Hatch pattern on ground fill causes small features below minimum clearance

When a copper zone with hatch pattern is filled, it can create sharp corners with widths far below the minimum clearance value. I suspect that this could cause issues in manufacturing, though I haven’t tried yet.

Is this known, or is it or is it not considered a problem in real life?

If it is a problem, does someone know whether a bug report already exists?

I might be seeing this wrong but the feature you are measuring looks like its not on the same layer where the hatch pattern fill is. Am I wrong ?

There are filled copper zones with hatch pattern on both sides, in this case I am measuring the blue side (back), and the red one with the larger “hole” is on the other side (front) which is not active.
I admit that it would have been clearer if I had switched off the view of the other layer.

Application: KiCad x86_64 on x86_64

Version: 7.99.0-1.20230524git76d6657.fc37, release build

wxWidgets 3.2.1
FreeType 2.12.1
HarfBuzz 5.2.0
FontConfig 2.14.1
libcurl/7.85.0 OpenSSL/3.0.8 zlib/1.2.12 brotli/1.0.9 libidn2/2.3.4 libpsl/0.21.1 (+libidn2/2.3.3) libssh/0.10.4/openssl/zlib nghttp2/1.51.0

Platform: Fedora release 37 (Thirty Seven), 64 bit, Little endian, wxGTK, KDE, wayland

Build Info:
Date: May 24 2023 17:17:15
wxWidgets: 3.2.1 (wchar_t,wx containers) GTK+ 3.24
Boost: 1.78.0
OCC: 7.6.3
Curl: 7.85.0
ngspice: 40
Compiler: GCC 12.3.1 with C++ ABI 1017

Build settings:

1 Like

Ah, I see better now, thanks. Not really a clearance thing is it ? all those features are on the same net ? it might be a sliver violation if you have that turned on. I think you have to fix those manually.

I might have missed the sliver violation thing… Where do I turn that on?

It is not strictly a clearance violation in the sense between different nets, but I have read that very small features are still a potential problem in manufacturing and can cause trapping of acid for example, which would cause a local deviation in the etch rate, or possibly during copper plating.
I am not an expert in manufacturing though, that’s why I ask for input from others who have experience and knowledge relevant to this case.

Neither am I :wink: everyday is a school day . . . sliver stuff is here: Board setup

I guess it could be argued it’s not a “sliver” either, more of the opposite of a copper sliver.

In my opinion it’s just a same net clearance thing and not considered by the normal standard DRC.

If we are pedantic, how would you technically separate this from a 90 degree inner corner? If you go close enough and choose two measuring points it will eventually violate any arbitrarily small copper-to-copper clearance, even 1 nm.

One could argue that “Allow fillets outside zone outline” should have effect here and help in some situations – maybe it even is a bug that it doesn’t – but it wouldn’t fix all minimum clearance situations.

I’ve never seen a manufacturer remark on these because they are same-net. At a microscopic level, the features will never come out the same shape as the gerber exactly, and in my experience, all that’s at risk here is that in this area there is potentially more deviation from the design than other areas. But these days with modern processes, “more” is relative and I can’t imagine this causing anyone actual problems.

1 Like

I have played around with it a little bit, and it seems that the hatch algorithm takes the smallest clearance into account in a situation like the one with 0,204mm in the screenshot below. If I move the via closest to the measurement any closer, the whole hatch gap where the measurement is located, will disappear.

But it doesn’t take sharp corners into account at all, and happily makes these kinds of sharp “needle shaped” bays that you can see near the upper right corner of R1 or below pad 1 of R1.

I would argue that a 90 degree corner is less concerning* than these kinds of deep and narrow bays, and it would look better if the algorithm just made a rounded corner where the bay width approaches the minimum clearance.

It might of course be that the software algorithms have a hard time taking into account whether the inner angle of the corner is 90 degrees or something considerably less.

  • I have seen some advice in a couple places that suggest a rounded 90 degree corner is better than a sharp one, for example on pads.

This is for surface-mount soldering process reliability, not for board manufacturability.

That could very well be where I have seen it. But we can disregard the 90 degree case for now, my main concern was about these narrow and long pointed “bays”.

I could imagine an algorithm which walks along the edge of the copper and looks if there is any other copper at a distance of [minimum clearance] perpendicular to the edge. In these pointed bays, it would then make a rounded corner where the bay gets too narrow.
90 degree and wider angle corners would not be considered a problem.

I can’t say if this is the code-wise most efficient way to implement it, but it is meant as a theoretical illustration of what I think could be done to achieve a more optimized shape of the hatch openings.

Here is one place where Siemens mentions this kind of a situation, the specifics are about in the middle of the page in this link:

An example where different parts belonging to the same net have low clearancearts, is shown in this figure:

Apparently someone has thought about it before me, so it is not just me making it up…