Schematic ERC: Conflict problem between pins, severity error (solved)

Application: kicad
Version: (2016-11-30 revision 6d6542e)-makepkg, release build
Libraries: wxWidgets 3.0.2
libcurl/7.46.0 OpenSSL/1.0.2d zlib/1.2.8 libidn/1.32 libssh2/1.6.0 librtmp/2.3
Platform: Windows 8 (build 9200), 64-bit edition, 64 bit, Little endian, wxMSW

  • Build Info -
    wxWidgets: 3.0.2 (wchar_t,wx containers,compatible with 2.8)
    Boost: 1.57.0
    Curl: 7.46.0
    KiCad - Compiler: GCC 5.2.0 with C++ ABI 1009
    Settings: USE_WX_GRAPHICS_CONTEXT=OFF
    USE_WX_OVERLAY=OFF
    KICAD_SCRIPTING=ON
    KICAD_SCRIPTING_MODULES=ON
    KICAD_SCRIPTING_WXPYTHON=ON
    BUILD_GITHUB_PLUGIN=ON
    KICAD_USE_SCH_IO_MANAGER=OFF
    KICAD_USE_OCE=ON

Recreating the sheets from scratch by pointing to sch file did not change anything.

There is a new stable release (4.0.5 since this week i think). You could try that.

I’ll try it but I doubt it, the release log says nothing of the kind related to what I’m experiencing.

I just looked at your file again.
Another very unlikely source could be your symbol.
You named your pins in unit a and b the same.
maybe name them ch1+,ch1-,ch1_out
(I’m clutching on straws here. I never experienced such a weird behavior.)

@kaitrek, please check your opamp symbol against pin overlaps. Someone might hide common pin 1 in unit B or common pin 7 in unit A, because didn’t know how to edit pins in the right way in multi unit components.
Load symbol to library editor and perform basic check (ladybug icon).

This component from Linear library, LT1492, looks the same way as mine.


Pin names don’t matter I think, as long as pin numbers are different.

Like I said, when I had one CH1, there were no errors, when I added CH2 and CH3, I started getting these errors.I just searched for the “net 4” and I keep finding the same net in different sheet names CH1/2/3, which is wrong, hence the error ERC throws out.

keruseykarya: I made the component, here it is, no bugs (duplicate or offgrid pins, etc) in library editor:

Could you provide with us a complete project?

And one very important question: Did you annotate full schematic after creating whole hierarchy with Eeschema’s Annotate tool?

Providing the whole project would be tough, have not played with ways to export custom components along with the project yet. I was just trying to reproduce the error with LT1492, just the input-preamp section with 2 channels, but I’m not getting the error. I’ll keep trying to narrow it down or reproduce.

Yes, I have re-annotated the entire schematic, otherwise I’d get multiple components error dialog before ERC is completed.

OK, SOLVED.
Let me point something out to you from the picture in my first post.
LOL, ffs damn it, couldn’t see that damned wire…

No errors now.

THANK YOU ALL!

2 Likes

Good catch! By the time that screen shot gets to my laptop screen, the colors are rather muted and smeared so the error was hiding in plain view.

I vaguely recall that somebody suggested an alternative color scheme about a year or 18 months ago. One of the objectives was to make connections (or lack of connections) more obvious, but I thought the bright, neon-like, colors created more eyestrain for the 99% of the time when you are NOT looking for errors. In EESchema you can experiment with color schemes using the “Preferences” > “Schematic Editor Options” > “Colors” menu.

I will give you credit for presenting a challenging case. Every few weeks somebody brings a problem that boils down to lines that appear to be connected, but actually aren’t. (Grid size mismatches, and lack of dots indicating junctions. are the usual suspects.) This is the first time I recall having an extra connection!

Dale

It would help if hierarchical sheet pins looked more like regular symbol pins.

I cannot think of a situation where a wire segment should be allowed to overlay a symbol box edge.
I have made this mistake once and its hard to spot, impossible in a black and white print

That’s quite true if you persist in the idea that a schematic must be human-readable and human-comprehensible. With this assumption information is lost when one line obscures another

If you define the schematic as a method of defining the components which make up a particular electronic circuit, and describing their interconnections, then colinear segments are a lesser problem. In the example we’re considering here, there is no problem because the line representing a wire (or connection) is an entirely electrical concept; while the line marking the edge of the functional block’s symbol has no electrical significance whatsoever.

I don’t know if it would be practical to identify this situation during drafting and issue a squawk since the condition occasionally happens as you move things around in the drawing. (And, hopefully, you promptly locate and correct all of these errors.) It would be helpful to include this type of fault in the DRC, with an error message something like, “Portions of the Net nxxxx connections are inside a component symbol at {X, Y}.”

Dale

If the schematic is not meant to be human readable, you might as well directly generate a netlist

1 Like

I’ve seen a few, mostly in the world of logic, that are essentially just that: many components, arranged neatly on a page, but with very few actual connections shown. Instead, every pin was labeled with a signal name. Essentially impossible to comprehend the relationships between those signals.

Dale

That sounds like it came from the wire wrapping era. When one of these boards was built it was almost impossible to trace the wiring by eye and all you could do was buzz it. I am really glad wire wrap has gone away

Also could control line width individually. Have global presets for wire, bus, polygons, and text. Then also have control over individual placed elements that would override global control, or sheet control.

I think the most recent builds include net highlighting in eeschema - selecting a net with the label tool will highlight the whole net. Not certain when this was added.

But you would have to be looking at a human readable schematic to see it :slight_smile:

1 Like