Electronics > Manufacturing & Assembly
Advice: assemble in house or not?
AlexI:
I'd like a sanity check on whether it makes sense to assemble boards in house in my scenario.
Background: I'm technical lead in a very small company which makes just one product. Annual volume is 50-100k for the next year. The electronics going into this is one board with fairly simple specs: it's 30x30mm, 25 bom line items, 60 placements, 0402 or larger, a few SOT323s, a few 0.5mm pitch QFPs. No 0201, no BGAs, etc.
We have been manufacturing these at a local assembly house in the SF bay area which charges about $10 per board for assembly with 1k batches. That's obviously not a long term solution since it adds a huge chunk of the total product cost. I've gotten quotes from some lower-cost offshore assemblers at around $1-$2; no idea on the quality. Macrofab would make this for around $3; quality seems consistently good there.
We have automated test for this; all functions of each board get 100% tested by us after receipt (but it wouldn't necessarily catch stuff that leads to delayed failures, of course). So the risk of letting a bad batch slip through unnoticed is smaller, I think we can to some extent just use the lowest cost supplier.
The question: does it make sense to set up an in-house production line for something like this?
This seems to be a pretty extreme case of "low mix"; the line would literally never have to be reconfigured as it will be cranking out an exact copy of one board forever. It doesn't have to run fast either; I think even a 2000cph machine running one shift/day would be fast enough.
The main reasons to do this is not even necessarily cost - I'm not at all sure that this will overall save money over the $1/board assembly house option - but rather risk; risk that a large batch of boards would come out defective, risk that we won't have product in time due to some type of screwup with the assembly house, risk that they use counterfeit parts, etc... If the assembly is local, they would own the problems (and they have in the past - we had low yields, they reworked the bad boards for free). If the assembly is a cheap overseas company, I really doubt they would do anything about it; we'd probably have to eat the cost of rework and/or parts if a batch is bad. That can become a pretty big problem if we run large batches (which we would want to for lowest cost). It is also nice to be able to stock parts rather than finished boards, and build them as needed to fill larger orders; less need to plan ahead. (* That said - if we don't save money over the current $10/board, we must be doing something very wrong)
Now the downsides: (a) We don't have any expertise with doing this sort of thing. We are a tiny company (3 people+consultants) and I'm the most technical person here, but much more on the R&D side than production side; I'm fairly familiar with SMT processes, and I've built a number of prototype boards myself (the hand pick and place and toaster oven way), but I haven't run a production line (and really wouldn't have time to). We'd have to hire staff (part-time or full time? how many?) We'd also need some good resources for fixing things when things go wrong (smt equipment dealers? some kind of smt line repair-person?) (b) We don't have a facility where this can be hosted yet (how much space? I'm guessing +/-1000 sq ft). It's not necessarily a huge production line, but it is a production line and needs some infrastructure around it.
So, with all that in mind: should I (a) shop around for a good mid to high volume manufacturer and develop a relationship with them, or (b) look at used SMT equipment, and folks to run it? Or, perhaps, considering how daunting it is setting something like this from scratch, am I insane for even thinking about it?
Gribo:
These risks are the same if you do it in house. but now you have to pay for them, not your CM. Fakes, bad batches, reworks happen everywhere, even on tier 1 CMs. Move out of the bay area, there are CMs in far cheaper places in the US, Canada and Mexico.
AlexI:
Hi Gribo - Of course, same risks everywhere, it just seems if they were in house we would be able to proactively do something about it.
Experience with one of the last batches - to give you flavor: CM said they'd do it in 3 weeks; at week 3 told us some parts are back ordered with 12 week lead time; we found the missing parts and consigned them, got actual product in 7 weeks; then 80% yield; took about 3 more months to fix the remaining 20% bad boards after we found the cause and showed them the fix. Every single build is like this :[
Cheaper CMs - certainly, I've done some shopping around. I don't really know that any of the folks that we'd want to work with (in terms of scale, capability etc) would be particularly interested in working with us; as a single low/mid volume customer we're just not that important to them. At our current CM we have no leverage at all, we're 0.1% of their business. At someplace like AllPCB, it would likely be much worse than that. As for Flex/Jabil...
I think you're saying different/better CM is the answer. Which one - any recommendations? More generally, what to look for?
OwO:
--- Quote from: AlexI on October 31, 2019, 02:13:54 am ---Experience with one of the last batches - to give you flavor: CM said they'd do it in 3 weeks; at week 3 told us some parts are back ordered with 12 week lead time; we found the missing parts and consigned them, got actual product in 7 weeks; then 80% yield; took about 3 more months to fix the remaining 20% bad boards after we found the cause and showed them the fix. Every single build is like this :[
--- End quote ---
Get a more competent assembly house. 80% yield is REALLY bad, and if there are soldering defects you can have latent failures in the field. Not to mention they are really overcharging you for a tiny 60-part board. I have a board with 300 placements, 60 BOM lines, double sided placement, fine pitch QFP and QFN, and I pay around $14/board at 50 units per batch. Last few batches had zero defects. They claim to do AOI, which you should expect at this price.
OwO:
If you decide to use an overseas assembly house make sure to plan the whole process - make sure they can test the boards themselves without having to ship it to you. I built a simple test jig which I shipped to the assembler along with sample boards so they can easily check for mistakes in parts orientation. Before every batch they will also manually assemble one prototype based on your design files. This will catch errors in the BOM (happened to me once). Make sure to pick an assembler with these services.
Navigation
[0] Message Index
[#] Next page
Go to full version