If I prove to myself that this device safe, I plan to install 12 of them.

As seen on the BOM (see above in files), excluding assembly and test costs, the Light Switch comes in at a whopping $83, quantity one. The top 5 parts cost items are the Particle Photon ($19), the Formlabs-printed front plastic piece ($15.80), the capacitive touch illuminated display ($7.10), the Wurth Flyback transformer (at my guess of $2.56), and the PCB ($2.15 from Elecrow). Sitting towards the bottom of the BOM sits a sneaky $10.00, my estimate for shipping where it is not already included in the parts cost. If for personal use in California, add $8.10 per unit in sales tax. For 12 units, the only substantial decline is the shipping per unit, so say it would cost $75 for 12.

Going to 100-unit quantity, the price drops to $47.91, as part costs drop and shipping costs/unit drop, and it is assumed the ultimate buyer pays the sales tax.

Where to cut costs? Well, sacrifice quality of the plastic front piece by using a filament 3D printer, and you save $15, dropping to quantity one cost at $68. The Cap Touch switch is kind of the aesthetic design center (though at 100 units, it drops to $4.42). And you are not going to do much about the flyback transformer – I had to order 50.

That obviously leaves the Particle Photon, which in a product as low end as this, represents a dominant 25% of the BOM. The obvious thing is to go to an ESP8266 for a couple dollars, dropping quantity one to $55, and quantity 100 to $36.91 (I assumed buying 100 Photons would be possible for $13 each versus $19, but I have no basis for that).

But when going to the ESP8266 (which I have not yet used), I am guessing I am going to spend more time on software development, as linking a Photon to some remote app is simple stuff. So, what is my time worth to write that software versus the lower costs. If developing just for myself, creating a cloud function on AWS or my own server, to connect, say, a dozen of these wall switches to remote applications, sounds like a lot of work, and the Particle Cloud seems worth the $17 adder (a total of about $200 in additional cost).

Now on the quantity 100 side, the savings would be $1100. So how much do I value my time? How much software maintenance/updates will I have to do over time on my “invented here” solution, and how many hours is that?

For simplicity, let’s say I could earn an engineer’s salary of $100K here in Silicon Valley, or about $50 an hour. That would allow me 22 hours to replace the Particle Cloud, excluding future maintenance. Even if I value my time at $20/hour, that would still only be 55 hours. For some of you, I imagine that would be a no brainer using AWS or some node.js application you can pull together quickly. But my software skills are insufficient to finish such a project in 22 or even 55 hours.

What if it were 1000 units? Well, using back of the envelope calculations, the BOM cost drops to either around $30 with the Photon, or around $20, using the ESP8266. And the total cost delta then becomes $10,000. At $50/hour, that means 200 hours for software development, at $20 hour 500 hours. Starts to look like a possibility. On the other hand, what about provisioning and fleet maintenance software?

Bottom line for me is the Particle Photon is way expensive versus alternatives like the ESP8266. But for my skill levels, and a project that will have Android and iPhone app’s, an Alexa skill, and a web page …. I should stick with what I know and can do in a reasonable amount of time.

How much is your time worth? How do you trade off lowering costs versus investing your time? Would you write your own server and maintain it?

We have not discussed the assembly and test side of this ramble - that will come in a future log entry.