Selecting Protection Devices: TVS Diodes vs. Metal-Oxide Varistors
An ideal protection device must limit the energy going to the load being defended, to a sufficient minimum level, such that the load is not damaged. The protection device must survive this energy burst to protect again another day. Energy at the load can be measured by Voltage × Current × Time (V × I × t). The best choice will combine low voltage clamping, low current passing, low dynamic resistance, and fast response time. Other factors such as longevity, repeatability, board space, cost, reliability, and a safe failure mechanism are also important. Laboratory tests and carefully correlated SPICE simulations were used to demonstrate and calculate the effect of 15kV transient events. In all cases, the standard 50 Ω resistance was used as the “load.” High-frequency (RF) response was not determined.
Designers should be aware of the differences between Device-Level and System-Level ESD Standards. Device-Level Standards such as Human Body Model (HBM), Machine Model (MM), and Charged Device Model (CDM) are used to define the handling conditions that the component can withstand. System-Level Standards, such as IEC61000-4-2 (Figure 1), address the conditions that the completed assembly must endure. Output current is very different for these tests, even at the same voltage.
For example, at 10kV the HBM waveform has a peak current of 6.67A, while IEC61000-4-2 waveform peaks at 37.5A. Likewise, the HBM peak occurs in 10 ns compared to 1ns for IEC61000-4-2. The dynamics are extremely different for this delta in di/dt. Understanding ESD device ratings is critical in selecting the best device. The rated specification from the manufacturer indicates the ESD level that the device can survive, without any correlation to system performance. Specifications indicated with the industry standard 8 µs/20 µs waveform do not correlate to 1ns/100ns performance. Additionally, the 1A industry standard rating is absolutely no indication of performance at 56.25A, which is the peak current level of the 15kV ESD strike. The best choice for each circuit application will provide the best protection for the load in question.
BASIC OPERATION
Protection devices operate in either “Normal” or “Protection” mode. During “Normal” mode (Figure 2), the system experiencing no unusual voltage or current surge events. Signaling is unaffected by the “ideal protection device.” No current flows in the branch containing the protection device. All of the current flows to the load, and no impact on signal integrity. Any current flowing through a practical protection device in “normal” mode is considered “leakage current.” This error term affects battery life in portable devices and signal integrity (when protecting communications lines, USB ports, HDMI lines, audio lines, etc.). While leakage currents can be quite small, these add up in battery-powered equipment and consume energy. Signal integrity is primarily influenced by the capacitance of the protection device. Compare these specifications carefully since not all manufacturers provide guaranteed rated maximum values.
Surges in voltage or current force the device into “Protection” mode (Figure 3). The ideal protection device then becomes a perfect short-circuit path to ground. Ideally, all of the energy flows through the protection device, defending the load from any damage. After the surge passes, the ideal device quickly returns to “Normal” mode, without any internal damage or changes in its performance.
MANUFACTURING DIFFERENCES
TVS semiconductor diodes are monolithic devices fabricated using standard semiconductor techniques. They can be easily implemented as arrays or incorporated into larger scale components, such as combination filter-protection devices. They feature very fast response time, low clamping voltage, and high reliability. When used within suggested design parameters, their specifications will not degrade over time or quantity of protection events. Charge carriers combine back and forth across the P-N junction during the different modes of operation. They are generally used to protect low voltage components.
MOV devices are ceramic masses composed of metal-oxide grains. Their structure is similar to that of a sugar cube. The boundary between grains forms a region with non-linear current and voltage performance, which behaves as a diode. These “diodes” arrange themselves in a random multitude of parallel and series combinations. This random structure leads to larger tolerances for specified parameters. MOV performance is affected by the volume of the device (Height × Length × Width). Large devices can handle very high voltage levels. For this reason, MOV devices are generally used to protect line powered circuitry.
Continue to next page
MOV devices are self-sacrificial. Some grain-to-ground boundaries fail after each over-voltage event, primarily due to localized heating. Laboratories measurements confirm the rise in leakage current (after each stress event) during “normal” mode. As these grain-to-grain boundaries fail, the device becomes more like a resistor than a varistor. Continuous over-stressing will eventually short the protection device. Degradation rate is inversely proportional to the volume of the device. Multi-layer Varistors (MLV) and other MOV techniques limit the current though themselves in an effort to degrade more slowly. Some varistors are intentionally designed with higher internal resistance, to limit the current passing through itself. While these devices will last longer in the circuit, they do not protect the load as well. This compromise between performance and reliability must be carefully weighed by the designer. Most MOV manufacturers consider the device to be “failed” when certain parameters have shifted by 10%. Refer to the manufacturers datasheets for details.
Both TVS diodes and MOVs can fail as an open-circuit. When the protection device has failed “open”, it provides absolutely no protection. The next ESD event will damage the load. TVS diodes tend to fail as an immediate short-circuit, nominally around 1-Ω.
MOV devices suffer from thermal runaway. As they become more resistive, continuous current flow causes further internal damage, and finally thermal runaway. Their ceramic structure is capable of much higher temperatures than semiconductor diodes. Through-hole MOV devices can reach temperatures over 400°C. Surface-mount MOV devices will typically melt their own solder joints before combustion. Care must be taken to limit the current with protection devices in any high-voltage application. Wire-wound resistors, intended to fail “open”, may need to be used. Some agency approvals require series fuses. MOV manufacturers generally provide some warnings or guidelines suggesting the necessary board space required around the device to accommodate this thermal issue.
LEAKAGE CURRENT
Each protection device is connected between signal and ground. Some systems may contain large quantities of protection components and their total additional required current becomes an issue. Low power, low voltage signaling techniques, are also highly susceptible to any additional current flow.
Laboratory measurements were performed in our ESD Testing Facility in Munich, Germany by Werner Simbuerger. For protecting 50-Ω signal lines below 5V, our tests swept bias voltages from -20VDC to +20VDC, and measured the current through the device under test. Prime area of interest (for 5V signaling) would be 5VDC. The leakage current of the TVS Diode was 10E-11A, while two MOV devices tested were 10E-09A. For low voltage applications, the leakage is 100x better for the TVS diode. Current on some devices will rise at elevated temperatures. Tests were performed at 25°C. Figure 4 shows the leakage current vs. applied voltage for MOVs and TVS.
LOW RESISTANCE PATH TO GROUND
During “Protection” Mode, the best choice will provide the lowest resistive path to ground. The ideal I-V curve would a vertical line at Vbr (breakdown voltage). Figure 5 plots the resistance path for a TVS and two MOS devices (Δresistance = ΔTLP voltage/ΔTLP current). Most manufacturers will specify the breakdown voltage at +/- 0.001A and clamping voltage at +/-1A. Resistance in the linear region of the I-V curve is calculated by the inverse slope of the I-V curve (delta V/delta I). Ironically, this slope is called “dynamic resistance” (Rdyn) but is actually used for static measurements, or IEC61000-4-2 calculations after the initial 10ns. During ESD events, the complex impedance of the protection devices changes dynamically, and calling this specification Rdyn often confuses design engineers. During the first 10ns, other methods are needed to predict the voltages. TVS diodes tend to have lower dynamic resistance (Rdyn) than MOV devices.
Test results revealed TVS diode dynamic resistance to be 10x lower than some MOV devices. According to IEC61000-4-2 ESD Test Conditions, 15kV discharge has a secondary peak surge current of 30A. This value is frequently used for both contact and air-discharge testing, because the effects of the initial transient have completely settled. During the 30A surge, the resistance of the respective protection device can be extracted from the lab results.
After the first 10ns of the IEC61000-4-2 waveform, voltage calculations are simplified since the rise and fall times become longer. The period from 25ns to 35ns, into the IEC61000-4-2 waveform, is often approximated by a rectangle that has amplitude 2A/kV. For 15kV, this is simply 30A. DC circuit analysis (Figure 6) quickly determines the voltage and current present at the load during the 30A time interval, by substituting the protection devices with their resistive value at 30A.
Continue to next page
Current through the load is given by:
Iload = (30 × Rpr) / (Rpr + 50)
ITVS = 21/50.7 = 0.414A
IMOV = 210/57 = 3.68A
The load resistance shown here is 50 Ω. During the current surge from 25ns to 35ns, the load being protected by the TVS Diode receives 10 times less current. The power (I2R) during the surge is determined by Iload × Iload × 50 Ω. Since the power is related to the square of the current (times the load resistance), is it easily recognized why current flow must be minimized.
CLAMPING VOLTAGE
Voltage peaks are limited at the load by the protection device. Laboratory test were performed by driving 300V pulses, 30ns width, through the various devices. Each device reacted quite quickly, however, the chart below indicates the relative voltage levels. The varistor granular structure does not allow for low clamping voltage. Areas of the array will force diodes in series, where their threshold voltages will sum. As seen in Figure 7, TVS diodes clamp at significantly lower voltages than their MOV counterparts, further reducing energy in low-voltage applications. As previously stated, these dynamic results are different than the Vclamp value that the datasheet would indicate. Where the current level of the input signal is known, Vclamp can be approximated as Vbreakdown + (Rdyn * Iknown) + L di/dt. For IEC61000-4-2 testing, after 10ns the L di/dt term approaches zero.
The requirement for dependable, repeatable, performance of the protection device will depend on the application. Leakage current is greatly impacted by device degradation, as shown in Figure 8. This parameter may be measured by repeating stress events and measuring the “normal” mode current. As expected, the varistor become more resistive after each over-voltage event. Carefully interpret the graph below, as the x-axis is logarithmic. TVS diodes do not degrade after each event. 10E-11A is approaching the limit of most laboratory equipment. When comparing two TVS devices, refer to the manufactures guaranteed specifications. Look for the lowest leakage current specification.
Figure 9 illustrates device degradation that also forces a shift in breakdown voltage (VBR). TVS diodes show no measurable shift in VBR over time. Some MOV devices show a clear reduction in VBR after each stress event. The failures of the MOV1 array (red) are creating a more conductive path through the varistor, seen by VBR heading towards zero. This structure (MOV1) will eventually become a short circuit. The failures of the MOV2 array (black) are creating a less conductive path through the varistor, seen by VBR heading towards infinity. Since MOV2 is less conductive initially (see the I-V curves), the ability to protect the load is further reduced with every stress event. Eventually, this structure (MOV2) will fail as an open circuit, providing no protection at all.
ENERGY LIMITING CAPABILITY
The best protection devices must limit voltage and current quickly. 15kV was used as the worst-case scenario. The common, simplified, 150pF and 330Ω , 15kV circuit was used to generate the input waveform. Peak current does occur earlier, but the same total power is contained in the input signal.
Figure 10 shows the power waveforms simulated at the load. The TVS diode combines low clamping voltage with low resistance and fast response time. Energy at the load is calculated by determining the area under the respective curves.
For this low-voltage application, the TVS diode allows 4.5µJ while the (red) MOV allows 18.0µJ at the load. Four time difference between these protection devices. This could easily be the difference between protection or failure, depending on the safe operating area (SOA) of the load. Select the protection device that offers the widest safety margin within the SOA of the load.
Some high-current, high-voltage applications will require either large MOV's or an array of TVS diodes. The designer needs to ensure system level protection against catastrophic failure. Stress beyond the specifications of most TVS diodes will result in sudden failure as a short circuit. This will result in the load not properly functioning, however, the system fails in a “safe” manner. The diode fails quickly and therefore does not have sufficient time to generate heat. Metal-oxide varistors fail in a different fashion. Their operating parameters shift with the number of stress events, even when used within specification. They become more conductive with use and thermal runaway occurs. Their ceramic construction can handle higher temperature than their silicon counterparts.
Some MOV devices will crack or explode if the temperature rises suddenly, possibly causing the device to fail as an open circuit. MOV devices that maintain their shape and form can reach temperatures above the combustion temperature of paper, which introduces the possibility of fire. Properly design your protection circuits to handle over-voltage and over-current conditions.
Many systems may have low-voltage supervisory microcontroller circuits or interface circuitry that is best protected by TVS diodes, while the AC mains or high-voltage DC stages may best be protected by MOV devices. Low-voltage signal paths are offered better protection from TVS diodes; however, some loads may operate within their SOA with either device.