ST72321Rx ST72321ARx ST72321Jx
186/193
15 KNOWN LIMITATIONS
15.1 ALL FLASH AND ROM DEVICES
15.1.1 External RC option
The External RC clock source option described in
previous datasheet revisions is no longer support-
ed and has been removed from this specification.
15.1.2 Safe Connection of OSC1/OSC2 Pins
The OSC1 and/or OSC2 pins must not be left un-
connected otherwise the ST7 main oscillator may
start and, in this configuration, could generate an
fOSC clock frequency in excess of the allowed
maximum (>16MHz.), putting the ST7 in an un-
15.1.3 Reset pin protection with LVD Enabled
160, when the LVD is enabled, it is recommended
not to connect a pull-up resistor or capacitor. A
10nF pull-down capacitor is required to filter noise
on the reset line.
15.1.4 Unexpected Reset Fetch
If an interrupt request occurs while a “POP CC” in-
struction is executed, the interrupt controller does
not recognise the source of the interrupt and, by
default, passes the RESET vector address to the
CPU.
Workaround
To solve this issue, a “POP CC” instruction must
always be preceded by a “SIM” instruction.
15.1.5 External interrupt missed
To avoid any risk if generating a parasitic interrupt,
the edge detector is automatically disabled for one
clock cycle during an access to either DDR and
OR. Any input signal edge during this period will
not be detected and will not generate an interrupt.
This case can typically occur if the application re-
freshes the port configuration registers at intervals
during runtime.
Workaround
The workaround is based on software checking
the level on the interrupt pin before and after writ-
ing to the PxOR or PxDDR registers. If there is a
level change (depending on the sensitivity pro-
grammed for this pin) the interrupt routine is in-
voked using the call instruction with three extra
PUSH instructions before executing the interrupt
routine (this is to make the call compatible with the
IRET instruction at the end of the interrupt service
routine).
But detection of the level change does not make
sure that edge occurs during the critical 1 cycle du-
ration and the interrupt has been missed. This may
lead to occurrence of same interrupt twice (one
hardware and another with software call).
To avoid this, a semaphore is set to '1' before
checking the level change. The semaphore is
changed to level '0' inside the interrupt routine.
When a level change is detected, the semaphore
status is checked and if it is '1' this means that the
last interrupt has been missed. In this case, the in-
terrupt routine is invoked with the call instruction.
There is another possible case i.e. if writing to
PxOR or PxDDR is done with global interrupts dis-
abled (interrupt mask bit set). In this case, the
semaphore is changed to '1' when the level
change is detected. Detecting a missed interrupt is
done after the global interrupts are enabled (inter-
rupt mask bit reset) and by checking the status of
the semaphore. If it is '1' this means that the last
interrupt was missed and the interrupt routine is in-
voked with the call instruction.
To implement the workaround, the following soft-
ware sequence is to be followed for writing into the
PxOR/PxDDR registers. The example is for for
Port PF1 with falling edge interrupt sensitivity. The
software sequence is given for both cases (global
interrupt disabled/enabled).
Case 1: Writing to PxOR or PxDDR with Global In-
terrupts Enabled:
LD A,#01
LD sema,A
; set the semaphore to '1'
LD A,PFDR
AND A,#02
LD X,A
; store the level before writing to
PxOR/PxDDR
LD A,#$90
LD PFDDR,A ; Write to PFDDR
LD A,#$ff
LD PFOR,A
; Write to PFOR
LD A,PFDR
AND A,#02
LD Y,A
; store the level after writing to
PxOR/PxDDR
LD A,X
; check for falling edge
cp A,#02
jrne OUT