Controlling USB connections using a BeagleBone

[David Frey]

Hi,

I'm trying to prototype an automated testing solution to plug in and
unplug
USB devices using software control. At a very high level, the system
would
look like this:

+------------+
| USB Device | +--------------------+
+------------+ | BeagleBone |
| | |
| | +----------------+ |
Some sort of --------> /|----------+-| GPIO as Output | |
switching circuit | | +----------------+ |
| +--------------------+
+----------+
| Computer |
+----------+

I'm pretty sure I can figure out the software part, but I don't have a
hardware
background, so I don't really know what my "switching circuit" will be.
I am
hoping that someone in the BeagleBone community can give some advice in
that
area.

This is how I have conceptualized the "switching circuit":


BeagleBone 3.3v
GPIO output
|
|
+-----+
| ??? |
+-----+
|
+-----+
| |
+-------+ |
USB Ground ----| Relay |------------
+-------+ |
|
+-----+
| |
+-------+ |
USB 5V -------| Relay |------------
+-------+ |
|
+-----+
| |
+-------+ |
USB D+ -------| Relay |-----------
+-------+ |
|
+-----+
|
+-------+
USB D- -------| Relay |-----------
+-------+

The idea is that flipping by switching on/off the GPIO, I will
simultaneously
connect/disconnect all 4 USB wires. I'm guessing that the 3.3V signal
from the
GPIO will not be the exact voltage required to activate the relays.
That is
why I have the "???" box. Presumably there is some circuitry required
in order
to create a suitable signal to control the relays. I would also
ideally like
to have an LED in the path of the GPIO so that I can see when it is
activated.
Any hints as to what the "???" box would contain would be greatly
appreciated.

The USB 5V line can draw up to 1000mA in USB3.0 (yes, I know there are
more
wires, but I am thinking ahead a bit), so I would like to choose relays
that
are capable of controlling that load. Are there components that
package a
bunch of relays together into a small-ish convenient package or am I
stuck
piecing together individual relays? Perhaps relays aren't even the
correct
solution. Let me know if there is some other technique that is more
appropriate for doing switching.

I realize that the BeagleBone is overkill for this application, but I
eventually intend to use the network capability of the BeagleBone for
switching
control and results reporting.

If this is not an appropriate place to discuss the design of circuits
connected
to a BeagleBone, please point me in the right direction.

Thanks,
David

[fanfan]

 why not try to use GATE like 4053 - SPDT 

[dave higton]

Quoting fanfan <fkp...@gmail.com>:

> why not try to use GATE like 4053 - SPDT

Because its resistance is too high for this job.

Dave

[E V]

Why not MOSFET? Rdson can be quite low... And power consumption is not very high. Does the Rdson (can be quite low) distort the data lines?

2012/5/18 dave higton <daveh...@dsl.pipex.com>

-- To join: http://beagleboard.org/discuss
To unsubscribe from this group, send email to:
beagleboard...@googlegroups.com
Frequently asked questions: http://beagleboard.org/faq

[dave higton]

Quoting E V <evia...@gmail.com>:

> Why not MOSFET? Rdson can be quite low... And power consumption is not very
> high. Does the Rdson (can be quite low) distort the data lines?

If you look at the USB spec, the signal levels on the two lines are
quite tightly defined, and the bus is terminated. The resistances
you can expect from those families of analogue switch are enough to
change the signal levels quite a lot, and put them out of spec.

MOSFETs are a possibility, but you have to remember that a discrete
MOSFET normally has a diode between source and drain, so it may not
behave quite the way you expect. The diode may or may not be shown
explicitly in the symbol in the data sheet, but it normally exists.

Dave