Y-cable and USB-OTG: mutually exclusive

Joachim Steiger roh at openmoko.org
Wed Apr 30 20:30:52 CEST 2008

Werner Almesberger wrote:
> Joachim Steiger wrote:
>> - i really like my mini-usb. (for mechanical, practical and spread in
>> the real world)
> Me too. Of course, if the market really switches to micro-USB, the
> scenario may be very different in a few months. If only Nokia use
> micro-USB, we'll be of course happier with mini.


> We may be able to design for one, and change our minds later. The
> problem here is mainly the case. Apparently, it's fiendishly
> expensive and time-consuming to make the molds, but some changes
> can apparently be accommodated without too much pain.
>> - the N8xx series is the only device i know i were near which does OTG
>> as far as i know.
> I don't know enough about OTG to tell whether it's actually good or bad.
> Of course, if it becomes popular, we'd look rather silly with our own
> hack.

indeed. i also see nothing which stands in the way of having the usb-b
do all what otg needs on a electrical/protocol level when the soc does
all that for us.

>> * i would rather like A due to the more widespread of cables and
>> devices, but its propably too big
> Yeah, I'd have to dig quite deep into my gadgets box to actually
> find something that does mini-A.
>> * put a single 2-positions switch below the battery flap to switch this
>>  whole port to jtag.
> That's where your proposal gets a bit messy. It means that we couldn't
> have JTAG/serial console and host mode. So if anything blows up there,
> we're blind. Not the end of the world (after all, we survive quite well
> on the PC, without having all these niceties), but unpleasant.

when usb HOST is failing you can still debug that via ssh over the
usb-client mode/charger port (or kernel-gadget from cdc-ethernet to
cdc-adm ;))
so only when usb is dead completely you cannot debug usb-host, only the
otg port via jtag

>> please let me know what you think
> The host vs. JTAG/console exclusion aside, it has a nice ring to it.
> One bit that may be troublesome is the USB switch, though. USB is very
> finicky about line impedance, reflections, an all that, so making a
> switch that doesn't distort the signal too much may be difficult to
> make.

yes. but its only usb1.1 12mbit, not the 480mbit otg capable usb which
will go through the switch/switched latch so this should be doable
without too much hassle. i bet even a stone-old cmos 4066 latch would do.


Joachim Steiger
developer relations/support

More information about the hardware mailing list