Catch me at RECON, Production Updates, Testing, Shipping Plans
First off - if you happen to be at RECON this weekend in Montreal, Canada, be sure to say hello! I'm giving a brief talk there around side-channel analysis and glitching attacks.
Anyway we received the boards from SUNSEL - here is a few boxes with ~460 ChipWhisperer-Lite boards:
What's interesting is to compare the received quality between my overseas fab (which assembled the prototype boards) and these Sunsel boards. In particular we can look at the microUSB connector (which is a tricky combination of SMD and through-hole):
You can notice that the overseas fab use a small amount of solder paste and never touched anything up - the result being insufficient soldering on the tab/shield connections. This translates into considerably less mechanical strength - I'm a lot happier with the Sunsel boards having a complete fill!
Sunsel already performed automatic optical inspection on the boards - that should have caught issues such as components tomb-stoning, solder bridges, or insufficient solder/bad joints.
We'd still like to ensure there are no other odd issues, and have set out to build a test jig for validating all the boards we ship out. This is based somewhat on the Adafruit tutorial on test jigs - there is a number of pogo pins that make contact with the PCB, the following shows a ChipWhisperer-Lite in the partially finished jig (I'm waiting on more pogo pins which should be here today):
The software side performs the following tests:
- Programs the SAM3U with the USB Firmware.
- Checks the current consumption with the FPGA configured and ADC sampling @ 105 MS/s, should be around 220mA.
- Tests the XMEGA programmer function.
- Tests the AVR programmer function.
- Tests the LEDs (requires a human to validate LED pattern).
- Tests the analog bandwidth from ~1-150MHz using a signal generator (because the XMEGA target is connected still, it's not a proper test of the -3dB bandwidth, but instead compares it to a known-good benchmark).
- Tests both of the glitching MOSFETs (high-power and low-power).
- Performs a CPA attack, tests how many traces are required to break the target.
- Tests I/O pins.
The result is a single-page test report that is automatically printed and will be included with your CW-Lite. Here's an example of what it looks like - it includes graphs of the measure bandwidth response, along with the glitch waveforms:
We are awaiting a few more parts before can start packaging and shipping. Lots of larger quantity orders - here's what 250 IC sockets + 500 LEDs + 1200 standoffs looks like:
We're also awaiting more boxes - we have about 300 of these smaller ones, and about 50 of slightly larger ones (with more on order), along with anti-static bags (ChipWhisperer-Lite for scale):
We are still finalizing shipping plans. It looks like the fastest way might be to ship directly ourselves, rather than shipping en-mass first and then distributing. We'll have more details in a later update, but currently our plan is:
- Plan on using standard airmail to all destinations (which is what our original Kickstarter shipping prices were based on). This should get everyone's reward in their hands well before the end of August.
- Find the cheapest courier rates - if anyone wants faster shipments, we'd offer this for the "incremental cost" based on destination. This would take into account the base shipping cost "built into" our Kickstarter rewards, along with the additional cost international backers have already paid.
- As Kickstarter rewards are not straight product sales, when rewards are sent the associated customs documentation will reflect this. Hopefully this means backers will not have any additional fees/duty/taxes due.