Data sheet – GE Industrial Solutions CP2725AC54TE User Manual
Page 16
GE
Data Sheet
CP2725AC54TE CPL High Efficiency Rectifier
100-120/200-277V
AC
input; Default Outputs: ±54V
DC
@ 2725W, 5V
DC
@ 4W
January 30, 2014
©2013 General Electric Company. All rights reserved.
Page 16
State Change Definition
A state_change is an indication that an event has occurred
that the MASTER should be aware of. The following events
shall trigger a state_change;
Initial power-up of the system when AC gets turned ON .
This is the indication from the rectifier that it has been
turned ON. Note that the master needs to read the status
of each power supply to reset the system_interrupt. If the
power supply is back-biased through the 8V_INT or the
5VSTB it will not issue an SMBALERT# when AC power is
turned back ON.
Whenever the power supply gets hot-plugged into a
working system. This is the indicator to the system
(MASTER) that a new power supply is on line.
Any changes in the bit patterns of the STATUS and ALARM
registers are a STATUS change which triggers the
SMBALERT# flag. Note that a host-issued command such
as CLEAR_FAULTS will not trigger an SMB
Hot plug procedures
Careful system control is recommended when hot plugging a
power supply into a live system. It takes about 15 seconds for
a power supply to configure its address on the bus based on
the analog voltage levels present on the backplane. If
communications are not stopped during this interval, multiple
power supplies may respond to specific instructions because
the address of the hot plugged power supply always defaults
to xxxx000 (depending on which device is being addressed
within the power supply) until the power supply configures its
address.
The recommended procedure for hot plug is the following: The
system controller should be told which power supply is to be
removed. The controller turns the service LED ON, thus
informing the installer that the identified power supply can be
removed from the system. The system controller should then
poll the module_present signal to verify when the power
supply is re-inserted. It should time out for 15 seconds after
this signal is verified. At the end of the time out all
communications can resume.
Predictive Failures
Alarm warnings that do not cause a shutdown are indicators
of potential future failures of the power supply. For example, if
a thermal sensor failed, a warning is issued but an immediate
shutdown of the power supply is not warranted.
Another example of potential predictive failure mechanisms
can be derived from information such as fan speed when
multiple fans are used in the same power supply. If the speed
of the fans varies by more than 20% from each other, this is an
indication of an impending fan wear out.
The goal is to identify problems early before a protective
shutdown would occur that would take the power supply out
of service.
External EEPROM
A 64k-bit EEPROM is provided across the I
2
C bus. This EEPROM
is used for both storing FRU_ID information and for providing a
scratchpad memory function for customer use.
Functionally the EEPROM is equivalent to the ST M34D64 part
that has its memory partitioned into a write protected upper ¼
of memory space and the lower ¾ section that cannot be
protected. FRU_ID is written into the write protected portion of
memory.
Write protect feature:
Writing into the upper 1/4 of memory
can be accomplished either by hardware or software.
The power supply pulls down the write_protect (Wp) pin to
ground via a 500Ω resistor between the ‘module_present’
signal pin and Logic_GRD (see the Module Present Signal
section of Input Signals). Writing into the upper ¼ of memory
can be accomplished by pulling HI the module_present pin.
An alternative, and the recommended approach, is to issue the
Enable_write command via software.
Page implementation:
The external EEPROM is partitioned into
32 byte pages. For a write operation only the starting address
is required. The device automatically increments the memory
address for each byte of additional data it receives. However,
if the 32 byte limit is exceeded the device executes a wrap-
around that will start rewriting from the first address specified.
Thus byte 33 will replace the first byte written, byte 34 the
second byte and so on. One needs to be careful therefore not
to exceed the 32 byte page limitation of the device.