I/O expansion header without voltage converters

Hi,

I'm talking with a subcontractor in order to build some beagle
expansion board for me. It's going to be connected to the expansion
header with a flat ribbon cable. After initial discussions it looks
like we are going to use some PIC microcontroller from Microchip which
runs on 1.8V. This means it looks like we can remove the typical
voltage converters and connect the microcontroller directly to the
header on the beagle board.

Do we need to add clamping diodes not to burn I/Os when both boards
run on 1.8V and the I/Os on both boards are set by mistake to outputs?

Is there anything else to take care of?

Your experiences and input are highly appreciated.

Regards,

Robert

Hi Robert,

Do we need to add clamping diodes not to burn I/Os when both boards
run on 1.8V and the I/Os on both boards are set by mistake to outputs?

How would a clamping diode save you for this. The short answer is: Avoid
having both output driving the same net simultaneously. In case needed due
to some kind of special hw-protocol use open-drain-buffers to the pins. Most
pins on the OMAP start in safe-mode (=high-impedance) and you can just mux
what pins are needed to input and output and thereby be able to totally
avoid the problem?

For further information about which pins start in which state please consult
the OMAP datasheet (omap3530.pdf)...

Best regards - Good luck
  Søren

Hi Søren,

Hi Robert,

> Do we need to add clamping diodes not to burn I/Os when both boards
> run on 1.8V and the I/Os on both boards are set by mistake to outputs?

How would a clamping diode save you for this. The short answer is: Avoid
having both output driving the same net simultaneously.

I know I should avoid it, but since the add on board will be used for
trainings I am afraid it's not entirely up to me;)
It would be desirable for the trainees not have a dead board after
making a software mistake and that's why I would like to find a good
way to protect it.

A solution might be to add series resistors between the PIC and the
OMAP to limit the current to something non fatal.
100 Ohms or 1K should work.
With 1K there is a danger to get noise coupled in from high speed
lines.
Placing the resistor as close as possible to inputs might limit this
problem.

I would rather use 1K close to inputs to limit the current to 1.8mA
*) avoid latch up effects in case the OMAP is powered, but the PIC is
not powered yet
*) all OMAP and PIC GPIOs are set as outputs and looking at each other

What do you think?

Regards,

Robert