[sdiy] PCB assembly house board failure rates?

[Colin f]

> I'd be interested in what people would consider an 'average' 
> or 'fair' failure rate as a percentage for boards assembled 
> by a professional PCB house - based on their own experiences.

'Average' and 'fair' are two very different things.
A lot of factors will affect the failure rate, even between different
batches of the same board done by the same people.
If you are going to get a factory to populate boards, I would strongly
recommended that you also give them the means to electronically test each
board.
It's all very well that they'll re-work boards that turn out to have faults
when you test them, but the hassle of having to do the fault finding and
ship back boards can put a large extra cost on you.
If they have a jig they can sit a board on, and ideally an automated test
circuit, or at least a written test procedure they can run through, then you
should receive 100% working boards.
Their failure rate before testing becomes irrelevant.
They may want to charge you a bit extra for each board to do the testing,
but at least that cost is known in advance.
If you get a run of 100 boards, and 50 of them don't work, well, it's not
fun.
There is also plenty scope for a company to screw up a whole run of boards -
it only takes the guy doing the setup for the pick and place machine to read
one digit wrong on a one tiny little SMD package, and you might get a whole
run of non-functional boards. Or worse... boards which seem to work, but
fail rapidly or misbehave in some inexplicable way.

Cheers,
Colin f