Is it reasonable to set up an assembly line for 2000 PCBAs per year?

[deadlock]

I designed a small PCB that is part of a product we sell.

Currently I buy all components from DigiKey, Mouser, and some other companies.
I order the custom PCB from a company abroad.
I pack it all in a big carton and send it to a contract manufacturer in the country and they use their pick and place machines and runs it thru their reflow oven. I get the PCBs back and we assemble the end product.

Now the local contract manufacturer have no time to do our assembly and they have also increased their prices to ~$8 per board with 10 components each, nothing fancy. (16 boards on a panel)

I think it is expensive and is very hard to plan when I need to wait for a time in their schedule.

So I am considering if we could do the assembly our selves. Would give us a huge flexibility. We are a small company renting a small office. I have found some good looking pick and place machines made in china and also reflow ovens. I guess we would also need a stencil printer for the solder paste. I think I could get a pick and place machine + reflow oven for about what we use to pay for assembling 1000 PCBs so financially I think it could be a good idea.

But the questions I have are many and I would be happy to receive any opinions, thoughts etc if you have any thoughts on any of my questions I would be happy to hear about them:

- Can this be done with quality or should I just continue to use the expensive contract manufacturer?
- Could pick and place machines that can be bought for this price range ~$5000 good for production use, small volume?
- Could a reflow oven with coneyer belt for ~$2000 be good for production use?
- What kind of ESD protection do we need to have? Using ESD shoes, ESD floor tiles, ESD clothes? any recommendations? How careful do we need to be?
- I see that the reflow ovens usually have some kind of exhaust, can this be sent out to the free air outside the building or does it need some filtering etc?
- How to dispose of old solder paste in the stencil printer in a responsible way?
- Would I generally need to have a controlled climate for running a pick and place machine or would it work if it is colder sometimes and warmer sometimes? Might have an option for a space to have the equipment but it is not fully wamed up during winter time.
- For depaneling the boards, any recommended equipment?

[bdunham7]

Now the local contract manufacturer have no time to do our assembly and they have also increased their prices to ~$8 per board with 10 components each, nothing fancy. (16 boards on a panel)

I understand the frustration, but can you really set all that up and run it for $16K per year?  Think about capital cost, maintenance, space, tax issues and the big one, staffing.  That said, assuming you really only have 10 components and they aren't all large BGA or something like that, you are obviously paying a pretty high low-volume premium.  Personally, as much as I like buying domestic, I'd be getting a quote from JLCPCB (and perhaps a few others) to make the whole board.

[Jackster]

You could probably pick up an old 2nd hand pick and place machine and 3-zone oven on that budget.
New machines from China, you are looking at the NeoDen YY1 at that price and I would stay far away from it going off the feedback from members on this forum.



- Can this be done with quality or should I just continue to use the expensive contract manufacturer?
You have to weigh up more than just the cost of the kit. There is the cost of transport getting it to you, power requirements, facility space, time learning the processes and how to use the machines and all the quirks of PCB assembly plus your time making the boards and all the handling equipment and accessories that go with the assembly line.
Trust me, it all adds up even on the budget end. And you will think that you don't need some of it but you really do need it.

- Could pick and place machines that can be bought for this price range ~$5000 good for production use, small volume?
Small volume yes. 120 boards per month is nothing for a production machine. Could do that in less than an hour. But on a budget machine, probably all day.

- Could a reflow oven with coneyer belt for ~$2000 be good for production use?
Reflow ovens don't hold their value. You can pick up some decent ovens on eBay for that amount. But finding parts for them becomes expensive. Especially if they are branded ones.

- What kind of ESD protection do we need to have? Using ESD shoes, ESD floor tiles, ESD clothes? any recommendations? How careful do we need to be?
Most electronics these days are "fine" without any major ESD safety considerations. That does not mean you are okay to wear clothing that creates a lot of static buildup and to not have grounding points scattered around. Should still be using ESD protection for personnel safety as well. Carrying PCBs around in ESD trays and boxes is also a good idea.

- I see that the reflow ovens usually have some kind of exhaust, can this be sent out to the free air outside the building or does it need some filtering etc?
Depends on your local authority. Ideally, fully filter that air and exhaust it.

- How to dispose of old solder paste in the stencil printer in a responsible way?
Recycling centres will take it all for you. You might even get a few pennies if you build up enough stock.

- Would I generally need to have a controlled climate for running a pick and place machine or would it work if it is colder sometimes and warmer sometimes? Might have an option for a space to have the equipment but it is not fully wamed up during winter time.
Humidity is what kills the electronics. I keep my facility at a safe humidity level throughout the year. Heating is not necessary but cold air = higher relative humidity so not ideal.
It really depends on what electronics you are making as you how you need to act around environmental conditions.

- For depaneling the boards, any recommended equipment?
Depends on your panel. V-groves require v-groove panelling machines while mouse bites require either CNC routers (not needed at your quantity) or a good pair of snips.


IMHO, unless if you have money and time to burn. Look for a better and cheaper PCBA company at this quantity.
Your time is probably more valuable making new productions and selling existing products than assembling them yourself.

[ddavidebor]

Just outsource it to a company iike PCBWay, and save yourself the stress.
Your time is better spent designing new products!

[loki42]

No. Do it in China, Mexico,  Vietnam, USA etc.   I spent around usd 100k total on my line (very second hand) and it took a lot of hours to get everything working properly.  A decent source of clean dry air will be more than your budget.  A proper manufacturer will have stuff like SPI, AOI, fresh paste, a calibrated oven with plenty of zones, etc.  If you need PTH cost either labour or gear go up a lot.  I was designing around a 150 boards a day and it's not a sensible investment for me at 150 / day.

[AgileE]

Or just look for a cheaper local assembler 

[deadlock]

Thanks a lot for all the feedback and thoughts!

[Feynman]

I would strongly discourage you from setting up your own assembly line. I've seen a lot of companies try that as well and every single one failed miserably, because they underestimated what it takes to setup and maintain an in-house assembly line.

I would advise you to look out for a different CM or have a chat with your current CM. Maybe, if you let your current CM handle buying parts and PCBs, you might end up with a cheaper price, since everything goes through your CM's regular processes. It's usually a pain in the ass for CMs to handle parts and boards that come from their customers. They happily deal with it, but also charge accordingly.

[SMTech]

If it was multiple designs and difficult to predict requirements/scheduling, then maybe. But you are talking about one simple product and like everyone else, I'd say not worth it.

$8/board is ludicrous, you could try elsewhere or try different batch sizes or as others suggest show them the whole job and see if its suddenly more tempting if they do more of the work or purchasing.

[jmelson]

I do small-scale production in-house (literally in the basement of my family home).  I made an adjustable frame to hold the PC board and stencil and manually apply the paste with a squeegee.  Then, I put the board into a used P&P machine that I got on auction.  Then, I reflow it in a GE toaster oven with a thermocouple temperature controller that does the ramp-and-soak temp profile.  I've done over 2000 boards this way, but over a MUCH longer time scale.  For 10 parts/board or 160 parts/panel, that is not very demanding.  What is the smallers part you use, and the finest lead pitch?  I am on my 2nd P&P machine, my first one was not accurate enough for 0.5mm lead pitch parts I was using.  the new machine, after some calibration, does just FINE for those parts.

Depending on the size and lead pitch you need to place, a relatively cheap used machine might be able to do what you need.  The Chinese machines look good on paper, but there are LOTS of software issues getting them to run reliably.  Feeders are the BIG issue, and feeders that can reliably feed tapes of parts without constant fiddling and clearing jams were a problem on my first machine, the newer one is better.
I have a Quad QSA30A (built by Samsung) and it is built like a tank.  But, the software is just a bit quirky, and the manuals are AWFUL.  Fortunately, I was put in touch with the guy who trained users at the factory, and he was an ENORMOUS help!

So, I started with a Philips CSM84 (made by Yamaha) and the Philips manuals were quite good.  Having NEVER SEEN a P&P before, I was assembling boards in ONE WEEK!  But, it was a very SIMPLE machine, no vision, no nozzle changer, etc.
The Quad machine has vision, auto nozzle changer and optical alignment.  So, it needs to know the XYZ size of every component type.  That takes a lot more programming time, at least when new component types show up.
My first machine came from a current production facility, and ran for 13 years in my shop with VERY few issues. My 2nd machine was tinkered with by monkeys and stored for years in bad conditions (hot/humid).  I had a ton of breakdowns on the machine.  But, I got it for only $500 at auction.  I ended up spending about $7500 total to get it totally equipped with feeders and repairs.
If you have any more questions, ask away!
Jon

[CatalinaWOW]

All of the above practical advice is useful, but may have missed two of the most important points.

1.  Do you want to do it? (if you are more than a one person operation, the question applies to someone in your company).  There will be a lot of time and effort learning the tools and setting up the process.  If that interests you, you will likely succeed in the end.  If you hate every moment, failure is almost assured.

2.  Are there legal/contractual limitations on the use of your office space?   If you don't mind being noncompliant this may not be a showstopper because you can conceal what you are doing, but if you are noncompliant it is a significant risk to you and/or your company.

[deadlock]

Thanks for those alternative views to my question.

I believe I want to do it and that is a problem in itself because I realise it will take time from other things i should rather focus on.

The interesting thing here is that we are apparently paying far too much for assembly today. After this discussion I asked for quotes from 2 other smaller assembly companies nearby and price is like one fifth of my current assembly house. And for this reason i cannot justify to myself to set this up in house.

I think the main advantage if having assembly in house for me would be the freedom of not having to plan in advance. Instead to assemble more boards immediately when required.

Thanks again everyone for valuable input

[JPortici]

I currently am in a simillar situation: we have received an extra big order (like 1/3 of our usual yearly quantity) of a product.
Lucky for us, it's a really simple board, all SMT components.
In these cases what i do is a production run at JLCPCB, it takes about 10 days (not 10 WD) from order to received. This time we had some component they would either not place or charge a HUGE premium because they were not available on LCSC so we tried the reflowing process. This time board is complete, minus the MCU, another IC and a pogo pin.
So we bought a hot plate (the blue cube one) and a syringe of solder paste. Placing/dragging very little paste on the pads, put component, put on hot plate. voila. It actually takes very little time with little practice to achieve a nice result, should have had the guts to try it sooner.
Final price (JLC+Shipping+Duty+Price of remaining components) in line with what our contractor asks for simillar quantities

[CatalinaWOW]

Thanks for those alternative views to my question.

I believe I want to do it and that is a problem in itself because I realise it will take time from other things i should rather focus on.

The interesting thing here is that we are apparently paying far too much for assembly today. After this discussion I asked for quotes from 2 other smaller assembly companies nearby and price is like one fifth of my current assembly house. And for this reason i cannot justify to myself to set this up in house.

I think the main advantage if having assembly in house for me would be the freedom of not having to plan in advance. Instead to assemble more boards immediately when required.

Thanks again everyone for valuable input

If you really want to do it, there is another variable to factor in.  You will undoubtedly end up with excess capacity.  The way to make the capital investment pay off is to utilize it, so you might want to take on fab jobs for others.

I know a couple of people who have followed this general path (facilitized to meet a need of their existing business) and ended up with a much larger and more profitable business doing the "side" job.

[Rat_Patrol]

What type of components? I have modded and tweaked my Charmhigh 48VB machines to be greater than 99% reliable, but they are not chip shooters averaging 1200 CPH using both heads. That said, it cannot do large chips, BGA, or anything over 5mm high (like electrolytic caps). I'm slowly working on a write-up to post here for these machines, along with the stl files for the mods and accessories I made for them.

So figure $10k for a bare minimum setup, you will want AgnosPCB inspection tool subscription, and then you need to run the machines. Depending what you pay your people per hour (or what you value your time at) this may/may not be worth it at 2k boards/yr. For me, I make my stuff when I need it on my schedule, 110% worth it.

[newto]

Is that 2000 panels with 16 boards, or 125 panels with 16 boards each?

I imagine the first would be rough to handle, and better off outsourcing, but 125 panels a year would be perfectly reasonable in a small production line, that's like 20-30 a week on average. With only 10 components, you could really dial in a cheap machine's feeders. You could probably even handle that in a toaster oven, but a proper oven would probably make your life a lot easier.

But for 128$ a panel right now, that gives you a lot of budget to save, that's 16000$ a year (assuming you mean 125 panels).

[jmelson]

I think the main advantage if having assembly in house for me would be the freedom of not having to plan in advance. Instead to assemble more boards immediately when required.

Yes.  The problem with contract assemblers is they scale prices by volume.  So, to have them make just a few panels is nearly as expensive as a long production run, most of the cost is the setup.  But, having 2000 boards sitting around for a year is expensive for a small outfit.  With in-house, especially for just one PRODUCT, you can leave the machine set up all the time, and buy components and make short runs when inventory runs low.  That's how i do things, it keeps inventory cost low.
Jon

[mcconkeyb]

Building 2000 CCA's a year works out to 40 per week and 8 per day. This should be able to be done by hand, even by a beginner. So to save yourself all of the problems that go into building a circuit card assembly line just hire someone to do it by hand. You could be up and running in under a week with a human assembler, the production line will take 3 months if you are lucky and a couple of years if you don't have extensive experience.
Well that's my 2 cents. 

[Jennifer J Walker]

You know, you can find a Chinese supplier, they are usually cheaper.

[jmelson]

If you really want to do it, there is another variable to factor in.  You will undoubtedly end up with excess capacity.  The way to make the capital investment pay off is to utilize it, so you might want to take on fab jobs for others.

WARNING!  I did this for an outfit I was friendly with.  I warned them about reducing aperture sizes on their stencils, they totally ignored me on that, and the stencils were too thick and the apertures were WAY too large.  Passives were OK, but there were solder balls all over.  The microcontrollers were floating on lakes of solder, and while perfectly positioned on the P&P they were massively (several mm) off after reflow.  It caused some hard feelings!
Jon