Benny Sj�strand wrote: >It's very difficult to find out what to do when do you dont have any >schematics on the board, and i supose Compaq will probably never leave >the schematics of any iPAQ H3xxxx in public. >And there is not any detailed specification of hw either left in public. >
The following document specifies where all the GPIO pins on the SA1110 are connected on H3600's and H3700's: http://www.handhelds.org/Compaq/iPAQH3600/iPAQ_H3600.html I updated and verified it with respect to schematics, which will not be released. We never received similar specs for the H3100 and H3800, so we have not posted a specs for those. >>All figures for these will be subject to a specific set of conditions. >>For example, it's well known that a floating IO pin causes a higher power >>drain than if it is permanently held high or low by some means. To get >>maximal lifetime out of the battery, you really have to be extremely >>careful about the setting of each and every single IO line. >> > >Here are muy big issue and question to Compaq, wich are all those IO pins >that i need to take care of ? While I (or we) dont know that, the iPAQ >will probably never get a sufficient low power consumption on sleep on >Linux. > See the above document. All the GPIO pins are attached to something. The ones that are attached to something that is turned off while the CPU is asleep should be left in a floating state, I believe. The PCMCIA signals are in that class -- we turn off power to PCMCIA when the CPU is put to sleep to save power. At least, that is what we intend to do -- it needs to be verified. Floating inputs are bad -- these can cause the input drivers of a chip to consume lots of power, but only if the chip is turned on. Floating outputs are fine and may save power as long as they do not cause problems with floating inputs. > >In the case of the iPAQ in arch/arm/mach-sa1100/h3600.c: > > /* Configure suspend conditions */ > PGSR = 0; > PWER = 0x1 | (1 << 31); > PCFR = PCFR_OPDE | PCFR_FP | PCFR_FS; > >So on Linux the PCMCIA control signals and static chip select are floating >during sleep, but dont know if this is good or bad, again, I supose that >Compaq have the right answer. > We probably should not be floating the static chip selects. It would be really great if someone with a multimeter had time to check some of these settings in the kernel and to make some measurements. Jamey _______________________________________________ http://lists.arm.linux.org.uk/mailman/listinfo/linux-arm Please visit the above address for information on this list.
