"Last year, when nobody had heard of the Raspberry Pi, we had been unable to find a British manufacturer whose prices per unit (especially at a point where we were thinking of sales in the tens of thousands, not the hundreds of thousands you’re seeing now) would work for us, and who believed that the project would be enough of a success for them to risk line space for us."
Who wouldn't take one project because they were afraid you wouldn't come back for more? Is business so hopping over there that people scoff at a $1M deal because you may not come back later for a $5M deal? Or, were you actually asking them to reserve production, but that you might cancel before it started?
The key point I think is line space. Getting set up for production is quite a bit of work, and given that their new contract is for 30,000/month, the manufacturers clearly didn't want to set up an entire production line if they may only get a couple of months of usage out of it (since the original order would've been only in the tens of thousands).
That is the key thing I question and makes me wonder about EMS in the UK. There are literally dozens of vendors who would kill for that job here in the States - even without any guarantee of repeat order. There just must not be very many there in the UK.
(I do understand some of what it takes to setup and run a manufacturing line, I work with several EMSes both in the US and China and run a [smaller scale] electronics manufacturer.)
Edit: Before my comment is taken the wrong way, what I am wondering is if the UK has much capacity for mid-scale EMS/CEM, or if the industry there has been largely tailored towards the very big players. But, I'm pretty sure I misread the statement, namely that they meant that the only places willing to take the jobs were charging too much, rather than that no one wanted the job because it was too small.
It depends on the design (PTH or SMD). A lot of PTH takes a lot of people, as many parts still can't be machine-placed.
The facility we use an hour from us runs about 600 PTH units (about 40 components/board) on one assembly line with 10 people in about 4 hours. I doubt one would always work them at that level. Lead-bending, wave soldering is all pretty much automated.
The bulk of an SMD process is done by pick-and-place, our stuff gets run on 10,000+ CPH machines. You need one person to load/de-load panels and switch out feeders - and one person can work multiple machines if everything is set up right. The machines in general are really good these days.
Printing and machining the PCBs takes some time, but this is a process that can flow continuously once you let a little backlog build up at each stage. Electrical testing can be slow at the back-end, as is polishing board edges, packaging, etc.
I could see it taking 30 people to do 10k a month, but the same 30 people possibly capable of 60k+ a month. It depends on where you load them in the process. The more SMD the more volume you can get up to the testing/packaging stage. rPi is mostly SMD, with some PTH - so they have to mix the processes, but its really hard to avoid...)
Notably, however, if you look at the Hackberry (which I like better, btw!) the vast majority of the connectors are SMD. I wonder if the rPI could see some labor savings (at higher BOM) by switching to them.
One thing we've been working on in-house is making big automated jigs to program and test entire panels of devices at once, before they're de-paneled. These are really expensive to make though (think $5k+ per jig) - but they save us thousands of hours of labor a year. We're moving to make machines that can do it without jigs now. Fun stuff =)
Margins. In other words if cost of sales is high enough (and it usually is for large/old b2b) then working with small time customers is net negative revenue.
"Last year, when nobody had heard of the Raspberry Pi, we had been unable to find a British manufacturer whose prices per unit (especially at a point where we were thinking of sales in the tens of thousands, not the hundreds of thousands you’re seeing now) would work for us, and who believed that the project would be enough of a success for them to risk line space for us."
Who wouldn't take one project because they were afraid you wouldn't come back for more? Is business so hopping over there that people scoff at a $1M deal because you may not come back later for a $5M deal? Or, were you actually asking them to reserve production, but that you might cancel before it started?