USB device and host on a secondary board
Casainho
secondary board, however here is a product doing it:
http://www.coolcomponents.co.uk/catalog/product_info.php?products_id=528
--
Cumprimentos,
Jorge Pinto
john cooper
> I thought it wasn't possible/good idea to root USB signals to a
> secondary board, however here is a product doing it:
> http://www.coolcomponents.co.uk/catalog/product_info.php?products_id=528
It's not impossible. It is just a question of where
you decide to spend the signal integrity budget. I think
that design was more concerned about cramming connectors
into a particular footprint vs. driving maximum specified
cable runs. I'd bet they weren't too worried about
maintaining a 90-ohm line impedance with the 100mil sockets.
FSL even pooh-poohs running the usb signal lines over a
ground plane break.
That said I thought FSL was a being a bit overly pessimistic
in their recommendations of usb trace routing. But it
is a 400Mb/s path and they did design the data separator
in the usb phy.. so maybe they have some concerns. ;)
-john
Opendous Inc.
Speed) of the i.MX233. Electrically this is completely different. HS
USB signals are very fragile and delicate and need extra care and
attention. FS USB signals can be routed using clothes hangers.
On Jun 23, 11:10 pm, john cooper <john.coo...@third-harmonic.com>
wrote:
john cooper
> That board's USB is 12Mbit/s (Full Speed) and not the 480Mbit/s (High
> Speed) of the i.MX233.
It is indeed. I hadn't dug that far into it.
> Electrically this is completely different. HS
> USB signals are very fragile and delicate and need extra care and
> attention. FS USB signals can be routed using clothes hangers.
I tend to err on the conservative side of design margin and
I'm inclined to agree. But empirically the vanilla PC mb
breaks out front panel USB connectors through 100mil headers
and less than carefully impedance controlled cables. Despite
those sins, somehow sufficient margin remains in practice to
sustain a usable 480Mb/s path. So I think in general a
substantial budget is possible though certainly not guaranteed.
On at least the lqfp128 version of the imx233 the usb data
lines are bonded out immediately adjacent to the 24Mhz xtal
which intuitively isn't the ideal location SI-wise. The
layout needs to juggle XTAL and usb traces which for their
own reasons want to remain on the component surface. This
may be one concern as could be whatever FSL conjured up for
the usb data separator/phy. It would be interesting to
measure the eye pattern daylight w/r/t error rate relative
to FSL's eval board to better weigh the criticality of their
routing precautions. Although frankly I've just followed
conventional guidelines hoping that exercise was avoidable.
-john