NCV7513B
www.onsemi.com
17
Fault Detection and Capture
Each channel of the NCV7513B is capable of detecting
shorted load faults when the channel is on, and short to
ground or open load faults when the channel is off. Each
fault type is uniquely encoded into two−bit per channel
fault data. A drain feedback input for each channel
compares the voltage at the drain of the channel’s external
MOSFET to several internal reference voltages. Separate
detection references are used to distinguish the three fault
types, and blanking and filter timers are used respectively
to allow for output state transition settling and for glitch
suppression.
Fault diagnostics are disabled when either enable input
is low. When both enable inputs are high, each channel’s
drain feedback input is continuously compared to
references appropriate to the channel’s input state to detect
faults, but the comparison result is only latched at the end
of either a blanking or filter timer event.
Blanking timers for all channels are triggered when
either enable input changes state from low to high while the
other enable input is high, or when both enable inputs go
high simultaneously. A single channel’s blanking timer is
triggered when its input state changes. If the comparison of
the feedback to a reference indicates an abnormal condition
when the blanking time ends, a fault has been detected and
the fault data is latched into the channel’s fault latch.
A channel’s filter timer is triggered when its drain
feedback comparison state changes. If the change indicates
an abnormal condition when the filter time ends, a fault has
been detected and the fault data is latched into the channel’s
fault latch.
Thus, a state change of the inputs (ENA
X
, IN
X
or G
X
) or
a state change of an individual channel’s feedback (DRN
X
)
comparison must occur for a timer to be triggered and a
detected fault to be captured.
Fault Capture, SPI Communication, and SPI
Frame Error Detection
The fault capture and frame error detection strategies of
the NCV7513B combine to ensure that intermittent faults
can be captured and identified, and that the device cannot
be inadvertently reprogrammed by a communication error.
The NCV7513B latches a fault when it is detected, and
frame error detection will not allow any register to accept
data if an invalid frame occurred.
When a fault has been detected, the FLTB flag is set and
fault data is latched into a channel’s fault latch. The latch
captures and holds the fault data and ignores subsequent
fault data for that channel until a valid SPI frame occurs.
Fault data from all channels is transferred from each
channel’s fault latch into the SPI shift register and the FLTB
flag is reset when CSB goes low at the start of the SPI
frame. Fault latches are cleared and re−armed when CSB
goes high at the end of the SPI frame only if a valid frame
has occurred; otherwise the latches retain the detected fault
data until a valid frame occurs. The FLTB flag will be set
if a fault is still present.
Fault latches for all channels and the FLTB flag can also
be cleared and re−armed by toggling ENA1 H−L−H. A full
I/O truth table is given in Table 9.
Fault Data Readback Examples
Several examples are shown to illustrate fault detection,
capture and SPI read−back of fault data for one channel. A
normal SPI frame returns 16 bits of data but only the two
bits of serial data for the single channel are shown for
clarity.
The examples assume:
• The NCV7513B is configured as in Figure 2
• Both enable inputs are high
• The channel’s flag mask bit is cleared
• Disable mode is set to auto−retry
• The parallel input commands the channel
• SPI frame is always valid
Shorted Load Detected
Refer to Figure 18. The channel is commanded on when
IN
X
goes high. GAT
X
goes high and the timers are started.
At “A”, the STAB flag is set as the DRN
X
feedback falls
through the V
OL
threshold. A SPI frame sent soon after the
IN
X
command returns data indicating “no fault.”
The blanking time ends and the filter timer is triggered
as DRN
X
rises through the FLTREF threshold. The STAB
flag is reset as DRN
X
passes through the V
OL
threshold.
DRN
X
is nearly at V
LOAD
when the filter time ends at “B”.
A shorted load fault is detected and captured by the fault
latch, GAT
X
goes low, the FLTB flag is set, and the
auto−retry timer is started.
An SPI frame sent soon after “B” returns data indicating
“shorted load”. The FLTB flag is reset when CSB goes low.
At “C” when CSB goes high at the end of the frame, the
fault latch is cleared and re−armed. Since IN
X
and the
DRN
X
feedback are unchanged, FLTB and the fault latch
are set and the fault is recaptured.
When the auto−retry timer ends at “D”, GAT
X
goes high
and the blanking and filter timers are started. Since IN
X
and
DRN
X
are unchanged, GAT
X
goes low when the blanking
time ends at “E” and the auto−retry timer is started.
Read−back data continues to indicate a “shorted load” and
the FLTB flag continues to be set while the fault persists.
