6

There's a spectrum of how you get PCBs assembled.

1) Have a human pick out all the parts, solder them all on by hand, and check them all by eye. Low cost of entry. Good for low-volume high-mix work. Surface mount is difficult/impossible, depending on the package.

N) A pick-and-place machine, wave soldering, and automated optical verification. Very expensive equipment. Only cost-effective at high volumes, 100+ at least. Through-hole is difficult/impossible for some pick-and-place machines.

What's in between? Reflow ovens? Automated solder paste dispensers? Some sort of automated part dispensing system that's between a wall full of bins and a pick-and-place machine? Stuff I haven't imagined?

The bigger question is this: what steps can be taken to increase the efficiency of an entirely hand-solder operation? In what stages can such an operation make the transition from all hand-soldered to fully automated production? And what is the cost range on the equipment involved?

Stephen Collings
  • 17,373
  • 17
  • 92
  • 180
  • 2
    Your #1 is wrong about surface mount. SMD is easier and simpler than thru hole. Problems occur with fine pitch, leadless packages lik QFN, and ball-grid arrays. SMD by itself is not a problem, in fact is preferred over thru hole. The only place for thru hole nowadays is where the extra mechanical strength is useful, like with connectors or large and heavy parts. – Olin Lathrop Aug 06 '13 at 13:48
  • @OlinLathrop Yeah, it clearly depends on the package, like I said. But assume you start at 1) with all through-hole parts. At some point you'll need to change to SMT in order to move towards automated production. Mass-producing large-pitch exposed-pad SMT devices may require some extra step that isn't involved in through-hole stuff, which I'm not quite clear on. Glue, perhaps? Solder paste instead of strand solder? – Stephen Collings Aug 06 '13 at 14:45
  • 1
    No, hand soldering large-pitch SMD parts requires *less* steps than for thru hole. You don't have to flip the board over and clip off the extra lead lengths. It's also easier to position a SMD part in the first place since the leads don't have to be threaded thru holes. This allows the board to sit flat on a table, for example. Thru hole is a pain. And no, you don't start with thru hole. Nobody uses thru hole anymore except for special purposes. – Olin Lathrop Aug 06 '13 at 14:49
  • @OlinLathrop I'm phrasing poorly (and maybe this is outside the scope of my original question as worded). Let me rephrase. Assume your process is already in place to do through-hole parts; since many parts only come in through-hole, that's a reasonable assumption. A through-hole-only process is (for many applications, not all) more flexible than an SMD-only process. Now assume you want to add in SMD parts. Are you saying that NO new equipment or techniques are involved? Just set the part down and solder it on with the same strand solder you'd use on through-hole parts? That's the info I want. – Stephen Collings Aug 06 '13 at 14:54
  • 3
    *"Many parts only come in thru-hole"*. Huh!!? Way more parts are available in modern packages. There is no demand for archaic thru hole, except for special cases where the extra mechanical strength is useful. As a result manufacturers generally don't make thru hole parts anymore. About the only mass produced thru-hole boards are old designs that aren't worth re-doing for economic or certification reasons. When thru-hole equivalents are available, they are more expensive due to the low volume and more complicated manufacture. – Olin Lathrop Aug 06 '13 at 15:02
  • 2
    @OlinLathrop You clearly live in a different realm of electronics than I do. Many terminal blocks and power converters I use simply can not be had in SMD packages. – Stephen Collings Aug 06 '13 at 15:05
  • 1
    Connectors and large things tend to be through-hole, ICs often SMD only. Often you end up with an entirely SMD board that needs a few TH things hand-soldered on the edge. – pjc50 Aug 06 '13 at 15:08
  • 1
    @Remiel But as Olin said, there are special cases where TH is king and those are mostly when mechanical strength is required (Connectors, large inductors, etc.) or high currents (some power supply components, etc.). Other than these special cases, professional EE's prefer SMT over TH. This has been the case for over 20 years. –  Aug 06 '13 at 15:10
  • @DavidKessner I'm not disagreeing. I'm trying to find out how I can gradually modernize an archaic production process, and we keep getting dragged off onto whether I'm technically correct about part availability in different packages. – Stephen Collings Aug 06 '13 at 15:11
  • @Olin "Nobody uses thru hole anymore except for special purposes" You clearly don't have in mind the electronics hobbyists who mostly use thru hole parts.. – m.Alin Aug 06 '13 at 15:29
  • 1
    @m.Alin: OK, then *nobody to reasonable approximation* uses thru-hole. Resistors, capacitors, microcontrollers, and just about every other component are not made for hobbyists since their demand amounts to about 0 to reasonable approximation. Bringing up hobbyists in this discussion just adds confusion since hobbyists aren't relevant. – Olin Lathrop Aug 06 '13 at 15:39
  • You may want to include an Electronics Manufacturing Service (EMS) as one of your intermediate steps. Many are very reasonably priced and willing to work with quantities as small as prototype quantities. They often also have the expertise to handle advanced packages (such as fine-pitch BGAs). You can then focus on repair/rework equipment on a smaller budget. – Tut Aug 06 '13 at 20:52

2 Answers2

6

As Olin said, SMT is normally easier than TH. I can solder SMT parts much better and faster (except for QFN's and BGA's, which I don't have the tools for).

Much of the remainder of your question is highly dependent on the situation and opinion. Let me give you some random data points for you to consider:

  • The company that I work has three PCB assembly lines. For prototypes we often run about 12 PCB's in a batch. But we have one assembly line that we call our "Build To Order" line, in which products are only built once they are ordered from the customer. That line is setup to be economical just running a batch of a SINGLE PCB.
  • Prototype runs on some assembly lines is considered a dry-run for production. So even if that run of boards is not cost effective in the short term, it might still need to be done to make sure that the assembly line is properly configured for volume production of that board.
  • Reflow ovens are the preferred method to solder SMT parts. Wave soldering was preferred for TH parts. But it is becoming increasingly popular to get rid of the wave soldering machines and do TH parts in a reflow oven. This is done by putting enough paste down the hole.
  • Hand soldering, even in volume production, is sometimes required for difficult parts. When this occurs, the other parts are normally soldered using the machines and only the difficult parts are hand soldered. An example of this might be a large inductor that is right next to a BGA. It could take a lot of heat to solder that inductor properly in an oven, enough heat to damage the BGA. In that case, the inductor would be hand-stuffed.
  • The costs of a professional level SMT assembly line varies depending on speed, volume, accuracy (for small parts), etc. I would guess you could do it for as little as US$100k, but $500k to more than $1 million is not uncommon. Places that do a lot of BGAs usually have an X-Ray machine that could cost $750k all by itself.
  • Most pick and place machines are separated into TH machines and SMT machines. I'm sure there is something out there that can do them both, but I haven't seen one. If you want to do them both, then you need two machines.
  • The best way to improve the efficiency of hand-soldering is to not do it. Seriously! Or to get things assembled in a country with very cheap labor.

The different levels of assembly are roughly this:

  1. Hand assemble a circuit on perf-board. Useful for making ONE unit, ever.
  2. Have the EE stuff a PCB by hand. Useful for 1 or 2 units.
  3. Have a technician stuff a PCB by hand. Useful for 1 to 20 units, depending on the number of parts and the complexity.
  4. Stuff some parts by hand, some by machine. Only useful if you can't do it all by machine.
  5. Do it all by machine. Useful for 1 to 1,000,000+, depending on the complexity of the circuit and how the factory is setup.

It is hard to generalize about assembly methods and cost-effectiveness without considering the factory/company that will be doing the assembly.

5

You certainly can hand-solder down to "0402" parts with strand solder and a suitably narrow chisel bit: apply flux, place part, get a pad-sized bead of solder on the iron, apply to one end of part while holding part with tweezers, dab more solder on iron, do the other end.

A head-mounted magnifier will help you see what you're doing at that scale.

The next stage of neatness is to hand-apply solder paste to the pads before putting the parts on. The paste is slightly sticky and includes flux, so you get better joints and less risk of the part pinging off never to be seen again.

If your run is more than a few, get a solder paste mask made which turns applying paste into a single action: align a panel of PCBs, wipe paste across, remove.

Once you're using paste you can consider reflow. Reflowing both sides (one after the other, not both at once!) can be fiddly as you need to make sure the components on the bottom don't fall off.

The next stage is to go pick-and-place; there are people doing cottage-industry or homebrew pick-and-place, sometimes with secondhand equipment. The actual soldering is still pastemask and reflow oven.

The final stage is to buy the fully automated line as a single unit: boards go in one end, component reels in the side, finished boards come out the end with automatic inspection. Worth it if you can keep it busy.

I've also seen partial wave soldering used to add TH parts to SMT boards. This involved extra production jigs to hold the TH parts in place and shield areas of the board from wave soldering.

pjc50
  • 46,540
  • 4
  • 64
  • 126
  • Can you elaborate on "Wipe paste across"? My first guess is that we're talking about a jar of solder paste and some large flat edge (like, say, a putty knife) to spread a thin layer with. I'm picturing it somewhat like applying thermal paste to a device and heat sink. Is there more to it than that? – Stephen Collings Aug 06 '13 at 15:20
  • It's almost exactly that. You apply a layer to the top of the stencil and scrape over it, and the amount of paste left on the pads when you remove the stencil is determined by its thickness. – pjc50 Aug 06 '13 at 15:34
  • The only part you have to be extra careful with is when you remove the stencil. You have to make sure that you do not smear that paste that you just applied. (this is of course assuming that you are trying to do this by hand and dont have all of the correct tools) – EE_PCB Aug 06 '13 at 16:38
  • +1. You might add a link [showing that reflow](http://www.opencircuits.com/Skillet_reflow) on only one side is relatively easy. – davidcary May 24 '14 at 17:03