
10 68HC912D60MSE4 Rev 2
March 14, 2001
in LHM since no crystal oscillations will be detected. This situation can arise with
short reset periods and/or crystals that exhibit slow start-up characteristics.
For the first 4096 cycles i.e. during the internal reset period, Limp Home mode will
be de-asserted if oscillator activity is detected by the clock monitor circuit -due to
the asserted Reset signal there can be no CPU activity during the Reset phase.
Following release of the external or internal POR RESET in LHM (which ever is
later) the crystal oscillator is sampled by the clock monitor circuit after another 4096
VCO clock cycles and at intervals of 8192 clock cycles thereafter until the crystal
is deemed to be operating. If the crystal oscillator is showing activity at the time it
is checked then it will be deemed to be good, even though it may not have fully
stabilized, and LHM will be de-asserted. This can cause the device to switch from
LH mode to normal mode with the CPU clocked from an unstable signal from the
crystal oscillator (see fig. 2) resulting in unpredictable function of the CPU.
The COP Reset doesn’t exhibit this behavior as, although the same reset
sequence is followed, the oscillator isn’t stopped.
When exiting Stop mode (DLY=1) a similar 4096 cycle delay is executed and
therefore this behavior could also show up at this time. In applications where this
is likely to be an issue, using pseudo-stop is recommended as an alternative.
Current draw will increase <100 ?A at 4MHz in pseudo stop versus stop mode.
Following a loss of external clock in normal operation, Limp Home mode will be
entered successfully but if the oscillator is reconnected for some reason a similar
situation may arise.
The Reset condition can be overcome by allowing the crystal oscillator circuit to
stabilize before releasing the external RESET line (see fig. 3). Operation is similar
to that shown in fig 2.
To determine if crystal is ’stable’ at the release of reset can be difficult and the time
can vary some from board to board. If the customer has special high impedance
probes, it is possible to monitor the amplitude of the voltage from XTAL to ground
(<2 pF scope probes are recommended). Please note that any loading on the
circuit can affect its operation. (Any resistance to ground or Vdd on the EXTAL pin
can greatly attenuate the amplifier gain and cause erroneous operation.)
A second way to measure the oscillator startup time is to monitor the XFC pin. This
method does not require a high impedance scope probe. The PLL will not lock until
the oscillator clock feeding it is present and stable. Remove the external reset
circuit and during power up watch the XFC pin. The voltage should start high (Vdd).
After the part releases internal reset it will drop to some stable voltage between
Vdd and Vss. If external reset (measured independent from this test) is held till this
’stable voltage’ time the oscillator will be stable. Please note the filter components
mounted on the XFC pin will affect this ramp (for evaluation purposes, alternative
components can be selected to provide a fast lock time). More than one board
should be measured because of pcb and crystal variability. It is also recommended
that the test be run over the operating temperature of the device.
Comentarios a estos manuales