Power leakage: save 40mW by turning GSM on

[Werner Almesberger]

I'm continuing this thread on the more appropriate hardware list.

Background: I'm trying to get the system into a very low-power
state (and very low power variation state) so that I can measure
the effect of the various WLAN power saving options. This yielded
a few surprises. Instead of a few mW, I ended up with suspend
power in the order of up to 0.4W :-(

I found one more funny today: activating GSM saves about 40mW.

					min	 avg	  max	    stddev
drain, power cycle, pull USB            242.40mA 243.17mA 243.59mA  0.370mA
suspend                                  81.90mA  82.96mA  85.91mA  1.043mA
resume                                  240.53mA 243.16mA 248.99mA  2.867mA
suspend                                  79.49mA  79.94mA  80.52mA  0.339mA
resume                                  241.32mA 241.80mA 242.48mA  0.376mA
echo 1 >/sys/bus/platform/devices/neo1973-pm-gsm.0/power_on
                                        229.18mA 230.48mA 233.02mA  1.343mA
suspend                                  67.31mA  69.08mA  71.82mA  1.553mA
resume                                  227.32mA 228.56mA 232.24mA  1.429mA
suspend                                  66.87mA  68.46mA  70.53mA  1.343mA

(This is "battery" current, voltage is 4.0V, each measurements
consists of ten samples within 4-5s, integration period for each
sample is 200ms. GSM introduces some noise, so when I hit a "busy"
period, I rejected the result and retried.)

This run is without all the other power cutting I did yesterday.
So things like BT and GPS are up and running.

My best-case suspend power is still about 80mW, so I haven't found
the black hole yet.

- Werner