As someone who's manufactured thousands of PCB designs, I couldn't disagree with this article/video more. Some tips he provides are aspects of good design, regardless of aesthetics, but a lot of what he suggests decreases manufacturability/yield. PCB designers should focus on making a robust, functional design and consult their manufacturer's manufacturing/process engineering team for advice on how to improve the manufacturability of the design. Almost never will those changes improve the aesthetics of the design.
??? IME soldermask is pulled back from the pad generally unless you're doing soldermask defined pads for things like BGA. Plus, the fab generally modifies the pullback themselves. My soldermask layer is pretty much just 1:1 with the pad outline, if I want a particular pullback I would specify that in the fab drawing.
I can't see how soldermask color would affect placement either, especially not if you place fiducials (or if the fab places them for you!)
Pigments can affect the properties and behavior of resins, if you're working with a sloppy fab I could see that causing issues (even if only not accounting for different curing times).
For actual production design like the OOP is talking about, you would likely go with whichever option is cheapest overall (which in my experience has universally been the standard green).
Exactly, if you're hand soldering through hole components, it's not a big deal. However, blue/black/white solder mask pigments were noticeably (>2mil) smaller in tests than the standard green for a high volume manufacturer. Most of the time a partially mangled pad reflow wouldn't be a problem, but when you're automatically stuffing the boards, you won't know and could have latent failures (temperature/time) after production test. This is especially the case for aQFNs and BGA ICs.
Fiducials are automatically placed in top copper, but we also noticed larger misalignments of paste and smaller 01005 passive components (pre-reflow) on non-standard colors. Again, unlikely to cause large yield problems unless you've got a lot of components, but why risk it.
We did still use red for test PCBs so that they wouldn't be shipped to customers. If we were making gaming video cards colors would have been more of a priority.
Let me counterexample where changing soldermask is good: In my experience, changing from Green Soldermask (everybody's default) generally gets you better soldermask coverage as they aren't trying to optimize usage to the micropenny.
If I want mega-cheap, green is good. If I want reliability, almost any other color is better.
I love seeing hardware posts on here. I often use colors other than green to differentiate revisions, however green typically has the smallest minimum web requirement which makes a difference on high density designs.
Hmmm, never thought about that. I don't think I've noticed a different minimum web requirement based on color for 0402-class and QFN stuff, but I could certainly believe it for extremely small footprints and high-density BGAs.
Where I seem to see issues with green soldermask coverage are tented vias underneath components.
The canonical example is a non-GND via underneath a QFN component with a pad. However, I have also seen issues with vias next to large pads of power transistors where the transistor can "float" around on the solder.
This seems counterintuitive since I would expect the main soldermask color to be used for more BGA-type components where the breakout vias would need to be covered properly. Nevertheless, experience suggests otherwise.
I finally got fed up once and ordered the same PCB with green soldermask for one batch and blue for another. The green ones had some via coverage fails; the blue ones didn't have any. Since then, I've just avoided green soldermask unless I'm going for uber-volume levels of PCBs.
> As someone who's manufactured thousands of PCB designs, I couldn't disagree with this article/video more.
I think you're vastly overstating how much "bad" advice there is in the video, and the degree to which aesthetics are opposed to function. It is possible to get silly (e.g. luridly coloured PCBs, "artistic" routing, and silkscreen-as-canvas), but this video doesn't push very hard.
Phil mostly stresses attention to consistency in placement, routing, and silkscreen, which is 100% solid. A PCB layout is like a schematic: it's read far more than it's written, and efforts to make it legible and coherent have an excellent payoff/effort ratio. Efforts here make bring-up, rework, debugging, and overall comprehension easier for the EEs involved. They also make the design more accessible to non-EEs, which fosters collaboration with mechanical/thermal, assembly, debugging, etc. They may have minor impacts on cost and manufacturability, but the right compromises are usually obvious if the knowledge is there.
There is a design sensibility for PCBs that combines form and function without compromising either. Most experienced EEs can tell with a single glance whether a board was laid out with care by an experienced practitioner or not. I find good PCB design to be beautiful by construction, in the same way that good architecture is beautiful by construction (and not by slapping lipstick on a pig).
I guess I agree in a sense that function of a PCB isn't necessarily related to aesthetics, but there's a strong correlation between sloppy layout and designs that are a pain to work with, both in ECAD and in the physical world.
That aside, I don't think there's any plainly bad advice in here. What recommendations do you see as having a negative effect on yield?
Edge plating/castellations/milling (not drilling) copper in general is one of the biggest causes of scrapped boards in the PCB manufacturing process as the the lateral movement of the drill bit causes the copper layer to delaminate. Most board houses will upcharge for this because all of the problems it causes.
Improving PCB manufacturing yield is a very complex discussion which the concept of aesthetics has actually been a major problem in the industry since the 80's when operators would endlessly reflow solder joints until 'they looked good'.
While the rest of the advice is not bad, there are objective reasons why it is better that way other than the subjective reason of aesthetics.
