# PCB assembly cost?

Status
Not open for further replies.

#### wannaBinventor

##### Member
I've been looking at the cost of PCB assembly and I have been absolutely shocked. I am working on something with about 20 different components, most of which are only used once or twice.

I'd be using a couple of LGAs, some SMT connectors, and then I was planning on using SMT capacitors. I'll use through hole for bigger resistors and what not.

The quick quotes I'm getting show prices that I just look at as ridiculous. Just doing things like adding two LGAs to each board boosts the cost by $15 dollars per board. To me that is insane. Insane. Insane! What's the solution here. I was figuring about$10/board for assembly in my cost calculations, and I seem to be dead wrong here. It's so expensive that I just want to scrap the entire idea.

What are my options here? Am I looking in the wrong places?

#### Diver300

##### Well-Known Member
A lot of the cost is setting up. How many boards are you wanting to get made?

Larger components are much more difficult to handle on pick and place machines, as the feeders are a lot more expensive and take up more space.

However, if the cost is going up by $15 for two LGAs makes me think that they are either doing it by hand, or having to move to a much more expensive machine. It could be that you are paying for a stencil that is needed for the LGAs. A small loss rate on the LGA components will put the price up a lot. #### wannaBinventor ##### Member I'll probably do 10-15 myself to start with and see how they sell (After lots of testing of course). If they sold quickly, I'd try and do a run of probably 100 or so and then go from there. I'd only "go all in" and scale up production after sales looked promising. #### carmusic ##### New Member you need quantity at least 100 to get 10$ per board assembly, at this low volume 50-80% of the cost is just setup time! Just to setup a board it takes hours to the company who makes assembly to setup their machines and production line.
For very small volume they solder it by hand!

#### wannaBinventor

##### Member
Thanks for the replies guys. I think I'll get some stencils made, get a hot plate and do it myself. I think in a weekend I could probably assemble and test 100 boards and save myself 3 grand.

#### Mr RB

##### Well-Known Member
20 components including fussy LGAs will probably take you 1-2 hours each when soldered by hand including soldering, testing cleaning and errors. That's for one person, it gets better with 2 or more people as you can optimise the system for efficiency, but it's also worse if you haven't done it before and don't have jigs and sytsems already set up.

As for doing 100 boards in a weekend, that should be fine if you can do 100-200 hours work in a weekend.

#### wannaBinventor

##### Member
Mr. RB:

I'm not talking about hand soldering. I'm talking about reflowing everything but the through hole devices, and then just running the iron to them. There will only be 7 or 8 of them anyway.

I picture it running like this:
Laying all the components out in individual boxes with everything labeled. I'll take the PCBs and run paste over a stencil that I'll have made. With practice, I believe that I could stencil a small board in a about 2 minutes, place components in about 5 minutes, and then lay it on the hot plate and start the process over for the next board as that one started to heat. I've seen pictures on spark fun electronics where they have made some very professional looking boards with this method.

When all of this is done, I can test each board. Perhaps I'm ignorant, but I'd think that this entire process wouldn't take more than 15 minutes per board once I got it going.

I think testing would be as easy as hooking it up to a computer through UART, having an ISR programmed for an interrupt on change, and then just bridging one of the INT pins that will be tied to ground with a resistor to Vcc to enter the ISR. The ISR could take a series of readings from the enternal components, and I could have something programed on the computer side that checks the received UART message for reasonability. From there, I think it's just a matter of close visual inspection and leaving it on for a while to make sure it doesn't get too hot. The actual hands on time per board for testing would be maybe 5 minutes.

Am I too optomistic here?

Last edited:

#### Mosaic

##### Well-Known Member
Commercial items ought to be ROHS compliant ....do u have a lead & cadmium free process?

#### wannaBinventor

##### Member
The components are all going to be ROHS compliant led-free, and I could just make sure that I only use led free solder-paste.

From the PCB manufacturing perspective, I think it would be cheaper/faster for assembly on my end to have a large board made with several copies of the board circuit.

IE: One PCB is 3x3 inches, so I put it on a 9x9 and get 9 actual units from one. If that's too big for the PCB manufacturer, I could probably do a 6x6 and get four from it.

Anyone know the best method to preforate the board so that I could assembly several circuits on one board, and then when everything was dry just break them apart into the individual units.

#### Mr RB

##### Well-Known Member
wannaBinventor-
Am I too optomistic here?
I think so.

Over the years I've done some small quantity manufacturing myself, I've helped others set up their small quantity manufacturing and worked with associates doing small-med quantities.

It always ends up chewing up more labour than you expected, especially if you want to sell a quality product that has had a lot of visual inspection and human supervised testing. And for projects that are new and unknown that labour blowout is always increased again.

I believe that I could stencil a small board in a about 2 minutes, place components in about 5 minutes, and then lay it on the hot plate and start the process over for the next board as that one started to heat.
That sounds like a two man job, and still optimistic. Hot plate SMD soldering on a start-up budget will probably require 100% concentration from an operator and 5 minutes to place 20 SMD parts with 2 LGAs... You might be able to do that in a race, but try it for a 10 hour shift.

Anyway I'n not trying to discourage you, there's no reason you can't efficiently produce good quality gizmos with budget start-up equipment and a one person crew. But as for doing 100 in a weekend on your first manufacting attempt? I think you might need a few friends there and LOT of coffee!

I'm curious how big are these LGA's that they want to charge you an extra $15 to place? And what does this gizmo do? Testing usually takes more time than you think, and the more complex the fucntion the more testing is needed. #### blueroomelectronics ##### Well-Known Member How many pins are the LGAs? #### Mosaic ##### Well-Known Member The components are all going to be ROHS compliant led-free, and I could just make sure that I only use led free solder-paste. Anyone know the best method to preforate the board so that I could assembly several circuits on one board, and then when everything was dry just break them apart into the individual units. I'd scribe it with a Plexi carbide cutter & a metal straight edge before the hot plate step, then snap it off after. I am actually modding a soldering station to work as a Hot air pencil tonite. This unit: Stahl Tools SSVT Variable Temperature Soldering Station | Parts-Express.com It's hollow and Ive sealed up all vents with JB weld. For the air flow I place a tiny 1/8" barbed Elbow in the tail of it with more JBweld and connect an aquarium air feed. To control airflow I have those handy super precise adjusters from a$2, Intravenous drip unit. I had those leftover from a Continuous ink supply mod.

So I should have both temp control with the Station and airflow control!.

For a tip I am using a metal cone tip from a disposable ballpoint pen. I apply threadseal tape to the end of the Soldering iron threads then JBweld on that , then set the Cone tip on that...let it set and I have a threaded cone tip!.

I plan to make other tip shapes with some heavy copper foil/fins I have left over from a PC radiator project I did.

#### RMMM

##### New Member
You should have ZERO through hole devices. That will drop your cost.

Through hole still has to be done mostly by hand. Switch all of your components to smd and see what the price changes to.

I would shop around.

Also, I dont see why you would be using a LGA and only having 19 other components on the board. I would look for a replacement.

They are the LEAST RELIABLE package when it comes to solder problems along with BGA.

I would take a good long look at your controller and see what you can do away with. Get to the easiest hacker friendly package you can.

The best thing you can do for your device is to allow it to be re-purposed by the consumer.

#### wannaBinventor

##### Member
Thanks for the replies guys.

Mr. RB:
What do you look at as reasonable in terms of assembly time then using the hot plate method?

For what it's worth, I think I'll be able to fit 9 circuits on one board, so I'd have the stencil made as such and apply the paste to all of those at one time, then place the components.

The LGAs consist of an LGA-14 and LGA-16 from Freescale and STMicro. All lands are around the outside perimeter of the chip.

Mosaic:
Please let me know how that works out for you.

RMMM:
Unfortunately these are only available in the LGA package.

As far as the through hole stuff: I'm having trouble coming across decently priced SMD resistors with good wattage values (as in at least half watt or more). Then again, maybe I just need to do the math for what they cost over a through hole versus what they cost to place.

Last edited:

#### Mosaic

##### Well-Known Member
Zero thru hole is not always an option. Especially for power parts that have thermal considerations or display parts like LCD/LED screens & digits.

#### atferrari

##### Well-Known Member
Maybe the OP won't even look at this thread again, which is more than 3 years old.

Status
Not open for further replies.