Ipod classic: disabling charging doesn't work

[momom]

Maybe its because of incomplete power managment: disabling charging in usb mode under settings doesn't work, it always charges the Ipod. Its not a problem on newer usb hub chips but on older, specially VIA, that can't handle in a right way consumption over 100/200 mA, it may cause disconnection or faulty connection during file transfer.
It's a pity this power problem - like Hamlet: to box or not to box? On one side the immense custumisations and codecs on the other a real short battery life - what a stupid little command line can do :-)
Mine is a 6th gen 160 GB

[saratoga]

Maybe its because of incomplete power managment: disabling charging in usb mode under settings doesn't work, it always charges the Ipod.

How are you verifying this?

[momom]

In settings I'll choose not to charge during usb connection. This setting has no effect on classic: if you connect it to pc it charges always (battery indicator and remaining power after disconnecting). It's only a problem with older devices with defect power managment, like the Via chips have and on some laptops. It would be a big problem if you have only a pc with those USB: file transfers would be diffucult if not impossible (disconnections, writing error etc.). On my old desktop it is impossible, it chrges the device (slowly) and disconnects as I try to tranfer files or after a few seconds/minutes in best case. I know that my mb and pci card (both with VIA chipset, I have 8 USB ports) can't handle in a right way 500 mA even if there is only one connection.
Post Merge: January 08, 2012, 11:02:14 AMI noticed another little problem if I set I set pause on headphone unplug (or pause and resume), and disconnect headphones auto shut off won't work, Ipod remains on.

[torne]

I don't know how this is implemented on the Classic, but I redesigned the USB charging system a while ago, so I know how it should work For the older ipods, setting charging to "off" limits current to 100mA; it does not prevent it from drawing power entirely. The charging indicator will still show up. If the driver for the Classic is implemented correctly then it should behave the same, but I didn't write the code for that device.

We don't currently implement drawing no power at all, because this can cause problems if the battery is very low; you may not be able to get far enough into Rockbox to re-enable charging before the battery gives out. We're open to patches to handle this - the driver for the older iPods does know how to stop drawing power, it's just not used.

Any USB port/hub *must* be able to supply 100mA to each connected device, otherwise it is way outside the spec and also going to have serious problems supporting any devices at all; devices are allowed to draw this much power without having completed a handshake with the host.

[momom]

you are, of course, right :-) As said it is a problem mainly with older VIA chipsets and frankly, who cares? My desktop is 8 years old, its time for a change :-)
The big problem on Classic is power managment, its really a mA sucker, without a headphone or line plugged its a miracle if it arrives to 7/8 hours, not to mention if you skip and search...

[torne]

you are, of course, right :-) As said it is a problem mainly with older VIA chipsets and frankly, who cares? My desktop is 8 years old, its time for a change :-)

Well, if the problem is that it's drawing 500mA when you don't want it to, then that's not "who cares", that's "the power management logic for the Classic is missing a feature we should support". If the Classic supports USB charging power control then it should work, and if not then it shouldn't be offering the option.

[dreamlayers]

Currently, the current limit is set using "GPIOCMD = 0xb060e | (on ? 1 : 0);" in usb_charging_maxcurrent_change() in firmware/target/arm/s5l8702/ipod6g/power-ipod6g.c.  Maybe it's instead necessary to use the usbdevstat bits in the MBCC7 register in the PCF50633 PMU? They're documented on page 79 of that PDF.

[prof_wolfff]

Hi!, i did some tests while developing the usb_charger_detect patch for Classic, have some conclusions, not 100% sure as usual:

Charge management is done by external LTC4066 IC instead of PMU, but dont know how the power paths are routed exactly. I think (not yet verified) OF zeroes registers 0x43..0x49, anyways it doesn't matter if the charge management is enabled or disabled on PMU, the battery is always charged in the same way, seems that writing to those registers has no effect, in addition status regs (0x4b..0x4d) are not being updated while the whole charge is in progress.

I also figure USB Vbus is only connected to PMU by EXTON2 pin because cannot observe coherent activity on other (documented) registers when plugging/unplugging USB.

If i am right, 100mA/500mA current selection could be done using HPWR pin on LTC4066, the GPIOCMD register is undocumented for me but i suppose it is connected to this pin, not tested yet but will check on the next battery charge cycle, its supposed @100mA charging should take more time.