[sdiy] I could use a few pointers on doing a board layout

[James Patchell]

To see what Harry is talking about...take a look at this file:

http://www.oldcrows.net/~patchell/synthmodules/vocalfilterpicture.html

It demonstrates exactly what Harry is talking about...more or less...

At 10:10 PM 9/7/2005 -0400, harrybissell wrote:
>OK I'm in...
>
>1)  YES by all means route power traces and grounds first, along with
>decoupling caps.  I prefer to route the +/- and ground beneath each
>IC between the .3" dimension  (assuming through hole) on the SOLDER
>side of the board.  Youc an fit three .040" traces underneath the IC on
>the
>solder side. Use small vias to change directions.
>
>2) once you set those power and ground traces, run all lines PARALLEL
>to those traces on the Solder side as well
>
>3) Run all traces on the COMPONENT side at 90 degree angle to the solder
>
>side traces.
>
>4) Use VIAS where ever you need to change direction from horizontal to
>vertical.
>(break this rule ONLY if you are so close to the target pad that you
>could not
>possibly fit a trace between the via and the pad.
>
>5) I run axial componentsoriented in the direction that pins 1 and 8
>would be on the
>ICs. This lets you route traces between component leads AND through IC
>pins on the
>COMPONENT side of the board.  I never run traces between pins on the
>solder side
>to avoid solder shorts
>
>6) Group all components that go 'with' a particular IC in line with the
>pins of that IC...
>Picture an 8 pin dual opamp... the feedback resistor goes from pin 1,
>the inverting
>input resistor goes to pin 2.  Run these on the COMP side.  Keep the
>traces on
>input pins short.  If you have a long run, run it from an output through
>a long trace...
>and keep the resistor at the input pin and SHORT. The long run will be
>tied to the
>output for most of the way and won't pick up noise.
>
>7) The name of the game is ROUTING CHANNELS.  The techniques I've
>described
>keep the maximum number of options open. DO NOT try to save 'vias'.
>They don't
>cosr much and the blocking of a routing channel to save one hole will
>cost time and
>money later.
>
>8) Do complete netlist checking if the program allows it. Make sure that
>EVERY
>complaint is accounted for (you might say, OK I know that run is close,
>or off grid)
>but NEVER accept any schematic errors. Resolve them all
>
>9) Fvck autorouting. Need I say more on the subject ???   :^P
>
>10) Spend a couple of hours when you are finished to do clean-up work.
>Its called
>ARTWORK... and it should be beautiful when its finished. I'll get a
>little 'mystic'
>on you now and say that if the layout is NOT beautiful when done, look
>out for stupid
>schematic or design errors.  Now is the time to remove excess vias.
>Make sure that you have all the testpoints you'll want... add some extra
>holes for power and ground
>(you might need to hook up some more panel stuff, right ???)
>
>Rock out !
>
>H^) harry
>
>Jeff Farr wrote:
>
> > While we're talking about eagle, I've still got a few pressing
> > questions.  I've got some nice schematics drawn up (like the ARP 4035
> > ladder filter module) and believe I'm ready to being making a board.
> > And here the process comes to a screeching halt.  I've now made
> > several attempts, but really can't get anything remotly exceptable.
> > Should I try to lay down traces for the supplies and ground first?
> > why don't the icons for supply show up as something on the board
> > screen?  I'm using the freeware version so I don't have the
> > autorouter, is this worth the money?

         -Jim
***************************************************************
http://www.oldcrows.net/~patchell

***************************************************************