[gta02-core] schematics symbol and drawing style

Werner Almesberger werner at openmoko.org
Fri Jun 19 22:00:06 CEST 2009

Rene Harder wrote:
> I learn today that the ANSI [1] specification also include most of the
> IEC symbols, same for logic symbols [2].

So it has both styles ? Unfortunately, the documents are members-only.

> We go for the traditional Symbols of the ANSI [1] specification because
> we already have lots of symbols drawn in this particular style. Also
> Openmoko is using this style which makes it easier for us to compare our
> drawings to Openmoko's.

Sounds good, yes.

> - Symbols: traditional US symbols [1], if a symbols is already used in a
>   different part of the schematics we'll use the same symbol, not an
>   alternative style!!

Yup, using different symbols for the same thing in our schematics
would be bad.

> - Sheet size is A3


> - Font size for labels, names etc. is 50mil

Sounds good. We almost use that consistently already, with the following

- dtc123.lib, emh4.lib, ht210.lib, k4m51323pe.lib, and sw_push_4.lib
  have pin text that differs from 50 mil (usually it's 40 and 60 mil)

- labels on signal default to 60 mil, which is what seems to be used
  consistently so far.

- except for the 2442 and the powers, all component names and references
  are set with 60 mil

If we're agreed on 50 mil, I'll fix the renegade libs and change the
signal names to 50 mil. Not sure what we want to do about component
names and references. Might as well leave them at 60 mil.

- some symbols use 60 mil fonts (e.g., k4m51323pe.lib and sw_push_4.lib)

> - no cross-shaped junctions
>   (see [3] for examples, style 3 is the preferred one)

Yup, with the exception that KiCad doesn't put junction circles
when a signal enters or leaves a bus.

> - we'll use Openmoko's component references. Removed symbol
>   references will not be reused. New symbols get unique references.


> Did i forgot anything or do you have a different opinion?  Please, let
> me know.

There's also spacing: the general rule of thumb is to keep things
at least 200 mil apart from each other, e.g., signals changing
direction after a pin, joints after a pin, etc. This is not a hard
rule. Sometimes, higher density still works, particularly in a
regular structure.

Parallel signals are separated by at least 100 mil.

- Werner

More information about the gta02-core mailing list