In the same experiment in which I connected our CMU200 instrument to a decased
Pirelli motherboard (see rssi) I also observed the Tx power levels put out by
this phone in the PCS band by commanding the phone (by way of the CMU200 acting
in the role of a BTS in signaling mode) to transmit at each PCL from 0 to 15
and observing the actual power measured by the CMU200.  Here are the findings:
the Pirelli puts out 29.2 dBm at the highest PCL 0 (spec number is 30 dBm),
whereas for all other PCLs the measured power was within 0.2 dB of the spec
number.  Thus it appears that Pirelli's Tx calibration was done like TI's
LoCosto, with only the highest power level shifted down by 0.8 dB while all
others are spec numbers.  This is a smaller fudging than that which was done by
Openmoko and which we currently replicate on our FCDEV3B modems.