From: Andrew Kohlsmith (mailing lists account) <aklists@mi...> - 2010-07-16 05:01:49
Built a new expansion module for the Overo... 5 port 10/100 switch, PoE, DVI,
4 port USB 2.0 switch, audio, MMC, the whole nine yards...
After some initial troubles with the 1.8V supply acting funny and tracking it
down to some bootstrap ferrite beads connecting the 1.8V and 3.3V supplies
together, I gather up the courage to plug the Overo Earth into the system.
Silently offering up a prayer to the silicon gods, I turn on the bench supply
... nothing. Or rather, all supplies come up and D1 on the COM lights, but
nothing. Sniff test is passing so far, and I'm poking around and decide to feel
the COM. It's hot. Real hot. Over 50C hot. (Rule-of-thumb from the
industrial world: if I can't keep my hand on it, it's over 50C.)
Immediately power it off and curse myself for not doing enough testing.
I forgot that I'd shorted SYS_EN to ground as it is used to sequence my 1.8V
and 5V supplies. After reading about the SYSEN signal (honestly guys, why do
you rename the signals? it makes it difficult to try to figure out what's
connected to where) and discovering that the TPS65950 will pull it down as
it's sequencing a power-on and removing my testing short, i try again.
The console came up, and I'm now sitting at a happy boot screen.
I'm not sure why the OMAP+DDR would get so hot if the TPS65950 was having
trouble with SYSEN, but I'm very relieved that I didn't fry my COM.
So yeah.. tough little buggers, these COMs. I am always nervous about the
power supply routing on new designs, but I'm thankful that either by Gumstix
design or TI's design, these things can take a bit of abuse on the power side
and not let out their magic smoke.
Good job, guys.