Why not just put a (mini?) USB B connector on your cape and use a very
short A to (mini?) B cable to loop the existing host port up onto the
cape? They you can put a hub on a cape and the form factor won't be
much larger. Since your requested modification would render the host
port unusable anyway when your hub cape is installed, it's the same
number of USB ports this way.
If you make your cape only usable on the top of a cape stack, you don't
even need an EEPROM or to have all the P8/P9 headers fully populated
(they'd just be there to support it on and provide 5V), which should
give plenty of room for USB host ports on the sides above the header
locations.
I have to say that USB connection accesible from the side headers would be really nice feature. I aggre that you can loop a regular usb cable, but this way makes ot more robust and that just doesnt look good. Imagine someone buildibg a device basing on bbb. Everything could be connected to the cape and all a worker has to do is place the bbb on the cape.
I have to say that USB connection accesible from the side headers would
be really nice feature. I aggre that you can loop a regular usb cable,
but this way makes ot more robust and that just doesnt look good. Imagine
someone buildibg a device basing on bbb. Everything could be connected to
the cape and all a worker has to do is place the bbb on the cape.
USB uses differential signaling with controlled impedance. The connectors
used for the capes would violate this specification and cannot support USB
signaling.
Even though this might violate the spirit of the specification, it is highly
likely this will work. Most PCBs that use differential signals are very loosely
coupled by the nature of the traces having to be side by side. The traces can
wander quite a bit and still maintain a reasonably close differential impedance.
The more important aspect is to maintain a matched distance. Depending on
which speed of USB that you want support for, this can easily be achieved even
going through a header - as even this doesn’t have to be that precise.
Consider this, just about every motherboard in existence has USB port headers
to go to a ISA/PCI/PICe slot connector. All of these go to a ribbon cable with two
insulation displacement connectors - one at each end. These surprisingly work
well for most PC users - go figure - even with all the changes in impedance.
I suspect a 1/2" or so header won’t be as much of a challenge.
Doesn’t sound like Gerald has any intention to make this change anyway, but
if someone wanted to try this as an option for their own board, with planning, it
should be workable.
