I was able to put together six spectrometers for sale at Tindie before I ran out of AS72653 ICs, which have been hard to come by. Arrow Electronics finally got some supply today and I ordered 100 of each of the AS7265X (X = 1, 2, 3) 6-channel spectrometers, so Tindie sales should begin again before the end of the month.
I think I will discontinue the all-in-a-row version of the board and start producing (and continue testing) the latest design which is smaller, a little simpler to assemble, and arranges the spectrometer ICs in the same kind of tri-foil pattern used by AMS in their evaluation board; this just makes the most sense to me since then each of the three is treated the same, has the same optical path from the target object, sees the same illumination from the white led, etc.
Here is what the first prototype looks like (still missing the AS72653):
Latest design iteration of the "Compact, $25 Spectrometer". Left is board bottom, right board top.
The 3V3 and GND silkscreen labels are mistakenly swapped on the back (since corrected and submitted to OSH Park as v.03a). The green indicator led is the small component on the front top left. The two 0603 led footprints for source leds 1 and 3 are left unpopulated in case the user wants to add a particular frequency source, otherwise by default the 5700 K 90 CRI white led at the center (source led 2) provides the illumination. Switching between I2C mode (default) and UART mode is done by cutting the thin trace on the back solder jumper (between the top two pads) and soldering together the bottom two pads. The 1 MByte SPI NOR flash comes loaded with the appropriate firmware to make the spectrometer work, but the user can also upgrade with the latest firmware if and when this becomes available.
The spectrometer is smaller than a US Nickel (five cent coin), meeting the size goal of the project. The total cost to produce the board turns out to be less than $20 meeting the cost goal of the project. Of course, I have to sell the assembled and tested spectrometers on Tindie for more than the cost. But the interested maker can order the pcbs from OSH Park and the components from their favorite electronics retailer and assemble one or more of their own to save money.
I consider the design phase of the project complete. This project has been a lot of fun as well as technically challenging and satisfying. And while the results of my experiments to date using the spectrometer have not been wholly satisfactory, my hope is that using the production AS7265X ICs and the new arrangement of same will produce better results. The final device is definitely more than a toy. Whether it can rise to the level of scientific instrument or be used to discriminate between and identify different materials, etc, as shown in the AMS "tricorder" video, remains to be demonstrated.