LTC4259A-1 Quad IEEE 802.3af Power over Ethernet Controller with AC Disconnect DESCRIPTIO U FEATURES ■ ■ ■ ■ ■ ■ ■ The LTC®4259A-1 is a quad –48V Hot SwapTM controller designed for use in IEEE 802.3af compliant Power Sourcing Equipment (PSE). It consists of four independent ports, each with output current limit, short-circuit protection, complete Powered Device (PD) detection and classification capability, and programmable PD disconnect using AC or DC sensing. Used with power MOSFETs and passives as in Figure 1, the LTC4259A-1 can implement a complete IEEE 802.3af-compliant PSE. Controls Four Independent – 48V Powered Ethernet Ports Each LTC4259A-1 Port Includes: – IEEE 802®.3af Compliant PD Detection and Classification – Output Current Limit with Foldback – Short-Circuit Protection with Fast Gate Pull-Down – PD Disconnect Using AC or DC Sensing – Improved AC Disconnect – Improved UVLO Operates Autonomously or Controlled by I2CTM Serial Interface 4-Bit Programmable Digital Address Allows Control of Up to 64 Ports Current and Duty Cycle Limits Protect External FETs IMPROVED UVLO Available in a 36-pin SSOP package. The LTC4259A-1 can operate autonomously or be controlled by an I2C serial interface. Up to 16 LTC4259A-1s may coexist on the same data bus, allowing up to 64 powered Ethernet ports to be controlled with only two digital lines. Fault conditions are optionally signaled with a programmable INT pin to eliminate software polling. External power MOSFETs, current sense resistors and diodes allow easy scaling of current and power dissipation levels and provide protection against voltage and current spikes and ESD events. U APPLICATIO S ■ ■ IEEE 802.3af Compliant Endpoint and Midspan Power Sources IP Phone Systems DTE Power Distribution Linear Technology also provides solutions for 802.3af PD applications with the LTC4257, LTC4257-1 and LTC4267. , LT, LTC and LTM are registered trademarks of Linear Technology Corporation. Hot Swap is a registered trademark of Linear Technology Corporation. All other trademarks are the property of their respective owners. U ■ TYPICAL APPLICATIO INT SCL SDAIN SDAOUT AD0 AD1 AD2 AD3 SHDN1 SHDN2 SHDN3 SHDN4 0.1µF 100V X7R 3.3V 0.1µF VDD OSCIN AUTO BYP RESET DETECT1 DETECT2 DETECT3 LTC4259A-1 DETECT4 CMPD3003 ×4 DGND AGND VEE SENSE1 GATE1 OUT1 SENSE2 GATE2 OUT2 SENSE3 GATE3 OUT3 SENSE4 GATE4 OUT4 Q1 0.1µF 10k 10k RS1 –48V 10k 10k 0.1µF 100V ×4 SMAJ58A ×4 1k ×4 0.47µF 100V ×4 X7R PORT1 RS2 PORT2 Q2 RS3 PORT3 Q3 RS1 TO RS4: 0.5Ω Q1 TO Q4: IRFM120A RS4 Q4 PORT4 S1B ×4 4259A F01 Figure 1. Complete 4-Port Powered Ethernet Power Source 4259a1fa 1 LTC4259A-1 W W W AXI U U ABSOLUTE RATI GS U U W PACKAGE/ORDER I FOR ATIO (Note 1) Supply Voltages VDD to DGND .......................................... – 0.3V to 5V VEE to AGND ......................................... 0.3V to – 70V DGND to AGND (Note 2) .................................... ±1V Digital Pins SCL, SDAIN, SDAOUT, INT, AUTO, RESET SHDNn, ADn ................. DGND – 0.3V to DGND + 5V Analog Pins GATEn (Note 3) ................... VEE – 0.3V to VEE + 12V DETECTn Peak Currents (Note 4) .................. ±80mA SENSEn ................................. VEE – 0.3V to VEE + 1V OUTn .................................... VEE – 70V to VEE + 70V OSCIN .......................... DGND – 0.3V to DGND + 5V BYP Current .................................................... ±1mA Operating Ambient Temperature Range LTC4259AC-1 .......................................... 0°C to 70°C LTC4259AI-1 ........................................ –40°C – 85°C Junction Temperature (Note 5) ............................ 150°C Storage Temperature Range ................ – 65°C to 150°C Lead Temperature (Soldering, 10 sec)................. 300°C ORDER PART NUMBER TOP VIEW RESET 1 36 OSCIN BYP 2 35 AUTO INT 3 34 OUT1 SCL 4 33 GATE1 SDAOUT 5 32 SENSE1 SDAIN 6 31 OUT2 AD3 7 30 GATE2 AD2 8 29 SENSE2 AD1 9 28 VEE AD0 10 27 OUT3 DETECT1 11 26 GATE3 DETECT2 12 25 SENSE3 DETECT3 13 24 OUT4 DETECT4 14 23 GATE4 DGND 15 VDD 16 LTC4259ACGW-1 LTC4259AIGW-1 22 SENSE4 21 AGND SHDN1 17 20 SHDN4 SHDN2 18 19 SHDN3 GW PACKAGE 36-LEAD PLASTIC SSOP TJMAX = 150°C, θJA = 80°C/W Order Options Tape and Reel: Add #TR Lead Free: Add #PBF Lead Free Tape and Reel: Add #TRPBF Lead Free Part Marking: http://www.linear.com/leadfree/ Consult LTC Marketing for parts specified with wider operating temperature ranges. ELECTRICAL CHARACTERISTICS The ● denotes the specifications which apply over the full operating temperature range, otherwise specifications are at TA = 25°C. AGND = DGND = 0V, VDD = 3.3V, VEE = –48V unless otherwise noted (Note 6). SYMBOL PARAMETER Power Supplies VDD VDD Supply Voltage VEE VEE Supply Voltage IDD VDD Supply Current IEE VEE Supply Current VDDMIN VEEMINON VEEMINOFF Detection IDET VDD UVLO Voltage VEE UVLO Voltage (Turning On) VEE UVLO Voltage (Turning Off) Detection Current VDET Detection Voltage Compliance RDETMIN Minimum Valid Signature Resistance RDETMAX Maximum Valid Signature Resistance Classification VCLASS Classification Voltage ICLASS Classification Current Compliance CONDITIONS ● To Maintain IEEE Compliant Output (Note 7) ● MIN TYP MAX UNITS 3 –48 3.3 4 –57 5 –5 100 V V mA mA mA V V V 300 190 –23 19 33 µA µA V kΩ kΩ –21 75 V mA ● Normal Operation Classification Into a Short (VDETECTn = 0V) (Note 8) 2.5 –2 ● ● 2.7 –31 –28 VEE – AGND VEE – AGND First Point, VDETECTn = –10V Second Point, VDETECTn = –3.5V Open Circuit, Measured at DETECTn Pin ● ● ● ● ● 0mA < ICLASS < 31mA Into Short (VDETECT = 0V) 235 145 ● ● 15.2 26.7 –16.4 55 –20 17 29 4259a1fa 2 LTC4259A-1 ELECTRICAL CHARACTERISTICS The ● denotes the specifications which apply over the full operating temperature range, otherwise specifications are at TA = 25°C. AGND = DGND = 0V, VDD = 3.3V, VEE = –48V unless otherwise noted (Note 6). SYMBOL ITCLASS PARAMETER Classification Threshold Current Gate Driver IGON GATE Pin Current IGOFF GATE Pin Current IGPD GATE Pin Short-Circuit Pull-Down ∆VGATE External Gate Voltage (VGATEn – VEE) Output Voltage Sense VPG Power Good Threshold Voltage IVOUT Out Pin Bias Current Current Sense VCUT Overcurrent Detection Sense Voltage VLIM Current Limit Sense Voltage VMIN DC Disconnect Sense Voltage VSC Short-Circuit Sense Voltage ISENSE SENSE Pin Bias Current AC Disconnect (Note 10) ROSCIN Input Impedance of OSCIN Pin AVACD Voltage Gain OSCIN to DETECT1, 2 Voltage Gain OSCIN to DETECT3, 4 IACDMAX AC Disconnect DETECTn Output Current IACDMIN Remain Connected DETECT Pin Current Digital Interface VOLD Digital Output Low Voltage VILD Digital Input Low Voltage VIHD Digital Input High Voltage RPU Pull-Up Resistor to VDD RPD Pull-Down Resistor to DGND AC Characteristics tDETDLY Detection Delay tDET tCLSDLY tCLASS tPON Detection Duration Classification Delay Classification Duration Power On Delay, Auto Mode CONDITIONS Class 0-1 Class 1-2 Class 2-3 Class 3-4 Class 4-Overcurrent MIN 5.5 13 21 31 45 TYP 6.5 14.5 23 33 48 MAX 7.5 16 25 35 51 UNITS mA mA mA mA mA –50 ● –20 30 –70 300 µA µA mA V ● ● ● ● ● Gate On, VGATEn = VEE Gate Off, VGATEn = VEE + 5V VGATEn = VEE + 1V IGATEn = – 1µA (Note 3) ● ● 10 50 13 VOUTn – VEE 0V > VOUTn > –10V –10V > VOUTn > –30V VOUTn = –48V ● 1 2 ● ● 15 3 –6 –18 V µA µA µA mV mV mV mV mV mV µA –20 VSENSEn – VEE, VOUTn = VEE (Note 9) VSENSEn – VEE, VOUTn = VEE VSENSEn – VEE, VOUTn = AGND – 30V VSENSEn – VEE, VOUTn = AGND – 10V VSENSEn – VEE 166 201 201 30.2 2.52 187.5 212.5 199 224 224 3.75 275 –50 4.97 200 –2.7 2.7 500 –3 3 190 180 230 230 VSENSEn = VEE 0.1V < VOSCIN < 3V, fOSCIN < 200Hz Port Powered, PD Not Present Port Powered, PD Not Present Port Powered, –6V < VDETECTn < 0V Port Powered, VDETECTn = –3.4V (C grade) (I grade) ● ISDAOUT = 3mA, IINT = 3mA ISDAOUT = 5mA, IINT = 5mA SCL, SDAIN, RESET, SHDNn, AUTO, ADn SCL, SDAIN, RESET, SHDNn, AUTO, ADn ADn, RESET, SHDNn AUTO ● ● From Detect Command or Application of PD to Port to Detect Complete Time to Measure PD Signature Resistance (Figure 2) From Successful Detect in Auto or Semiauto Mode to Class Complete From Classify Command in Manual Mode (Figure 2) (Figure 2) From Valid Detect to Port On in Auto Mode (Figure 2) From Port On Command to GATE Pin Current = IGON (Note 10) ● ● ● ● ● ● ● V V V V kΩ kΩ 170 590 ms 170 10.1 230 52 ms ms 10.1 10.1 420 13 130 1 ms ms ms ms 2.4 50 50 ● ● ● ● ● kΩ V/V V/V µA µA 0.4 0.7 0.8 ● ● –3.3 3.3 ±600 260 260 4259a1fa 3 LTC4259A-1 ELECTRICAL CHARACTERISTICS The ● denotes the specifications which apply over the full operating temperature range, otherwise specifications are at TA = 25°C. AGND = DGND = 0V, VDD = 3.3V, VEE = –48V unless otherwise noted (Note 6). SYMBOL tSTART PARAMETER Maximum Current Limit Duration During Port Start-Up tICUT Maximum Current Limit Duration After Port Start-Up DCCLMAX tDIS Maximum Current Limit Duty Cycle Disconnect Delay tVMIN DC Disconnect Minimum Pulse Width Sensitivity CONDITIONS tSTART1 = 0, tSTART0 = 0 (Figure 3) tSTART1 = 0, tSTART0 = 1 (Figure 3) tSTART1 = 1, tSTART0 = 0 (Figure 3) tSTART1 = 1, tSTART0 = 1 (Figure 3) tICUT1 = 0, tICUT0 = 0 (Figure 3) tICUT1 = 0, tICUT0 = 1 (Figure 3) tICUT1 = 1, tICUT0 = 0 (Figure 3) tICUT1 = 1, tICUT0 = 1 (Figure 3) Reg16h = 00h tDIS1 = 0, tDIS0 = 0 (Figures 4, 5) tDIS1 = 0, tDIS0 = 1 (Figures 4, 5) tDIS1 = 1, tDIS0 = 0 (Figures 4, 5) tDIS1 = 1, tDIS0 = 1 (Figures 4, 5) VSENSEn – VEE > 5mV, VOUTn = –48V (Figure 4) (Note 11) Clock Frequency Bus Free Time Start Hold Time SCL Low Time SCL High Time Data Hold Time Data Set-Up Time Start Set-Up Time Stop Set-Up Time SCL, SDAIN Rise Time SCL, SDAIN Fall Time Fault Present to INT Pin Low Stop Condition to INT Pin Low ARA to INT Pin High Time (Note 11) Figure 6 (Notes 11, 12) Figure 6 (Notes 11, 12) Figure 6 (Notes 11, 12) Figure 6 (Notes 11, 12) Figure 6 (Notes 11, 12) Figure 6 (Notes 11, 12) Figure 6 (Notes 11, 12) Figure 6 (Notes 11, 12) Figure 6 (Notes 11, 12) Figure 6 (Notes 11, 12) (Notes 11, 12, 13) (Notes 11, 12, 13) (Notes 11, 12) I2C Timing fSCLK t1 t2 t3 t4 t5 t6 t7 t8 tr tf tFLTINT tSTOPINT tARAINT Note 1: Stresses beyond those listed under Absolute Maximum Ratings may cause permanent damage to the device. Exposure to any Absolute Maximum Rating condition for extended periods may affect device reliability and lifetime. Note 2: DGND and AGND should be tied together in normal operation. Note 3: An internal clamp limits the GATE pins to a minimum of 12V above VEE. Driving this pin beyond the clamp may damage the part. Note 4: When a port powers on or off, the transient voltage on the port couples through CDET (Figure 18). The LTC4259A-1 contains internal protection circuitry to withstand transient currents of up to 80mA for 5ms. As long as the absolute value of the current remains below 80mA, the LTC4259A-1 will keep the voltage at the DETECTn pin within the absolute maximum voltage range. A properly sized RDET should limit the current to less than 60mA. Note 5: This IC includes overtemperature protection that is intended to protect the device during momentary overload conditions. Junction temperature will exceed 125°C when overtemperature protection is active. Continuous operation above the specified maximum operating junction temperature may impair device reliability. ● ● ● ● ● ● ● ● ● ● ● ● ● MIN 50 25 100 200 50 25 100 200 5.8 300 75 150 600 ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● 1.3 600 1.3 600 150 200 600 600 20 20 20 60 20 TYP 60 30 120 240 60 30 120 240 6.3 360 90 180 720 0.02 MAX 70 35 140 280 70 35 140 280 6.7 400 100 200 800 1 UNITS ms ms ms ms ms ms ms ms % ms ms ms ms ms 400 kHz µs ns µs ns ns ns ns ns ns ns ns ns ns 300 150 150 200 300 Note 6: All currents into device pins are positive; all currents out of device pins are negative. All voltages are referenced to ground (AGND and DGND) unless otherwise specified. Note 7: The LTC4259A-1 is designed to maintain a port voltage of –46.6V to –57V. The VEE supply voltage range accounts for the drop across the diode, MOSFET and sense resistor. Note 8: VEE supply current, while classifying a short, is measured indirectly by measuring the DETECTn pin current while classifying a short. Note 9: The LTC4259A-1 implements overload current detection per IEEE 802.3af. The minimum overload current (ICUT) is dependent on port voltage; ICUT_MIN = 15.4W/VPORT_MIN. An IEEE compliant system using the LTC4259A-1 should maintain port voltage above –46.6V. Note 10: Unless otherwise specified, AC disconnect specifications require the following conditions: the DETECT pin is connected to the port as shown in Figure 1, a valid sine wave is applied to OSCIN, the OSCFAIL bit is cleared and the AC Disconnect Enable bits are set. Note 11: Guaranteed by design, not subject to test. Note 12: Values measured at VILD and VIHD. Note 13: If fault occurs during an I2C transaction, the INT pin will not be pulled down until a stop condition is present on the I2C bus. 4259a1fa 4 LTC4259A-1 U W TYPICAL PERFOR A CE CHARACTERISTICS Power On Sequence in Auto Mode PORT 1 VDD = 3.3V VEE = –48V Powering On a 180µF Load POWER ON PORT VOLTAGE 20V/DIV GND PORT VOLTAGE 10V/DIV VDD = 3.3V VEE = –48V GND VEE DETECTION DETECTION PHASE 1 PHASE 2 CLASSIFICATION VEE GATE +14V VOLTAGE 10V/DIV VEE PORT CURRENT 0mA 500mA/DIV VEE FET ON FOLDBACK 50ms/DIV LOAD FULLY CHARGED 425mA CURRENT LIMIT 5ms/DIV 4259 G01 4259 G02 INT and SDAOUT Pull Down Voltage vs Load Current 225 450 2.0 200 400 1.8 175 350 150 300 125 250 100 200 75 150 50 100 25 VDD = 3.3V VEE = –48V TA = 25°C 0 –48 PULL-DOWN VOLTAGE (V) ILIMIT WITH RSENSE = 0.5Ω (mA) VSENSEn (mV) Current Limit Foldback –32 –24 –16 –8 1.6 1.4 1.2 1.0 0.8 0.6 0.4 50 0.2 0 0 –40 VDD = 3.3V TA = 25°C 0 5 0 VOUTn-AGND (V) 15 20 10 LOAD CURRENT (mA) 4259 G04 4259 G03 Classification Transient Response to 40mA Load Step –18V 40mA PORT CURRENT 20mA/DIV 0mA 0 VDD = 3.3V –2 VEE = –48V T = 25°C –4 A –8 –10 –12 –14 PORT VOLTAGE WITH TYPICAL CMPD3003 –16 0 2.0 1.5 1.0 0.5 DETECTn PIN VOLTAGE –20 4258 G05 2.5 –6 –18 50µs/DIV 3.0 SUPPLY CURRENT (mA) PORT VOLTAGE 1V/DIV VEE DC Supply Current vs Supply Voltage Classification Current Compliance CLASSIFICATION VOLTAGE (V) VDD = 3.3V VEE = –48V TA = 25°C 25 10 20 30 40 50 60 CLASSIFICATION CURRENT (mA) 70 4258 G06 VDD = 3.3V REG 12h = 00h 0 –70 –60 –50 –40 –30 –20 –10 VEE SUPPLY VOLTAGE (V) 0 4258 G07 4259a1fa 5 LTC4259A-1 UW TEST TI I G PD INSERTED VPORTn 0V tDET VCLASS VGATEn VT VEE PORT TURN ON (AUTO MODE) INT tCLASS tCLSDLY 4259A F02 tDETDLY tPON Figure 2. Detect, Class and Turn-On Timing in Auto or Semiauto Modes VLIM VCUT VSENSEn TO VEE 0V tSTART, tICUT INT 4259A F03 Figure 3. Current Limit Timing VOSCIN VOUTn VSENSEn TO VEE VMIN IACDMIN IDETECTn PD REMOVED INT tVMIN tDIS INT 4259A F04 tDIS Figure 4. DC Disconnect Timing 4259A F05 Figure 5. AC Disconnect Timing t3 tr t4 tf SCL t2 t5 t6 t7 t8 SDA t1 4259A F06 Figure 6. I2C Interface Timing 4259a1fa 6 LTC4259A-1 W UW TI I G DIAGRA S SCL SDA 0 1 AD3 AD2 AD1 AD0 R/W ACK A7 0 START BY MASTER A6 A5 A4 A3 A2 A1 ACK BY SLAVE FRAME 1 SERIAL BUS ADDRESS BYTE A0 ACK D7 D6 D5 D4 D3 D2 ACK BY SLAVE D1 D0 ACK STOP BY MASTER ACK BY SLAVE FRAME 2 REGISTER ADDRESS BYTE FRAME 3 DATA BYTE 4259A F07 Figure 7. Writing to a Register SCL SDA 0 1 0 AD3 AD2 AD1 AD0 R/W ACK A7 START BY MASTER A6 A5 A4 A3 A2 FRAME 1 SERIAL BUS ADDRESS BYTE A0 ACK A1 ACK BY SLAVE 0 1 0 AD3 AD2 AD1 AD0 R/W ACK D7 REPEATED START BY MASTER ACK BY SLAVE D6 FRAME 1 SERIAL BUS ADDRESS BYTE FRAME 2 REGISTER ADDRESS BYTE D5 D4 D3 ACK BY SLAVE D2 D1 D0 ACK NO ACK BY MASTER FRAME 2 DATA BYTE STOP BY MASTER 4259A F08 Figure 8. Reading from a Register SCL SDA 0 1 0 AD3 AD2 AD1 AD0 R/W ACK D7 START BY MASTER D6 D5 D4 D3 ACK BY SLAVE D2 D1 D0 ACK STOP BY MASTER NO ACK BY MASTER FRAME 1 SERIAL BUS ADDRESS BYTE FRAME 2 DATA BYTE 4259A F09 Figure 9. Reading the Interrupt Register (Short Form) SCL SDA 0 0 0 1 1 0 0 R/W ACK START BY MASTER FRAME 1 ALERT RESPONSE ADDRESS BYTE 0 1 ACK BY SLAVE 0 AD3 AD2 AD1 AD0 1 NO ACK BY MASTER FRAME 2 SERIAL BUS ADDRESS BYTE ACK STOP BY MASTER 4259A F10 Figure 10. Reading from Alert Response Address 4259a1fa 7 LTC4259A-1 U U U PI FU CTIO S RESET (Pin 1): Chip Reset, Active Low. When the RESET pin is low, the LTC4259A-1 is held inactive with all ports off and all internal registers reset to their power-up states. When RESET is pulled high, the LTC4259A-1 begins normal operation. RESET can be connected to an external capacitor or RC network to provide a power turn-on delay. Internal filtering of the RESET pin prevents glitches less than 1µs wide from resetting the LTC4259A-1. Pull RESET high with ≤10k or tie to VDD. BYP (Pin 2): Bypass Output. The BYP pin is used to connect the internally generated – 20V supply to an external 0.1µF bypass capacitor. Use a 100V rated 0.1µF, X7R capacitor. Do not connect the BYP pin to any other external circuitry. INT (Pin 3): Interrupt Output, Open Drain. INT will pull low when any one of several events occur in the LTC4259A-1. It will return to a high impedance state when bits 6 or 7 are set in the Reset PB register (1Ah). The INT signal can be used to generate an interrupt to the host processor, eliminating the need for continuous software polling. Individual INT events can be disabled using the Int Mask register (01h). See Register Functions and Applications Information for more information. The INT pin is only updated between I2C transactions. SCL (Pin 4): Serial Clock Input. High impedance clock input for the I2C serial interface bus. The SCL pin should be connected directly to the I2C SCL bus line. SDAOUT (Pin 5): Serial Data Output, Open Drain Data Output for the I2C Serial Interface Bus. The LTC4259A-1 uses two pins to implement the bidirectional SDA function to simplify optoisolation of the I2C bus. To implement a standard bidirectional SDA pin, tie SDAOUT and SDAIN together. See Applications Information for more information. SDAIN (Pin 6): Serial Data Input. High impedance data input for the I2C serial interface bus. The LTC4259A-1 uses two pins to implement the bidirectional SDA function to simplify optoisolation of the I2C bus. To implement a standard bidirectional SDA pin, tie SDAOUT and SDAIN together. See Applications Information for more information. AD3 (Pin 7): Address Bit 3. Tie the address pins high or low to set the I2C serial address to which the LTC4259A-1 responds. This address will be (010A3A2A1A0)b. Pull AD3 high or low with ≤10k or tie to VDD or DGND. AD2 (Pin 8): Address Bit 2. See AD3. AD1 (Pin 9): Address Bit 1. See AD3. AD0 (Pin 10): Address Bit 0. See AD3. DETECT1 (Pin 11): Detect Sense, Port 1. The LTC4259A1 Powered Device (PD) detection, classification and AC disconnect hardware monitors port 1 with this pin. Connect DETECT1 to the output port via a 0.47µF 100V X7R capacitor in series with a 1k resistor, both in parallel with a low leakage diode (see Figure 1). The resistor and capacitor may be eliminated if AC disconnect is not used. If the port is unused, the DETECT1 pin can be tied to DGND or allowed to float. DETECT2 (Pin 12): Detection Sense, Port 2. See DETECT1. DETECT3 (Pin 13): Detection Sense, Port 3. See DETECT1. DETECT4 (Pin 14): Detection Sense, Port 4. See DETECT1. DGND (Pin 15): Digital Ground. DGND should be connected to the return from the 3.3V supply. DGND and AGND should be tied together. VDD (Pin 16): Logic Power Supply. Connect to a 3.3V power supply relative to DGND. VDD must be bypassed to DGND near the LTC4259A-1 with at least a 0.1µF capacitor. SHDN1 (Pin 17): Shutdown Port 1, Active Low. When pulled low, SHDN1 shuts down port 1, regardless of the state of the internal registers. Pulling SHDN1 low is equivalent to setting the Reset Port 1 bit in the Reset Pushbutton register (1Ah). Internal filtering of the SHDN1 pin prevents glitches less than 1µs wide from reseting the LTC4259A-1. Pull SHDN1 high with ≤10k or tie to VDD. SHDN2 (Pin 18): Shutdown Port 2, Active Low. See SHDN1. SHDN3 (Pin 19): Shutdown Port 3, Active Low. See SHDN1. SHDN4 (Pin 20): Shutdown Port 4, Active Low. See SHDN1. 4259a1fa 8 LTC4259A-1 U U U PI FU CTIO S AGND (Pin 21): Analog Ground. AGND should be connected to the return from the – 48V supply. AGND and DGND should be tied together. SENSE4 (Pin 22): Port 4 Current Sense Input. SENSE4 monitors the external MOSFET current via a 0.5Ω sense resistor between SENSE4 and VEE. Whenever the voltage across the sense resistor exceeds the overcurrent detection threshold VCUT, the current limit fault timer counts up. If the voltage across the sense resistor reaches the current limit threshold VLIM (typically 25mV/50mA higher), the GATE4 pin voltage is lowered to maintain constant current in the external MOSFET. See Applications Information for further details. If the port is unused, the SENSE4 pin must be tied to VEE. GATE4 (Pin 23): Port 4 Gate Drive. GATE4 should be connected to the gate of the external MOSFET for port 4. When the MOSFET is turned on, a 50µA pull-up current source is connected to the pin. The gate voltage is clamped to 13V (typ) above VEE. During a current limit condition, the voltage at GATE4 will be reduced to maintain constant current through the external MOSFET. If the fault timer expires, GATE4 is pulled down with 50µA, turning the MOSFET off and recording a tICUT or tSTART event. If the port is unused, float the GATE4 pin or tie it to VEE. OUT4 (Pin 24): Port 4 Output Voltage Monitor. OUT4 should be connected to the output port through a 10k series resistor. A current limit foldback circuit limits the power dissipation in the external MOSFET by reducing the current limit threshold when the port voltage is within 18V of AGND. The port 4 Power Good bit is set when the voltage from OUT4 to VEE drops below 2V (typ). A 2.5MΩ resistor is connected internally from OUT4 to AGND. If the port is unused, the OUT4 pin can be tied to AGND or allowed to float. SENSE3 (Pin 25): Port 3 Current Sense Input. See SENSE4. GATE3 (Pin 26): Port 3 Gate Drive. See GATE4. OUT3 (Pin 27): Port 3 Output Voltage Monitor. See OUT4. VEE (Pin 28): – 48V Supply Input. Connect to a – 48V to – 57V supply, relative to AGND. SENSE2 (Pin 29): Port 2 Current Sense Input. See SENSE4. GATE2 (Pin 30): Port 2 Gate Drive. See GATE4. OUT2 (Pin 31): Port 2 Output Voltage Monitor. See OUT4. SENSE1 (Pin 32): Port 1 Current Sense Input. See SENSE4. GATE1 (Pin 33): Port 1 Gate Drive. See GATE 4. OUT1 (Pin 34): Port 1 Output Voltage Monitor. See OUT4. AUTO (Pin 35): Auto Mode Input. Auto mode allows the LTC4259A-1 to detect and power up a PD even if there is no host controller present on the I2C bus. The voltage of the AUTO pin determines the state of the internal registers when the LTC4259A-1 is reset or comes out of VDD UVLO (see the Register map in Table 1). The states of these register bits can subsequently be changed via the I2C interface. The real-time state of the AUTO pin is read at bit 0 in the Pin Status register (11h). Pull AUTO high or low with ≤10k or tie to VDD or DGND. OSCIN (Pin 36): Oscillator Input. Connect to an oscillating signal source, preferably a sine wave, of approximately 100Hz with 2V peak-to-peak amplitude, negative peaks above – 0.3V and positive peaks below 2.5V. When a port is powered and AC disconnect is enabled, this signal is amplified and driven onto the appropriate DETECT pin to determine the AC impedance of the PD. 4259a1fa 9 10 RO RO RO RO RO 0Eh Port 3 Status 0Fh Port 4 Status 10h Power Status 11h Pin Status R/W R/W R/W R/W R/W 14h Detect/Class Enable 15h Reserved 16h Timing Config 17h Misc Config WO 1Ah Reset PB Change 4 Pwr Good Change 3 Pwr Good Mask 6 Change 2 Pwr Good Mask 5 Change 1 Pwr Good Mask 4 Class Complete Change 4 Pwr Enable Mask 3 Detect Complete BIT 3 Change 3 Pwr Enable Mask 2 Disconnect BIT 2 Change 2 Pwr Enable Mask 1 Pwr Good Event BIT 1 Change 1 Pwr Enable Mask 0 Pwr Enable Event BIT 0 Global 4321 4321 Global Global 4321 4321 4321 Global 4321 4 3 2 1 Global 4321 4321 Clear All Interrupts Power Off 4 Restart Class 4 Interrupt Pin Enable Reserved Reserved Class Enable 4 AC Discon En 4 Port 4 Mode 1 Reserved Power Good 4 Reserved Reserved Reserved Reserved Over Temp Reserved Disconnect 4 Reserved Power Off 2 Restart Class 2 Osc Fail Mask tSTART1 Reserved Class Enable 2 AC Discon En 2 Port 3 Mode 1 AD3 Pin Status Power Good 2 Class Status 1 Class Status 1 Class Status 1 Class Status 1 VDD UVLO Reserved Disconnect 2 Power Good 1 Class Status 0 Class Status 0 Class Status 0 Class Status 0 VEE UVLO Reserved Disconnect 1 Reset Port 4 Power On 4 Restart Detect 4 Reserved tICUT1 Reserved Detect Enable 4 DC Discon En 4 Port 2 Mode 1 AD1 Pin Status Power Enable 4 Reserved Reserved Reserved Reserved Reserved tSTART Fault 4 tICUT Fault 4 Reset Port 3 Power On 3 Restart Detect 3 Reserved tICUT0 Reserved Detect Enable 3 DC Discon En 3 Port 2 Mode 0 AD0 Pin Status Power Enable 3 Detect Status 2 Detect Status 2 Detect Status 2 Detect Status 2 Reserved tSTART Fault 3 tICUT Fault 3 Power Off, Detection and Class Off Will Not Advance Between States Detect and Class But Wait to Turn On Power Detect, Class and Power Automatically Reset All Power Off 1 Restart Class 1 Reserved tSTART0 Reserved Class Enable 1 AC Discon En 1 Port 3 Mode 0 AD2 Pin Status MODE BIT ENCODING 00 Shutdown 01 Manual 10 Semiauto 11 Auto Clear Interrupt Pin Power Off 3 Restart Class 3 Reserved Reserved Reserved Class Enable 3 AC Discon En 3 Port 4 Mode 0 Reserved Power Good 3 Class Status 2 Class Status 2 Class Status 2 Class Status 2 Reserved Reserved Disconnect 3 RESET STATE 0000,0000 0000,0000 0000,0000 1010,0000 0000,0000 0000,0000 0000,0000 0000,0000 0000,0000 00A3A2,A1A000 0000,0000 0000,0000 0000,0000 0000,0000 0000,0000 0011,0010* 0000,0000 0000,0000 0000,0000 0000,0000 1000,0000 0000,0000 0000,0000 0000,0000 1010,0000 0000,0000 0000,0000 1111,1111 1111,0000 1111,1111 00A3A2,A1A001 0000,0000 0000,0000 0000,0000 0000,0000 0000,0000 0011,0010* 0000,0000 0000,0000 0000,0000 0000,0000 1110,0100 1000,0000 Auto Pin High RESET STATE * The start-up state of the VEE UVLO and Osc Fail bits depend on the order in which the VDD and VEE supplies are brought up. The VDD UVLO bit is not set by the RESET pin or the reset all push button. Reset Port 1 Power On 1 Restart Detect 1 Reserved tDIS0 Reserved Detect Enable 1 DC Discon En 1 Port 1 Mode 0 Auto Pin Status Power Enable 1 Detect Status 0 Detect Status 0 Detect Status 0 Detect Status 0 Reserved tSTART Fault 1 tICUT Fault 1 Key: RO = Read Only R/W = Read/Write CoR = Clear on Read WO = Write Only Reset Port 2 Power On 2 Restart Detect 2 Reserved tDIS1 Reserved Detect Enable 2 DC Discon En 2 Port 1 Mode 1 Reserved Power Enable 2 Detect Status 1 Detect Status 1 Detect Status 1 Detect Status 1 Osc Fail tSTART Fault 2 tICUT Fault 2 4321 Class Complete 4 Class Complete 3 Class Complete 2 Class Complete 1 Detect Complete 4 Detect Complete 3 Detect Complete 2 Detect Complete 1 4321 Mask 7 Encoding CLASS STATUS DETECT STATUS 000 Class Status Unknown 000 Detect Status Unknown 001 Class 1 001 Short Circuit (<1V) 010 Class 2 010 Reserved 011 Class 3 011 RLOW 100 Class 4 100 Detect Good 101 Undefined—Read as Class 0 101 RHIGH 110 Class 0 110 Open Circuit 111 Overcurrent 111 Reserved WO WO 18h Det/Class Restart PB 19h Power Enable PB Pushbuttons R/W 12h Operating Mode 13h Disconnect Enable Configuration RO 0Ch Port 1 Status 0Dh Port 2 Status Status RO CoR 0Ah Supply Event 0Bh Supply Event CoR RO CoR 08h tSTART Event 09h tSTART Event CoR RO CoR 06h Fault Event 07h Fault Event CoR RO CoR 04h Detect Event 05h Detect Event CoR RO CoR 02h Power Event 03h Power Event CoR Events Global tICUT Fault BIT 4 R/W tSTART Fault BIT 5 01h Int Mask Supply Event BIT 6 1000,0000 Global BIT 7 RO PORT 00h Interrupt R/W Auto Pin Low ADDRESS REGISTER NAME TABLE 1. REGISTER W Interrupts LTC4259A-1 AP 4259a1fa LTC4259A-1 U U REGISTER FU CTIO S Interrupt Registers Interrupt (Address 00h): Interrupt Register, Read Only. A transition to logical 1 of any bit in this register will assert the INT pin (Pin 3) if the corresponding bit in the Int Mask register is set. Each bit is the logical OR of the corresponding bits in the Event registers. The Interrupt register is Read Only and its bits cannot be cleared directly. To clear a bit in the Interrupt register, clear the corresponding bits in the appropriate Status or Event registers or set bit 7 in the Reset Pushbutton register (1Ah). Int Mask (Address 01h): Interrupt Mask, Read/Write. A logic 1 in any bit of the Int Mask register allows the corresponding Interrupt register bit to assert the INT pin if it is set. A logic 0 in any bit of the Int Mask register prevents the corresponding Interrupt bit from affecting the INT pin. The actual Interrupt register bits are unaffected by the state of the Int Mask register. Event Registers Power Event (Address 02h): Power Event Register, Read Only. The lower four bits in this register indicate that the corresponding port Power Enable status bit has changed; the logical OR of these four bits appears in the Interrupt register as the Pwr Enable Event bit. The upper four bits indicate that the corresponding port Power Good status bit has changed; the logical OR of these four bits appears in the Interrupt register as the Pwr Good Event bit. The Power Event bits latch high and will remain high until cleared by reading from address 03h. Power Event CoR (Address 03h): Power Event Register, Clear on Read. Read this address to clear the Power Event register. Address 03h returns the same data as address 02h and reading address 03h clears all bits at both addresses. Detect Event (Address 04h): Detect Event Register, Read Only. The lower four bits in this register indicate that at least one detection cycle for the corresponding port has completed; the logical OR of these four bits appears in the Interrupt register as the Detect Complete bit. The upper four bits indicate that at least one classification cycle for the corresponding port has completed; the logical OR of these four bits appears in the Interrupt register as the Class Complete bit. In Manual mode, this register indicates that the requested detection/classification cycle has completed and the LTC4259A-1 is awaiting further instructions. In Semiauto or Auto modes, these bits indicate that the Detect Status and Class Status bits in the Port Status registers are valid. The Detect Event bits latch high and will remain high until cleared by reading from address 05h. Detect Event CoR (Address 05h): Detect Event Register, Clear on Read. Read this address to clear the Detect Event register. Address 05h returns the same data as address 04h, and reading address 05h clears all bits at both addresses. Fault Event (Address 06h): Fault Event Register, Read Only. The lower four bits in this register indicate that a tICUT fault has occurred at the corresponding port; the logical OR of these four bits appears in the Interrupt register as the tICUT Fault bit. The upper four bits indicate that a Disconnect event has occurred at the corresponding port; the logical OR of these four bits appears in the Interrupt register as the Disconnect bit. The Fault Event bits latch high and will remain high until cleared by reading from address 07h. Fault Event CoR (Address 07h): Fault Event Register, Clear on Read. Read this address to clear the Fault Event register. Address 07h returns the same data as address 06h and reading address 07h clears all bits at both addresses. tSTART Event (Address 08h): tSTART Event Register, Read Only. The lower four bits in this register indicate that a tSTART fault has occurred at the corresponding port; the logical OR of these four bits appears in the Interrupt register as the tSTART Fault bit. The tSTART Event bits latch high and will remain high until cleared by reading from address 09h. The upper four bits in this register are reserved and will always read as 0. tSTART Event CoR (Address 09h): tSTART Event Register, Clear on Read. Read this address to clear the Fault Event register. Address 09h returns the same data as address 08h and reading address 09h clears all bits at both addresses. Supply Event (Address 0Ah): Supply Event Register, Read Only. Bit 1, Osc Fail, sets when the signal at Pin 36, OSCIN, is absent or does not have the required amplitude and AC disconnect cannot operate properly. The Osc Fail bit latches high and will remain high until cleared by reading at 0Bh. The Osc Fail bit is set after power on or reset unless the VEE supply is not present. Power is removed on ports with AC 4259a1fa 11 LTC4259A-1 U U REGISTER FU CTIO S disconnect enabled independently of the state of the Osc Fail bit. See AC Disconnect under Applications Information for more details. Bit 4 indicates that VEE has dropped below the VEE UVLO level (typically –28V). Bit 5 signals that the VDD supply has dropped below the VDD UVLO threshold. Bit 7 indicates that the LTC4259A-1 die temperature has exceeded its thermal shutdown limit (see Note 5 under Electrical Characteristics). The logical OR of bits 1, 4, 5 and 7 appears in the Interrupt register as the Supply Fault bit. See the Misc Config register for information on masking the Osc Fail bit out of the Supply Fault interrupt. The remaining bits in the register are reserved and will always read as 0. The Supply Event bits latch high and will remain high until cleared by reading from address 0Bh. power good bits are latched high and are only cleared when a port is turned off or the LTC4259A-1 is reset. Supply Event CoR (Address 0Bh): Supply Event Register, Clear on Read. Read this address to clear the Fault Event register. Address 0Bh returns the same data as address 0Ah, and reading address 0Bh clears all bits at both addresses. Operating Mode (Address 12h): Operating Mode Configuration, Read/Write. This register contains the mode bits for each of the four ports in the LTC4259A-1. See Table 1 for mode bit encoding. At power-up, all bits in this register will be set to the logic state of the AUTO pin (Pin 35). See Operating Modes in the Applications Information section. Status Registers Port 1 Status (Address 0Ch): Port 1 Status Register, Read Only. This register reports the most recent detection and classification results for port 1. Bits 0-2 report the status of the most recent detection attempt at the port and bits 4-6 report the status of the most recent classification attempt at the port. If power is on, these bits report the detection/ classification status present just before power was turned on. If power is turned off at the port for any reason, all bits in this register will be cleared. See Table 1 for detection and classification status bit encoding. Port 2 Status (Address 0Dh): Port 2 Status Register, Read Only. See Port 1 Status. Port 3 Status (Address 0Eh): Port 3 Status Register, Read Only. See Port 1 Status. Port 4 Status (Address 0Fh): Port 4 Status Register, Read Only. See Port 1 Status. Power Status (Address 10h): Power Status Register, Read Only. The lower four bits in this register report the switch on/off state for the corresponding ports. The upper four bits (the power good bits) indicate that the drop across the power switch and sense resistor for the corresponding ports is less than 2V (typ) and power start-up is complete. The Pin Status (Address 11h): External Pin Status, Read Only. This register reports the real time status of the AUTO (Pin 35) and AD0-AD3 (Pins 7-10) digital input pins. The logic state of the AUTO pin appears at bit 0 and the AD0-AD3 pins at bits 2-5. The remaining bits are reserved and will read as 0. AUTO affects the initial states of some of the LTC4259A-1 configuration registers at start-up but has no effect after start-up and can be used as a general purpose input if desired, as long as it is guaranteed to be in the appropriate state at start-up. Configuration Registers Disconnect Enable (Address 13h): Disconnect Enable Register, Read/Write. The lower four bits of this register enable or disable DC disconnect detection circuitry at the corresponding port. If the DC Discon Enable bit is set the port circuitry will turn off power if the current draw at the port falls below IMIN for more than tDIS. IMIN is equal to VMIN/ RS, where RS is the sense resistor and should be 0.5Ω for IEEE 802.3af compliance. If the bit is clear the port will not remove power due to low current. The upper four bits enable or disable AC disconnect on the corresponding port. When a port’s AC disconnect bit is set, the LTC4259A-1 senses the impedance of that port by forcing an AC voltage on the port’s DETECT pin and measuring the AC current. If the DETECT pin sinks less than IACDMIN for more than tDIS, the port will turn off power. If the bit is clear, the port will not remove power due to high port impedance (AC current below IACDMIN). The DC and AC disconnect signals that reset tDIS are ORed together and either sensing method (if they are both enabled) will keep the port powered. A port with neither DC or AC disconnect enabled will not power off automatically when the PD is removed. 4259a1fa 12 LTC4259A-1 U U REGISTER FU CTIO S Detect/Class Enable (Address 14h): Detection and Classification Enable, Read/Write. The lower four bits of this register enable the detection circuitry at the corresponding port if that port is in Auto or Semiauto mode. The upper four bits enable the classification circuitry at the corresponding port if that port is in Auto or Semiauto mode. In manual mode, setting a bit in this register will cause the LTC4259A-1 to perform one classification or detection cycle on the corresponding port. Writing to the Detect/Class Restart PB (18h) has the same effect without disturbing the Detect/Class Enable bits for other ports. Timing Config (Address 16h): Global Timing Configuration, Read/Write. Bits 0-1 program tDIS, the time duration before a port is automatically tuned off after the PD is removed. The LTC4259A-1 can be programmed to monitor whether port current is below IMIN (DC connect) or port impedance is high (AC disconnect). Bits 2-3 program tICUT, the time during which a port’s current can exceed ICUT without it being turned off. If the current is still above ICUT after tICUT, the LTC4259A-1 will indicate a tICUT fault and turn the port off. Bits 4-5 program tSTART, the time duration before an overcurrent condition during port power-on is considered a tSTART fault and the port is turned off. Note that using the tICUT and tSTART times other than the default is not compliant with IEEE 802.3af and may double or quadruple the energy dissipated by the external MOSFETs during fault conditions. Bits 6-7 are reserved and should be read/written as 0. See Electrical Characteristics for timer bit encoding. Also see the Applications Information for descriptions of tSTART, tICUT, DC and AC disconnect timing. Misc Config (Address 17h): Miscellaneous Configuration, Read/Write. Bit 5 is the Osc Fail Mask; it is set by default. When the Osc Fail Mask bit is clear, it prevents a failure on the OSCIN pin from setting the Osc Fail bit and causing a Supply Event Interrupt. Setting bit 7 enables the INT pin. If this bit is reset, the LTC4259A-1 will not pull down the INT pin in any condition nor will it respond to the Alert Response Address. This bit is set by default. Pushbutton Registers Note Regarding Pushbutton Registers: “Pushbutton” registers are specialized registers that trigger an event when a 1 is written to a bit; writing a 0 to a bit will do nothing. Unlike a standard read/write register, where setting a single bit involves reading the register to determine its status, setting the appropriate bit in software and writing back the entire register, a pushbutton register allows a single bit to be written without knowing or affecting the status of the other bits in the register. Pushbutton registers are writeonly and will return 00h if read. Det/Class Restart PB (Address 18h): Detection/Classification Restart Pushbutton Register, Write Only. Writing a 1 to any bit in this register will start or restart a single detection or classification cycle at the corresponding port in Manual mode. It can also be used to set the corresponding bits in the Detect/Class Enable register (address 14h) for ports in auto or semiauto mode. The lower 4 bits affect detection on each port while the upper 4 bits affect classification. Power Enable PB (Address 19h): Power Enable Pushbutton Register, Write Only. The lower four bits of this register set the Power Enable bit in the corresponding Port Status register; the upper four bits clear the corresponding Power Enable bit. Setting or clearing the Power Enable bits via this register will turn on or off the power in any mode except shutdown, regardless of the state of detection or classification. Note that tICUT, tSTART and disconnect events (if enabled) will still turn off power if they occur. The Power Enable bit cannot be set if the port has turned off due to a tICUT or tSTART fault and the tICUT timer has not yet counted back to zero. See Applications Information for more information on tICUT timing. Clearing the Power Enable bits with this register also clears the detect and fault event bits, the Port Status register, and the Detection and Classification Enable bits for the affected port(s). Reset PB (Address 1Ah): Reset Pushbutton, Write Only. Bits 0-3 reset the corresponding port by clearing the power enable bit, the detect and fault event bits, the status register and the detection and classification enable bits for that port. Bit 4 returns the entire LTC4259A-1 to the power-on reset state; all ports are turned off, the AUTO pin is reread and all registers are returned to their power-on defaults, except VDD UVLO, which remains cleared. Bit 5 is reserved; setting it has no effect. Setting bit 6 releases the Interrupt pin if it is asserted without affecting the Event registers or the Interrupt register. When the INT pin is released in this 4259a1fa 13 LTC4259A-1 U U REGISTER FU CTIO S way, the condition causing the LTC4259A-1 to pull the INT pin down must be removed before the LTC4259A-1 will be able to pull INT down again. This can be done by reading and clearing the event registers or by writing a 1 into bit 7 of this register. Setting bit 7 releases the Interrupt pin, clears all the Event registers and clears all the bits in the Interrupt register. U W U U APPLICATIO S I FOR ATIO OVERVIEW Over the years, twisted-pair Ethernet has become the most commonly used method for local area networking. The IEEE 802.3 group, the originator of the Ethernet standard, has defined an extension to the standard, known as 802.3af, which allows DC power to be delivered simultaneously over the same cable used for data communication. This promises a whole new class of Ethernet devices, including IP telephones, wireless access points, and PDA charging stations, which do not require additional AC wiring or external power transformers, a.k.a. “wall warts.” With about 13W of power available, small data devices can be powered by their Ethernet connections, free from AC wall outlets. Sophisticated detection and power monitoring techniques prevent damage to legacy data-only devices, while still supplying power to newer, Ethernetpowered devices over the twisted-pair cable. A device that supplies power is called Power Sourcing Equipment (PSE); a device that draws power from the wire is called a Powered Device (PD). A PSE is typically an Ethernet switch, router, hub, or other network switching PSE RJ45 4 CAT 5 20Ω MAX ROUNDTRIP 0.05µF MAX 5 0.1µF 100V DGND BYP I2C A PSE is required to provide a nominal 48V DC between either the signal pairs or the spare pairs (but not both) as shown in Figure 11. The power is applied as a voltage between two of the pairs, typically by powering the centertaps of the isolation transformers used to couple the differential data signals to the wire. Since Ethernet data is transformer coupled at both ends and is sent differentially, a voltage difference between the transmit pairs and the receive pairs does not affect the data. A 10base-T/ 100base-TX Ethernet connection only uses 2 of the 4 pairs in the cable. The unused or spare pairs can be powered directly, as shown in Figure 11, without affecting the data. However, 1000base-T uses all 4 pairs and power must be connected to the transformer center taps if compatibility with 1000base-T is required. PD RJ45 4 5 1N4002 ×4 SPARE PAIR GND 3.3V INTERRUPT equipment that is commonly found in the wiring closets where cables converge. PDs can take many forms: digital IP telephones, wireless network access points, PDA or notebook computer docking stations, cell phone chargers, and HVAC thermostats are examples of devices that can draw power from the network. 0.1µF AGND CMPD3003 VDD DETECT INT 1/4 SCL LTC4259A-1 SDAIN SDAOUT VEE 1 3 0.5Ω –48V IRFM120A S1B 2 3 Rx SENSE GATE OUT 10k DATA PAIR 0.1µF Tx 6 DATA PAIR 6 5µF ≤ CIN ≤ 300µF SMAJ58A 58V Rx 2 1k 0.47µF 100V X7R 1 Tx 1N4002 ×4 GND RCLASS SMAJ58A 58V PWRGD LTC4257 7 7 8 8 –48VIN –48VOUT DC/DC CONVERTER + VOUT – SPARE PAIR 4259A F11 Figure 11. Power over Ethernet System Diagram 4259a1fa 14 LTC4259A-1 U W U U APPLICATIO S I FOR ATIO The LTC4259A-1 provides a complete solution for detection and powering of PD devices in an IEEE 802.3af compliant system. The LTC4259A-1 consists of four independent ports, each with the ability to detect, classify, and provide isolated –48V power to a PD device connected to it. The LTC4259A-1 senses removal of a PD with IEEE 802.3af compliant AC or DC methods and turns off –48V power when the PD is removed. An internal control circuit takes care of system configuration and timing, and uses an I2C interface to communicate with the host system. Regardless of which mode it is in, the LTC4259A-1 will remove power automatically from any port that generates a tSTART or tICUT overcurrent fault event (see tICUT Timing and tSTART Timing sections). It will also automatically remove power from any port that generates a disconnect event if the appropriate Disconnect Enable bit is set in the Disconnect Enable register. The host controller may also remove power at any time by setting the appropriate Power Off bit in the Power Enable PB register. Power-On RESET OPERATING MODES Each LTC4259A-1 port can operate in one of four modes: Manual, Semiauto, Auto or Shutdown. The operating mode for a port is set by the appropriate bits in the Operating Mode register. The LTC4259A-1 will power up with all ports in Shutdown mode if the external AUTO pin is tied low; if AUTO is high, all ports will wake up in Auto mode. The operating mode can be changed at any time via the I2C interface, regardless of the state of the AUTO pin. • In Manual mode, a port will wait for instructions from the host system before taking any action. It will run single detection or classification cycles when commanded, and will report results in the Port Status registers. When the host system decides it is time to turn on or off power to a port, it can do so by setting the appropriate Power On/Off bits in the Power Enable PB register regardless of the current status of detection or classification. • In Semiauto mode, the port will repeatedly attempt to detect and classify a PD device attached to the link. It will report this information in its Port Status register, and wait for the host system to set the appropriate Power On bit in the Power Enable PB register before applying power to the port. • In Auto mode, the port will detect and classify a PD device connected to it, then immediately turn on the power if detection was successful regardless of the result of classification. • In Shutdown mode, the port is disabled and will not detect or power a PD. Also, the detect and fault event bits, status bits and enable bits for the port are reset to zero. At turn-on or any time the LTC4259A-1 is reset (either by pulling the RESET pin low or writing to the global Reset All bit), all the ports turn off and all internal registers go to a predefined state, shown in Table 1. Several of the registers assume different states based on the state of the AUTO pin at reset. The default states with AUTO high allow the LTC4259A-1 to detect and power up a PD in Automatic mode, even if nothing is connected to the I2C interface. SIGNATURE DETECTION The IEEE defines a specific pair-to-pair PD signature resistance that identifies a device that can accept Power over Ethernet in accordance with the 802.3af specification. When the port voltage is below 10V, an 802.3af compliant PD will have a 25k signature resistance. Figure 12 illustrates the relationship between the PD signature resistance (white box from 23.75k to 26.25k) and required resistance ranges the PSE must accept (white box) and reject (gray boxes). According to the 802.3af specification, the PSE may or may not accept resistances in the two ranges of 15k to 19k and 26.5k to 33k. Note that the black box in Figure 12 represents the 150Ω pair-to-pair termination used in legacy 802.3 devices like a computer’s network interface card (NIC) that cannot accept power. RESISTANCE 0Ω PD PSE 20k 10k 150Ω (NIC) 15k 30k 23.75k 26.25k 19k 26.5k 33k 4259A F12 Figure 12. IEEE 802.3af Signature Resistance Ranges 4259a1fa 15 LTC4259A-1 U W U U APPLICATIO S I FOR ATIO The LTC4259A-1 checks for the signature resistance by forcing two test currents on the port (via the DETECTn pins) in sequence and measuring the resulting voltages. It then subtracts the two V-I points to determine the resistive slope while removing voltage offset caused by any series diodes or current offset caused by leakage at the port (see Figure 13). The LTC4259A-1 will typically accept any PD resistance between 17k and 29k as a valid PD and report Detect Good (100 binary) in the Detect Status bits (bits 2 through 0) of the corresponding Port Status register. Values outside this range, including open and short circuits, are also reported in the Detect Status bits. Refer to Table 1 for a complete decoding of the Detect Status bits. The first test point is taken by forcing a test current into the port, waiting a short time to allow the line to settle and measuring the resulting voltage. This result is stored and the second current is applied to the port, allowed to settle and the voltage measured. Each point takes about 100ms to measure, and an entire detection cycle takes 230ms (max). CURRENT (µA) 165 VALID PD 0V-2V OFFSET The port’s operating mode controls if and when the LTC4259A-1 runs a detection cycle. In manual mode, the port will sit idle until a Restart Detection (register 18h) command is received. It will then run a complete 200ms detection cycle on the selected port, report the results in the Detect Status bits in the corresponding Port Status register and return to idle until another command is received. In Semiauto mode, the LTC4259A-1 autonomously tests valid PDs connected to the ports but it will not apply power until instructed to do so by the host controller. It repeatedly queries the port every 320ms and updates the Detect Status bits at the end of each cycle. If a Detect Good is reported, it will advance to the classification phase and report that result in the Port Status register. Until instructed to do otherwise, the LTC4259A-1 will continue to repeat detection on the port. Behavior in Auto mode is similar to Semiauto; however, after a Detect Good is reported, the LTC4259A-1 performs the classification phase and then powers up the port without further intervention. The signature detection circuitry is disabled when the port is in Shutdown mode, powered up or the corresponding Detect Enable bit is cleared. 275 25kΩ SLOPE The LTC4259A-1 will not report Detect Good if the PD has more than 5µF in parallel with its signature resistor. FIRST DETECTION POINT CLASSIFICATION SECOND DETECTION POINT A PD has the option of presenting a “classification signature” to the PSE to indicate how much power it will draw when powered up. This signature consists of a specific constant current draw when the PSE port voltage is between 15.5V and 20.5V, with the current level indicating the power class to which the PD belongs. Per the IEEE 802.3af specification, the LTC4259A-1 identifies the five classes of PD VOLTAGE 4259A F13 Figure 13. PD Detection Table 2. IEEE 802.3af Powered Device Classes IEEE 802.3af CLASS CLASSIFICATION CURRENT AT PSE MAXIMUM PD POWER MINIMUM PSE OUTPUT POWER 0 0mA to 5mA 12.95W 15.4W 1 8mA to 13mA 3.84W 4W Low Power PD Medium Power PD CLASS DESCRIPTION PD Does Not Implement Classification, Unknown Power 2 16mA to 21mA 6.49W 7W 3 25mA to 31mA 12.95W 15.4W High or Full Power PD 4 35mA to 45mA 12.95W 15.4W Reserved, Power as Class O 4259a1fa 16 LTC4259A-1 U W U U APPLICATIO S I FOR ATIO listed in Table 2. During classification, the LTC4259A-1 controls and measures the port voltage through the DETECTn pin. Note that class 4 is presently specified by the IEEE as reserved for future use. Figure 14 shows a PD load line, starting with the shallow slope of the 25k signature resistor below 10V, then drawing the classification current (in this case, class 3) between 14.5V and 20.5V. The LTC4259A-1’s load line for classification is also shown in Figure 14. It has low impedance until current limit is reached at 55mA (min). The LTC4259A-1 will classify a port immediately after a successful detection cycle in Semiauto or Auto modes, or when commanded to in Manual mode. It measures the PD classification signature current by applying 18V (typ) to the port and measuring the resulting current. It reports the detected class in the Class Status bits in the corresponding Port Status register. Note that in Auto mode, the port will power up regardless of which class is detected. The classification circuitry is disabled when the port is in Shutdown mode, powered up, or the corresponding Class 60 PSE LOAD LINE OVER CURRENT 50 CURRENT (mA) 48mA 40 CLASS 4 30 CLASS 3 33mA 23mA 20 TYPICAL CLASS 3 PD LOAD LINE 10 0 0 5 controlled manner that satisfies the PD’s power needs while minimizing disturbances on the –48V backplane. Gate Currents Once the decision has been made to turn on power to a port, the LTC4259A-1 uses a 50µA current source to pull up on the GATE pin. Under normal power-up circumstances, the MOSFET gate will charge up rapidly to VT (the MOSFET threshold voltage), the MOSFET current will rise quickly to the current limit level and the GATE pin will be servoed to maintain the proper IINRUSH charging current. When output charging is complete, the MOSFET current will fall and the GATE pin will be allowed to continue rising to fully enhance the MOSFET and minimize its on resistance. The final VGS is nominally 13V. When a port is turned off, a 50µA current source pulls down on the GATE pin, turning the MOSFET off in a controlled manner. No External Capacitors No external capacitors are required on the GATE pins for active current limit stability, lowering part count and cost. This also allows the fastest possible turn-off under severe overcurrent conditions, providing maximum safety and protection for the MOSFETs, load devices and board traces. Connecting capacitors to the external MOSFET gates can adversely affect the LTC4259A-1’s ability to respond to a shorted port. CLASS 2 Inrush Control 14.5mA CLASS 1 CLASS 0 10 15 VOLTAGE (VCLASS) 6.5mA 25 20 4259A F14 Figure 14. PD Classification Enable bit is cleared. POWER CONTROL The primary function of the LTC4259A-1 is to control the delivery of power to the PSE port. It does this by controlling the gate drive voltage of an external power MOSFET while monitoring the current via a sense resistor and the output voltage at the OUT pin. This circuitry serves to couple the raw isolated –48V input supply to the port in a The 802.3af standard lists two separate maximum current limits, ILIM and IINRUSH. Because they have identical values, the LTC4259A-1 implements both as a single current limit using VLIM (described below). Their functions are differentiated through the use of tICUT and tSTART, respectively (see tICUT Timing and tSTART Timing sections). To maintain consistency with the standard, the IINRUSH term is used when referring to an initial tSTART power-up event. When the LTC4259A-1 turns on a port, it turns on the MOSFET by pulling up on the gate. The LTC4259A-1 is designed to power up the port in current limit, limiting the inrush current to IINRUSH. The port voltage will quickly rise to the point where the PD reaches its input turn-on threshold and begins to draw 4259a1fa 17 LTC4259A-1 U W U U APPLICATIO S I FOR ATIO current to charge its bypass capacitance, slowing the rate of port voltage increase. Dual-Level Current Limit A PD is permitted to draw up to 15.4W continuously and up to 400mA for 50ms. The LTC4259A-1 has two corresponding current limit thresholds, ICUT (375mA typ) and ILIM (425mA typ). These are given by the equations: ICUT = VCUT/RS, ILIM = VLIM/RS RS is the sense resistor and should be 0.5Ω for IEEE 802.3af compliance. While the LTC4259A-1 allows the port current to exceed ICUT for a limited time period (see tICUT timing below), it does not allow the current to exceed ILIM. The current limit circuit monitors the port current by monitoring the voltage across the sense resistor and reduces the MOSFET gate voltage as needed to keep the current at or below ILIM. When the current drops below ILIM, the gate voltage is restored to the full value to keep the MOSFET resistance to a minimum. enters current limit but then exits before the timer expires, the count will decrease slowly, giving the ICUT timer the ability to turn off sooner in the case of a repetitive fault. If the overcurrent duty cycle is less than 6.3% the tICUT timer will be fully reset. If the tICUT timer expires and causes the port to shut off, the timer will continue to run, counting down at the slow 1/16th rate and preventing the port from being repowered until the count returns to zero. This protects the MOSFET from damage due to a faulty PD that may still have a valid signature, or from errant software that repeatedly writes to the Power On bit. The port will not repower until after the tICUT counter returns to zero. In manual and semiauto modes the power enable command must be received after the tICUT counter reaches zero. In auto mode the LTC4259A-1 must complete a valid detection cycle after the tICUT counter reaches zero. tSTART Timing tICUT Timing Whenever more than ICUT = VCUT/RS flows through a port, the port’s sense voltage is above VCUT and the tICUT timer counts up. The tICUT timer also counts up when the port’s OUT pin voltage is above VPG. If either of these conditions persists and the tICUT timer expires, the LTC4259A-1 will turn off power to the port immediately and set the appropriate tICUT fault bit in register 06h/07h. The tICUT duration can be programmed via register 16h, bits 2 and 3 (Table 1). The tICUT timer is an up/down counter that is designed to protect the external MOSFET from thermal stress caused by repeatedly operating in current limit. The counter counts up whenever the current is above ICUT and counts down at 1/16th the rate when it is not. The counter will bottom out at zero to prevent underflow. Full count indicates that the tICUT timer has expired and the port will be turned off. This count up/count down behavior implements duty cycle protection, preventing intermittent current limit faults from causing cumulative thermal stress in the MOSFET. If the port To distinguish between normal turn-on current limit behavior and current limit faults which occur after power-up is complete, the LTC4259A-1 starts a timer (the tSTART timer) whenever a power-up sequence begins. The tSTART timer serves three functions. First and foremost, it allows the user to specify a different current limit timeout (tSTART instead of tICUT) during turn-on (current limit duty cycle protection remains functional). Second, the DC disconnect timer is disabled during this period and can only begin counting up after the tSTART timer has expired. Together, these two features let the PD draw the maximum current IINRUSH to charge its input capacitance, boot up and begin drawing power without triggering a tSTART fault. Finally, if the device is in current limit for the entire tSTART period, a tSTART fault will be generated instead of a tICUT fault. This can be useful for tracking down the cause of an overcurrent fault. As long as the PD draws less than ICUT at the end of tSTART and begins drawing the minimum current within tDIS after tSTART expires (if DC disconnect is enabled), no faults will be indicated. 4259a1fa 18 LTC4259A-1 U W U U APPLICATIO S I FOR ATIO The tSTART timer also implements the duty cycle protection described under tICUT timing and its duration can be programmed via register 16h, bits 5 and 4 (Table 1). tICUT timer to count up. During a short circuit, ILIM will be reduced by the foldback feature to 1/7th of the nominal value. See Figures 15 and 16 for examples. Foldback Choosing External MOSFETs Foldback is designed to limit power dissipation in the MOSFET during power-up and momentary short-circuit conditions. At low port output voltages, the voltage across the MOSFET is high, and power dissipation will be large if significant current is flowing. Foldback monitors the port output voltage and reduces the VLIM current limit level linearly from its full value (212.5mV typ) at a port voltage of 18V to approximately 1/7th of the full value (30mV typ) at a port voltage of 0V. With 0.5Ω sense resistors, this limits the short-circuit current to 60mA (typ) instead of the full 425mA (typ) current limit. When the LTC4259A-1 is in foldback, the tICUT timer is active. Power delivery to the ports is regulated with external power MOSFETs. These MOSFETs are controlled as previously described to meet the IEEE 802.3af specification. Under normal operation, once the port is powered and the PD’s bypass capacitor is charged to the port voltage, the external MOSFET dissipates very little power. This suggests that a small MOSFET is adequate for the job. Unfortunately, other requirements of the IEEE 802.3af mandate a MOSFET capable of dissipating significant power. When the port is being powered up, the port voltage must reach 30V or more before the PD turns on. The port voltage can then drop to 0V as the PD’s bypass capacitor is charged. According to the IEEE, the PD can directly connect a 180µF capacitor to the port and the PSE must charge that capacitor with a current limit of 400mA to 450mA for at least 50ms. Short-Circuit Protection If a port is suddenly shorted out, the MOSFET power dissipation can rise to very high levels, jeopardizing the MOSFET even before the normal current limit circuit can respond. A separate short-circuit current limit circuit watches for significant overcurrent events (VSENSE >275mV, >550mA with a 0.5Ω sense resistor) and pulls the GATE pin down immediately if such an event occurs, shutting off the MOSFET in less than 1µs (with no external capacitor on GATE). Approximately 100µs later, GATE is allowed to rise back up and the normal current limit circuit will take over, allowing ILIM current to flow and causing the An even more extreme example is a noncompliant PD that provides the proper signature during detection but then behaves like a low valued resistor, say 50Ω, in parallel with a 1µF capacitor. When the PSE has charged this noncompliant PD up to 20V, the 50Ω resistor will draw 400mA (the minimum IEEE prescribed ILIM current limit) keeping the port voltage at 20V for the remainder of tSTART. The external MOSFET sees 24V to 37V VDS at 400mA to 450mA, dissipating 9.6W to 16.7W for 60ms (typ). GND VDD = 3.3V VEE = –48V VEE VEE GATE +15V VOLTAGE 10V/DIV VEE PORT CURRENT 0mA 20A/DIV VDD = 3.3V VEE = –48V GND PORT VOLTAGE 20V/DIV PORT VOLTAGE 20V/DIV VEE FAST PULL-DOWN ACTIVATED GATE VOLTAGE 10V/DIV FET OFF VEE +15V VEE PORT CURRENT 0mA 500mA/DIV SHORT APPLIED 250ns/DIV CURRENT LIMIT FAST PULL-DOWN SHORT REMOVED SHORT APPLIED 100µs/DIV 4258 G04 Figure 15. Rapid Response to 1Ω Short 4258 G05 Figure 16. Rapid Response to Momentary 100Ω Short 4259a1fa 19 LTC4259A-1 U W U U APPLICATIO S I FOR ATIO The LTC4259A-1 implements foldback to reduce the current limit when the MOSFET VDS is high; see the Foldback section. Without foldback, the MOSFET could see as much as 25.7W for 60ms (typ) when powering a shorted or a noncompliant PD with only a few ohms of resistance. With foldback, the MOSFET sees a maximum of 18W for the duration of tSTART. The LTC4259A-1’s duty cycle protection enforces 15 times longer off time than on time, preventing successive attempts to power a defective PD from damaging the MOSFET. System software can enforce even longer wait times. When the LTC4259A-1 is operated in semiauto or manual mode—described in more detail under Operating Modes—it will not power on a port until commanded to do so by the host controller. By keeping track of tSTART and tICUT faults, the host controller can delay turning on the port again after one of these faults even if the LTC4259A1 reports a Detect Good. In this way the host controller implements a MOSFET cooling off period which may be programmed to protect smaller MOSFETs from repeated thermal cycling. The LTC4259A-1 has built-in duty cycle protection for tICUT and tSTART (see tICUT Timing and tSTART Timing sections) that is sufficient to protect the MOSFETs shown in Figure 1. Before designing a MOSFET into your system, carefully compare its safe operating area (SOA) with the worst case conditions (like powering up a defective PD) the device will face. Using transient suppressors, polyfuses and extended wait times after disconnecting a PD are effective strategies to reduce the extremes applied to the external MOSFETs. Surge Suppressors and Circuit Protection IEEE 802.3af Power over Ethernet is a challenging Hot Swap application because it must survive the (probably unintentional) abuse of everyone in the building. While hot swapping boards in a networking or telecom card cage is done by a trained technician or network administrator, anyone in the building can plug a device into the network. Moreover, in a card cage the physical domain being powered is confined to the card cage. With Power over Ethernet, the PSE supplies power to devices up to 100 meters away. Ethernet cables could potentially be cut, shorted together, and so on by all kinds of events from a contrac- tor cutting into walls to someone carelessly sticking a screwdriver where it doesn’t belong. Consequently, the Power over Ethernet power source (PSE) must be designed to handle these events. The most dramatic of these is shorting a powered port. What the PSE sees depends on how much CAT-5 cable is between it and the short. If the short occurs on the far end of a long cable, the cable inductance will prevent the current in the cable from increasing too quickly and the LTC4259A-1’s built-in short-circuit protection will take control of the situation and turn off the port. Some energy is stored in the cable, but the transient suppressor on the port clamps the port voltage when the cable inductance causes the voltage to fly back after the MOSFET is turned off. Because the cable only had 600mA or so going through it, an SMAJ58A or equivalent device can easily control the port voltage during flyback. With no cable connected at all, a powered port shorted at the PSE’s RJ-45 connector can reach high current levels before the port is shut down. There is no cable inductance to store energy so once the port is shut down the situation is under control. A short—hence low inductance—piece of CAT-5 will not limit the rapid increase of current when the port is shorted. Even though the LTC4259A-1 short-circuit shutdown is fast, the cable may have many amps flowing through it before the MOSFET can be turned off. Due to the high current, this short piece of cable flies back with significant energy behind it and must be controlled by the transient suppressor. Choosing a surge suppressor that will not develop more than a few volts of forward voltage while passing more than 10A is important. A positive port voltage may forward bias the detect diode (DDETn), bringing the LTC4259A-1’s DETECTn pin positive as well and engaging the DETECTn clamps. This will generally not damage the LTC4259A-1 but extreme cases can cause the LTC4259A-1 to reset. When it resets, the LTC4259A-1 signals an interrupt, alerting the host controller which can then return the LTC4259A-1 to normal operating mode. A substantial transient surge suppressor can typically protect the LTC4259A-1 and the rest of the PSE from these faults. Placing a polyfuse between the RJ-45 connector and the LTC4259A-1 and its associated circuitry can provide additional protection. To meet safety requirements, place the polyfuse in the ground leg of the PSE’s output. 4259a1fa 20 LTC4259A-1 U W U U APPLICATIO S I FOR ATIO DC DISCONNECT DC disconnect monitors the sense resistor voltage whenever the power is on to make sure that the PD is drawing the minimum specified current. The disconnect timer counts up whenever port current is below 7.5mA (typ). If the tDIS timer runs out, the corresponding port will be turned off and the disconnect bit in the fault register will be set. If the undercurrent condition goes away before the tDIS timer runs out, the timer will reset. The timer will start counting from the beginning if the undercurrent condition occurs again. The undercurrent circuit includes a glitch filter to filter out noise. The DC disconnect feature can be disabled by clearing the corresponding DC Discon Enable bits in the Disconnect register (13h). The tDIS timer duration can be programmed by bits 1 and 0 of register 16h. The LTC4259A-1 implements a variety of current sense and limit thresholds to control current flowing through the port. Figure 17 is a graphical representation of these thresholds and the action the LTC4259A-1 takes when currrent crosses the thresholds. 300mV 600mA 250mV 500mA 200mV 400mA 150mV 300mA 100mV 200mA 50mV 100mA 0mV SENSEn VOLTAGE 0mA CURRENT DC DISCUT RS = 0.5Ω CONNECT (ICUT) CURRENT LIMIT IN 1µs PORT OFF IN tICUT OR tSTART CURRENT LIMIT NORMAL OPERATION PORT OFF IN tDIS LIMIT (ILIM) SHORT CIRCUIT EFFECT 4259A F17 Figure 17. LTC4259A-1 Current Sense and Limits AC DISCONNECT AC disconnect is an alternate method of sensing the presence or absence of a PD by monitoring the port impedance. The LTC4259A-1 forces a signal, amplified from the OSCIN pin, out of the DETECT pins and onto the Power over Ethernet connection. It calculates the connection impedance from ohm’s law, ZPORT = VAC/IAC. Like DC disconnect, the AC disconnect sensing circuitry controls the disconnect timer. When the connection impedance rises (AC current falls below IACDMIN) due to the removal of the PD, the disconnect timer counts up. If the impedance remains high (AC current remains below IACDMIN), the disconnect timer counts to tDIS, the port is turned off and the port’s disconnect bit in the Fault Register is set. If the impedance falls (AC current rises above IACDMIN) before the maximum count of the disconnect timer, the timer resets and the port remains powered. Like DC disconnect, AC disconnect can also be disabled by clearing the corresponding AC Discon Enable bits in the Disconnect register (13h). AC disconnect is also affected by the tDIS duration programmed in register 16h. Unlike DC disconnect, AC disconnect has no continuous time output to the timer. Rather, AC disconnect will reset the timer once every cycle, 1/fOSCIN, of the OSCIN signal if the port draws more than IACDMIN during that period. Because of this behavior, the time to turn off the port after PD removal, tDIS, may vary by up to one cycle of OSCIN (1/fOSCIN) from the delay programmed with the tDIS1 and tDIS0 bits. Note that AC disconnect and DC disconnect signals that reset the tDIS timer are ORed together. Thus on a port where both disconnect modes are enabled, either disconnect sensing method can keep the port powered even if the other reports that there is no PD connected. The LTC4259A-1 provides a higher AC disconnect threshold, IACDMIN (see Electrical Characteristics) than the LTC4259A. This reduces the port impedance that the LTC4259A-1’s AC disconnect circuitry senses as the absence of a PD. Consequently, in the same application circuit the LTC4259A-1 will have less sensitivity to parasitics like leakage and stray capacitance than an LTC4259A. Both the LTC4259A and LTC4259A-1 (used with the recommended application circuit) meet all the AC disconnect requirements of 802.3af standard. Because the LTC4259A-1 offers improved performance, Linear Technology recommends the LTC4259A-1 for new designs and as a replacement for the LTC4259A in existing designs. The AC disconnect circuitry senses the port and Power over Ethernet connection from the DETECT pins. Connect a 0.47µF 100V X7R capacitor (CDET) and a 1k resistor (RDET) from the port’s DETECT pin to the port’s output as shown in Figure 18. This provides an AC path for sensing 4259a1fa 21 LTC4259A-1 U W U U APPLICATIO S I FOR ATIO the port impedance. The 1k resistor, RDET, limits current flowing through this path during port power on and power off. Sizing of capacitors is critical to ensure proper function of AC disconnect. CPSE (Figure 18) controls the connection impedance on the PSE side. Its capacitance must be kept low enough for AC disconnect to be able to sense the PD. For operation near 100Hz, use a CPSE of 0.1µF. On the other hand, CDET has to be large enough to pass the signal at the frequency of OSCIN. For fOSCIN ≈ 100Hz, use at least a 0.47µF 100V X7R capacitor. The sizes of CPSE, CDET, RDET and the frequency, fOSCIN, are chosen to create an economical, physically compact and functionally robust system. Moreover, the complete Power over Ethernet AC disconnect system (PSE, transformers, cabling, PD, etc.) is complex; deviating from the recommended values of CDET, RDET and CPSE is discouraged. Contact the LTC Applications department for additional support. When choosing CDET and CPSE, carefully consider voltage derating of the capacitors. Capacitors built around an X7R dielectric will have about 60% of the specified capacitance at their rated voltage. Operated at half their rated voltage, X7R capacitors exhibit more than 80% of their specified capacitance. With other ceramic dielectrics commonly used in 50V and 100V chip capacitors, capacitance falls much more dramatically with voltage. At their rated voltage, Y5V or Z5U capacitors exhibit less than 30% of their zero-bias capacitance. Ceramic capacitors can also have significantly less capacitance at elevated temperatures. In order to produce the desired capacitance at the operating bias, 100V or 250V X7R capacitors should be used with the LTC4259A-1. As illustrated in Figure 19, the Power over Ethernet connection between the PSE and PD includes a large amount of capacitance. Cable capacitance is particularly troubling because CAT-3 and CAT-5 pair-to-pair capacitance is not tightly specified by the IEEE 802.3 standard or well controlled by cable manufacturers. Considering that patch panels, additional connectors, old wiring, etc. are likely to be placed between the PSE and PD, pair-to-pair capacitance is a pretty nebulous quantity. Consequently, the cable’s contribution to the port impedance (at the frequency used for AC disconnect) can be a concern. PD GND OSCIN DGND AGND DETECT4 LEVEL 1/4 LTC4259A-1 SHIFT VEE CPSE4 0.1µF 100V X7R SENSE4 GATE4 OUT4 DDET4 CMPD3003 ROUT4 10k –48V Q4 RS4 0.5Ω 1% CPD_D ≥0.05µF RPD_D ≤26.25k RDET4 1k CDET4 0.47µF 100V X7R DAC4 S1B 4259A F18 Figure 18. AC Disconnect Single Port Application Circuit (Port 4 Shown) OSCILLATOR INPUT CURRENT SENSE ~7k ~16k RDET 1k CDET 0.47µF CPSE 0.10µF ZLINK < 14k ZCABLE < 32k ZPD < 14k CCABLE ≤ 0.05µF <32k PD CPD_D ≥ 0.05µF RPD_D < 26.25k 4259A F19 Figure 19. Simplified AC Disconnect Circuit with Impedances at 100Hz 4259a1fa 22 LTC4259A-1 U W U U APPLICATIO S I FOR ATIO Assumimg that fOSCIN is 100Hz, the 0.1µF of CPSE plus 0.05µF of cable capacitance gives a port impedance of 10k at 100Hz. The PD AC signature resistance is about 25k. Connecting a PD with the maximum allowed resistance of 26.25k brings the connection impedance to about 8k. The presence of a PD only makes a 20% reduction in the port impedance requiring the AC disconnect circuitry to be quite sensitive. When the OSCIN pin is driven with a sine wave, the LTC4259A-1 is able to distinguish between capacitive impedance and resistive impedance on the Power over Ethernet connection. AC disconnect is reliable for cable capacitance up to about 0.2µF, nearly an order of magnitude greater than worst case for a long CAT-3 or CAT-5 cable. OSCIN Input and Oscillator Requirements AC disconnect depends on an external oscillator source applied to the OSCIN pin. The LTC4259A-1 measures port impedance by applying an amplified version of the OSCIN signal to the port’s DETECT pin (see Figure 18). The oscillator should be well-controlled because errors in this signal become errors in the measured port impedance. As shown in Figure 19, the load being sensed by AC disconnect has a resistive and a large reactive component. Current through the PD’s signature resistor depends on the amplitude of the AC signal while current into the capacitors depends on the slew rated: I = C • dV/dt. Consequently, the LTC4259A-1 is sensitive to the amplitude and slew rate of the OSCIN signal, but is more tolerant of frequency and offset errors. Internal limits prevent the LTC4259A-1 from being adversely affected by OSCIN signals with excessive amplitude. There are many ways to build oscillators with controlled amplitudes and slew rates, especially since the frequency of the oscillator does not have to be well-controlled. Contact the LTC Applications department for oscillator circuits. As alluded to previously, AC disconnect is complicated and redesigning for different component sizes is a difficult task. For optimum performance, use the recommended component values and drive OSCIN with a 100Hz 2VP-P, 1.2V offset sine wave. Keep in mind that the IEEE 802.3af specification places upper limits of 100V/ms on the slew rate and 500Hz on the frequency of the AC signal at the port. Voltage gain, AVACD, from OSCIN to DETECTn increases the slew rate by the voltage gain. Since AVACD has a maximum absolute value of 3.3V/V (±3V typ), the slew rate at the OSCIN pin must be less than 30V/ms. A slew rate around 0.6V/ms at OSCIN will work with the recommended values of CDET, RDET and CPSE. The LTC4259A-1’s OSCIN input amplifier will accept signals between DGND – 0.3V and VDD + 0.5V. This amplifier has a gain of –1 and is referenced to 1.2V above DGND. An OSCIN voltage greater than 2.2V will cause the amplifier’s output to clip against DGND. Clipping will not affect the performance of AC disconnect until the clipping becomes so severe that even the midrange (where the controlled slew rate occurs) of the signal is clipped. Keep the midrange or average voltage of the OSCIN signal between 0.9V and 1.5V to avoid severe clipping. OSCIN signals below DGND can interact with the ESD protection circuitry on the pin and are not recommended. Also, meeting the IEEE 802.3af specification for maximum AC amplitude on the port just after the PD is removed depends on the OSCIN input peakto-peak amplitude. Clipping by LTC4259A-1’s OSCIN input circuitry will generally ensure that this specification is not exceeded. Note that under normal operation, the AC disconnect output on the DETECTn pin will have an amplitude near 6V peak-to-peak. The combination of RDET, CDET and CPSE attenuate the signal so roughly half this amplitude is seen at the port when the port is powered and the PD has just been removed. When the PD is still connected there will be almost no AC signal at the port. The LTC4259A-1 monitors Pin 36 for the presence of an oscillating signal. If no signal is present and the Osc Fail Mask bit is set, then Osc Fail (bit 1 of the Supply Event register) is set, triggering an interrupt. As the LTC4259A1’s AC disconnect circuitry self-checks the OSCIN signal, the Osc Fail bit is intended as a fault indicator to alert the PSE host controller. The Osc Fail bit has no effect beyond triggering the interrupt. A clear Osc Fail bit indicates that the OSCIN signal goes below 0.6V and above 1.8V at least once every 250ms. It does not necessarily guarantee that AC disconnect will function properly. However, AC disconnect itself is a more thorough test of the OSCIN signal. When the OSCIN signal is either absent or corrupted, 4259a1fa 23 LTC4259A-1 U W U U APPLICATIO S I FOR ATIO powered ports with AC disconnect enabled (and DC disconnect not enabled) will automatically disconnect. After the LTC4259A-1 is reset (by power on, Reset All bit or the RESET pin) the Osc Fail bit is set. Once the Osc Fail bit is cleared, it will only be set by an invalid signal on the OSCIN pin or another reset. SERIAL DIGITAL INTERFACE The LTC4259A-1 communicates with a host (master) using the standard 2-wire interface as described in the SMBus Specification Version 2.0 (available at http:// smbus.org). The SMBus is an extension of the I2C bus, and the LTC4259A-1 is also compatible with the I2C bus standard. The Timing Diagrams (Figures 6 through 10) show the timing relationship of the signals on the bus. The two bus lines, SDA and SCL, must be high when the bus is not in use. External pull-up resistors or current sources, such as the LTC1694 SMBus accelerator, are required on these lines. If the SDA and SCL pull-ups are absent, not connected to the same positive supply as the LTC4259A1’s VDD pin, or are not activated when the power is applied to the LTC4259A-1, it is possible for the LTC4259A-1 to see a START condition on the I2C bus. The interrupt pin (INT) is only updated between I2C transactions. Therefore if the LTC4259A-1 sees a START condition when it powers up because the SCL and SDA lines were left floating, it will not assert an interrupt (pull INT low) until it sees a STOP condition on the bus. In a typical application the I2C bus will immediately have traffic and the LTC4259A-1 will see a STOP so soon after power up that this momentary condition will go unnoticed. Isolating the Serial Digital Interface IEEE 802.3af requires that network segments be electrically isolated from the chassis ground of each network interface device. However, the network segments are not required to be isolated from each other provided that the segments are connected to devices residing within a single building on a single power distribution system. For simple devices such as small powered Ethernet switches, the requirement can be met by using an isolated power supply to power the entire device. This implementation can only be used if the device has no electrically conducting ports other than twisted-pair Ethernet. In this case, the SDAIN and SDAOUT pins of the LTC4259A-1 can be connected together to act as a standard I2C/SMBus SDA pin. If the device is part of a larger system, contains serial ports, or must be referenced to protective ground for some other reason, the Power over Ethernet subsystem including the LTC4259A-1s must be electrically isolated from the rest of the system. The LTC4259A-1 includes separate pins (SDAIN and SDAOUT) for the input and output functions of the bidirectional data line. This eases the use of optocouplers to isolate the data path between the LTC4259A-1s and the system controller. Figure 20 shows one possible implementation of an isolated interface. The SDAOUT pin of the LTC4259A-1 is designed to drive the inputs of an optocoupler directly, but a standard I2C device typically cannot. U1 is used to buffer I2C signals into the optocouplers from the system controller side. Schmitt triggers must be used to prevent extra edges on transitions of SDA and SCL. Bus Addresses and Protocols The LTC4259A-1 is a read-write slave device. The master can communicate with the LTC4259A-1 using the Write Byte, Read Byte and Receive Byte protocols. The LTC4259A-1’s primary serial bus address is (010A3A2A1A0)b, as designated by pins AD3-AD0. All LTC4259A-1s also respond to the address (0110000)b, allowing the host to write the same command into all of the LTC4259A-1s on a bus in a single transaction. If the LTC4259A-1 is asserting (pulling low) the INT pin, it will also acknowledge the Alert Response Address (0001100)b using the receive byte protocol. The START and STOP Conditions When the bus is idle, both SCL and SDA must be high. A bus master (typically the host controller) signals the beginning of communication with a slave device (like the LTC4259A-1) by transmitting a START condition. A START condition is generated by transitioning SDA from high to low while SCL is high. A REPEATED START condition is functionally the same as a START condition, but used to extend the protocol for a change in data transmission 4259a1fa 24 LTC4259A-1 U W U U APPLICATIO S I FOR ATIO 0.1µF VDD INT SCL SDAIN SDAOUT AD0 LTC4259A-1 AD1 AD2 AD3 DGND AGND BYP 0.1µF 0.1µF VDD CPU U1 SCL 2k 200Ω 0100000 0.1µF VDD INT SCL SDAIN SDAOUT AD0 LTC4259A-1 AD1 AD2 AD3 DGND AGND BYP 2k U2 200Ω I2C ADDRESS 0100001 0.1µF SDA 0.1µF HCPL-063L TO CONTROLLER U3 VDD INT SCL SDAIN SDAOUT AD0 LTC4259A-1 AD1 AD2 AD3 DGND AGND BYP 200Ω 200Ω SMBALERT 0.1µF 0.1µF GND CPU U1: FAIRCHILD NC7WZ17 U2, U3: AGILENT HCPL-063L 0.1µF HCPL-063L 0100010 • • • VDD INT SCL SDAIN SDAOUT AD0 LTC4259A-1 AD1 AD2 AD3 DGND AGND BYP 0101110 0.1µF 0.1µF ISOLATED 3.3V + 10µF ISOLATED GND VDD INT SCL SDAIN SDAOUT AD0 LTC4259A-1 AD1 AD2 AD3 DGND AGND BYP 0.1µF 0101111 4258A F20 Figure 20. Optoisolating the I2C Bus 4259a1fa 25 LTC4259A-1 U W U U APPLICATIO S I FOR ATIO direction. A STOP condition is not used to set up a REPEATED START condition, for this would clear any data already latched in. When the master has finished communicating with the slave, it issues a STOP condition. A STOP condition is generated by transitioning SDA from low to high while SCL is high. The bus is then free for communication with another SMBus or I2C device. Acknowledge The Acknowledge signal is used for handshaking between the master and the slave. An Acknowledge (active LOW) generated by the slave lets the master know that the latest byte of information was received. The corresponding SCL clock pulse is always generated by the master. The master releases the SDA line (HIGH) during the Acknowledge clock pulse. The slave must pull down the SDA line during the Acknowledge clock pulse so that it remains a stable LOW during the HIGH period of this clock pulse. When the master is reading from a slave device, it is the master’s responsibility to acknowledge receipt of the data byte in the bit that follows unless the transaction is complete. In that case the master will decline to acknowledge and issue the STOP condition to terminate the communication. Write Byte Protocol The master initiates communication to the LTC4259A-1 with a START condition and a 7-bit bus address followed by the Write Bit (Wr) = 0. If the LTC4259A-1 recognizes its own address, it acknowledges and the master delivers the command byte, signifying to which internal LTC4259A-1 register the master wishes to write. The LTC4259A-1 acknowledges and latches the lower five bits of the command byte into its Register Address register. Only the lower five bits of the command byte are checked by the LTC4259A1; the upper three bits are ignored. The master then delivers the data byte. The LTC4259A-1 acknowledges once more and latches the data into the appropriate control register. Finally, the master terminates the communication with a STOP condition. Upon reception of the STOP condition, the Register Address register is cleared (see Figure 7). Read Byte Protocol The master initiates communication from the LTC4259A1 with a START condition and the same 7-bit bus address followed by the Write Bit (Wr) = 0. If the LTC4259A-1 recognizes its own address, it acknowledges and the master delivers the command byte, signifying which internal LTC4259A-1 register it wishes to read from. The LTC4259A-1 acknowledges and latches the lower five bits of the command byte into its Register Address register. At this time the master sends a REPEATED START condition and the same 7-bit bus address followed by the Read Bit (Rd) = 1. The LTC4259A-1 acknowledges and sends the contents of the requested register. Finally, the master declines to acknowledge and terminates communication with a STOP condition. Upon reception of the STOP condition, the Register Address register is cleared (see Figure 8). Receive Byte Protocol Since the LTC4259A-1 clears the Register Address register on each STOP condition, the interrupt register (register 0) may be read with the Receive Byte Protocol as well as with the Read Byte Protocol. In this protocol, the master initiates communication with the LTC4259A-1 with a START condition and a 7-bit bus address followed by the Read Bit (Rd) = 1. The LTC4259A-1 acknowledges and sends the contents of the interrupt register. The master then declines to acknowledge and terminates communication with a STOP condition (see Figure 9). Alert Response Address and the INT Pin In a system where several LTC4259A-1s share a common INT line, the master can use the Alert Response Address (ARA) to determine which LTC4259A-1 initiated the interrupt. The master initiates the ARA procedure with a START condition and the 7-bit ARA bus address (0001100)b followed by the Read Bit (Rd) = 1. If an LTC4259A-1 is asserting the INT pin, it acknowledges and sends its 7-bit bus address (010A3A2A1A0)b and a 1 (see Figure 10). While it is sending its address, it monitors the SDAIN pin to see if another device is sending an address at the same 4259a1fa 26 LTC4259A-1 U W U U APPLICATIO S I FOR ATIO time using standard I2C bus arbitration. If the LTC4259A1 is sending a 1 and reads a 0 on the SDAIN pin on the rising edge of SCL, it assumes another device with a lower address is sending and the LTC4259A-1 immediately aborts its transfer and waits for the next ARA cycle to try again. If transfer is successfully completed, the LTC4259A1 will stop pulling down the INT pin. When the INT pin is released in this way or if a 1 is written into the Clear Interrupt pin bit (bit 6 of register 1Ah), the condition causing the LTC4259A-1 to pull the INT pin down must be removed before the LTC4259A-1 will be able to pull INT down again. This can be done by reading and clearing the event registers or by writing a 1 into the Clear All Interrupts bit (bit 7 of register 1Ah). The state of the INT pin can only change between I2C transactions, so an interrupt is cleared or new interrupts are generated after a transaction completes and before new I2C bus communication commences. Periodic polling of the alert response address can be used instead of the INT pin if desired. If any device acknowledges the alert response address, then the INT line, if connected, would have been low. System Software Strategy Control of the LTC4259A-1 hinges on one decision, the LTC4259A-1’s operating mode. The three choices are described under Operating Modes. In Auto mode the LTC4259A-1 can operate autonomously without direction from a host controller. Because LTC4259A-1s running in Auto mode will power every valid PD connected to them, the PSE must have 15.4W/port available. To reduce the power requirements of the –48V supply, PSE systems can track power usage, only turning on ports when sufficient power is available. The IEEE describes this as a power allocation algorithm and places two limitations: the PSE shall not power a PD unless it can supply the guaranteed power for that PD’s class (see Table 2) and power allocation may not be based solely on a history of each PD’s power consumption. In order for a PSE to implement power allocation, the PSE’s processor/controller must control whether ports are powered—the LTC4259A-1 cannot be allowed to operate in Auto mode. Semiauto mode fits the bill as the LTC4259A-1 automatically detects and classifies PDs, then makes this information available to the host controller, which decides to apply power or not. Operating the LTC4259A-1 in Manual mode also lets the controller decide whether to power the ports but the controller must also control detection and classification. If the host controller operates near the limit of its computing resources, it may not be able to guide a Manual mode LTC4259A-1 through detect, class and port turn-on in less than the IEEE mandated maximum of 950ms. In a typical PSE, the LTC4259A-1s will operate in Semiauto mode as this allows the controller to decide to power a port without unduly burdening the controller. With an interrupt mask of F4h, the LTC4259A-1 will signal to the host after it has successfully detected and classed a PD, at which point the host can decide whether enough power is available and command the LTC4259A-1 to turn that port on. Similarly, the LTC4259A-1 will generate interrupts when a port’s power is turned off. By reading the LTC4259A-1’s interrupt register, the host can determine if a port was turned off due to overcurrent (tSTART or tICUT faults) or because the PD was removed (Disconnect event). The host then updates the amount of available power to reflect the power no longer consumed by the disconnected PD. Setting the MSB of the interrupt mask causes the LTC4259A-1 to communicate fault conditions caused by failures within the PSE, so the host does not need to poll to check that the LTC4259A-1s are operating properly. This interrupt driven system architecture provides the controller with the final say on powering ports at the same time, minimizing the controller’s computation requirements because interrupts are only generated when a PD is detected or on a fault condition. The LTC4259A-1 can also be used to power older powered Ethernet devices that are not 802.3af compliant and may be detected with other methods. Although the LTC4259A1 does not implement these older detection methods automatically, if software or external circuitry can detect the noncompliant devices, the host controller may command the LTC4259A-1 to power the port, bypassing IEEE compliant detection and classification and sending power to the noncompliant device. 4259a1fa 27 LTC4259A-1 U W U U APPLICATIO S I FOR ATIO B1100 100µH ISOLATED GND 910k + 1µF 100V 510Ω 4.7µH 3.32k 1% Si2328DS + 47µF 10V 10µF 16V 8 VIN 7 ISOLATED GND 6 DRV GATE 5 SENSE LT1619 FB S/S GND VC 1 4 3 FMMT593 FMMT593 2 0.100Ω 1% 1W 47k 1.24k 1% 100k 100pF 4700pF VEE ISOLATED –48V 47µF 10V 100Ω Si2328DS CMPZ4702B + VDD 3.3V 300mA 10Ω 4259A F21 Figure 21. –48V to 3.3V Boost Converter LOGIC LEVEL SUPPLY In additon to the 48V used to source power to each port, a logic level supply is required to power the digital portion of the LTC4259A-1. To simplify design and meet voltage isolation requirements, the logic level supply can be generated from the isolated – 48V supply. Figure 21 shows an example method using an LT®1619 to control a –48V to 3.3V current mode supply. This boost converter topology uses the LT1619 current mode controller and a current mirror which reflects the 3.3V output voltage to the –48V rail, improving the regulation tolerance over the more traditional large resistor voltage divider. This approach achieves high accuracy with a transformerless design. IEEE 802.3af COMPLIANCE AND EXTERNAL COMPONENT SELECTION The LTC4259A-1 is designed to control power delivery in IEEE 802.3af compliant Power Sourcing Equipment (PSE). Because proper operation of the LTC4259A-1 may depend on external signals and power sources, like the –48V supply (VEE) or the OSCIN oscillator source, external components such as the sense resistors (RS), and possibly software running on an external microprocessor, using the LTC4259A-1 in a PSE does not guarantee 802.3af compliance. Using an LTC4259A-1 does get you most of the way there. This section discusses the rest of the elements that go along with the LTC4259A-1 to make an 802.3af compliant PSE. Each paragraph below addresses a component which is critical for PSE compliance as well as possible pitfalls that can cause a PSE to be noncompliant. For further assistance please contact Linear Technology’s Applications department. Sense Resistors The LTC4259A-1 is designed to use a 0.5Ω sense resistor, RS, to monitor the current through each port. The value of the sense resistor has been minimized in order to reduce power loss and as a consequence, the voltage which the LTC4259A-1 must measure is small. Each port may be drawing up to 450mA with this current flowing through the sense resistor and associated circuit board traces. To prevent parasitic resistance on the circuit board from obscuring the voltage drop across the sense resistor, the LTC4259A-1 must Kelvin sense the resistor voltage. One way to achieve Kelvin sensing is “star grounding,” shown pictorially in Figure 1. Another option is to use a – 48V power plane to connect the sense resistor and the LTC4259A-1 VEE pin. Either of these strategies will prevent voltages developed across parasitic circuit board resistances from affecting the LTC4259A-1 current measurement accuracy. The precision of the sense resistor directly affects the measurement of the IEEE parameters IINRUSH, 4259a1fa 28 LTC4259A-1 U W U U APPLICATIO S I FOR ATIO ILIM, ICUT and IMIN. Therefore, to maintain IEEE compliance, use a resistor with 0.5% or better accuracy. voltages that may damage circuitry (in both the PSE and PD) connected to the ports. Power MOSFETs Detect, AC Blocking and Transient Suppressor Diodes The LTC4259A-1 controls power MOSFETs in order to regulate current flow through the Ethernet ports. Under certain conditions these MOSFETs have to dissipate significant power. See the Choosing External MOSFETs section for a detailed discussion of the requirements these devices must meet. During detection and classification, the LTC4259A-1 senses the port voltage through the detect diodes DDET in Figure 18. Excessive voltage drop across DDET will corrupt the LTC4259A-1’s detect and classification results. Select a diode for DDET that will have less than 0.7V of forward drop at 0.4mA and less than 0.9V of forward drop at 50mA. Common Mode Chokes Both nonpowered and powered Ethernet connections achieve best performance (for data transfer, power transfer and EMI) when a common mode choke is used on each port. In the name of cost reduction, some designs share a common mode choke between two adjacent ports. Even for nonpowered Ethernet, sharing a choke is not recommended. With two ports passing through the choke, it cannot limit the common mode current of either port. Instead, the choke only controls the sum of both ports’ common mode current. Because cabling from the ports generally connects to different devices up to 200m apart, a current loop can form. In such a loop, common mode current flows in one port and out the other, and the choke will not prevent this because the sum of the currents is zero. Another way to view this interaction between the paired ports is that the choke acts as a transformer coupling the ports’ common modes together. In nonpowered Ethernet, common mode current results from nonidealities like ground loops; it is not part of normal operation. However, Power over Ethernet sends power and hence significant current through the ports; common mode current is a byproduct of normal operation. As described in the Choosing External MOSFETs section and under the Power Supplies heading below, large transients can occur when a port’s power is turned on or off. When a powered port is shorted (see Surge Suppressors and Circuit Protection), a port’s common mode current may be excessive. Sharing a common mode choke between two ports couples start-up, disconnect and fault transients from one port to the other. The end result can range from momentary noncompliance with 802.3af to intermittent behavior and even to excessive When the port is powered, the detect diode is reverse biased. Any leakage through the detect diode prevents the LTC4259A-1 from sensing all the current coupled through the CDET capacitor. At high temperature with 70V of reverse bias, a typical switching diode like the 1N4148 may have more than 50µA of leakage. Such leakage can interfere with AC disconnect because it is a large fraction of the LTC4259A1’s IACDMIN threshold. Using a low leakage detect diode like the CMPD3003 is recommended. The AC blocking diodes can interfere with AC disconnect sensing if they become leaky. If the AC blocking diode (DAC in Figure 18) begins leaking, it contributes to the Ethernet port impedance, potentially bringing the impedance low enough to draw IACDMIN from the DETECT pin and keep the port powered. More likely, leakage through the AC blocking diode will cause shifts in the AC disconnect threshold that are not large enough to make the PSE noncompliant. Generally, diode leakage is caused by voltage or temperature stress. Diodes that are rated to 100V or more and can handle dissipating at least 0.5W should be acceptable in this application. Other component leakages can have a similar affect on AC disconnect and even affect DC disconnect if the leakage becomes severe. Among components to be wary of are the transient surge suppressors. The devices shown in Figure 1 are rated for less than 5µA of leakage at 58V. However there is a potential for stress induced leakage, so healthy margins should be used when selecting diodes for these applications. Capacitors Sizing of both the CDET and CPSE capacitors is critical to proper operation of the LTC4259A-1’s AC disconnect sensing. See the AC Disconnect section for more information. 4259a1fa 29 LTC4259A-1 U W U U APPLICATIO S I FOR ATIO Also, CPSE may be important to the voltage stability of a powered port. Port voltage instability is generally not a problem if VEE, the –48V supply, is well bypassed. For both of these reasons be aware that many ceramic dielectrics have dramatic DC voltage and temperature coefficients. A 0.22µF ceramic capacitor is often nowhere near 0.22µF when operating at 50VDC or 100VDC. Use 100V or higher rated X7R capacitors for CDET and CPSE as these have reduced voltage dependance while also being relatively small and inexpensive. Power Supplies The LTC4259A-1 must be supplied with 3.3V (VDD) and –48V (VEE). Poor regulation on either of these supplies can lead to noncompliance. The IEEE requires a PSE output voltage between 44V and 57V. When the LTC4259A1 begins powering an Ethernet port, it controls the current through the port to minimize disturbances on VEE. However, if the VEE supply is underdamped or otherwise unstable, its voltage could go outside of the IEEE specified limits, causing all ports in the PSE to be noncompliant. This scenario can be even worse when a PD is unplugged because the current can drop immediately to zero. In both cases the port voltage must always stay between –44V and –57V. In addition, the 802.3af specification places specific ripple, noise and load regulation requirements on the PSE. Among other things, disturbances on either VDD or VEE can adversely affect detection, classification and the AC disconnection sensing. Proper bypassing and stability of the VDD and VEE supplies is important. Another problem that can affect the VEE supply is insufficient power, leading to the supply voltage drooping out of the specified range. The 802.3af specification states that if a PSE powers a PD it must be able to provide the maximum power level requested by the PD based on the PD’s classification. The specification does allow a PSE to choose not to power a port because the PD requires more power than the PSE has left to deliver. If a PSE is built with a VEE supply capable of less than 15.4W • (number of PSE’s Ethernet ports), it must implement a power allocation algorithm that prevents ports from being powered when there is insufficient power. Because the specification also requires the PSE to supply 400mA at up to a 5% duty cycle, the VEE supply capability should be at least a few percent more than the maximum total power the PSE will supply to PDs. Finally, the LTC4259A-1s draw current from VEE. If the VDD supply is generated from VEE, that power divided by the switcher efficiency must also be added to the VEE supply’s capability. Fast VEE transients can damage the LTC4259A-1. Limit the VEE slew rate to 50mV/µs. In most applications, existing VEE bypass capacitors (described above) will cause the VEE supply to slew much slower than 50mV/µs. OSCIN Input AC disconnect also relies on an oscillating signal applied to the OSCIN pin. Requirements for this signal are provided in the OSCIN Input and Oscillator Requirements section. Out-of-band noise on the OSCIN pin will disrupt the LTC4259A-1’s ability to sense the absence of a PD. Any noise present at the OSCIN pin is amplified by the LTC4259A-1 and driven out of the DETECT pins (of powered ports with AC disconnect enabled). Due to the amount of capacitance connected to the DETECT pins, driving this noise can easily require more than IACDMIN, tripping the DETECT pin current sense and keeping the port powered. During circuit board layout, keep wiring from the oscillator to the OSCIN pin away from noise sources like digital clock and data lines. A single-stage RC lowpass filter (shown in Figure 22) will attenuate out-of-band noise. 4259a1fa 30 LTC4259A-1 U PACKAGE DESCRIPTIO GW Package 36-Lead Plastic SSOP (Wide .300 Inch) (Reference LTC DWG # 05-08-1642) 36 19 1.40 ±0.127 10.804 MIN 15.291 – 15.545* (.602 – .612) 36 35 34 33 32 31 30 29 28 27 26 25 24 23 22 21 20 19 7.75 – 8.258 10.11 – 10.55 (.398 – .415) 1 0.520 ±0.0635 18 0.800 BSC RECOMMENDED SOLDER PAD LAYOUT 7.417 – 7.595** (.292 – .299) 0.254 – 0.406 × 45° (.010 – .016) 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 2.286 – 2.388 (.090 – .094) 2.44 – 2.64 (.096 – .104) 0.355 REF 0° – 8° TYP 0.231 – 0.3175 (.0091 – .0125) 0.40 – 1.27 (.015 – .050) NOTE: 1. CONTROLLING DIMENSION: MILLIMETERS MILLIMETERS 2. DIMENSIONS ARE IN (INCHES) 0.800 (.0315) BSC 0.28 – 0.51 (.011 – .02) TYP 0.1 – 0.3 (.004 – .0118) GW36 SSOP 0204 *DIMENSION DOES NOT INCLUDE MOLD FLASH. MOLD FLASH SHALL NOT EXCEED 0.152mm (0.006") PER SIDE **DIMENSION DOES NOT INCLUDE INTERLEAD FLASH. INTERLEAD FLASH SHALL NOT EXCEED 0.254mm (0.010") PER SIDE 4259a1fa Information furnished by Linear Technology Corporation is believed to be accurate and reliable. However, no responsibility is assumed for its use. Linear Technology Corporation makes no representation that the interconnection of its circuits as described herein will not infringe on existing patent rights. 31 LTC4259A-1 U TYPICAL APPLICATIO 2VP-P, 100Hz 1.2V OFFSET OSCILLATOR ISOLATED 3.3V 1k ISOLATED GND 0.1µF 0.1µF 100V X7R OSCIN DGND AGND 2k U2 200Ω VDD BYP DETECT U1 SCL 2k 200Ω VEE SENSE GATE OUT 0.1µF TO CONTROLLER RS 0.5Ω SDA HCPL-063L –48V ISOLATED U3 0.1µF 100V X7R RDET 1k SCL 1/4 SDAIN LTC4259A-1 SDAOUT INT VDD CPU 1µF DDET CMPD3003 L1 CDET 0.47µF 100V X7R DTSS 58V SMAJ58A 10k Q1 IRFM120A DAC S1B RJ45 CONNECTOR 1/2 PULSE H2009 200Ω 0.01µF 200V 75Ω 75Ω 0.01µF 200V 0.01µF 200V 75Ω 75Ω 200Ω SMBALERT PHY 0.1µF (NETWORK PHYSICAL LAYER CHIP) GND CPU HCPL-063L DAC: DIODES INC OR FAIRCHILD S1B DDET: CENTRAL SEMI CMPD3003 DTSS: DIODES INC SMAJ58A CDET: TDK C3225X7R2A474K L1: PULSE ENG PO473 Q1: FAIRCHILD IRFM120A RS: VISHAY WSL2010 0.5Ω 0.5% T1: PULSE ENG H2009 U1: FAIRCHILD NC7WZ17 U2, U3: AGILENT HCPL-063L T1 1:1 1 2 3 4 5 6 7 8 0.01µF 200V 4258 F22A T1 1:1 1000pF 2000V Figure 22. One Complete Isolated Powered Ethernet Port RELATED PARTS PART NUMBER DESCRIPTION COMMENTS LT1619 Low Voltage Current Mode PWM Controller –48V to 3.3V at 300mA, MSOP Package LTC1694 SMBus/I2C Accelerator Improved I2C Rise Time, Ensures Data Integrity LTC4255 Quad Network Power Controller Non-IEEE 802.3af Compliant Current Levels LTC4257 IEEE 802.3af PD Interface Controller 100V, 400mA Internal Switch, Programmable Class LTC4257-1 IEEE 802.3af PD Interface Controller 100V, 400mA Internal Switch, Dual Current Limit, Programmable Class LTC4258 Quad IEEE 802.3af Power Over Ethernet Controller DC Disconnect Only LTC4267 IEEE802.3af PD Interface with Integrated Switching Regulator 100V, 400mA UVLO Switch, Dual Inrush Current, Programmable Class 4259a1fa 32 Linear Technology Corporation LT/LWI 1006 REV A • PRINTED IN USA 1630 McCarthy Blvd., Milpitas, CA 95035-7417 (408) 432-1900 ● FAX: (408) 434-0507 ● www.linear.com © LINEAR TECHNOLOGY CORPORATION 2005