Cost optimization - PCB Area vs. Double-sided load

Question

If one has an opportunity to choose between laying out a PCB as either a single-side load (components all on one side, so it goes through reflow only once and requires only one stencil) or a double-side load but using half the PCB area (so twice the number of boards per panel), which cost optimization tends to be more economical?

I've read this question, but there is no accepted answer and the two answers that are present seem to be at odds with each other. One answer shows that single vs. double-sided for the same board resulting in a small savings ($0.161/board), but that answer didn't include a comparison for board area reduction. It did seem to imply that the single vs. double side load cost different is rather small.

The other answer seemed to imply that single-sided assembly was probably the larger cost driver, but then states that it impossible to know without specifics.

Presume that no components require special treatment (i.e. no gluing or hand-soldering - just a bunch of SMD components).

Answer 1

There are a lot of moving parts when making this comparison, and there's not an easy answer. A double-sided assembly can reduce board area, but can also increase the cost of the board per unit area.

One thing to consider is that a double-sided design doesn't actually give you twice the space:

• Each via takes up real estate on every layer of the board, which impedes routing efficiency.
• IC's often cannot be routed "back-to-back", because the via pads from one IC interfere with the lands of the second IC.
• Even if you can make it work, it is considered bad practice to mount IC's (especially BGA's) back-to-back, as it makes reflow soldering (and also subsequent x-ray inspection) more difficult.

These problems can be overcome by using HDI techniques (e.g. blind/buried vias, microvias), but this significantly adds to the cost of the boards. Not to mention an increase in design complexity.

---

To compare assembly quotes is easy enough; they only care about how many parts, how many layers, how many reflows. They can quote you for both of these options, before the layout is completed.

With the PCB quotes, on the other hand, it can be difficult to get a board house to make estimates unless you have a completed design to review. I would recommend making two mock-ups for yourself to see how much space you predict you'll need. Also figure out your stackup, your required trace/space, etc, and then call your board house. Be polite, ask for favors, and they may be able to give you an estimate on the two designs :)

---

Finally, in my experience, it isn't a good idea to choose double-sided assembly for cost optimization. For size optimization, though, a tight, double-sided multi-layer HDI design can be a great (if more expensive) solution.

Answer 2

In scenarios like this, I usually ask the fab/ CM to quote both. Even if the solution isn't totally routed, you can place components in an approximate location and change the board size to create an "A" and a "B" option. To them, it's just another quote.

Ultimately it comes down to the cost model of the company you're working with, and the size of your components/ overall solution (including how well it fits into a frame, etc.). If you have one large MCU and a bunch of tiny discretes, a 10% board size reduction may not matter in the face of labor costs for a second SMT process.

However, if you have lots of large-ish components and a double sided board gives a massive size reduction, the fab can order fewer frames and potentially save you quite a bit.

Answer 3

You can do your own estimates by using board assembly houses online estimating tools. Example, https://www.screamingcircuits.com/ Enter the basics of your project, board size, if you want to create a PL in their requested format they'll even get prices from Digikey for parts (make sure you use Tape/Reel part numbers, or the Digireel part numbers). You can supply your own board, or the tool can use prices (from Sunstone I think), supply your own parts, etc.