A couple people suggested I go the crowd-funding direction with Pi Power.
I'm still not entirely sure why I should bother.
I tried a Kickstarter for the AD8495 breakout board. It didn't meet its goal, and so I wound up just going with a slightly higher price for it and selling it myself. It hasn't done very well, and the coup de grace was that AdaFruit started selling their own <sad trombone>.
I've set the price for a Pi Power at $15. How did I arrive at that price?
Well, what does it cost to build 1000 of them?
It turns out that in researching the crowd-funding angle for Pi Power - this time at IndieGoGo - I went through that whole exercise. I went with a US based PCB fab and assembly house. The assumption was that I'd ask them to build 1000 units and I was not in a hurry, so they'd take about two months from the time I said "Go!"
For the DigiKey BOM, I selected full reels of parts where the quantities and/or pricing made sense (why anyone would buy less than a reel of 0805 1% 1/8W resistors for $12 is beyond me). For the rest, I chose DigiReels (you get the cut tape price plus a $7 reeling fee).
And the answer? Around $8500. So $8.50 each. The 'doubling law' says I'm supposed to charge $17, but I'll go with $15. And I'll even throw in the 26 pin stacking header for free (1000 of those is less than 50¢ each).
Now, if you do the math for smaller quantities, what you quickly run into is that the cost of the fab making the PCBs starts to exceed the cost of an OSH Park medium run order ($1/in^2 each), and if I'm going to do that, then it just makes more sense to place, paste and bake the batches myself on demand.
I have very, very serious doubts that a crowd-funding effort for Pi Power can come up with $8500 in 60 days. Yes, there are projects that do that and even better. But they're the exception, not the rule.
So what does it take to get there? 1000 units is a lot. At one order a day (including weekends and holidays), it would take nearly 3 years to sell them. And one order a day would be a miraculous level of success.
Crowd funding makes sense for some things, certainly. But when the per-piece costs of what you're doing dominate the equation, I think it doesn't bring a lot to the table.
Am I wrong? Missing something? I'd love to hear from and/or discuss this in the comments.