At 09:56 AM 9/5/01 +0200, Florian Finsterbusch wrote:
>On our multilayer board the top and bottom layer should be connected to GND.
>For that purpose we have placed polygons on both layers.
>The polygons are connected to the GND net.
>The pads should be surrounded by arcs.
>Grid Size = 0.2 mm, Track Width = 0.22 mm

First of all, set the grid to zero. I also recommend using imperial units 
for the track width, though I am not sure that this will make a difference; 
it's just that the Protel internal database is imperial so you might get a 
slightly better pour.

>When protel is pouring the polygon, we have rectangles around some pads.
>Also we have rectangular openings in the polygon itself!

Something like this is to be expected under some conditions. For some 
reason the pour routine is unable to place the fill tracks; if an arc is 
missing, any opening left will be rectangular, if one has 90 degree 
hatching selected. Mr. Finsterbusch did not state his setting for the 
minimum primitive size. If this is too large there will likely be missing 
primitives. This would only get worse with a fixed grid size.

A minimum length of zero seems to work fine. However, under some conditions 
this could result in too many pour tracks and I would not be terribly 
surprised if Protel crashed. I leave it at 1 mil. One could make it smaller 
than that.

Try setting hatching style to "No Hatching" and turn off "Remove Dead 
Copper." This will show you only the pad clearance outlines and the outline 
of the polygon. With this setting, polygon pour will surround each pad with 
an arc or octagon (octagons may reduce plot size if software arcs are used) 
*if* the clearance rules will allow it. The grid size has no effect on 
this. If you are not getting an outline around a pad, there are two 

(1) your clearance rules will not allow it.
(2) there is a bug. I think I have seen some circumstances where the pour 
outlines are incomplete, but it is difficult to reproduce and I don't have 
an example handy.

Number (1) is the most likely cause. Try placing a line or arc primitive 
where you think a missing primitive would be. Assign it the GND net. Does 
this create a clearance violation? If so, no wonder the pour does not 
complete the fill!

Then, if hatching is turned on, fill track will be added. This track is *on 
grid*. If your grid setting does not meld well with the pad placements, 
some fill tracks will be missing, causing rectangular holes in your pour. 
For this reason, set the grid to zero. Protel properly interprets this as 
meaning "fill gridless." This is generally recommended, it should be the 
default setting!

There is little reason to use cross-hatching (90 degrees or 45 degrees) 
when grid is set to zero and a very small primitive length is used. It will 
just add extra lines. Note, however, that lines which are precisely butted 
up next to each other can display a very fine gap, either in PCB or in some 
gerber viewers. That is not real, it is a display artifact. But if one is 
not willing to tolerate the appearance of this false gap -- it should not 
be on the film -- then using cross-hatching will eliminate it. That may be 
a better solution than using a pour grid and a slightly oversize track.

There are other possible causes for missing track. For example, there might 
be a layer-specific keepout primitive that is invisible, or some other rule 
interaction. Note that there are design rule clearance settings for 
polygons which are in addition to the other clearance rules. (Most 
designers would prefer larger clearances on polygons than elsewhere on a 
layer because the polygon clearances are everywhere and thus more likely to 
cause fabrication or soldering problems.)

If the cause of the problem is not found, I recommend creating a small file 
that shows the problem. (Edit down your existing file). This will help 
Protel, but before sending it to Protel, submit it to me or to another user 
who has indicated a willingness to look at it -- don't attempt to send it 
to the list!)

If there turns out to be a bug here, we can then add this to the bug database.

Abdulrahman Lomax
Easthampton, Massachusetts USA

* * * * * * * * * * * * * * * * * * * * * * * * * * * * * *
* To post a message: mailto:[EMAIL PROTECTED]
* To leave this list visit:
* Contact the list manager:
* Forum Guidelines Rules:
* Browse or Search previous postings:
* * * * * * * * * * * * * * * * * * * * * * * * * * * * * *

Reply via email to