10K$ SMT line suggestion

[ali_asadzadeh]

Hi,
This is the First time I want to Buy SMT Equipment, so my budget is around 10K$-15K$, And I need the line for manufacturing my own projects which could be around 20-30 PCB design each year, around 5-1000 units each PCB, the main requirements are as follow

1- Smallest part size is 0402 (But I prefer to go down till 0201)
2- The biggest part is 35mX35mm BGA with 1mm ball pitch (Also Prefer to reserve room for bigger BGA parts)
3- My boards vary on complexity, Usually on the higher side of complexity with low number of production QTY,  But they usually range from 60 different components to 185 different parts, so the more feeders are welcome
4- All of my boards are double side loaded parts, So I Should be able to do Top and Bottom assembly.

So any suggestions and feedback is highly appreciated.

[SMTech]

Oh come on now. ADD A ZERO! Even then you won't buying new kit. Also read the 5k other threads with the same silly budget.

1MM BGA Pitch, not really a challenge, neither is double sided assembly although that can involve some workaround jigs or compromises if you choose the wrong very entry level equipment.

35mm will exclude most/all old "chipshooter" machines as they have narrow FOV's

0402 isn't a big challenge in 2022, 0201 needs decent feeders and a decent pick and place as well as good printing and PCB design, As far as I'm concerned 0201 means your printer is fully automatic.

185 different parts on 1 board is a lot and suggests poor design choices, cheap Chinese machines often top out at 48 or 60 8mm lanes. Even with more conventional equipment you would typically approach this many parts & feeders with two or more machines inline because 60 front 60 rear is a very "standard" arrangement.

You can load one set of feeders build, and then run again (shortest parts first) but that approach doesn't scale, fine for a batch 5 & completely unworkable for any volume at all (exactly how unworkable would depend exactly how much time and money you put in own extra feeders and kitting all 185 parts in advance)as well as how many partially built boards you want racked up in magazines drying out.

1000 boards also means a real oven not some tiny benchtop thing, they're not small.

If you genuinely want and need an SMT line and your product make you money, buying the correct equipment will pay off, spend more than a shitty second hand car or outsource.

Just as an example, here is a good 2nd user machine that is designed for exactly the type of work you describe. It can't quite hold enough parts, but I happen to know you can do a live feeder cassette switch on every board while its building which would be fine for very occasional highly complex jobs but a nightmare in any other scenario. https://www.shawline.co.uk/product.php?id_product=42, the price is slightly inflated as it has a new head on it and is advertised by a broker. However it's still £23K GBP and doesn't come with feeders, hunting them down will cost as much again, at least, because they are forwards compatible with new models and are highly sought after, whoever is selling this machine clearly kept them for their new one and this is pretty normal.

[mikeselectricstuff]

You need to take a hard look at exactly why you want to go in-house vs. subcontract, as well as explore as many subcontractors as you can. 
With such a wide range of volumes, it may well be that a mix of subcontract for high-volume and in-house for low-volume/urgent makes the most sense.
If you can sensibly sub out the hundreds to thousands jobs, you may well be able to get a perfectly good in-house setup to suit the lower volumes within your budget.

This is not just about equipment, but staffing/your time as well.

[ali_asadzadeh]

185 different parts on 1 board is a lot and suggests poor design choices, cheap Chinese machines often top out at 48 or 60 8mm lanes. Even with more conventional equipment you would typically approach this many parts & feeders with two or more machines inline because 60 front 60 rear is a very "standard" arrangement.

The Designs are simply just complex, since you did not see any let me show you some pictures!



The QTY is usually in the lower end, and 1000 units are not in hurry too, I thought maybe by adding a NeoDen YY1 Pick And Place Machine to the line, The price would not go that far, also I have checked a handfull of our last designs, they usually have under 120 different parts, so I think NeoDen YY1 can come in handy,

If you can sensibly sub out the hundreds to thousands jobs, you may well be able to get a perfectly good in-house setup to suit the lower volumes within your budget.

Thanks for the tip, I can do that, so do you suggest equipment with in my budget?

[Mangozac]

My SMT line cost several times more than $10k and I would hate to try to build boards like that

Too many component types, critical, fine pitch parts and BGAs and very long SMT connectors all in the one board would be a massive pain. Just saying "120 parts" doesn't mean a whole lot - many of those parts if supplied on tape will take up more than one feeder location.

The Neoden will be sloooow and you still have to pay someone to mind it while operating. I don't see the point for production.

[mikeselectricstuff]

For boards that complex in qny more than a few tens, I really think you should find a subcontractor with experience, in-house x-ray etc.
Think about how much time and wasted parts you could spend fault finding and optimising your process.

[nctnico]

For boards that complex in qny more than a few tens, I really think you should find a subcontractor with experience, in-house x-ray etc.
Think about how much time and wasted parts you could spend fault finding and optimising your process.

I agree. Doing in-house production doesn't make sense. Time is better spend on finding a subcontractor that is price competitive while doing a good job.

[Jackster]

My line is 2-3x your budget (I got half of it 2nd hand at a stupidly low price) and even I would somewhat struggle with this.

Cheaper machines without feeders are fine for prototypes doing Rs and Cs but not anything more. They take forever to change reels over and set up.
That and you will want a half decent over and stencil printer, minimum.

Don't forget all the other costs. I am just fitting fume extraction to my unit and well now I don't have any money for a ski holiday 

Your time + running costs might also be more than you save farming this work out don't forget.
Sometimes you are better off spending $50 on paying someone to do it for you when you could make $100 for the same time doing higher-level work. We have all been there.

[ali_asadzadeh]

Thanks guys for the tips and feedbacks, actually being able to manufacture the boards locally has very good advantages for us, Like the lead time, the lower price and the most important one, not compromising the designs!

Also since the production QTY is mostly on the lower side, I guess the machines would be 90% off in a year, so It can be used to manufacture other customer projects too and generate some side income, and I think it can pay itself back soon enough.

Though the challenging part is, the majority of the parts are smaller parts like Resistors, Capacitors, Inductors, LED's, Transistors, Oscillators and crystals and smaller IC's under 10mm x 10mm, so maybe having smaller machines like lumenpnp or NeoDen YY1  can come in handy.
Do we have very competitively priced Ovens that can do both side parts and you are aware of?

[mikeselectricstuff]

Also since the production QTY is mostly on the lower side, I guess the machines would be 90% off in a year, so It can be used to manufacture other customer projects too and generate some side income, and I think it can pay itself back soon enough.

Being a subcontract manufacturer for other people's jobs is a COMPLETELY different thing to doing your own. You will probably need to have a dedicated staff member for this. 

[SMTech]

Thanks guys for the tips and feedbacks, actually being able to manufacture the boards locally has very good advantages for us, Like the lead time, the lower price and the most important one, not compromising the designs!

Also since the production QTY is mostly on the lower side, I guess the machines would be 90% off in a year, so It can be used to manufacture other customer projects too and generate some side income, and I think it can pay itself back soon enough.

Though the challenging part is, the majority of the parts are smaller parts like Resistors, Capacitors, Inductors, LED's, Transistors, Oscillators and crystals and smaller IC's under 10mm x 10mm, so maybe having smaller machines like lumenpnp or NeoDen YY1  can come in handy.
Do we have very competitively priced Ovens that can do both side parts and you are aware of?

I think perhaps you're not getting just how basic these cheap machines are and how very different they are from production machines. They are literally toys. If you were making tiny little controllers for hobbyists such as the many many Arduino derived little projects that people like to make where you have a BOM of maybe 30 lines and you run it now and again, sure it works, provided you accept you will sometimes spend a bit of time prodding stuff back into position that isn't placed accurately, fiddling with a feeder because its not working properly and other annoyances. In addition setting up a pick and place is not a fast process and its slower for low volume as you are very likely to not have all your parts in production packaging (trays, long tapes/reels, tubes), instead you'll have annoying little packets of parts that will need cover extension tape, a custom tray, careful handling labelling and storage so they don't get mixed up once they are out of their packets. This takes TIME, time is money and you're time and skills are needed design, testing & verifying those 20-30 complex designs a year.

I run the SMT of a small sub contractor, we do small batches of boards all the time, some of it really quite complex & expensive. I run equipment considerably better than what you are contemplating and I would decline to manufacture the board pictured if its an example of  one of your 185 lines BOMs and I would probably direct you elsewhere because my line is still missing several expensive pieces of equipment ideally needed to make it that our existing client base do not yet make necessary. Complex boards are expensive to make for all the reasons in the previous paragraph, the Neoden would be more help than just a pair of tweezers but not help enough, from your description even if it was just putting down all the little stuff, its too small and not accurate enough, it will drive you up the wall. The sad reality however is that if you need designs this complex and the associated technology they are simply not low volume friendly, and the most common workaround is to use modules like Pi compute module, Arduino Vidor etc to reduce your manufacturing complexity. (and a whole bunch of impedance matching, and complex design work those parts would need if implementing yourself)

Two sided assembly doesn't require much from an oven other than clearance, you just need to have a fixture of some kind that keeps the PCB suspended off the Mesh/Tray the PCB sits on. I used to use titanium clips traditionally used to strengthen/support a PCB during wave soldering, but these days I have a large Oven with a pin chain conveyor which is an option on most convection reflow ovens for production.

Of course it does depend what you can get over your border too....

[48X24X48X]

$10K probably can get you a 2nd hand Dek stencil printer. That's about it.

[ali_asadzadeh]

Thanks SMTech for the tips,As you just said, we have lot's of complexities with complex boards, they make profit and I should find a way to solve these challenges, so they would not be that hard for the coming years, either I should wait to make more money or should I find a clever solution.