[Fwd: Re: GTA02 USB and battery current profile] FYI

Andy Green andy at openmoko.com
Sun Mar 9 15:38:07 CET 2008


-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1


> >>> In any case, keep in mind usb hw changes, without them Neo won't be usb
> >>> certified and that's a really big issue for a sdk-like device on
> >>> manufacturers point of view. We can chat about it later ;-)
> >> Definitely the first step is putting that in front of Wolfgang to see
> >> how critical it is all things considered.  Even if we can't
> >> conform on this first run we need to as soon as possible.
> > 
> > afaict, I didn't see an otg transceiver on neo's hw. Am I right ?
> > If that's true, we won't ever have otg support.
> > 
> > The connector is piece of cake, just change from mini-b to micro-ab
> > (mini shouldn't be used anymore), but the transceiver involves much more
> > expenses and changes to the hw design. Is there any space in the pcb for
> > a new chip? And what about the i2c bus ? (otg transceivers are mostly
> > connected through i2c and/or ulpi interfaces).
> 
> I don't know much about OTG, but we currently have some discretes
> allowing to control 15K pulldowns on to the USB +/- and 1K5 to 3.3V.  We
> have a separate pump to make 5V for being host power that can be enabled
> and an overcurrent detect.
> 
> The CPU has a USB host controller and device controller logic that can
> operate on the same phy pins.
> 
> Is this adequate?  If it isn't and the transceiver replaces some or all
> of the existing discretes, then there is probably room.

This sounds adequate... but I'm concerned about id pin. Device should
switch to host role automaticaly upon detecting id-pin grounded (micro-a
connector attached to micro-ab receptacle) and switch to peripheral upon
detecting id pin floating (micro-b connector attached to micro-ab
receptacle).

D+/D- terminations will probably be inside the otg transceiver, so as
the charge pump, just gotta look what is in the market. Make some
ISPXXXX is enough.

if I could take a look at neo's schematics, I could at least wonder if
an otg transceiver would be necessary of not. The
datasheet/trm/whatever-doc from samsung cpu would also be handy.


- --------------


Somebody in the thread at some point said:

>> >> Is this adequate?  If it isn't and the transceiver replaces some
or all
>> >> of the existing discretes, then there is probably room.
> >
> > This sounds adequate... but I'm concerned about id pin. Device should
> > switch to host role automaticaly upon detecting id-pin grounded (micro-a
> > connector attached to micro-ab receptacle) and switch to peripheral upon
> > detecting id pin floating (micro-b connector attached to micro-ab
> > receptacle).

I think this detection will be OK since we can discover it from an ADC
on the PMU via a resistor network (ID is in the middle of 100K to 0V +
39K to switched 3.3V).  But it means polling this which is fine, I don't
have much idea how to handle that in terms of role switching... a daemon
/ kernel thread?  But it sounds possible.

> > D+/D- terminations will probably be inside the otg transceiver, so as
> > the charge pump, just gotta look what is in the market. Make some
> > ISPXXXX is enough.

If we are okay with what we have, we would leave it alone probably.  So
far I understood only the connector change is mandatory.

...

- -Andy
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.7 (GNU/Linux)
Comment: Using GnuPG with Fedora - http://enigmail.mozdev.org

iD8DBQFH0/ZPOjLpvpq7dMoRAh71AJ9vtp5gfreJoXle/f/DooU8d4cUTwCeNjqW
SX0dVlvBNHg6XsrEIwEn1PI=
=EmG2
-----END PGP SIGNATURE-----




More information about the openmoko-kernel mailing list