Datasheet

REV. B
ADMC401
–42–
WATCHDOG TIMER
OVERVIEW
The watchdog timer is used as a protection mechanism against
unintentional software events causing the DSP to become stuck
in infinite loops. It can be used to cause a complete DSP and
peripheral reset in the event of such a software error. The watch-
dog timer consists of a 16-bit timer that is clocked at the CLKIN
rate, t
CKI
.
The watchdog timer is disabled after a master reset (RESET =
LO). This also resets the WDFLAG bit in the SYSSTAT regis-
ter. The watchdog timer is enabled by writing a TIMEOUT value
to the WDTIMER register. Once the watchdog timer has been
initialized, the timer is decremented at the CLKIN rate. In
order to prevent a watchdog timer trip, it is necessary to write
again to the WDTIMER register. For all writes to the WDTIMER
register (subsequent to the initial write), it is unimportant which
value is written. The act of writing to the WDTIMER register
automatically reloads the initial TIMEOUT value. If the watch-
dog timer is not rewritten to after an interval:
T
WDT
= WDTIMER × t
CKI
the watchdog timer will decrement to zero and a watchdog trip
will be generated. In this case, a complete reset of the DSP core
and motor control peripherals (except the watchdog timer
itself) is initiated and Bit 1 of the SYSSTAT register (WDFLAG)
is set. Following a reset, the DSP core can determine if the reset
was caused by a watchdog trip (and if so take appropriate ac-
tion) or if it was due to the normal reset sequence. The watchdog
timer remains disabled while the WDFLAG is set to prevent
continuous watchdog trips. The watchdog timer can be restarted
and the WDFLAG reset by writing a nonzero TIMEOUT value
to the WDTIMER register. The WDFLAG will be reset, but
the watchdog timer will remain disabled if 0x0000 is written to
the WDTIMER register.
The watchdog timer is only reset by a low input on the RESET
pin. The watchdog circuit is not reset by a software controlled
Peripheral Reset.
PROGRAMMABLE INTERRUPT CONTROLLER
OVERVIEW
The ADMC401 uses the IRQ2 pin of the DSP core to generate
a peripheral interrupt. There are multiple sources of peripheral
interrupts, e.g., the ADC block, PIO block, EIU block, ETU
block and PWM block. A Programmable Interrupt Controller
(PIC) is used to avoid a software latency in determining the
source of the interrupt. With the occurrence of an interrupt
from the peripheral blocks, the PIC block generates an address
that points to the corresponding vector address in the DSP vector
table. The PIC consists of an output register, PICVECTOR,
that contains a pointer to an entry in the DSP vector table.
During normal operation, an interrupt service routine (ISR)
located at vector address 0x0004 (or the IRQ2/peripheral inter-
rupt) jumps to the address pointed to by the PICVECTOR
register. The necessary code to perform this jump from address
0x0004 is automatically placed there by the internal ROM code
when MMAP = BMODE = 1.
The vector addresses between 0x00 and 0x2C are reserved
for the DSP core interrupts. The vector table addresses from
PM(0x30) to PM(0x58) are reserved for use by peripheral inter-
rupt service routines. Each vector address occupies four addresses
of PM. The priority of the peripheral interrupts is fixed in hard-
ware. The ISR at address PM(0x30) has the highest priority
whereas the ISR at address PM(0x58) has the lowest.
In the case of multiple simultaneous interrupts, the PIC will
load the PICVECTOR register with the interrupt that has the
highest priority. Between reads of the PICVECTOR register
(while the DSP is servicing other interrupts for example)
PICVECTOR is updated with the highest priority of any periph-
eral interrupts. This ensures that when the IRQ2 is reasserted,
the highest priority interrupt that occurred since the last reading
of the PICVECTOR register is now waiting to be serviced.
When PICVECTOR is read, if another interrupt is pending in
the PIC, then the IRQ2 line to the DSP remains LO and no edge
will be seen. In order to catch all interrupts, IRQ2 interrupts
should be configured as level sensitive in the ICNTL register.
The four least significant PIO pins are assigned unique vector
addresses. An interrupt on any of the remaining eight lines
(PIO4 to PIO11) will trigger a separate fifth PIO interrupt that
has its own vector address. The PIOFLAG register can be read
to determine the exact source of this fifth interrupt. An 11-bit
PICMASK register can be used to enable or disable any or all of
the eleven peripheral interrupt sources. The program memory
address reserved for each of the interrupts is summarized in
Table VII.
Table VII. Interrupt Vector Addresses
Function Vector Address
RESET Startup (or Power Up with
PUCR = 1) 0x00 (Highest Priority)
Power-Down (Nonmaskable) 0x2C
ADC End-of-Conversion Interrupt 0x30
PWMSYNC Interrupt 0x34
EIU Loop Timer Timeout Interrupt 0x38
PIO4 to PIO11 Interrupt 0x3C
EIU Counter Error Interrupt 0x40
ETU Interrupt 0x44
PIO0 Interrupt 0x48
PIO1 Interrupt 0x4C
PIO2 Interrupt 0x50
PIO3 Interrupt 0x54
PWM Trip Interrupt 0x58
SPORT0 Transmit 0x10
SPORT0 Receive 0x14
Software Interrupt 1 0x18
Software Interrupt 0 0x1C
SPORT1 Transmit (or IRQ1) 0x20
SPORT1 Receive (or IRQ0) 0x24
Interval Timer Interrupt 0x28 (Lowest Priority)
Interrupt Masking
Interrupt masking (or disabling) is controlled by the IMASK
register of the DSP core and the PICMASK register. These
registers contain individual bits that must be set to enable the
various interrupt sources. It is important to remember that if
any peripheral interrupt is to be enabled both the IRQ2
interrupt enable bit (Bit 9) of the IMASK register and the
appropriate bit of the PICMASK register must be set. The
configuration of both the IMASK and PICMASK registers of
the ADMC401 is shown at the end of the data sheet.