TISP® Types Introduction This section covers the overvoltage protection functions and Bourns TISP® (Totally Integrated Surge Protector) thyristor SPDs (Surge Protective Devices) in terms of evolution, function, silicon structure, electrical characteristics, electrical rating and device variants. SPD Voltage Limiting Voltage Limiting Basic Protection Function SPDs have a non-linear voltage-current characteristic which limits overvoltages by diverting the current caused by an overvoltage. Figure 1 shows how a twoterminal SPD is applied to limit the voltage between two conductors and one conductor and ground. There are two basic types of SPD characteristics; clamping and switching. SPD Figure 1. SPD Circuit Application OCTOBER 2000 - REVISED AUGUST 2001 TISP® Origination In 1980, British Telecom announced the development of a new generation of telephone exchanges called “System X” which would employ the latest semiconductor devices. Lightning, a.c. power system faults and switching produce voltages on the telephone line. The ICs which directly interfaced to the telephone line would be +i exposed to such voltages and needed Overaccurate and fast voltage Clamping overvoltage protecSPD tion to prevent Voltage System failure. It had been Protection Level Current at Rated Rated Voltage established that the Voltage +v best type of SPD for System SPD Voltage this application was Rated Voltage thyristor based. Figure 2. Clamping SPD Characteristic and Voltage Limiting Waveshape However, at that time, the thyristor SPDs Switching SPDs have a discontinuous voltage-current available would only protect in one voltage polarity. Proteccharacteristic, the discontinuity being caused by the tion in both voltage polarities and both line conductors switching action between high voltage and low-voltage would require four devices. System X designers asked several regions. The TISP® device has a switching characteristhyristor manufacturers if the protection for one line could tic as it is a thyristor device. Before the TISP® device be integrated into one package. switches into a low voltage state, there is a small clamping action, caused by the breakdown region - see Bourns’ Bedford semiconductor power device facility Figure 3. In limiting an overvoltage, the downstream met this challenge. Being an established thyristor manufacturer meant +i that the basic thyristor structure already Overexisted. However, voltage Breakdown how to integrate four Region devices, achieve a 30 Voltage System year life and accurate Protection Level Current at Rated Thyristor voltage control reRated Voltage Voltage SPD +v quired the initiation System SPD Voltage of a major research Rated and development Voltage Figure 3. Switching SPD Characteristic and Voltage Limiting Waveshape program. The need for integration and load will be exposed to a high voltage for the brief long life was solved by the adoption of a planar structime period that the TISP® device is in the breakdown ture, and accurate voltage control was achieved through region before switching into a low-voltage state - see the use of a buried ion-implanted layer (UK patent Figure 3 waveshape. When the diverted current falls 2113907, 1981). Production of these devices started in below a critical value, the TISP® device switches off 1982 and, today, both fixed voltage and gated prodand allows normal system operation to resume. ucts are produced. The following clauses give more detail on the make up of a fixed voltage TISP® device. Voltage Protection Level Voltage Protection Level SPD Current SPD Current Clamping SPDs have a continuous voltage-current characteristic - Figure 2. In limiting an overvoltage, the downstream load will be exposed to a high voltage for the time period that the overvoltage exceeds the system rated voltage level - see Figure 2 waveshape. Obviously it is important that normal system voltages are not clamped. OCTOBER 2000 - REVISED AUGUST 2001 Basic Thyristor Structure Figure 4 shows the simplified structure of a unidirectional thyristor protector. The switching action occurs when the top contact is negative with respect to the bottom contact. Also shown are the equivalent circuit elements created by the semiconductor layers. Typically, the protector would be manufactured starting with an n- wafer. Layers of p material would then be diffused into the top and bottom. Next, n+ regions would be diffused into the top surface. Finally, the top and bottom would be metallized to provide contacts. Junction Termination The technology used for junction termination at the chip surface was decided by future needs and performance history. Three common ways of controlling junction breakdown at the surface are shown in Figure 5. Glassed Mesa technology terminates the high voltage junction on the edge of a mesa created by grooving the silicon wafer. The mesa is sealed with a layer of glass to prevent contamination and reduce the surface field strength. Planar brings a lateral junction up to the top surface. The surface is sealed with silicon dioxide in the junction area. Additional stability is provided by a channel stopper diffusion. Transistor TR1 Although the is formed by the double sided glass n+pn- layers. Similarly, mesa gives the transistor TR2 is formed by smallest chip size, the pn-p layers. Avalanche diode the double mesa D1 is formed by the central n p layers. grooves weaken the wafer Resistance R1 is the lateral resistance of the p and this limits the maximum layer. Resistor R2, together with resistor R1, Figure 4. wafer size that can be processed shunt the base-emitter junction of transistor without excessive breakage. TR1 to define the value of switch-off current. Simplified Protector Structure Single glass mesa and planar Resistor R2 has a relatively low value of resistance solve these problems at the expense of increased chip and is considered as a local base short to the emitter. size used to bring the lower junction up to the top In manufacture, the emitter diffusion is perforated surface. The mesa groove still limits the maximum with a series of dots to create these shorts. In Figure 4, wafer size. Planar does not have any wafer size limitasome of the top metallization has been omitted to tion. Both these technologies were established in show the p-type dots. The practical implementation production at Bedford. of this structure requires technology decisions on how the protection voltage is set and the surface termination of the voltage blocking junctions. OCTOBER 2000 - REVISED AUGUST 2001 Final Junction Current - nA 15%. Planar was In the previous chosen as it had illustrations, the bulk better junction voltage breakdown would have current characteristics. been controlled by the resistivity of the This is shown in Figure 6 which silicon used for the wafer. The precise value of is the result of a high temperature Figure 5. resistivity will vary across a wafer, reverse biased (HTRB) life test on Junction Termination Technologies from wafer to wafer and from lot to high voltage glass mesa and planar lot. Using starting wafer resistivity for determining the transistors. An advantage of the planar structure breakdown voltage could create yield consistency approach is that multiple protectors may be easily problems on tight voltage windows. made on a single chip. Each protector is formed in a wraparound of p-type silicon, which serves to isolate An alternative approach is to use a high resistivity the protector sections from each other. silicon wafer which would give a higher breakdown voltage than required, and diffuse Transistor Junctions, 1000 hours, 720 V at 150°C 100000 in a lightly doped layer which lowers the breakdown voltage to the required value. 10000 In order to be successful, the doping of this additional layer needs to be highly 1000 accurate. Ion-Implantation is needed to give sufficient precision in controlling the level of dopant. Lowering the silicons’ Planar 10 Glass Mesa breakdown voltage by diffusion allows No Change Line selective breakdown areas. A pad of 1 dopant can be implanted at the centroid 1 10 100 1000 10000 100000 of each emitter. Breakdown will occur at Initial Junction Current - nA Figure 6. Junction Current Stability this pad, which will be deep in the bulk of the silicon and well away from the chip surface. By defining the breakdown Voltage Accuracy region, the switch-on and current spreading perforProtectors have to be supplied to a specified voltage mance should be more consistent. window; that is a maximum and minimum value of breakdown voltage. This window might be as little as 100 OCTOBER 2000 - REVISED AUGUST 2001 The use of a breakdown pad considerably reduces base junction of transistor TR2. The current path is the breakdown voltage sensitivity to through transistor TR2, and resistor RH. surface states. Chip surface voltage The current levels are very low control will still be required, during this condition as shown but the chances of premature in Figure 6. surface breakdown are now considerably reduced. Figure 7 The breakdown region initiates shows a typical TISP® device when the applied voltage is structure which uses a buried sufficient to cause the diode D1, breakdown pad and planar to avalanche. As the diode D1 is junction termination. This structure integrated with transistor TR2, the shows a 1988 development of the original breakdown region is a classic transistor Figure 7. ion-implanted patent and is covered by BVCEO shape. After the initial breakdown, Typical TISP ® Structure US patent number 4,967,256. the characteristic is re-entrant as the transistor gain increases and then the characteristic shows a positive slope characteristic at higher current levels. Simple Equivalent Circuit The current path is through the emitter of transistor The array of circuit elements shown in Figure 4 can TR2, then through the parallel combination of the be simplified to a lumped equivalent circuit containtransistor collector-base and diode D1, and exits ing a transistor pair, TR1 and TR2, an avalanche through resistor RH. diode, D1, and a shunt resistor, RH. Figure 8 shows these two circuits together with the graphical symbol When the voltage developed across resistor RH, is for the protection structure. sufficient to cause the conduction of transistor TR1, its collector draws extra current TR2 from the base of transistor TR2 D1 TR2. The extra base current D1 TR1 causes transistor TR2, to pass R1 TR1 RH more base current into tranR2 sistor TR1. A regenerative transistor pair is formed. Lumped Equivalent Graphical Array of Circuit Elements in Silicon Structure The breakdown characteristic Circuit Symbol develops a negative slope and Figure 8. Protector Circuits and Symbol the characteristic terminates when the transistors pass enough current to switch Protection Characteristics into a low voltage on state. The voltage-current characteristic of the protector can be explained using the lumped equivalent circuit of In the low voltage state and at high current, the Figure 8. Figure 9 shows the protector voltage-current transistors will pass most of the current. As the characteristic together with the parts of the lumped current reduces, the current taken by resistor RH, equivalent circuit that are carrying significant current shunting the base emitter of transistor TR1, becomes during each of the various operating modes. significant. Switch off occurs when the current level falls to a point where the voltage developed across the In the off-state condition, when the voltage is inresistor RH, the transistor TR1, base-emitter voltage, creased from zero in the positive polarity, most of the is less than the value needed to cause transistor voltage is supported by the (reversed biased) collectconduction and the transistor pair switches off. OCTOBER 2000 - REVISED AUGUST 2001 Switching Quadrant On State +i TR2 TR1 TR2 D1 TR1 RH Switch On Switch Off Breakdown TR2 TR1 RH RH D1 TR2 -v TR2 D1 RH Blocking +v Off State Blocking Quadrant -i TR2 RH Figure 9. Protector Equivalent Circuits and VI characteristic In the opposite voltage polarity, most of the voltage is supported by the reverse biased base-emitter of transistor TR2. In this blocking condition, the current path is through transistor TR2, and resistor RH. the blocking characteristics of the paralleled sections for information only. The switching and diode conduction characteristics will mask out the blocking characteristics on the actual protector characteristic. This protector has a switching characteristic in the positive polarity and a blocking characteristic in the negative polarity. For this reason it is called a unidirectional protector - the switching characteristic only occurs in one polarity. It can be further qualified as a reverse blocking unidirectional protector. Figure 10 shows the characteristic and symbol for the reverse blocking unidirectional protector. Note that the reverse blocking capability is higher than the positive polarity breakdown voltage. Three protection structures are the building blocks for fixed voltage TISP® SPDs; the reverse blocking unidirectional, the forward conducting unidirectional and the bidirectional. A bidirectional protector can be made by integrating a similar unidirectional protector in antiparallel - Figure 11. When one unidirectional protector is in its switching polarity, the other will be blocking. Similarly, by integrating a diode in antiparallel, a forward conducting unidirectional protector can be made Figure 11. Although this protector passes high currents in both voltage polarities, it is still called unidirectional as there is only one switching quadrant. The bottom pair of characteristics in Figure 11 shows +i +v -v -i Figure 10. Reverse Blocking Unidirectional Thyristor OCTOBER 2000 - REVISED AUGUST 2001 7 +i +i +v -v +v -v Unidirectional Unidirectional -i -i Plus Plus +i +i +v -v +v -v Unidirectional Diode -i -i Gives Gives +i +i +v -v +v -v Forward Conducting Unidirectional Bidirectional -i -i Figure 11. Formation of Bidirectional and Forward Conducting Unidirectional Protectors by Antiparallel Connection TISP® Electrical Requirements and Terms Figure 12 shows the switching characteristic of Figure 9 annotated with parameter letter symbols. In service, the TISP® device must meet requirements in three areas: Transparency - The normal system operation must not be degraded Protection - Under the specified overstress conditions, the protected equipment performs as required without creating hazards Durability - Provide reliable long term transparency and protection Transparency The protector’s off-state current and the capacitance must not affect the signals normally occurring on the telephone wires. This covers the part of the characteristic from zero volts to the rated maximum repetitive off-state voltage, VDRM. Also, after an overvoltage that causes the protector to switch, switch off must occur to restore normal operation. Switch off occurs at the low current termination point of the on-state region called the holding current, IH. Two off-state current conditions are normally specified, one for standby and the other at the system normal maximum voltage level. In POTS (Plain Old OCTOBER 2000 - REVISED AUGUST 2001 +i ITSP Switching Characteristic On-state Region ITSM IT V(BO) VT IH Switching Region I(BO) Breakdown Region IDRM ID -i VD Off-state Region VDRM +v Figure 12. Switching Characteristic Regions and Parameters Telephone Service) lines, in standby, the line normally has a battery voltage applied. This battery voltage is typically about -50 V. At standby, it is important that the protector draws little current. To ensure this, a d.c. off-state voltage, VD, equal to the battery voltage, is applied and the resulting off-state current, ID, measured. Increases in ambient temperature increase ID, so it is important to specify ID to be measured at the highest expected ambient temperature. As the standby is the normal line condition, durability testing often uses this as the electrical bias on the device. The second off-state test comprehends the maximum line voltage resulting from the system signal or testing or ringing conditions. Under the normal system maximum voltage condition, the protector should not clamp and distort the voltage. To ensure this, the protector’s repetitive off-state voltage rating, VDRM, is set to be equal or greater than the maximum system voltage. For the rating verification test, the voltage is ramped up across the device and the repetitive offstate current, IDRM, is measured when the voltage reaches VDRM. The low-current breakdown voltage region is temperature dependent at a rate of about 0.13 %/°C. Characteristic curves of this are often included in TISP® data sheets. This means that any clipping will be most severe at the lowest ambient temperature. Therefore, VDRM should be measured at the lowest expected equipment ambient temperature, or at 25 °C with an appropriate increase in the VDRM value. Device junction capacitance decreases with increasing applied voltage. Capacitance is measured with a specified d.c. and the a.c. bias voltage levels, e.g. -5 V and 1 V r.m.s. The a.c. voltage adds and subtracts to the d.c. voltage and at any moment the net voltage applied is the sum of the d.c. and instantaneous a.c. voltage values. The instantaneous capacitance value will be set by this voltage sum. Thus, as capacitance testing normally uses 1 V r.m.s a.c., the capacitance value measured will be an average value resulting from an applied voltage that varies ±1.4 V on a mean (d.c.) value. When the d.c. voltage bias is much greater than the a.c. voltage bias, the variation in capacitance caused by the a.c. is negligible. The capacitance is not frequency sensitive, but both circuit and package inductance can resonate with the capacitance and this is a more important frequency consideration. The capacitance measurement conditions should reflect the expected d.c. voltage bias, VD, and, if possible, the appropriate a.c. voltage signal level, Vd. TISP® data sheets contain capacitance values for several d.c. voltage bias levels and often have graphs of capacitance versus voltage. When the protector switches into the on-state, it diverts the line d.c. feed current as well as the current resulting from the overvoltage. After the overvoltage current stops, the protector must switch off to restore normal operation (transparency). This means that the protector’s switch off current, called holding current, IH, must be higher than the line d.c. feed; otherwise the protector would stay on. As the value of IH decreases with increasing temperature, IH should be specified to be equal to or higher than the line d.c. feed at the highest expected ambient temperature. Most TISP® data sheets contain a graph of IH versus temperature. The holding current value is determined by switching the protector on to a specified on-state current level, I T, then ramping down the current until the protector switches off. The current level where switch off occurs is the device IH. Protection There are three areas concerned with the protection function; protection voltage, rated current capability and safety performance. OCTOBER 2000 - REVISED AUGUST 2001 9 The protector must limit the voltage to a level that the following circuits can withstand (Figure 3) under both a.c. and impulse conditions. In the breakdown region, the peak voltage developed is called the breakover voltage, V(BO), and the current at this point is I(BO), the breakover current. AC conditions can range from a few cycles of high current, terminated by the operation of the overcurrent protection, to 900 s or more of low level current. To cover the a.c. condition, the TISP® V(BO) is normally measured with a voltage ramp of 250 V/ms and 750 V/ms and a source resistance of 300 Ω. Under a.c. conditions, the values of V(BO) and I(BO) are major factors in the breakdown region power loss. So, for a.c. conditions, both V(BO) and I(BO) are measured. Under impulse conditions, V(BO) will strongly depend on the impulse di/dt in the breakdown region. (Note: this is the dual of Gas Discharge Tubes, GDTs, whose peak limiting voltage depends on dv/dt.) To cover the impulse condition, the TISP® V(BO) is normally measured at 20 A/µs. The unclamped voltage rise is 1000 V/µs from a source resistance of 50 Ω. Other impulse voltage alternatives are: V(BO) for a specific impulse wave shape (e.g. 2/10), the limiting voltage-time envelope for a specific impulse wave shape (TISPPBL3) or a graph of V(BO) versus di/dt. Standard GR-1089-CORE (February 1999) Peak Voltage Setting V Voltage Waveform data sheet rating table. TISP® data sheets will often show a graph of ITSM versus time, as well. As the current through the protector is quasi-sinusoidal, the current is expressed as a peak value rather than an r.m.s. value. Overcurrent protector operating time is expressed for d.c. or r.m.s current values. The overcurrent and overvoltage protectors are coordinated if, at a given current level, the overcurrent protector operates before the overvoltage protector reaches its time limit. To make this comparison, the respective current values must be converted to a common format, either peak or r.m.s. To cover international and national standards and recommendations, the peak impulse non-repetitive on-state current, ITSP, rating must be given for a wide range of waveshape designations. Common impulse waveshape designations are shown in Table 1 below. Impulse current ratings vary with temperature. Designers should check that the ITSP rating is adequate for the expected temperature range. Depending on the TISP® Series, the ITSP temperature range can be -40 °C to +85 °C or 0 °C to +70 °C, or expressed as a graph of ITSP versus temperature. Sometimes there will be a series resistance before the TISP® device which will reduce the impulse current levels. The reduction in impulse level may allow the use of a lower Peak Short Circuit Current Current Waveform current rated A protector. 2500 2/10 2 x 500 2/10 1000 10/1000 2 x 100 10/1000 1500 10/160 200 10/160 The on-state voltage, VT, at 800 10/560 100 10/560 FCC Part 68 a specified on1000 25 (March 1998) 9/720 † 5/320 † state current, 1500 37.5 IT, is often ITU-T covering: 1000 25 K.20 (02/2000) 1500 37.5 quoted for 10/700 5/310 K.45 (02/2000) 100 4000 historical K.21 (10/2000) 150 6000 thyristor † FCC Part 68 terminology for the waveforms produced by the ITU-T recommendation K.21 10/700 impulse generator reasons. In Table 1. practice, a VT value is not particularly useful as it is only a given current power loss, which is already comprehended in the ITSM and ITSP ratings. As with V(BO), the current ratings need to be specified for a.c. and impulse conditions. AC ratings such Safety standards often specify tests that will cause as peak non-repetitive on-state current, ITSM, for protector failure to monitor the resultant fault mode and given times, are usually given as spot values in the OCTOBER 2000 - REVISED AUGUST 2001 Fixed Voltage TISP® Function Range check that there are no safety hazards. This type of requirement is covered in the TISP4350L3BJ data sheet. Test Description Conditions Duration High Temperature Reverse Bias 150 °C, 80 % RV 1000 h Biased Humidity 85 °C, 85 % RH, -50 V Temperature Cycle -85 °C To +150 °C 1000 h 200 cycles Solder heat 260 °C Solvent Resistance 3 Solvents Solderability 245 °C, After 8 h Steam Age 10 s 5s Table 2. Durability Operated with their rated a.c. and impulse values, TISP® devices have an extremely long operating life, e.g. 105 impulses. Environmental and assembly reliability are covered by the TISP® qualification procedures. All TISP® devices are qualified in accordance with the standards listed in Table 2 prior to production release. Accelerated stress tests are chosen to ensure new products will meet the reliability requirements of the telecom industry. Each quarter, samples representative of the function and technology of the TISP® range are monitored to ensure that reliability remains satisfactory over the production life of the product. Additional TISP® Electrical Requirements and Terms Conducting unidirectional protectors will have a diode characteristic in one voltage polarity - Figure 13. The diode forward voltage, VF, at a specified forward current, IF, is the limiting voltage for the protected circuit. To cover a.c. conditions, the VF should be given at a series of IF values or as a graph. Under impulse conditions, the diode conduction may be delayed, leading to a short term increase in conduction voltage called forward recovery voltage, VFRM. Like impulse V(BO), VFRM is a function of impulse initial di/dt and can be given for a specific impulse wave shape (i.e. 2/10) or a graph of VFRM versus di/dt. Although the diode has its own rating letter symbols, the thyristor ones are often substituted; ITSM for peak non-repetitive forward current, IFSM and ITSP for peak impulse non-repetitive forward current, IFSP. Table 3 lists the current range of fixed voltage devices. Two terminal LTPD % Standard Method devices are basic 3 MIL STD 750 1048 building blocks. 3 JEDEC STD 22 A101 These are the 3 MIL STD 883 1010 bidirectional 10 MIL STD 750 2031 20 MIL STD 883 2015 TISP4xxx and the 7 MIL STD 883 2003 unidirectional conducting TISP5xxx Series (unidirectional blocking devices are also made for specific customers needs). A single device can be used to limit the voltage between wires or between several points if placed on the d.c. output of a diode bridge. Two devices can be used on a 2-wire telecom line to limit the conductor voltage to protective ground by connecting each device between a wire and ground. There are three protection points involved here; the two wires and ground. This protector arrangement gives two modes of protection, which will be where the protector is directly connected between each wire and ground. The voltage between the wires will be limited by operation of the two protectors in series and this will be higher than the IFSP (ITSP) +i Diode Forward Conduction Characteristic IFSM (ITSM) IF VF -v +v -i Figure 13. Diode Characteristic and Parameters limited voltages to ground. Connecting a third protector between the wires gives direct inter-wire voltage limiting and 3 modes of protection. Modes of protection are classified by the protector being directly connected between those terminals. Indirect routes, via other terminals, do not count. OCTOBER 2000 - REVISED AUGUST 2001 11 Fixed Voltage TISP ® Configurations Class Bidirectional Symmetrical Asymmetrical Protection Protection Integrated Points Modes Protection Elements 2 1 1 TISP4xxx Series 3 2 2 TISP3xxx Series 3 3 3 TISP7xxx Series TISP5xxx Series +i -i +i +v -v +v -v -v Custom TISP1xxx Series Custom +i Protection Mode Terminal Pair VI Characteristic Unidirectional Conducting Blocking -i +i +v -v -i +v -i Table 3. ® The original design of the TISP device allowed the integration of multiple protective elements. The current product range integrates the 3-point protection function into a single silicon chip. The TISP3xxx and TISP1xxx Series have 2 mode protection. As a standard, the TISP3xxx has a symmetrical protection characteristic in both voltage polarities. Special variants can be made where the protection characteristic is asymmetrical. The TISP7xxx Series is a 3-point protector with 3 mode protection. As a standard, the TISP7xxx has the same protection voltage between any terminal pair; that is each wire to ground and wire to wire. Special variants can be made where the wire to wire protection voltage is lower than the wire to ground voltage. Graphical symbols used in this table represent the protective characteristic measured at the device terminals. For ease of integration or to meet certain performance requirements, the integrated protection elements might not be of the form shown in the graphical symbol. This is similar to the graphical symbol of a triangle used to represent an operational amplifier which actually consists of many transistors and other components. The final row of the table is a graphical reminder of the VI characteristic of the various protection classes. Gated TISP® Protection Function These TISP® devices have a gate electrode, G, which controls the TISP® switching threshold. When the voltage between the gate and its adjacent electrode reaches a threshold value, TISP® current conduction is initiated. Further voltage increase causes the TISP® device to switch. There are two main applications for gated TISP® devices: protection voltage level setting by using an external voltage reference, and current triggering by the overstress current. Normally, TISP® devices work as overvoltage protectors by diverting current. In the current triggered mode, the gated TISP® device still diverts current, but the current diverted was previously taken by the protected load. Gated TISP® Structure Gates are formed by making connections to the intermediate silicon layers of a fixed voltage TISP® OCTOBER 2000 - REVISED AUGUST 2001 transistor base-emitter. If RH had been left across the NPN transistor base-emitter, it would have shunted the reverse blocking characteristic. device. Gates are named by the type of silicon layer they connect to: N-gate or P-gate. Figure 14 is the fixed voltage structure of Figure 7 with gates added. The p layer, adjacent to the cathode n+ layer, has a P-gate electrode connection. The thyristor can be switched on by a positive voltage applied to the P-gatecathode electrode pair. Current Triggered Operation In the normal circuit-current triggered mode, the voltage measurement reference terminal is the non-adjacent electrode to The n- layer, adjacent to the gate: anode for the P-gate the anode p layer, has an and cathode for the N-gate. additional n+ diffusion for The ground connected reference the N-gate electrode contact. terminal passes the diverted The thyristor can be current to ground. The gated switched on by a negative devices of Figure 15 are unidirecvoltage applied to the N-gatetional and will only switch in the anode electrode pair. appropriate voltage polarity. P-gate SCRs only switch in the negative voltage polarity Section (a) of Figure 15 reproduces the and N-gate SCRs switch in the positive lumped equivalent circuit and graphic Figure 14. voltage polarity. This P-gate and N-gate symbol of Figure 8. The gates of Figure Gated Protector Cross Section SCR pair is called a complementary pair as 14 connect to the bases of the NPN, TR1, their terminal current and voltage polarities are and PNP, TR2, transistors in the lumped equivalent opposites. circuit. The P-gate connection to the base of the NPN transistor creates a P-gate SCR (Silicon Controlled Figure 16 shows the appropriate bias conditions for Rectifier) or simply an SCR, as this is the most comthe complimentary SCR types. The P-gate SCR is mon form of SCR ((b) of Figure 15). The N-gate operated with the anode grounded, the cathode conconnection to the base of the PNP transistor creates nected to the incoming wire and the gate connected an N-gate SCR ((c) of Figure 15). In the N-gate SCR to the protected load. At a negative voltage, -v, a lumped equivalent circuit, shunt resistor RH, which current -i is drawn from the protected load. As the sets the thyristor holding current, has been moved current flows though the parallel combination of the from the NPN transistor base-emitter to the PNP Anode Anode RH TR2 TR1 TR2 TR2 D1 TR1 RH D1 P-Gate N-Gate D1 TR1 RH Lumped Equivalent Graphical Circuit Symbol (a) Cathode P-Gate Lumped (P-Gate) Equivalent Circuit SCR Cathode N-Gate Lumped N-Gate Equivalent Circuit SCR (b) (c) Figure 15. Fixed Voltage, P-gate and N-gate SCR Circuits and Symbols OCTOBER 2000 - REVISED AUGUST 2001 N-Gate SCR + +i Anode vGA - RGA N-Gate +v (P-Gate) SCR Protected Load Cathode Ground Anode -v P-Gate Cathode -i Protected Load RGK - vGK non-adjacent electrode connection to the gate. In addition to complimentary SCRs, this bidirectional function is given by a device called a TRIAC (Triode for Ac Control), which is used in 50 Hz/60 Hz applications. In the current-triggered mode, the SCRs provide current limiting to the protected load. By incorporating a defined voltage breakdown region in the SCR structures, the SCRs will limit the maximum voltage to the load as well, by operating as fixed voltage protectors. When operating in the fixed voltage mode, the current from the protected load is not needed to cause switching. + Figure 16. Current-Triggered SCR Bias Conditions SCR gate-cathode resistance and the resistor RGK , a voltage, VGK, is developed that is positive with respect to the SCR cathode. To switch the SCR into the on state, the value of VGK must reach the gate triggering voltage, VGT, which is about 700 mV. The circuit current, iS, to cause switching depends on the SCR triggering current, IGT, and the value of RGK. If I GT was 25 mA and the required circuit current to trigger was 200 mA, the value of R GK becomes 700/(200 - 25) = 4 Ω. Thus, when the incoming negative overvoltage draws 200 mA from the protected load, the SCR will switch on, diverting the current caused by the overvoltage from the protected load to ground. The operation of the N-gate SCR in the positive polarity is the same as the P-gate SCR in the negative polarity. The complimentary description requires polarity reversal of the voltages and currents, plus the exchange of anode and cathode references. A complimentary SCR pair is required to protect a single wire against positive and negative voltages, Figure 17. Referencing the composite protector electrodes is a problem as the P-gate SCR anode joins to the N-gate cathode and the P-gate SCR cathode joins to the N-gate anode. To solve this problem the electrodes are named MT1 and MT2. Electrode MT1 (Main Terminal 1) is the electrode connection that is adjacent to the gate and electrode MT2 is the Voltage Reference (Tracking) Operation As in the current-triggered mode, the voltage measurement reference terminal is the non-adjacent electrode to the gate. For voltage tracking operation, the gate is connected to an external voltage reference which controls the working and protection voltage. A simplified circuit is shown in Figure 18. In Figure 18, the P-gate SCR is operated with the anode grounded, the cathode connected to the incoming wire and the gate connected to a power supply which biases the gate at voltage -VG. For the SCR to trigger and switch on, the negative voltage, -v, on the wire must be equal or greater than the sum of the gate bias, VG, and the gate-cathode voltage for SCR triggering. The triggering voltage, VGT, is about 0.7 V, so the protection voltage is (-VG -0.7). If VG varies, the negative protection voltage will track the changes with an offset of -0.7 V. Similarly, the MT2 Ground ±v G Protected Load RGMT1 ±i MT1 vGMT1 Figure 17. Bidirectional Current-Triggered Protection OCTOBER 2000 - REVISED AUGUST 2001 N-Gate SCR protection voltage of the N-gate SCR will track the positive gate bias, +VG, with an offset of 0.7 V. These voltage tracking protectors are sometimes called programmable protectors. + Anode -iG N-Gate +v In practice, the gate bias voltage comes from the supply powering the SLIC (Subscriber Line Interface Circuit) used to drive the telephone line. This form of tracking protection ensures that the SLIC overvoltages are limited close to the power supply voltage. A fixed voltage protector cannot give the protection voltage precision of a tracking protector. The simplified circuit of Figure 18 has two fundamental problems. One is that SCR input resistance, which can be in the region of 30 Ω, will create an excessive leakage current between the gate power supply and the wire. The second problem is when the SCR switches on, it will also try to pull the gate power supply voltage to ground as well as the wire voltage. Both these problems can be prevented by inserting a series gate diode block reverse gate current as shown in Figure 19. N-Gate Gate Blocking SCR Diode + Anode -iG N-Gate +v (P-Gate) SCR vGA - + (P-Gate) SCR P-Gate Cathode Cathode +iG vGK - Protected Load Gate Blocking Diode Figure 19. Use of Gate Diodes to Block Reverse Current An IC based protection thyristor incorporating this blocking diode was proposed in 1986. By 1989, several parts were commercially available. This still left one problem, which was caused by the nature of the power supply. Typically, the power supply used +iG + -vG vGK - Protected Load Figure 18. Simplified Voltage Tracking Protection Circuit - + -vG PS P-Gate +vG PS - + -v Ground -v +vG PS Cathode Anode Protected Load + Anode - + Protected Load Ground PS Cathode vGA would be a switching mode. Switching mode power supplies that rectify and smooth pulses can only supply current in one direction. Unlike a battery, these switching mode power supplies cannot accept reverse current, i.e. they cannot be charged like a battery. The problem this causes is shown in Figure 20. AC induction testing at low levels of VAC gives a cathode currents, IK, too low to cause switching of the SCR, Th1. The peak negative voltage is clamped just below the gate supply voltage of -50 V. A proportion of the cathode current appears as gate current, IG. The reason for the gate current dipping in the middle of the negative a.c. cycle is SCR regeneration, which reduces the gate current value needed to support the cathode current. As the gate current does not fall to zero, the SCR never regenerates sufficiently to switch on. The gate current peaks at 80 mA and averages out to 21 mA. Unless the SLIC is drawing at least 21 mA via diode D2, the net current of the negative voltage power supply will be positive, i.e. a proportion of the gate current will be charging the power supply. Once the voltage starts to lower, most switching mode power supply loops will stop producing charging pulses. This reverse biases the rectifier diode D1, leaving the gate current to charge the voltage lower and lower. This situation often results in the SLIC failing due to overvoltage. The SLIC illustrated does not need to produce positive voltages. The overvoltage protection in the OCTOBER 2000 - REVISED AUGUST 2001 VG 5 10 Time - ms 15 IK IG = zero 100 D3 D5 (Added for VG Limiting) Tx IG(AV) Charges V G C1 D1 VG Switching Mode Power Supply IG(AV) Loads VG I - Average Gate Current - mA G(AV) 25 20 15 10 5 0 -5 -10 G 15 IG IK 100 80 60 40 20 0 -20 -40 -60 -80 -100 200 SCR D4 Th1 D4 Th1 TR1 20 IG = +80 mA Some solutions 0 to this problem are shown in -100 Figure 21. Figure IK -200 21 is a simplified -300 form of the 0 5 10 Figure 20 circuit. Time - m s Two SCR variants Figure 20. are shown: one as in Figure 20 and the other with a transistor TR1 replacing the gate diode D3. The graph shows the average current produced by the two variants. The SCR average current peaks just before the SCR starts to switch. Higher levels of a.c. test voltage, VAC, reduce the voltage clamping time of the SCR and the average gate current. Breakdown diode D5 is added to divert current when the gate current starts to charge the power supply. This diode is often sacrificial, TransistorBuffered SCR SLIC D3 300 VG D4 Th1 VK RAC I G - Gate Current - mA Voltage - V 10 0 -10 -20 -30 -40 -50 -60 0 I - Cathode Current - mA K positive polarity is given by diode D4, which clips the voltage just above ground. Gate current during diode conduction is zero. VK VBat D2 VAC C2 Switching Mode Power Supply VG Tx D1 AC Induction Test C1 20 AC Induction Testing overheating and going short circuit under adverse power induction conditions. The transistor buffered SCR shows an unexpected characteristic - the average gate current is negative, so always correctly loading the power supply. The gain of the transistor will reduce the average gate current during cathode current flow, but this will still be a positive value even though it is lower than the unbuffered SCR gate current. Figure 22 provides the answer; RAC = 600 Ω Protector there is negative Clamping current flow during the diode D4 conduction. By careful integration, Protector Starting to Switch some of the diode current diverts to the SCR transistor base causing a negative gate current flow during diode conduction. The peak positive gate current is 1.6 mA and the peak Transistor Buffered SCR negative gate current is a larger -7 mA. 200 300 400 500 0 100 VAC - Open-circuit RMS AC Voltage to Cathode - V Figure 21. Switching Mode Charging Currents OCTOBER 2000 - REVISED AUGUST 2001 IK 200 Transistor Buffered IG = +1.6 mA 100 0 -100 Transistor Buffered IG = -7 mA -200 IK -300 0 5 10 Time - ms 15 Figure 22. Transistor Buffered SCR Currents Protectors like the TISP61089B are designed on this principle. Transistor Buffered Thyristor Structure Figure 23 shows how the NPN transistor is integrated with the P-gate SCR. The NPN transistor is formed from the starting silicon n- material and diffusion of p+ and an n- conversion. Low resistance contact areas for the collector and emitter electrodes are given by further n+ diffusion. A metal strap connects the transistor emitter to the SCR P-gate. This type of structure is used in the TISP8200M devices and a complimentary version for the TISP8201M devices (PNP transistor and an N-gate SCR). Gated TISP® Function Range Gated devices can be made in the same classes as fixed voltage devices: conducting unidirectional, blocking unidirectional, symmetrical bidirectional and asymmetrical bidirectional. Likewise, the protection modes can be single or multiple in a single package. The extra dimensions gated devices bring are P-gate, N-gate or both, plus buffered or unbuffered. Adding these gate options to the fixed voltage classes and modes of Table 3 would make a complex table. Table 4 on the following page is a subset showing the catalog parts currently offered by Bourns. 100 80 60 40 20 0 -20 -40 -60 -80 -100 20 IG - Gate Current - mA I - Cathode Current - mA K 300 The current product range integrates several 3-point protection functions into a single silicon chip. The TISP6xxxx Series, TISPPBL3, TISP8200M and TISP8201M have 2 mode protection. All are transistor buffered to reduce triggering current. The TISP6xxxx Series and TISPPBL3 are conducting unidirectional protectors intended for protecting negative battery voltage SLICs. The TISP8200M and TISP8201M are used as a pair to protect ringing SLICs which have plus and minus battery supplies. The TISP6NTP2x series are 5-point protectors with 4 modes of protection. These devices integrate 4 conducting unidirectional protectors in a single package. Figure 23. Transistor Buffered Thyristor Structure OCTOBER 2000 - REVISED AUGUST 2001 Gated TISP ® Configurations Class - Unidirectional Conducting Blocking Buffered P-gate Buffered P-gate Buffered N-gate Points Modes Integrated Elements Protection 3 2 2 TISP61xxx Series TISPPBL3 5 4 4 TISP6NTP2x Series +i +i Protection Mode Terminal Pair VI Characteristic TISP8201M TISP8200M -v +v -v +v -v -i -i +i +v -i Table 4. TISP® Packaging When the TISP® device was first introduced in 1982, it used the standard TO-220AB power device package. This package, together with the smaller SOT82, was intended to be secured to a heatsink to allow high dissipation. Generally, the dissipation of a TISP® device does not require the use of a heatsink. This led to the introduction of various TISP® products in smaller, lower height through-hole mounting packages in the 1987 to 1998 period. Recognizing the trend to surface mount technology, in 1993 TISP® devices were introduced in 8-pin small outline packages, D008, and followed in 1997 by the 2-pin DO-214, SMB package. This path of migrating from power device packaging to smaller size through-hole package and the introduction of surface mount technology is shown in Figure 24. OCTOBER 2000 - REVISED AUGUST 2001 D008 (SOIC) Surface-Mount Packages SMB (DO-214AA) TO-92 P008 (PDIP) Through-Hole Isolated Packages TO-220 SL003 (SIP) DO-92 (MOD. TO-92) SOT82 Through-Hole Power Packages 1990 1985 1980 2000 1995 Figure 24. Package Introductions Applications Information TISP® data sheets are comprehensive and contain applications information relevant to that device. The Area Circuits Topic following table lists some of the topics found in TISP® series data sheets. TISP ® Data Sheet Area Systems TISP ® Data Sheet Topic BOD replacement ’ 3700F3 ADSL Diode bridge multi-point protection ’5xxxH3BJ ’4360H3BJ HDSL ’40xxL1BJ, ’40xxH1BJ LCAS (Line Card Access Switch) ’4125/’219H3BJ Home phone networking ’40xxL1BJ, ’40xxH1BJ Low capacitance 3-point protection ’5xxxH3BJ ISDN - d.c. feed ’5xxxH3BJ, ’6NTP2B SLICs - Ericsson ’PBL3 ISDN-S/U ’40xxL1BJ, ’40xxH1BJ SLICs - Ringing, negative supply ’61089B LAN and MAN ’3700F3 SLICs - Ringing, positive/negative supply ’8200M, ’8201M Pair Gain ’40xxL1BJ, ’40xxH1BJ SLICs - Negative supply ’5xxxH3BJ xDSL ’6NTP2A System insulation protection ’3700F3, ’4700F3 Transformer winding protection ’40xxL1BJ, ’40xxH1BJ POTS intra-building, ’4xxxL3BJ, ’4360H3BJ - Cable MODEMs, Standards FCC Part 68 Equipment ADSL MODEMs ’4360H3BJ ’6NTP2A GR-1089-CORE ’61089B, ’6NTP2A - Router, IEC 61000-4-5 ’6NTP2A - Set top boxes, WLL IEE E Std 802.3 ’3700F3 ITU-T K.20, K.21 ’6NTP2A Impulse GR-1089-CORE ’61089B Generators IEC 61000-4-5 generator ’7xxxF3 UL 1950/60950, CSA 22.2 No.950 ’4xxxL3BJ, ’3700F3 ITU-T generators Table 5. OCTOBER 2000 - REVISED AUGUST 2001