LINER LTC4259A-1

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