Searching \ for '[EE] Question: groundplanes and apertures in Eagle' in subject line. ()
Make payments with PayPal - it's fast, free and secure! Help us get a faster server
FAQ page: piclist.com/techref/pcbs.htm?key=eagle
Search entire site for: 'Question: groundplanes and apertures in Eagle'.

Exact match. Not showing close matches.
PICList Thread
'[EE] Question: groundplanes and apertures in Eagle'
2005\08\02@174923 by PicDude

flavicon
face
Hi all,

I am generating the gerbers for a relatively large PC Board using Eagle, and
it is taking a *very* long time.  This is really a bunch of projects put
together, totalling to ~14" x 11.5", 2 layers, no soldermask or silkscreen,
and ~4,000 holes.  Machine is 1.6Ghz w/256Mb RAM.  I'm trying to find out why
so long, and what I can do about it.

My guess is the groundplane is the reason -- from past experience I have
noticed that as soon as I add a groundplane to a board, it takes considerably
longer to process (I don't have actual numbers to quantify that), and approx
half of this board has a groundplane on one side.

Currently (over 2 hrs into the processing), it is up to "Printing aperture
D41".  I understand that an "aperture" is an individial element on the board.  
Not sure if that's just components or also includes pads and signals, etc,
but either way I'm sure this board will have much more than 41 apertures and
I'm concerned that this will not be completed in the near future.

Any idea what I'm in for time-wise and what I could do about this to speed it
up?

Thanks,
-Neil.


2005\08\02@225126 by William Chops Westfield

face picon face

On Aug 2, 2005, at 2:58 PM, PicDude wrote:

> I am generating the gerbers for a relatively large PC Board using
> Eagle, and
> it is taking a *very* long time.

> My guess is the groundplane is the reason...

What are you using as the line width for the polygons that make up
the ground plane?  Eagle essentially does a space-filling serpentine
of that linewidth to make up the polygon, so the smaller the line
width, the longer it takes.  (OTOH, increasing the line width will
prevent the ground plane from going some places...)

BillW

2005\08\03@022226 by Vasile Surducan

face picon face
On 8/3/05, William Chops Westfield <spam_OUTwestfwTakeThisOuTspammac.com> wrote:
>
> On Aug 2, 2005, at 2:58 PM, PicDude wrote:
>
> > I am generating the gerbers for a relatively large PC Board using
> > Eagle, and
> > it is taking a *very* long time.
>
> > My guess is the groundplane is the reason...
>
> What are you using as the line width for the polygons that make up
> the ground plane?  Eagle essentially does a space-filling serpentine
> of that linewidth to make up the polygon, so the smaller the line
> width, the longer it takes.  (OTOH, increasing the line width will
> prevent the ground plane from going some places...)

 And keep the ground plane off till the final inspection.

Vasile

2005\08\03@072858 by olin piclist

face picon face
PicDude wrote:
> I am generating the gerbers for a relatively large PC Board using
> Eagle, and it is taking a *very* long time.
>
> ...
>
> My guess is the groundplane is the reason

Probably.  What width parameter did you use for it?  Small width values can
make photoplotting (and generation of photoplotter files) take a long time.

> Currently (over 2 hrs into the processing),

Wow, that really is "long".  Usually the whole process takes a few seconds.
The longest it ever took for me was on the QuickProto-01 board because it
had a sea of holes with a ground polygon around them.  But that only took
maybe a minute or two on several of the arpetures, maybe 5-10 minutes total.

> it is up to "Printing
> aperture D41".  I understand that an "aperture" is an individial
> element on the board.

No.  It is one of the various shapes use to draw with.  Gerber plotting is
really a line drawing process with various shape and size "pens".  Large
areas are drawn by going over them in a raster pattern, just like you would
if you were trying to fill in an area with a real pen.  Of course the finer
the pen tip, the longer this takes.

Photoplotters draw with light onto film, hence the "pens" are apertures that
result in a beam of specific size and shape.  In the old days, there were a
few standard apertures on an aperture wheel that had fixed holes pre-punched
into it.  Now the equipment can do variable aperture on the fly.


*****************************************************************
Embed Inc, embedded system specialists in Littleton Massachusetts
(978) 742-9014, http://www.embedinc.com

2005\08\03@190955 by PicDude

flavicon
face
I used the defaults -- 0.016 width and 0.05 spacing.  Not sure how the spacing
affects this, but I'll have to dig up the manuals and re-learn some things.

By the time I got back, the board was done, and from the time on the file it
appears that it took ~6-7 hours.  Since then, I ran one test with the
groundplane removed and it generated the gerbers in ~2 minutes!

I will experiment with various settings for line width and spacing to see what
it does and how long it takes.

Thanks,
-Neil.


On Tuesday 02 August 2005 09:51 pm, William "Chops" Westfield scribbled:
> What are you using as the line width for the polygons that make up
> the ground plane?  Eagle essentially does a space-filling serpentine
> of that linewidth to make up the polygon, so the smaller the line
> width, the longer it takes.  (OTOH, increasing the line width will
> prevent the ground plane from going some places...)
>
> BillW


2005\08\03@191132 by PicDude

flavicon
face
Yes -- I keep this until the very end as it takes a long time to render, and
when I make my manual changes to the board, the groundplane would get in the
way if present.

Cheers,
-Neil.

On Wednesday 03 August 2005 01:22 am, Vasile Surducan scribbled:
>   And keep the ground plane off till the final inspection.
>
> Vasile


2005\08\03@201112 by PicDude

flavicon
face
On Wednesday 03 August 2005 06:29 am, Olin Lathrop scribbled:
> Probably.  What width parameter did you use for it?  Small width values can
> make photoplotting (and generation of photoplotter files) take a long time.

Default of 0.016".

Just read a bit on polygons in the manual.  Found out that spacing is only
applicable to hatch patterns.


> > Currently (over 2 hrs into the processing),
>
> Wow, that really is "long".  Usually the whole process takes a few seconds.
> The longest it ever took for me was on the QuickProto-01 board because it
> had a sea of holes with a ground polygon around them.  But that only took
> maybe a minute or two on several of the arpetures, maybe 5-10 minutes
> total.

This board is a *bit* bigger at 162 sq in.  But yes, I agree it is excessively
long.


{Quote hidden}

Interesting.  Never knew this.

I just got a chance to experiment with this a bit... no groundplane took about
~2 minutes.  Groundplane with 0.016" width took ~6-7 hours.  Groundplane with
0.05" width took ~4 minutes.  Not a lot of datapoints, but enough to prove
that was indeed the problem.  I'll have to experiment with various settings
next time.

Thanks,
-Neil.



2005\08\04@073937 by olin piclist

face picon face
PicDude wrote:
> I just got a chance to experiment with this a bit... no groundplane
> took about ~2 minutes.  Groundplane with 0.016" width took ~6-7 hours.
> Groundplane with
> 0.05" width took ~4 minutes.

A width of .050 is quite large, and may prevent the ground plane from
flowing between many adjacent pads and other areas, so check it carefully.
I'm surprised it took that long at .016, that's not really that small.  Are
you really sure it defaulted to what you thought it did?  If you've still
got the old version, do a VIEW > INFO on the polygon to check.  Otherwise
try setting it explicitly to .016.

The time should be roughly proportional to the inverse of the square of the
width.  If it took 2 minutes at .050, then it should take about 20 minutes
at .016.  Actually it will take a bit longer if the plane now flows into
more spaces, and there are probably some other inefficiencies, but I still
don't see how it should take 6 hours.  Is the smaller setting possibly
causing RAM usage to exceed the real memory size?  Once the OS starts
thrashing things can slow down by an order of magnitude or two.


*****************************************************************
Embed Inc, embedded system specialists in Littleton Massachusetts
(978) 742-9014, http://www.embedinc.com

More... (looser matching)
- Last day of these posts
- In 2005 , 2006 only
- Today
- New search...