Looks like I spoke too soon.

I'm playing with the _40h_test.mxp, and it looks like it works fine with
only a couple LEDs lit, but starts to get a lot of noise with lots
turned on. If most of the board is turned on, then I start getting lots
of key press signals, even though nothing is being pressed (there's not
even a light panel attached).

So I must have a short or something, but I can't see anything. Can
anyone tell me where I should look, what might cause this?

Edit: further experimentation makes it look like this is only happening 
when the ribbon cables are plugged in. Maybe I have bad ribbon cables, 
or there's interference or something?

check the solder joints on the cable sockets, make sure there's solder 
all the way around the metal.  you can also re-heat the joints to 
'reflow' the solder, or add more solder.

it could also be that your ribbon cable connectors are loose, check it 
visually, there shouldn't be a gap between the cable and the plastic 
header.

sounds like a flaky connection or two.

But if that's the case, why would it only do it when the cable is 
plugged in? Seems like it would still be flaky when there's no cable.

Oh dear. This is so stupid I just have to share.

I forgot a part.

While I was waiting for the iron to heat up again, I absentmindedly 
shook the parts envelope... A ceramic cap fell out. So I bet once this 
is on, no more ghost signals...

Kudos Monome, for making a device so robust it'll still work with one of 
the parts missing!

Edit: nope, still nothing. It acts just the same as before.

I really don't think it's the connector. I've looked really carefully at 
those joints, none of them seem to be bridged or loose or anything.

It would have to be a bridge between the two rows of pins, since one of 
the rows is power for lights and the other is button presses.

So now what can I do? I don't know how I can take the connector back off 
and try again.

one thing about those connectors is that they lie very flat against the 
LED board and if soldered down don't leave much room for the connector 
to squeeze onto, does the ribbon cable fit all the way down onto the 
connector?

maybe some pics could help.

This is the connector on the logic board side, I dont have an LED board 
yet.

The more I think about this the more I think it's just interference, and 
will go away once this connector is actually plugged into something.

It's only the one cable (the one next to the 165), only when a lot of 
lights are on, and you can stop it by holding the cable tightly between 
your hands, or putting the other cable next to it.

hey, you'll get crazy noise if the pins aren't pulled either high or low 
on the button side, which is why you're getting weird data. it'll all be 
fine once you hook it up to a led matrix board.

Exactly my thoughts. Half those pins have current on them going nowhere, 
the other half are listening for current. Of course they'll radiate a 
bit.

I'm just glad I didn't break anything trying to fix this imaginary 
problem.

One question remains though. Board worked fine, minus one cap (next to 
the big chip with the firmware). What's that cap good for, if the board 
still works without it?

you can't run the logic board without the LED board, glad that got 
sorted.

the cap is a 'bypass cap' they're placed next to the power supplies of 
ICs to help smooth out noise from power and ground that the IC and 
others around it can generate.  they're helper parts and not always 
critical it's just standard design practice to put them there.

Ah, that makes sense.
The assembly instructions say to use a wire and an LED to simulate the 
button pad for testing, which seems to work for fewer than 16 or so LEDs 
at once.

I am used to software, where noise and interference isn't even a 
consideration. : )