (not including ringing and other artifacts). The rise and fall times of this drive pulse are less than 50 nS, and the resulting signal spectrum is extended to 20 MHz.
In the example in Figure 2, the length of the wires in the flexible wiring harness is 3m. The capacitance between two adjacent wires is about 63pF [12], which forms an impedance of 125 ohms at 20MHz. The rough equivalent schematic is similar to the schematic in Figure 5a.
Due to the characteristics of capacitive coupling, the higher the frequency, the higher the induced voltage. Correspondingly, the sharper the pulse edge, the higher the induced voltage.
Figure 5a: Induces high-frequency voltage into the cable assembly in the flexible conduit of Figure 2
Figure 5b: Ground induced current
The long wires connected in parallel form a distributed transformer. Without iron core and winding turns, it can only work at higher frequencies, which is the problem. Figure 5b shows how a current in a wire can apply a corresponding current to a nearby wire. Due to the characteristics of this parasitic transformer, only high-frequency signals are transmitted from one wire to another, resulting in a waveform similar to that shown in Figure 5a.
People can easily immerse themselves in the simulation and calculation of induced voltage and current. However, in our example, it is unlikely to produce realistic results due to the number of variables not considered in the equivalent schematic and the variability of parameters between tools. But the measurement has a more practical purpose. The measurement methods and techniques are described in detail in this article [13], which was previously published on In Compliance.
Figure 6 shows the typical voltage between the nozzle of the robot arm and the chassis in the IC processor. The spikes correspond to the rising and falling edges of the interference signal.
Figure 6: Voltage between the nozzle of the robot arm and the chassis in the IC processor
Figure 7 depicts the current between the robot arm and the chassis in different tools. The current is measured with a Tektronix CT1 probe, with a ratio of 5mV/mA and a peak current of 76.8mA. Ringing is just a product of unbalanced impedance matching, and manufacturing equipment is far from a perfectly matched radio frequency instrument.
Figure 7: The current between the robotic arm and the chassis
What could be the problem with the small voltage between different grounded parts? In many tools and processes, this is not a problem. If your equipment is not sensitive to electrical overload (EOS) and you don't care about data integrity and measurement accuracy, there is no need to worry. However, since you are reading this article, you must be interested in keeping the ground voltage and current as low as possible.
The grounded surface should provide a "safe space" for sensitive components without any possibility of overvoltage exposure. But if we actually do the measurement, the situation may be very different, and it is usually "unsafe".
For example, consider the common handling of ICs in an IC processor or SMT placement machine (Figure 8). Compared with the chassis we described above, the actuator/nozzle at the end of the robotic arm has sufficient high-frequency voltage. The silicon chip of the IC is capacitively coupled to the nozzle next to it. At high frequencies, this capacitive coupling exhibits very low impedance. When this IC is placed on a test socket or shuttle (a metal bracket used to move the IC on a horizontal surface), excessive current may flow through the device, weakening its structure and causing field failures, or even complete failure.
Figure 8: EOS' mechanism in automatically processing IC
This is just an example. Any metal contact with the device, such as welding [14], wire bonding [15] or other methods, will expose the device to unwanted voltage and current.
There are many documents on controlling the resistance/impedance of ground connections. However, the SEMI standard E.176 "Guidelines for Evaluating and Minimizing Electromagnetic Interference (EMI) in the Semiconductor Manufacturing Environment" is the only relevant industry document, which actually specifies the maximum allowable ground EMI voltage and current according to the characteristics of the equipment used . process.
Although SEMI E.176 is mainly written for semiconductor manufacturing, it has a direct impact on all semiconductor applications, including most of today's equipment. After all, once the semiconductor device is shipped to the PCB assembly plant, its sensitivity will not change. I have published several articles in the previous issues of In Compliance [16] [17], discussing SEMI E.176 in detail.
As a reference point, today’s ordinary ICs with a 10nm geometry should not withstand voltages higher than 0.1V in the unpowered state (that is, in IC manufacturing and processing, such as PCB and product assembly), and the peak grounding of this geometric structure The current should not exceed 10mA (level 3 in SEMI E.176).
Unless you can measure and quantify ground voltage and current, you cannot control it. Another article I previously published on In Compliance [13] provides detailed guidance on the methods, instruments, and techniques of this type of measurement, and I encourage you to read it before performing any measurement.
High-frequency signals can interfere with data and measurements in many ways. The induced EMI voltage can present itself as an effective signal because it may be close to the real signal in both amplitude and waveform. This can lead to data corruption [18] and measurement errors [19], [6]
Electrical engineers are familiar with the "ground bounce" effect in semiconductors (for example, see [20]). Ground bombs are generally considered to occur at the IC level, but the physical principles of ground bombs also apply to the system level. Figures 9 and 10 show examples of how it happens.
Figure 9: System-level ground bomb
Figure 10: Ground bombs cause "extra pulses"
Figure 9 shows how current spikes from sources such as motor operation are transmitted to the facility ground, creating a voltage drop on the tool's ground wire. The voltage generated on the tool ground is no longer the same as the facility ground, and is different from the ground of another tool (USB in this case) that the tool is trying to communicate with. In this case, the logic level is no longer valid. As shown in Figure 10, the next logic gate can easily mistake "1" for "0", and vice versa, depending on the timing and amplitude of such interference. Worst of all, there is no record of such events in the system, and they often cannot be reproduced.
Simply understanding the problem is only the first step in solving the problem. There are many ways to alleviate ground EMI problems. All of this revolves around the same three basic principles:
Depending on whether you are a device designer or a device user, your choice may be different.
The two biggest sources of EMI in equipment are pulse width modulation (PWM) motors (such as servos, steppers, and VFD) and switch-mode power supplies. If we manage to reduce the dV/dt at the edge of its pulse (in other words, "slow down" the signal transition), the ground-induced EMI will be reduced. Designers of PWM drivers and SMPS are trying to make these edges as sharp as possible so that the output transistor driver does not overheat and the circuit is simpler. The typical rise/fall time of the drive pulse in a servo motor is about 50nS, which translates into a frequency spectrum up to 20MHz.
Our job now is to make these drives and SMPS work for us the way we want. The only practical way to increase the rise and fall times of pulse edges is filtering. For SMPS, the more filtering applied to its DC output, the better. The PWM driver requires a more cautious approach, because trying to filter the pulse drive signal can easily make the motor perform poorly or not work at all.
Figure 11 shows the original rising edge of the servo motor drive pulse and the modified edge after applying the servo motor filter. Figures 12a and 12b show the result of this edge modification, the ground current is reduced by about 50 times.
Figure 11: Use SF20101 motor filter to modify the rise time
Figure 12a: Ground current without filter
Figure 12b: Ground current of the filter
In order to reduce the EMI of switch-mode power supplies, DC filters are often used, such as the filters shown in Figure 13, because they can remove high-frequency components in the DC power supply.
EMI filtration is like filtering sewage, blocking pollutants and allowing clean water to pass through. Our readers may already be familiar with the concept of filtering EMI on wires and cables, even if they have never considered filters. The ubiquitous ferrite clamp (usually a black block on a computer cable) is actually an EMI filter for the cable. From a technical point of view, a ferrite clamp is a current transformer with a short-circuited secondary that converts the high-frequency signal in the cable into heat (no, you will not be able to check it by touching it-the energy is too low And can't notice this road). And the ferrite clamp is cheap and easy to implement.
The problem is that their performance is limited. Most ferrite clamps are only effective at the higher end of the spectrum, that is, above 50MHz (a lot of EMI energy in manufacturing is less than 1MHz), and the attenuation they provide is mainly limited to 10dB at these frequencies. Ferrite clamping is usually the first method of controlling EMI propagation. But using ferrite clamps is no different from using band-aids. It will prevent minor bleeding and cover a small scratch, but it is not enough in cases involving more serious injuries.
Grounded EMI filters, such as the one shown in Figure 14, provide better performance by greatly attenuating broadband signals while also providing low impedance to the power supply frequency (don’t forget that grounding is a safety element). One of the applications of grounding filters is shown in Figure 15. It solves the EOS exposure problem caused by EMI, as shown in Figure 8. The modification is simple, including an insulating board made of mechanical hard material, such as FR4, Bakelite or equivalent material, clamped on the robot arm and the end piece is grounded through the filter in Figure 14. (For a detailed description of implementing this type of filtering in an IC processor in production, see [24]).
Figure 14: Equipment grounding EMI filter[25]
Figure 15: The grounding filter on the robot arm prevents EMI on the nozzle
Figures 16a and 16b show the ground current between the robot arm and the corresponding chassis, without and with filters. This kind of grounding filter inserted into the wire is used for the ESD grounding inside the device, which will prevent the spread of EMI throughout the tool, while complying with all relevant ESD and safety standards. Similar methods with similar results can also be used on the facility ground, especially in facilities that use separate grounding. In this case, inserting a ground filter every few meters can prevent EMI from spreading from noisy tools to tools that require a low-noise environment.
Figure 16a: Ground current without filter
Figure 16b: Ground current when GLE04-01 is installed
The key point about grounding filters is to remember that grounding is a safety element, and the use of grounding filters should not affect compliance with relevant ESD standards and practices.
Correct grounding is more than just running a green wire. Good grounding helps ensure uninterrupted operation of the equipment and data integrity, while poor grounding is counterproductive. Whether you are an electrician, an ESD practitioner or an EMC engineer, you should consider and solve not only the grounding aspects that match your profession, but also all grounding precautions, including safety, ESD, EMI, and data integrity. In most cases, a single standard cannot fully consider all the requirements in the process. Pay special attention to the EMI on the ground, because it connects all equipment and is a pipe through which EMI is transmitted. Comprehensive, high-quality ground is a solid foundation to help ensure the smooth and efficient operation of your processes and equipment.
Vladimir Kraz is the president and founder of OnFILTER, a California manufacturer of high-performance EMI filters and instruments made in the United States. Vladimir holds a number of US patents on EMI and ESD topics. He is the leader of the SEMI EMC Standards Working Group, a member of the ESD Standards Association, and has published many technical papers in publications and international seminars.
Your email address will not be published.
Save my name, email, and website in this browser for the next time you comment.
In Compliance is the main source of news, information, education and inspiration for electrical and electronic engineering professionals.
Aerospace Automobile Communication Consumer Electronics Education Energy and Power Industry Information Technology Medical Military and National Defense