LTC4222
16
4222fb
APPLICATIONS INFORMATION
Setting the CONFIG pin high allows the two channels to
start up and turn off independently. When both ON signals
are brought high sequentially, the channel turned on first
immediately begins to start up and the second channel
has a 200ns window to assert it’s ON signal in order to
start up in the same timer period. If the second ON signal
is asserted after the 200ns window but before the end of
the first channel’s start-up time, the second channel start-
up is delayed. The second channel will then start 100ms
after the first channel’s start-up timer has expired and the
TIMER pin, if used, reaches its 200mV low threshold.
When an external TIMER capacitor is used, the TIMER
capacitor voltage ramps up with a 100µA current. Once the
TIMER pin reaches its 1.235V threshold the TIMER begins
to discharge. While the TIMER capacitor is discharging, the
ON signal for the second channel should not be asserted
for 2ms/µF of TIMER capacitance. This allows the TIMER
capacitor to return to its low state and ensures that the next
channel to start receives a full timer cycle. This wait time
is unnecessary when using the internal 100ms timer.
Board Present Change of State
The EN pins may be used to detect the presence of one or
two downstream cards. Whenever an EN pin toggles, FAULT
bit 4 is set to indicate a change of state. When the EN pin
goes high, indicating board removal, the corresponding
GATE turns off immediately (with a 1mA current to ground)
and the board present STATUS bit 4, is cleared. If the EN
pin is pulled low, indicating a board insertion, all fault bits
for that channel except FAULT bit 4 are cleared and enable
STATUS bit 4, is set. If the EN pin remains low for 100ms
the state of the ON pin is captured in ‘FET On’ CONTROL
bit 3. This turns the switch on if the ON pin is tied high.
There is an internal 10µA pull-up current source on the
EN pin. If the CONFIG pin is tied low, both EN pins must
be low for 100ms for the two channels to be enabled and
if either EN pin goes high both channels will turn off.
If a channel shuts down due to a fault, it may be desirable
to restart that channel simply by removing and reinserting
the related load card. In cases where the LTC4222 and the
switch reside on a backplane or midplane and the load
resides on a plug-in card, the EN pin detects when the
plug-in card is removed. Figure 4 shows an example where
the EN pin is used to detect insertion. Once the plug-in card
is reinserted the fault register is cleared except for FAULT
bit 4. After 100ms the state of the ON pin is latched into
bit 3 of the CONTROL register. At this point the channel
starts up again.
If a connection sense on the plug-in card is driving an EN
pin, insertion or removal of the card may cause the pin
voltage to bounce. This results in clearing the fault register
when the card is removed. The pin may be debounced
using a filter capacitor, C
EN
, on the EN pin as shown in
Figure 4. The filter time is given by:
t
FILTER
= C
EN
• 123 (ms/µF)
Figure 4. Plug-In Card Insertion/Removal
+
1.235V
GND
MOTHERBOARD
CONNECTOR PLUG-IN
CARD
SOURCE
OUT
LTC4222
10µA
EN
C
EN
LOAD
4222 F04
FET Short Fault
A FET short fault is reported if the data converter measures
a current sense voltage greater than or equal to 2mV while
the corresponding GATE is turned off. This condition sets
FET short bit, Fault bit 5.
Power-Bad Fault
A power-bad fault is reported if a FB pin voltage drops
below its 1.235V threshold for more than 2µs when the
corresponding GATE is above the 4.3V gate to source
threshold. This pulls the GPIO pin low immediately in the
default power good configuration, and sets power-bad
present bit, STATUS bit 3, and power-bad bit, FAULT bit 3.
A circuit prevents power-bad faults if the GATE-to-SOURCE
voltage is low, eliminating false power-bad faults during
power-up or power-down. If the FB pin voltage subsequently
rises back above the threshold, a power good configured
GPIO pin returns to a high impedance state and STATUS
bit 3 is reset.
LTC4222
17
4222fb
APPLICATIONS INFORMATION
Fault Alerts
When any of the fault bits in a FAULT register (see Table 4)
are set, an optional bus alert is generated if the appropri-
ate bit in the ALERT register has been set. This allows
only selected faults to generate alerts. At power-up the
default state is to not alert on faults and the ALERT pin
is high. If an alert is enabled, the corresponding fault
causes the ALERT pin to pull low. After the bus master
controller broadcasts the Alert Response Address, the
LTC4222 responds with its address on the SDA line and
releases ALERT as shown in Table 7. If there is a collision
between two LTC4222s responding with their addresses
simultaneously, then the device with the lower address
wins arbitration and responds first. The ALERT line is also
released if the device is addressed by the bus master if
ALERT is pulled low due to an alert.
Once the ALERT signal has been released for one fault, it
is not pulled low again until the FAULT register indicates a
different fault has occurred or the original fault is cleared
and it occurs again. Note that this means repeated or
continuing faults do not generate alerts until the associ-
ated FAULT register bit has been cleared.
Resetting Faults
Faults are reset with any of the following conditions on a
given channel. First, a serial bus command writing zeros
to the FAULT register bits 0 to 5 clears the associated
faults. Second, FAULT register bits 0 to 5 are cleared when
the corresponding switch is turned off by the ON pin or
STATUS bit 3 going from high to low, if the corresponding
UV pin is brought below its 0.4V reset threshold for 2µs,
or if INTV
CC
falls below its 2.64V undervoltage lockout
threshold. Finally, when EN is brought from high to low,
only corresponding FAULT bits 0-3 and 5 are cleared, and
bit 4, which indicates a EN change of state, is set. Note
that faults that are still present, as indicated in the STATUS
registers, cannot be cleared.
The FAULT registers are not cleared when auto-retrying.
When auto-retry is disabled the existence of an overvoltage,
undervoltage, or overcurrent fault keeps the switch off.
As soon as the fault is cleared, the switch turns on. If
auto-retry is enabled, then a high value in STATUS register
bits 0 or 1 holds the switch off and the fault register is
ignored. Subsequently, when STATUS register bits 0 and
1 are cleared by removal of the fault condition, the switch
is allowed to turn on again. The LTC4222 will set FAULT
bit 2 and turn off in the event of an overcurrent fault,
preventing it from remaining in an overcurrent condition.
If configured to auto-retry, the LTC4222 will continually
attempt to restart after cool-down cycles until it succeeds
in starting up without generating an overcurrent fault. Note
that if a switch is on after an auto-retry and the FAULT bit
has not been reset, clearing the corresponding auto-retry
bit will turn the channel off.
Data Converter
The LTC4222 incorporates a 10-bit A/D converter that
continuously scans six different voltages. The SOURCE
pins have a 1/24 resistive divider to monitor a full-scale
voltage of 32V with 31.25mV resolution. The ADIN pins are
monitored with a 1.28V full scale and 1.25mV resolution,
and the voltage between the V
DD
and SENSE pins is moni-
tored with a 64mV full scale and 62.5µV resolution.
Results from each conversion are stored, left justified, in
registers as seen in Tables 7 and 8, and are updated 15
times per second. Setting ADC_CONTROL register bit 0
invokes a test mode that halts the data converter so that
the data converter result registers may be written to and
read from for software testing.
The data converter also has a direct address mode that
allows the user to take a specific measurement at a spe-
cific time and hold that value for later readback. Direct
address mode is entered by setting the Halt bit, bit 0, in
the ADC_CONTROL register (see Table 9). Then when
the channel address bits, ADC_CONTROL bits 1 to 3, are
written to, the ADC will make a single measurement on
the channel indicated by those bits, then stop. Setting the
ADC Alert bit, ADC_CONTROL bit 4, will enable an interrupt
when the data converter finishes the conversion, result-
ing in the ALERT pin pulling low when the data is ready.
Alternately, the ADC Busy bit, ADC_CONTROL bit 5, can
be polled to check for the end of the conversion, after a
direct address conversion the ADC Busy bit will go low. In
normal mode ADC Busy is always high. Resetting the Halt
bit returns the data converter to the scan mode.
LTC4222
18
4222fb
APPLICATIONS INFORMATION
Configuring the GPIO Pins
Table 3 describes the possible states of the GPIO pins using
the CONTROL registers bits 6 and 7. At power-up, the default
state is for a GPIO pin to go high impedance when power is
good (FB pin greater than 1.235V). Other applications for a
GPIO pin are to pull down when power is good, a general
purpose output and a general purpose input.
A simple application of the GPIO pin in the power good
configuration is to connect it to the UV pin of the other
channel with the CONFIG pin high. This will result in the
second channel being turned on after the first channel has
started up and signaled power good.
Current Limit Stability
For many applications the LTC4222 current limit will be
stable without additional components. However there are
certain conditions where additional components may be
needed to improve stability. The dominant pole of the cur-
rent limit circuit is set by the capacitance and resistance at
the gate of the external MOSFET, and larger gate capaci-
tance makes the current limit loop more stable. Usually
a total of 8nF gate to source capacitance is sufficient for
stability and is typically provided by inherent MOSFET C
GS
,
however the stability of the loop is degraded by increasing
R
SENSE
or by reducing the size of the resistor on a gate RC
network if one is used, which may require additional gate
to source capacitance. Board level short-circuit testing
is highly recommended as board layout can also affect
transient performance, for stability testing the worst-case
condition for current limit stability occurs when the output
is shorted to ground after a normal start-up.
There are two possible parasitic oscillations when the
MOSFET operates as a source follower when ramping
at power-up or during current limiting. The first type of
oscillation occurs at high frequencies, typically above
1MHz. This high frequency oscillation is easily damped
with R5 as shown in Figure 1. In some applications, one
may find that R5 helps in short-circuit transient recovery
as well. However, too large of an R5 value will slow down
the turn-off time. The recommended R5 range is between
5Ω and 500Ω.
The second type of source follower oscillation occurs at
frequencies between 200kHz and 800kHz due to the load
capacitance being between 0.2µF and 9µF, the presence
of R5 resistance, the absence of a drain bypass capacitor,
a combination of bus wiring inductance and bus supply
output impedance. To prevent this second type of oscillation
avoid load capacitance below 10µF, alternately connect an
external capacitor from the MOSFET gate to ground with
a value greater than 1.5nF.
Supply Transients
The LTC4222 is designed to ride through supply transients
caused by load steps. If there is a shorted load and the
parasitic inductance back to the supply is greater than
0.5µH, there is a chance that the supply collapses before
the active current limit circuit brings down the GATE pin.
If this occurs, the undervoltage monitors pull the corre-
sponding GATE pin low. The undervoltage lockout circuit
has a 2µs filter time after V
DD
drops below 2.35V. The UV
pin reacts in 2µs to shut the GATE off, but it is recom-
mended to add a filter capacitor, C
F
, to prevent unwanted
shutdown caused by a transient. Eventually either the UV
pin or undervoltage lockout responds to bring the current
under control before the supply completely collapses.
Supply Transient Protection
The LTC4222 is safe from damage with supply voltages up
to 35V. However, spikes above 35V may damage the part.
During a short-circuit condition, large changes in current
flowing through power supply traces may cause induc-
tive voltage spikes which exceed 35V. To minimize such
spikes, the power trace inductance should be minimized
by using wider traces or heavier trace plating. Also, a
snubber circuit dampens inductive voltage spikes. Build
a snubber by using a 100Ω resistor in series with a 0.1µF
capacitor between V
DD
and GND. A surge suppressor, Z1
in Figure 1, at the input can also prevent damage from
voltage surges.
Design Example
As a design example, take the following specifications for
channel 1: V
IN
= 12V, I
MAX
= 5A, I
INRUSH
= 1A, dI/dt
INRUSH
= 10A/ms, C
L
= 330µF, V
UV(RISING)
= 10.75V, V
OV(FALLING)
= 14.0V, V
PWRGD(UP)
= 11.6V, and I
2
C ADDRESS = 1000111.
This completed design is shown in Figure 1.

LTC4222IUH#PBF

Mfr. #:
Manufacturer:
Analog Devices Inc.
Description:
Hot Swap Voltage Controllers Dual Hot Swap Controller w/ADC and I2C
Lifecycle:
New from this manufacturer.
Delivery:
DHL FedEx Ups TNT EMS
Payment:
T/T Paypal Visa MoneyGram Western Union