>> >Making the Schematic Library part is only a 5 minute job,
>> >and has the benefit of letting you arrange the
>> >pins.  For example all the inputs on the left
>> >and the outputs on the right often makes the
>> >most understandable schematic.
>
>Yes.


In simple systems I'd agree with you, but strictly adhering to
that rule will make a mess in larger systems.


>> It's supposed to be a logical flow of information designed to
>> take you quickly to the part of the circuit you're interested in,
>> and to tell you what signals are involved in that portion of the
>> circuit, and where they come from, and go to.
>
>I *stronly* prefer the kind of schematic that matches the layout or at
>least physical pin-out. It makes troubleshooting on the physical thing
>much faster. For this, I have to redo the libraries for most gates in
>Orcad for example.


IMHO, this results in a mess, since the mfgrs aren't usually kind enough
to put the pins for a given section of the chip in the same place.
Power pins are on opposite sides or ends, which would mean that
the bypass cap has to leap across many other signals, giving the
impression that lead length isn't important. (!)

I move power and ground together, such that the CAP NP just fits,
and then the schematic strongly suggests that this is a critical connection.

The osc pins move near the ground pin(s) for the same reason.

Everything else is grouped by function, so that I have, at a glance, every
signal needed for a given function called out.

If my repair techs can't remember how to count pins, then they need to
find a new job.  I also include dupes of the datasheets on anything that
has more than two legs, so they really have all the info that's needed.

What I'm after is to quickly convey to them, if a given section of the
product isn't working, what signals does that section need to function,
and what signals (if any) does it need to produce.  Once they've got
that info, they find what's missing, and then the sub-sheets detail where
any given signal comes from.

Schematics aren't terribly useful in pulling out manufacturing defects
anyway, since you can get shorts between entirely unrelated sections
of the board simply because the two tracks routed through the same
area.

>In fact, I think that the logical functionality of a unit should emerge
>from a block diagram, not from the schematic. That one is for other
>purposes (such as manufacturing and troubleshooting).


The top level sheet of all my schematics is a block diagram, but I want
to carry that concept all the way through. There's a lot more information
to be conveyed than "This hooks up to that".

Doing it the other way leaves you with these big bricks that have no clear
relationship with each other, and it's not obvious where a signal goes, once
it leaves chip A..