|
|
| Static Fire Stories | Articles & Technical Papers | Current News |
|
First Published in EOS/ESD Technology April/May 1992 Evaluating Semi-Rigid Plastic Trays Developing a test method that evaluated ESD protection A.E. Warnecke Cecil W. Deisch Suppose a board (circuit pack) is enclosed in a static-protective package and this package is sitting on a conductive or static-dissipative surface. Someone who is electrically charged reaches to pick up this package and as his fingers approach, a spark discharge to the package occurs (Fig 1). Are any of the electronic circuits on the board inside harmed?
Several boards of one kind in a specific package can be tested and a known range of discharge voltages can be used to come up with meaningful results for this one kind of board. But how do you design tests where you can draw generalized conclusion and make meaningful predictions? With such a variety of circuit packs (each with different circuits and different components of varying sensitivities), the number of combinations of all the variables makes it nearly impossible to test. This does not include the task of thoroughly instrumenting the board to accurately measure harmful voltages on or around sensitive devices. It also does not include the variability of the dissipative package or the effects of varying the physical separation between the package and the high points of sensitive components on the board. Designing A Test Setup Instead of tackling this challenge head-on, we designed a test setup which closely mimics the electrical environment of a sensitive component in a worst-case ESD situation. Because ESD events contain high voltages and high-frequency components, we went to great lengths to thoroughly design the test fixture and the measuring probe to give an accurate time vs. voltage scope readout without being sensitive to stray pickup. Then we took data on many varied parameters and came up with a lower bound on the predicted effectiveness of ESD protection provided by a package. This data gives valuable insight into how to design the package in general, and allows a prediction of the minimum protection level. Different Packaging Materials There are three classes of material from which most packages are made. Metal. A well-made metal package can give near-perfect ESD and electrical overstress (EOS) protection. This is because there is virtually no resistive voltage drop anywhere on the surface. Furthermore, no flux lines can penetrate the metal at the high frequencies of a spark discharge so there can be no induced voltage between any two points on the inside of the package. If there is no voltage differential anywhere inside the package, then there can be no harmful voltages to damage sensitive parts. Metal containers, however, are generally impractical because they are expensive and their contents cannot be seen. Insulator. A package made of an insulator (e.g., untreated polyethylene) also gives excellent protection to the enclosed board but only for the experiment previously outlined. A charged person who touches the package will not generate a spark. Further, polyethylene can sustain several kilovolts per mil of thickness, so there is no breakdown from the outside to the inside surface. However, a charged person who picks up this package will remain fully charged because the package does not bleed away any charge during the time the fingers touched the package while it was still on the conductive surface. If the charged person now opens the package and touches the board, there will be a spark discharge, possibly harming the electronics. Static-dissipative. The ESD performance of a package made of static-dissipative material falls somewhere between the two materials above. A charged person can discharge a spark to this material. However, even though its surface resistivity (105> to 1011 ohms/sq.) is not nearly as low as metal, a properly-designed static-dissipative package can still give excellent EOS and ESD protection. In contrast to the insulator above, a static-dissipative package can quickly bleed away the charge on a person, lessening the chance of subsequent ESD damage. Many varieties of static-dissipative material are transparent, making the contents of the package clearly visible. Spark Discharge to a Dissipative Material
Consider the static-dissipative package (Fig 2) which sits on a conductive surface (earth ground) and totally encloses a circuit pack. As a charged body approaches the surface of the package, a thin column of air becomes ionized and a spark discharge is limited. This thin column of highly-conductive ionized gas forms exceedingly fast (in a nanosecond or less) and, once formed, it behaves electrically very mush like an ordinary small diameter wire. In other words, most all the voltage across it is caused by inductive voltage (L di/dt). Very quickly, electric charge flows through this ionized column and is deposited in a small area on the surface of the package. However, as the charge accumulates on this surface spot, the voltage of the spot is rapidly rising and approaches that o fthe charged body. Once this voltage difference gets small enough (about several hundred volts), then the ionized column can no longer support itself and it extinguishes. This whole sequence of events, from initiation to quenching, happens very quickly (approximately a few nanoseconds). The main parameter that limits the speed is the inductance of the thin column of ionized gas. High surface resistivity. What happens on the surface of the package during this short event is strongly dependent on its surface resistivity. For a material with surface resistivity near the high end (1011 ohms/sq.), there are many tiny, faint sparks during a discharge. For each tiny discharge, the point on the surface directly beneath the spark has very high voltage; however, the voltage just a short distance away may be many hundreds of volts less during the actual spark discharge. This is because the high surface resistivity slows down the diffusion of the spark-deposited charge. The point where the spark hits rises to several thousand volts, but just a short distance away (e.g. 1 in.) the voltage may be in the low hundreds or less. Ripple-type effect. Once the spark extinguishes, the concentrated charge on the surface quickly spreads out through surface conduction and the voltage peak quickly decays. These events are quite analogous to dropping a pebble into a pool of water. A column of water of substantial height forms right at the center, but the height of the rest of the water quickly drops off as the radius increases. As time passes, the height of the center column and the water immediately around it are quickly restored to the average surface height of the pool. Low surface resistivity. Packages with lower surface resistivity can have fatter and more aggressive spark discharges to the surface; but lower resistivity actually improved ESD protection. Since any discharge to the surface of the package has to charge a larger area, more charge must flow for a longer time to attain the same surface voltage. The longer it takes to charge up this larger area, the more time is available for the charge to bleed away because of the lower resistivity material. The more charge removed from the charged human body during the spark, the more body voltage is reduced. The net result is that, although a larger area of the package rises in voltage, the peak voltage at any point on the surface is less. Effects on Board Components As pointed out above, when a spark discharge occurs, a small area on the surface of the package around the point of discharge rapidly rises to several thousand volts. Most of the rest of the surface of the package distant from the spark rises very little. For all practical purposes, the worst-case scenario is to assume all the rest of the package remains at ground. Since the board inside takes on a voltage which is the average of all voltages on the surface of the package, and since only a small post rises to several thousand volts while the rest is (worst-case) ground, the board takes on the average voltage of nearly ground. In many situations, a sensitive component may sit right beneath the high-voltage spot on the surface of the package while the board has an average voltage near ground. The danger is analogous to a person standing in an open field (grounded) while there is a highly charged thunder cloud overhead.
Worst-case model. The worst-case equivalent circuit model is shown here in Fig. 3. Here the circuit board is replaced by a large grounded metal plate; one or more leads of the sensitive device are connected to this ground plane. Another lead of this sensitive device is directly beneath the high-voltage spot overhead which is momentarily caused by the spark. The parasitic capacitance to ground represents not only the capacitance of the sensitive device itself, but also the parasitic capacitance of routing conductor(s) connected to it. Voltage Coupling Mechanism At low spark discharge voltages, the "hot spot" on the surface of the package rises to perhaps a thousand or more volts. This hot spot behaves much like a plate of a capacitor suddenly yanked to a high voltage. An isolated conductor between the hot spot and the ground plane sees a fraction of the hot spot rise in voltage. This fraction depends on the size of the exposed conductor and how close it is to the hot spot; the larger the conductor and/or the closer it is to the hot spot, the larger the fraction. For a fixed configuration, as the discharge voltage increases, so does the capacitively-coupled voltage on the sensitive lead. This holds true up to a certain voltage when another phenomenon takes over. Spark Discharge: Package Surface to Board As the spark discharge voltage is raised, the coupled voltage also rises almost linearly. Often a point is reached where, if the spark voltage is increased just slightly more, the coupled voltage suddenly jumps up (often by a hundred or more volts). Once the discharge voltage gets high enough and/or the surface of the package gets close enough to a sensitive part on the circuit board, a secondary spark occurs between the inner surface of the package and a component on the board. This secondary spark is the hazard to avoid because it can be extremely damaging. (Note: This secondary spark effect applied only to materials that have relatively homogeneous volume resistivity properties. The volume resisitivity allows charge to freely flow between the outer and inner surface. In particular, this internal spark phenomenon probably does not apply to sandwich type materials such as dissipator-insulator-dissipative or dissipator-metal-film-dissipator.) Setting off the hazard. Since the internal spark discharge is the hazard to avoid, what are the parameters that are most likely to aggravate this hazard? Not surprisingly, the smaller the physical spacing between the package surface and the sensitive part, the more likely an internal spark. Also, a sharp point (such as a sheared end of a component lead) is much more likely to initiate a spark because of the ionizing voltage gradients at the surface of a small-radius conductor.
Scope Probe Instead of trying to accurately instrument a sensitive device to measure the voltage transient during a spark event, we let the tip of a scope probe represent the electrical equivalent of a sensitive device. The probe tip also allows direct and accurate measurements of transient voltage waveforms (Fig 4).
Probe tip. The scope probe tip is an excellent model of a worst-case sensitive part because it has only 12.5 pF between the tip and ground. An actual sensitive part may have as little as six to ten pF of input capacitance, but when it is mounted on the board it is connected to at least one other component through several inches of printed copper conductor. The actual total capacitance at a sensitive node is likely to be at least 15 pF. When considering immunity to ESD events, generally the higher the capacitance to ground, the lower the transient voltages will be. This is because it takes more coupling charge to raise the voltage on a larger capacitance (V=Q/C). Test Parameters Based on the discussion above, the most important parameters of the package that needed to be tested are:
Designing the Low-Inductance Probe Ground As mentioned above, spark discharges are rich in very high frequency components. Most o the energy in sparks falls in the region of 1MHz up to about 1 gHz. This implies that to get accurate measurements: 1. Inadvertent coupling and crosstalk must be held to a minimum 2. Ground inductance to the measuring probe must be of the order or 1 nH or less. This last requirement absolutely rules out ground clip leads because they may have inductance in the 100 NH or less. At the high frequencies being measured, any transient currents through the ground lead can cause voltages that thoroughly overwhelm any voltage being measured. (See Sanity Check below). Since we wanted to adjust the position of the probe in the hole in the ground plane, we machined a brass barrel (or sleeve) to fit over the ground ring of the probe end. This tight-fitting, low- inductance barrel provided a smooth outer surface, allowing the probe to be inserted or withdrawn over a range of about 0.8 in. To complete the low ground inductance connection to the plate, near its center a hole was punched which has six flexible metal tangs that make a multi-point ground connection to the barrel. This allows the probe to be moved in and out of the plate while maintaining a low grounding connection. Parts of the Test Fixture The metal plate is a piece of galvanized sheet steel about 3 ft. square. For convenience of use, it was mounted vertically. A small metal frame about 4-in. square was attached to the metal plate, with the probe tip in the middle. The outer surface of the frame is about 0.75 in. above the metal plate and allows flat samples of static-dissipative materials to be laid on top and held down with strip magnets (Fig 5). The metal frame has a low-inductance connection to the plate and the static-dissipative sample also has multi-point contact to the frame because of the compliant force of the strip magnets. With this frame and the special ground barrel on the probe, the spacing can be adjusted between the end of the probe and the inner surface of the static-dissipative sample from zero (touching) to about 0.7 in. (Fig. 6).
The Test Layout Mounting the ground palate vertically also provided the benefit of shielding. The scope and the probe lead were set on one side of the plate and all the high-voltage spark tests were performed on the other side. The large metal plate shielded the sensitive measuring scope (Fig. 7).
Sanity Check When measuring high frequencies, especially in the presence of high voltage spikes, it is best to perform a "null measurement." This sets up the test fixture in such a way that the expected voltage is zero. In never will be zero, but the results give an indication of how much background noise is contaminating the measurements. For the null test, the metal frame was covered with aluminum foil and then sparks discharged directly to the foil above the probe tip. Scope-measured voltage peaks were typically a fraction of a volt, but were always less than two volts even for 25 kV discharges. Since we are interested in measuring voltages above approximately 10 V, the low null voltage showed we had a good test setup. Tests, Results, and Conclusions About 10 static-dissipative materials of different composition from several manufacturers were tested over the full spark range of 5 to 25 kV. Each material was tested after having been prepared in two ways: 1) Aged for several days at 20 deg.
C. and 50% RH For each test coupon, the surface resistivity and charge bleed-down were first measured and then placed on the test fixture with a tip spacing of 0.5 in. Several discharges were performed at each voltage and the maximum peak voltage at the probe tip was recorded. The spacing was lowered to 0.25 in. and the same was reduced to 0.125 in. If peaks of more than approximately 100V were measured or if the secondary internal spark event was observed, we stopped the test and went on to the next test material. Results. After taking hundreds of measurements of many different materials with various spacings to the probe tip, we found that there was great variation in the data because of the randomness of spark discharges. Nonetheless, we can break the results down to some simple guidelines assuming:
If these guidelines are met, then a package designed with a 0.25-in. clearance from all components on the top of the board and from anything on the bottom of the board will give adequate protection at least 99.9% of the time. This 0.25-in. clearance can be modified, depending on circumstances. For example, if your circuit devices are much less sensitive to ESD, if the humidity is kept high, and/or any ESD is likely to be much less than 12 kV, then the package might be designed with an 0.125-in. Clearance all around. On the other hand, circuit packs with exceedingly sensitive devices, which are shipped or handled in very dry environments and which maybe subjected to substantial ESD, may need packages with as much as 0.5-in. clearances. Since the clearance guidelines are based on worst-case test measurements, the calculation of the probability of adequate protection is based on extremely conservative estimates. 1) The probability that the spark to the package occurs directly above the most sensitive device(s) is no more than 10% (more likely, the probability is 1% even for a small 4 x 6-in. board). 2) The probability that the minimum clearance occurs directly over the most sensitive occurs directly over the most sensitive device(s) is no more than 25% (more likely it is less than 5%). 3) The probability that the most sensitive device(s) has a sharp point is no more than 25%. Sharp points are more susceptible than rounded surfaces. (It is more likely that the probability of the most sensitive devices having sharp points less than 10%). 4) The probability that the most sensitive device has a node of only 12.5 pF (the tested value), is no more than 50%. The most sensitive node is likely to have at least 15-20 pF. The higher the node capacitance, the lower the susceptibility to ESD. (The probability of a 12.5 pF node is more likely less than 20%. 5) The probability that the package will have a surface resistivity as high as 1 x 1011 ohms/sq. is less than 70%. Spark protection increases as surface resistivity goes down. (A more reasonable assumption is that surface resistivity will be at its upper limit of 1 x 10 11 ohms/sq. less than 25% of the time). 6) The probability that the spark will be its most intense at 12 kV is no more than 50%. (Most discharges will be in the low-kilovolt region and the chances of a 12 kV spark is more likely less than 10%) Taking the very conservative probabilities of failure above and multiplying them gives an overall probability of failure of about 0.10 of 1% or an overall probability of protection of 99.9%. Using the more reasonable probabilities, the estimate is about 1 failure in 4,000,000 or a protection rate of 99.9999+%. A dissipative plastic tray with the correct resisitivity (surface and volume) and spacing from the tray outside to critical components will provide an excellent shield as well as physical and contamination protection. Based on published data it is reasonable to expect that more than 50% of early field failures could be eliminated by using this type of tray. |
The
discharge model and the maximum discharge voltage. Since human
handling will be the source of any discharge, the human body model
source was also used. Also, if reasonable ESD precautions were
taken by the handler, there should never be a discharge. However,
once in a while mistakes are made and someone will become inadvertently
charged. Although we tested to 25 kV, we chose 12 kV as the maximum
inadvertent charge.
The damage level of a sensitive device is 100V.The ESD Journal is not affiliated with any trade organization, Association or Society
ESD Journal & esdjournal.com are Trademarks of Fowler Associates, Inc. - All Rights Reserved
The content & Look of the ESD Journal & esdjournal.com are Copyrighted by Fowler Associates, Inc. - All Rights Reserved Copyright 2011
The YouTube name and logo are copyright of YouTube, LLC.