Pirelli fw behaviour on incoming calls: when the vibrating alert mode is
selected, the vibrator runs in pulses, each pulse appears to be 1 s in total
period, duty cycle too difficult to determine.  If the alert mode is set to
vibe and ring, then these vibrator cycles run in parallel with the ringer on/off
cycles - the vibrator cycle is faster, and the two cycles appear to be
independent, as in not explicitly synchronized.  If the alert mode is set to
vibe then ring, in that case only one vibration pulse is produced, followed by
ringing only, with no more vibration.

Pirelli fw behaviour on incoming SMS: when the alert mode is vibrate only, then
a total of 4 vibration pulses are produced.  If the alert mode is set to vibe
and ring, then only 3 vibration pulses are produced instead of 4, and this
vibration cycle coincides with SMS alert ringtone playing.  If the alert mode
is set to vibe then ring, in that case only one vibration pulse is produced,
followed by the SMS alert ringtone.

Pirelli's fw also has a user-accessible debug mode for the vibrator: *#2003#
puts the vibrator into the same state as for incoming calls, where it pulses
continuously until stopped with another *#2003# command.  The 1 s pulsing feels
exactly the same whether it's an actual incoming call or *#2003# test mode.

At FreeCalypso HQ we have a decased Pirelli motherboard that has the LCD and
keypad still attached, but no other peripherals.  Turning on *#2003# test mode
and putting an oscilloscope probe on an accessible test point in the vibrator
driving circuit has revealed the vibrator pulse sequence to be 500 ms on, 500 ms
off, endlessly repeating.