NCP1615
www.onsemi.com
28
SYSTEM FAILURE DETECTION
When manufacturing a power supply, elements can be
accidentally shorted or improperly soldered. Such failures
can also occur as the system ages due to component fatigue,
excessive stress, soldering faults, or external interactions. In
particular, a pin can be grounded, left open, or shorted to an
adjacent pin. Such open/short situations require a safe
failure without smoke, fire, or loud noises. The NCP1615
integrates functions that ease meeting this requirement.
Among them are:
•GND connection fault. If the GND pin is properly
connected, the supply current drawn from the positive
terminal of the VCC capacitor, flows out of the GND
pin and returns to the negative terminal of the VCC
capacitor. If the GND pin is disconnected, the internal
ESD protection diodes provides a return path. An open
or floating GND pin is detected if current flows in the
CS/ZCD ESD diode. If current flow is detected for
200 ms, a fault is acknowledged and the controller stops
operating.
•Open CS/ZCD Pin: A pull-up current source,
ICS/ZCD(bias1), on the CS/ZCD pin allows detection of
an open CS/ZCD pin. ICS/ZCD1, is typically 1 mA. If the
pin is open, the voltage on the pin will increase to the
supply rail. This condition is detected and the controller
is disabled.
•Grounded CS/ZCD Pin: If the CS/ZCD pin is
grounded, the circuit cannot detect a ZCD transition,
activating the watchdog timer (typically 200 ms). Once
the watchdog timer expires, a pull-up current source,
ICS/ZCD2, sources 250 mA to pull-up the CS/ZCD pin.
The driver is inhibited until the CS/ZCD pin voltage
exceeds the ZCD arming threshold, VZCD(rising),
typically 0.75 V. Therefore, if the pin is grounded, the
voltage on the pin will not exceed VZCD(rising) and
drive pulses will be inhibited. The external impedance
should be above 3.9 kW to ensure correct operation.
•Boost or bypass diode short. The NCP1615 addresses
the short situations of the boost and bypass diodes (a
bypass diode is generally placed between the input and
output high-voltage rails to divert this inrush current).
Practically, the overstress protection is implemented to
detect such conditions and forces a low duty ratio
operation until the fault is removed.
FAULT INPUT
The NCP1615 includes a dedicated fault input accessible
via the Fault pin. The controller can be latched by pulling up
the pin above the upper fault threshold, VFault(OVP),
typically 3.0 V. The controller is disabled if the Fault pin
voltage, VFault, is pulled below the lower fault threshold,
VFault(OTP_in), typically 0.4 V. The lower threshold is
normally used for detecting an overtemperature fault. The
controller operates normally while the Fault pin voltage is
maintained within the upper and lower fault thresholds.
Figure 26 shows the architecture of the Fault input.
The lower fault threshold is intended to be used to detect
an overtemperature fault using an NTC thermistor . A pull up
current source IFault(OTP), (typically 45.5 mA) generates a
voltage drop across the thermistor. The resistance of the
NTC thermistor decreases at higher temperatures resulting
in a lower voltage across the thermistor. The controller
detects a fault once the thermistor voltage drops below
VFault(OTP_in). Versions A and C latch-off the controller after
an overtemperature fault is detected. In versions B and D the
controller is re-enabled once the fault is removed such that
VFault increases above VFault(OTP_out) and VCC reaches
VCC(on). Figure 27 shows typical waveforms related to the
latch version where−as Figure 28 shows waveforms of the
auto-recovery version.
An active clamp prevents the Fault pin voltage from
reaching the upper latch threshold if the pin is open. To reach
the upper threshold, the external pull-up current has to be
higher than the pull-down capability of the clamp (set by
RFault(clamp) at VFault(clamp)). The upper fault threshold is
intended to be used for an overvoltage fault using a Zener
diode and a resistor in series from the auxiliary winding
voltage, VAUX. The controller is latched once VFault.exceeds
VFault(OVP).
The Fault input signal is filtered to prevent noise from
triggering the fault detectors. Upper and lower fault detector
blanking delays, tdelay(OVP) and tdelay(OTP) are both typically
30 ms. A fault is detected if the fault condition is asserted for
a period longer than the blanking delay.
The controller bias current is reduced during power up by
disabling most of the circuit blocks including IFault(OTP).
This current source is enabled once VCC reaches VCC(on). A
bypass capacitor is usually connected between the Fault and
GND pins and it will take some time for VFault to reach its
steady state value once IFault(OTP) is enabled. To prevent
false detection of an OTP fault during power up, a dedicated
timer, tblank(OTP), blanks the OTP signal during power up.
The t blank(OTP), duration is typically 5 ms. In versions B and
D, IFault(OTP) remains enabled while the lower fault is
present independent of VCC in order to provide temperature
hysteresis. IFault(OTP) is disabled once the fault is removed.
The controller can detect an upper fault (i.e. overvoltage)
once VCC exceeds VCC(reset).
Once the controller is latched, it is reset if a brownout
condition is detected or if VCC is cycled down to its reset
level, VCC(reset). In the typical application these conditions
occur only if the ac voltage is removed from the system. The
internal latch also resets once the controller enters power
saving mode. Prior to reaching VCC(reset) Vfault(clamp) is set
at 0 V.