Seriously, the new boards look amazing; my pictures can't do them justice, so I may actually try flatbed scanning a few when I get home. Needless to say, professionally pasted, placed, reflowed, cleaned, and depanelized boards look a tiny bit better than those I assemble in a toaster oven.

Sadly, the best contract manufacturer around can't escape the power of an incompetent designer [me]. In this case, I (a) misrepresented the rotation angle of the RGB LED on the placement file, and (b) didn't put a clear dot marker on the PCB to indicate orientation. Okay, I didn't put _any_ dot on the PCB, which I will fix shortly. So after driving a few miles (so convenient) and picking the boards up this morning, all wrapped up in their adorable little anti-static bags and safely packaged between some anti-static foam... :

... I did this:

Ouch, painful. Fortunately, I was able to rescue every single LED, clean up each board a tiny bit, and resolder the LEDs in the proper orientation in a few hours. I've only tested 30 or so of the 120-piece run, but every single one works perfectly:

I also never got around to laser cutting a polycarbonate guide for the v0.91 pogo rig, so programming required a steady hand and careful aim (as the pogo pads are only 1.2mm in diameter):

However, it all went pretty smoothly now that I finally picked up a fresh cable for my USBtinyISP.

So. Lesson learned? We'll see. Due to an unexpected global shortage of JST GH headers, KeytronicEMS was only able to do a partial run, and that was only because I swung by the plant and dropped off my partial reel left over from v0.8 production (did I mention how much I love having my CM in the same metro area?). It sounds like the balance of the boards will hit the line at the end of the month, and I've already confirmed that they're able to change the machine setup so this problem should be put to bed.