ADM1168 データシートの表示（PDF） - Analog Devices

部品番号

コンポーネント説明

メーカー

ADM1168

Super Sequencer and Monitor with Nonvolatile Fault Recording

Analog Devices 

ADM1168

Each fault record contains eight bytes, with each byte taking

typically about 250 μs to write to EEPROM, for a total write

time of about 2 ms. After the black box begins to write a fault

record into EEPROM, the ADM1168 ensures that it is complete

before attempting to write any additional fault records. This means

that if consecutive sequencing engine states are designated as

black box write states, then a time delay must be used in the first

state to ensure that the fault record is written before moving to

the next state.

When the ADM1168 powers on initially, it performs a search

to find the first fault record that has not been written to. It does

this by checking the flag bit in each fault record until it finds

one where the flag bit is 1. The first fault record is stored at

Address 0xF980, and at multiples of eight bytes after that, with

the last record stored at Address 0xF9F8.

The fault recorder is only able to write in the EEPROM. It is not

able to erase the EEPROM prior to writing the fault record.

Therefore, to ensure correct operation, it is important that the fault

record EEPROM is erased prior to use. When all the EEPROM

locations for the fault records are used, no more fault records are

written. This ensures that the first fault in any cascading fault is

stored and not overwritten and lost.

To avoid the fault recorder filling up and fault records being

lost, an application can periodically poll the ADM1168 to

determine if there are fault records to be read. Alternatively, one

of the PDOx outputs can be used to generate an interrupt for a

processor in the fault record write state to signal the need to

come and read one or more fault records.

After reading fault records during normal operation, two things

must be done before the fault recorder will be able to reuse the

EEPROM locations. First, the EEPROM section must be erased.

The fault recorder must then be reset so that it performs its search

again for the first unused location of EEPROM that is available to

store a fault record.

Data Sheet

BLACK BOX WRITES WITH NO EXTERNAL SUPPLY

In cases where all the input supplies fail, for example, if the card

has been removed from a powered backplane, the state machine

can be programmed to trigger a write into the black box EEPROM.

The decoupling capacitors on the rail that power the ADM1168

and other loads on the board form an energy reservoir. Depending

on the other loads on the board and their behavior as the supply

rails drop, there may be sufficient energy in the decoupling

capacitors to allow the ADM1168 to write a complete fault record

(eight bytes of data).

Typically, it takes 2 ms to write to the eight bytes of a fault record. If

the ADM1168 is powered using a 12 V supply on the VH pin, then

a UV threshold at 6 V can be set and used as the state machine

trigger to start writing a fault record to EEPROM. The higher the

threshold is, the earlier the black box write begins, and the more

energy available in the decoupling capacitors to ensure it completes

successfully.

Provided the VH supply, or another supply connected to a VPx pin,

remains above 3.0 V during the time to write, the entire fault record

is always written to the EEPROM. In many cases, there should be

sufficient decoupling capacitors on a board to power the ADM1168

as it writes into the EEPROM.

In cases where the decoupling capacitors are not able to supply

sufficient energy after the board is removed to ensure a complete

fault record is written, the value of the capacitor on VDDCAP

may be increased. In the worst case, assuming that no energy is

supplied to the ADM1168 by the external decoupling capacitors,

but that VDDCAP has 4.75 V on it, then a 47 μF is sufficient to

guarantee that a single complete black box record can be written

to the EEPROM.

Rev. B | Page 18 of 27

Share Link: