Re: [Paparazzi-devel] Lisa PCB pool

[Chris Gough]

Hi David,

I'm very glad you are open to this sort of thing.

> I can speak for PPZUAV in that if you don't order 100+ it's expensive. 10k
> for Umarim BOM cost alone. Just cost not shipping to get them in hand and
> get them to the assembly. You need 2% greater than BOM to account for PnP
> drops and losses. Stencil fee, prog fee, risk of assembly being wrong (they
> give guarantees but actually getting it exchanged is another matter with
> them)... How many out there have 15k to risk? I can't sell YAPA2 at my cost
> and have about 75 left. What assurance is there to sell 100 Umarims?

Yes exactly, 100-150 units is the minimum quantity for decent prices,
but much more than a convenient quantity for keeping a shop stocked.

> Yes, without any coordination that a big flaw in Umarims exists or that
> another version isn't due soon it's too much risk. My experience is as soon
> as I could have 100 of them something new comes out.

And pinning hardware-continuity on the prosperity of a high-tech
startup with audacious goals is just a different kind of dangerous.

> I envy DIY where the guy planning assembly can be sure a new model isn't due
> or that BOM parts can be shared. And that they do in-house PnP and reflow.

3DRobotics seem to be going well, good luck to them. But I think this
community is capable of developing a superior system (as usual :-)

DIYDrones have achieved a scale where they can support in-house
production. A fabless system could be significantly more efficient
though, maybe investors are impressed by assembly robots, blinking
lights and busy people. Ardu* are obviously limited by AVR, but don't
have much in the way of hardware abstraction (yet). A major rewrite
could cause a major fork (remember ArduPirates?), all that capital
intensity could really bite them if they took a wrong turn.

> I am not saying I'm not going to. I however think we all need to learn from
> DIY and not compete with each other, share costs, communicate, support and
> grow together is required. I think JobyRobotics, Quadshot are included. Im
> in California and if they want to make Lisa and I want to make Umarim maybe
> we can share BOM costs, assembly costs. Some risks. We all need to
> communicate and act like a single entity. Please realize I'm very open and
> if you have ideas (all current vendors) contact me. I want to see Paparazzi
> become all it can be. Let's work together.

DIY/Ardu*/3DRobotics stage production based on predictions about
what's coming down the development pipeline. Their predictions are
quite reliable because they orchestrate development and marketing.
They are a lot more Cathedral than we are, significantly less Bazaar.

Paparazzi hardware and software development are brilliant, but
financially viable manufacturing is a real challenge. The bazaar is
broken, un-met demand is queueing up for some products, while other
products are stockpiled to high. This critical performance bottleneck
is just an engineering problem, constraint-optimisation - a bit of
structural refactoring, a bit of focused refinement, surely we can get
the hardware flowing.

Adopting a centralised control model would be one way to attack the
manufacturing bottleneck; setting hardware release schedules,
coordinating investments in manufacturing, etc. But in some ways that
would be a backward step, it could kind of get in the developers' way.
Who likes being told what to do? What's the incentive for
contributing?

The best kind of orchestration would be an emergent phenomena that
didn't require furious peddling. Decentralise risk and reward so
developers, users, vendors, and manufacturers all act in a way that
suits them best in their current situation, but at the same time work
together so everyone gets what they need efficiently. Why can't
commerce be commons? Why can't commerce be social?

I think this conversation should head off list, so set this up:

http://mkswarm.org/

Forgive the cross-post, everyone's invited.

Chris Gough