www.SiliconStandard.com

11 of 16

SS8017

Rev.2.01 6/06/2003

Table 3. Data Format (Twos-Complement)

DIGITAL OUTPUT

DATA BITS

TEMP.

(

°C)

ROUND

TEMP.

(

°C)

SIGN

MSB

LSB

+130.00

+127

0

111

1111

+127.00

+127

0

111

1111

+126.50

+127

0

111

1111

+126.00

+126

0

111

1110

+25.25

+25

0

001

1001

+0.50

+1

0

000

0001

+0.25

+0

0

000

0000

+0.00

+0

0

000

0000

-0.25

+0

0

000

0000

-0.50

+0

0

000

0000

-0.75

-1

1

111

1111

-1.00

-1

1

111

1111

-25.00

-25

1

110

0111

-25.50

-25

1

110

0110

-54.75

-55

1

100

1001

-55.00

-55

1

100

1001

-65.00

-65

1

011

1111

-70.00

-65

1

011

1111

Alarm Threshold Registers

Four

registers

store

alarm

threshold

data,

with

high-temperature (THIGH) and low-temperature (TLOW)

registers for each A/D channel. If either measured tem-

perature equals or exceeds the corresponding alarm

threshold value, an ALERT interrupt is asserted.

The power-on-reset (POR) state of both THIGH regis-

ters is full scale (0111 1111, or +127°C). The POR

state of both TLOW registers is 1100 1001 or -55°C.

Diode Fault Alarm

There is a continuity fault detector at DXP that detects

whether the remote diode has an open-circuit condition.

At the beginning of each conversion, the diode fault is

checked, and the status byte is updated. This fault

detector is a simple voltage detector; if DXP rises abo-

fault is detected. Note that the diode fault isn't checked

until a conversion is initiated, so immediately after

power-on reset the status byte indicates no fault is

present, even if the diode path is broken.

If the remote channel is shorted (DXP to DXN or DXP to

GND), the ADC reads 0000 0000 so as not to trip ei-

ther the THIGH or TLOW alarms at their POR settings.

In applications that are never subjected to 0°C in nor-

mal operation, a 0000 0000 result can be checked to

indicate a fault condition in which DXP is accidentally

short circuited. Similarly, if DXP is short circuited to

device alarms.

ALERT

Interrupts

The ALERT interrupt output signal is latched and can

only be cleared by reading the Alert Response address.

Interrupts are generated in response to THIGH and

TLOW comparisons and when the remote diode is dis-

connected (for continuity fault detection). The interrupt

does not halt automatic conversions; new temperature

data continues to be available over the SMBus inter-

face after ALERT is asserted. The interrupt output pin

is open-drain so that device can share a common in-

terrupt line. The interrupt rate can never exceed the

conversion rate.

The interface responds to the SMBus Alert Response

address, an interrupt pointer return-address feature

(see Alert Response Address section). Prior to taking

corrective action, always check to ensure that an in-

terrupt is valid by reading the current temperature.

Alert Response Address

The SMBus Alert Response interrupt pointer provides

quick fault identification for simple slave devices that

lack the complex, expensive logic needed to be a bus

master. Upon receiving an ALERT interrupt signal, the

host master can broadcast a Receive Byte transmis-

sion to the Alert Response slave address (0001 100).

Then any slave device that generated an interrupt at-

tempts to identify itself by putting its own address on

the bus.