Advice on setting up a low-mid volume SMT line for <$130k?

[rx8pilot]

Pretty much any oven that is not totally ancient can get over 300C.

Sent from my horrible mobile....

[SVFeingold]

2 months from dead start to 1000 boards completed?  OK, now I know you are nuts!

Probably! But, I am committed and I have no choice but to make it work Even if it takes 3 months that's not the end of the world. But yes it will be a crazy amount of work. All I know is if I lose the drive and chicken out I will regret it forever.

The ducting and power should be in place before the machines arrive. I should be getting a dedicated 100-200A 3-phase breaker. Air conditioning, well yes that may be a challenge. It's solvable.

Getting the printer to run is what I am most worried about. I am putting faith in the guys at K&S that they can train me well in a week, and I think with the training I can manually program the board in another week even without the proper CAD format conversion tools.

Thankfully none of the parts on this board are too challenging. All standard components, no passives smaller than 0603. There is that 8-WSON 3x3mm buck converter. Aside from that and the PCB assembly that will be installed by the PnP, all should be standard footprints.

Don't forget the through-hole PSU assembly that I will need to hand solder. I am confident I can install a panel's worth in <5 minutes however. To date I have never had an obvious defect just doing hand stenciling, hand assembly, and toaster reflow. Even with lead-free BGA. Knock on wood...I know it's different when you're running thousands of boards but I hope that at least that means my stencil apertures are decent. I've had tombstoning issues doing lead-free but only with 0201s on a particular board. Probably due to the hand-stenciling and placement more than the oven. I have even done wafer-level and microBGA by hand successfully, and I have to imagine that the results should be more consistent with a proper reflow oven but as you say profiling may be a challenge.

Don't get me wrong I am not brushing off these issues. I'm expecting many issues to pop up but I have to believe I will solve them in time.

For the panels, with a 9 array panel that's about 650 components per panel. With a somewhat conservative placement rate of 5k that comes out to ~8 minutes per panel to place.

A lot of these questions and issues I simply won't properly understand or have answers to until I do it, but I feel ready for it. I know that sounds like hopeless naivety to many of you and I'm 100% sure I would say the same. Nevertheless I am excited for the challenge.


Speaking of naivety, what exactly is involved in setting up a stencil? Aside from software. As far as registration I would think that if the stencil works properly and can recognize fiducials on the board and stencil, alignment *should* be taken care of. That leaves paste selection, print angle/speed, aperture design, and stencil quality.

Hey if you really want a laugh, here is the "loading dock" that will need to be used...

http://imgur.com/a/2hjc2

Once inside the hallway is concrete floor, ~12 feet wide. Old industrial machine shop used for heavy machinery. Will be a game of tetris all the same...

[ar__systems]

and run a vent for the reflow oven. I had considered using a filter for the reflow exhaust but after consideration of the comments in this thread, that would be dumping 20+kW of heat into a 500 sq. ft. room. No thank you.

It is a lot less than 20kW. Once heated the power is not constantly on. Nevertheless, you do need the vent.

[SVFeingold]

An AC unit that can remove ~10kW of heat and an industrial dehumidifer will cost about $5k. This is preliminary search, I can probably find something cheaper. Just getting an idea of the costs...

[mikeselectricstuff]

2 months from dead start to 1000 boards completed?  OK, now I know you are nuts!  Maybe an experienced guy could pull this off, setting up a new shop and cranking out boards.  Let's list the issues:

If you are getting a machine with support and training, and are not trying to do anything else at the same time, and you don't have a problem with late nights and/or weekends I don't think it's craziliy unrealistic.
Just allow for plenty of spare PCBs and parts to get things up & running.
But definitely try to talk to people who are running lines, even if you have to pay for their time/buy lunch etc.

[rx8pilot]

What Mike said.

I sacrificed more than a few nights, weekends, parts, PCB as well as dispatched a stream of profanity that would offend a navy sailor.

Then, it all worked.

Sent from my horrible mobile....

[mikeselectricstuff]

... as well as dispatched a stream of profanity that would offend a navy sailor...

Oh yes, that.... a LOT of that...

[mairo]

Have you also considered space and funds for good rework station if you do not have already one - and by good I mean one that have the right tools for these particular boards so that for example if you have to replace the TQFP part on 100 boards you can do that quickly and accurately without "...dispatched a stream of profanity that would offend a navy sailor...''

Did you said that you will be doing the THD connectors manually? Maybe something like this jig will help,or some kind of solder pot ..

[SVFeingold]

The through-hole connectors will probably be replaced with SMT, however the SMT parts are 2x the cost. The through-hole connectors can be placed with the MG-8 machine, which would require a PiP method. However...they only come in bulk so a custom tray would need to be made, loaded, and programmed. If the through-hole parts can be used then we save ~$4,500 on connectors. Worth the effort? I'm not sure...will need to get a quote direct from JST before a decision happens.

There is also a secondary board that is very simple, 2"x4" with only a single connector and 2x rotary switches. No other parts. Problem is the rotary switches are not reflow safe so will either need to be soldered with a desktop wave machine (if one exists that's not trash) or manually with a soldering iron. 10k switches by hand...woof...but I can do it in a couple weeks or hire a few EE friends to briefly help for beer money. If I did them manually I would pick up a JBC solder feed station. For the PSU (also not reflow safe I believe) it would need to be done by hand. 4 pins per PSU, but they are large and connected to planes. I am considering the JBC heavy duty station (250W) to make this quick. My Weller WD1 does OK with a large tip but it will add a lot of time to the whole lot.

In terms of other equipment I have a pretty comprehensive setup for the kinds of things I would encounter. Tons of good tools, proper optical stereo microscope, ESD bench, hot air, flux, solvents, solder of all types, etc... I've been hand-assembling 0402, 0201, wafer/chip-scale, BGA, etc... up to now. Manual rework isn't a problem. The point is just to do less of it.

Now if by good rework station you mean a proper station with IC lifters, pre-heater, etc... I have a little work to do on that. But those things can be found cheap.

And don't worry I'm sure there will be many streams of profanity all the same!

[coppice]

Think very very carefully before replacing any through hole connectors with SMD ones. The majority of SMD connectors are so damned fragile, once the mating parts are inserted, that they are huge liability.

[jmelson]

We are doing dip soldering with a large Chinese solder pot (5 x 7").  After reflow of the SMT parts (100+ per board) we mask some holes that need to stay open, flux the board, put in the 40 through hole parts and dip it in the pot for 10 seconds.  We pre-heat the board on a hot plate first.  I've gotten this process down pretty well.  I'd like to find a gadget to spray flux on the bottom of the board AFTER the parts are dropped in, thereby fluxing the pins as well as the board.  That might help.  There are some fuse holders that have to be hand fluxed before dropping in the board or they don't solder at all.

Anyway, this actually works, takes up very little space, and the gear can be put away when not needed.  So, this is one option for through-hole parts.

Jon

[SVFeingold]

Think very very carefully before replacing any through hole connectors with SMD ones. The majority of SMD connectors are so damned fragile, once the mating parts are inserted, that they are huge liability.

Agreed, I'd prefer not to, however once assembled they will be sealed in a box and ideally never touched again or exposed to any external stresses. So I think for this application it's not an issue.

[jmelson]

Think very very carefully before replacing any through hole connectors with SMD ones. The majority of SMD connectors are so damned fragile, once the mating parts are inserted, that they are huge liability.

Yes, especially with lead-free, the solder joints can fracture easily, and especially if the connector is going to be mated by the end user more than a few times, it could be trouble.

Jon

[coppice]

Think very very carefully before replacing any through hole connectors with SMD ones. The majority of SMD connectors are so damned fragile, once the mating parts are inserted, that they are huge liability.

Agreed, I'd prefer not to, however once assembled they will be sealed in a box and ideally never touched again or exposed to any external stresses. So I think for this application it's not an issue.

What about transport? I've seen SMD connectors with a bit of leverage attached to them break off inside the equipment box when they receive a good bump. I guess if everything is supported well this might be mitigated,

[Kjelt]

I'd like to find a gadget to spray flux on the bottom of the board AFTER the parts are dropped in, thereby fluxing the pins as well as the board.  That might help.  There are some fuse holders that have to be hand fluxed before dropping in the board or they don't solder at all.

You can buy liquid no clean flux in gallon containers, so why not use a plastick bin filled with flux and dip the pcb in as a first step let it drip off a bit and then dip it in the solder?
That is how they tin copper wires.

[SVFeingold]

All of the connectors are JST PH series, an example of one of them here.

Lowest is 2 pin, highest 6 pin. The SMD version seen here.

I've used these before and they've been reasonable sturdy. There is not much leverage on the connectors, they're small and low profile. Packed well for shipping I don't foresee any issues. If reasonable I'd prefer to use the through-hole version provided the machine can place them and they are safe to reflow. Otherwise, they'll be mated 1-2 times and then installed in the product. I already have to manually solder 20,000 through hole power supply pins and 115,000 through hole switch pins. If I can do each of those switch boards in 20 seconds that's 27 full hours of soldering. So, I prefer not to add another 120k pins to that...

EDIT: Those SMT connectors are $0.11 each for 6k from JST direct. Pays to shop around! Over the whole assembly run that's over $7k saved on those connectors alone. Enough for a good dry box!

[SVFeingold]

Do any of you have experience with auto-dipping machines like these from Manncorp:

https://www.manncorp.com/wave-soldering-dip-soldering/auto-dip-3530-ts-soldering-system.html

What do you think, trash or useful?

If a cycle time of ~1-2 minutes can be achieved with these, that's theoretically the full lot of 5000 switch boards (50 per panel, 100 panels) in just a few hours - assuming I can get the machine to place them. Hell, 5x that is still way less time (and swearing) than manually soldering.

The reception to the desktop wave solder machines I've read here tends to be "meh." Not sure if this is better. Can't seem to find details on whether there's a solder pump involved, if not it might be difficult to solder through the constant layer of dross...

[ar__systems]

I already have to manually solder 20,000 through hole power supply pins and 115,000 through hole switch pins. If I can do each of those switch boards in 20 seconds that's 27 full hours of soldering. So, I prefer not to add another 120k pins to that...

2+10 pins in 20 sec? No way. You are off by a factor of 3-5. Even if you time just soldering when everything is already prepared in front of you and you just need to solder and nothing else, 12 pins in 20 sec is already pushing it. It maybe possible to do it once, but not for hundreds boards in a row. But more importantly you are forgetting time to unpack the connectors, insert them into the board, and just handling the boards - take them out of the box, put them back in the box, take another box etc... 

[SVFeingold]

Yes you are correct. If we say, 15s to pick up a board with pre-placed components and then place it elsewhere after soldering, + 2s per pin, + 5s for various checking and fiddling around, this is ~92 hours of hardcore fast soldering. Oh my.

Certainly I will need some sort of small dip/wave machine that will let me do a panel of these at a time...

[Kjelt]

Hire a student to help out.

[ar__systems]

Yes you are correct. If we say, 15s to pick up a board with pre-placed components and then place it elsewhere after soldering, + 2s per pin, + 5s for various checking and fiddling around, this is ~92 hours of hardcore fast soldering. Oh my.

Certainly I will need some sort of small dip/wave machine that will let me do a panel of these at a time...

I don't understand why you insist on doing it all by yourself Hire some help

[rx8pilot]

I don't understand why you insist on doing it all by yourself Hire some help

Speaking from experience....

When I have tackled big and brand new challenges, low-level help has rarely proven to be helpful. Students and others with little practical knowledge need a lot of hand holding and guidance. They have mostly slowed me down.  If I am attacking something that I don't have any experience myself, it is not helpful. When the project is mundane like running cables, assembling shelves, etc - sure, but that is a tiny fraction of the challenges that lie ahead. Every few minutes, you have a person asking 'What should I do?' as you are staring at a bunch of stuff that you are clueless about.

On the flip side, I have periodically brought in higher-level help. That has its own issues that have to be managed.
1. Experienced people do it their way. This is good in the sense that stuff gets done, but when the end product needs to be my own personal knowledge and experience - it is not good.
2. Experienced people hide the details of what it really takes. In P&P, I needed to learn the process myself to have a chance at long term success. I needed to know the details of the equipment, software, solder paste, maintenance, etc. If I just had someone do it each time, my costs would be higher and I would be beholden to the schedule of others. I could not justify a full-time person up-front and I wanted to dial in the process to the point where lower-cost help would be able to safely build PCB's

Just getting help is not an easy or straightforward decision. There is a long term benefit to the business owner to actually know the details of how things work. I have done that with all of my businesses and it has allowed me to get the most out of the people I hire down the road. I never have an employee that can pull the wool over my eyes - I know what can and can't be done and how long it should take. I can see wasted time and materials from a mile away - but only because I know the process really well.

[Kjelt]

@rxpilot: we were talking about the 92 hours of TH component soldering,  not about the P&P stuff.
TH component soldering can be learned in a day esp when you get an EE student that has soldered before, in China 10 year old kids do it.

[rx8pilot]

@rxpilot: we were talking about the 92 hours of TH component soldering,  not about the P&P stuff.
TH component soldering can be learned in a day esp when you get an EE student that has soldered before, in China 10 year old kids do it.

Ahhhhh...missed the critical details, lol! 

[mikeselectricstuff]

@rxpilot: we were talking about the 92 hours of TH component soldering,  not about the P&P stuff.
TH component soldering can be learned in a day esp when you get an EE student that has soldered before, in China 10 year old kids do it.

Ahhhhh...missed the critical details, lol! 

An answer to that might be to find someone who works ( or worked)  at an assembly place and wants some evening/weekend/part-time work.