Mon Oct 22 16:46:26 EDT 2012

Next: board powered from USB

- Power board form USB and disconnect PK2/FTDI.
- Drive the interpreter

Power from USB works with serial loopback.  However, I need to unplug
it for about 5-10 seconds before replug.  Maybe needs brown-out reset?

Hmm... it worked for a bit now not at all.  Ok it does work when I
wait 5 seconds and plug it into a different hub, facing down.  Maybe
solder issue, or a voltage issue.  Maybe I should try with an LF?

It's very picky about the port it goes into.  Plugging it directly
into the tower doesn't work either..  Only the powered hub.

I don't think it's the voltage.  Very unpredictable behavior.
Probably some soldering issue.

It all works fine as long as its powered by the PK2.  It seems really
just to crash.  Maybe a pause at bootup would work?

I checked on the scope.  Don't see much on the outside except for the
3.3V VUSB signal.  It stays on for about a second, then discharges to
about 1V over a little over 2 seconds.  This cycle repeats
indefinitely.  I don't see anything happening to the supply voltage or
the reset.  The device descriptor is received OK.

Very strange.  

This might be related to the thing not working with just a serial
console attached..

EDIT: Trying again using interactive PK2 it works for a bit, then it
disconnects again.

Maybe it's time to switch back to uart and see what is actually going

EDIT: Trying some more things.
Added a startup delay: 0 for 0 for 0 for next next next, which seems to work.

EDIT: When I connect just the ground from PK2, it seems to run without

Disconnecting the ground then keeps it running.  Nope, this is not
consistent.  I just switch the hub's switch (unplug/replug) and it
starts OK.  What can this be?

Touching the solder joints makes it crash.
If I don't send any data it lives longer.

Running from with PK2 with USB power disconnected works fine.
I can only make it crash by touching the oscillator connections.

Can this be caused by a floating input pin?

More weirdness.  PK2 is now supplying 3V.  I noticed this when trying
to close the USB 5V jumper while running. 

After programming, pk2 give 4.3V
Running with that, after a while it gives up.

Stopping PK2, then starting again runs it at 3V.
First time no proper comm (bad-reply) 2nd time it works.

Could it be that the regulator is enabled, but the signalling voltage
is not 3.3V?  EDIT: It is.

So everything seems to work at 2.7 and 3.6 V, but 4.3V doesn't work.
Maybe the power is just too unstable?