I have been working up a special adapter for connecting an Atrix lapdock
to a Raspberry Pi.
Basically a power switch with a high-side P-channel MOSFET feeding USB
power to the Pi, plus there is a mechanical adapter for the HDMI channel.
The switch seems to work perfectly when the power source is a "normal"
5V wall wart. My question has to do when the actual lapdock and HDMI
connections are set up.
I have measured 5.09V at the output of the switch when turned on. So
far so good. The HDMI has a 5V supply rail, too. Both of these are
routed and shorted together inside the Pi (also confirmed on the
Beaglebone Black). However, for some reason when I measure the power
rail on the Pi with just the HDMI connected I read about 0.8V. When
both are connected something weird happens.
It would seem there is an overvolt or overcurrent protection switch
inside the lapdock which shuts down the 5V supply rail after about
100ms, followed by what appears to be a capacitive discharge for the
next 150ms, lather rinse repeat. Ok, dandy that would be fine, but
because the lapdock has the DDC/CEC/Ground pin separated from the other
grounds (and is used to detect the HDMI connection) both USB and HDMI
must be connected for the lapdock to turn on the USB power and video.
The other funny thing, is that if I connect the USB with a straight
cable none of the protection features are activated, and the Pi starts
properly.
As far as the lapdock and Pi are concerned the USB should only see a
MOSFET source and drain in series, with a pretty good Rds(on). The Vth
value of the FET is -1V, and the gate voltage at activation is about 1V,
which is plenty enough to put the PFET into saturation (confirmed by the
voltage at the source).
I am probably not the sharpest pencil in the box, and am pretty new at
these things, but the system is "almost" there. If I knew how I would
post a schematic here.
So what gives. I have thought that the gate was a little bit too
"floaty" so I tried putting a 1uF cap to ground on the gate. The cap
inrush current causes the switch to briefly activate out of turn when
power is initially applied, so that is a no-go. I am considering if I
must put a series diode after the source pin to ensure only the 5V is
present at the USB lapdock end. The diode drop would be counteracted by
the combined USB/HDMI voltage inside the Pi, but I would really prefer
to not rely on that "feature" for the general use case.
Ideas?
J.R. Stoner
_______________________________________________
dorkbotpdx-blabber mailing list
[email protected]
http://music.columbia.edu/mailman/listinfo/dorkbotpdx-blabber
