Eagle for drawing electrodes

Community PCB Design
1

Hi all,

I would like to use Eagle to define electrodes on a small PCB which will be used for an experiment. I happen to need an elliptical electrode, which I had some difficulty in drawing in Eagle. In the end, I chose to implement the ellipse as a large number (~32) of polygon wedges produced by script, as in the following:

http://web.mit.edu/kimt/www/eagleroutin … olygon.PNG

http://web.mit.edu/kimt/www/eagleroutin … lygon2.PNG (after ratsnest)

http://web.mit.edu/kimt/www/eagleroutin … lygon3.PNG

Now, in order to get electrical access to the ‘center’ and ‘ring’ electrodes, I used vias and traces from the corners of the device, as follows:

http://web.mit.edu/kimt/www/eagleroutin … lygon4.PNG

If I now produce the gerber files, I find that the via/backtrace and ring electrodes are considered to be electrically independent, and a finite amount of area around the via is cut away according to the DRC (I believe). So, I went to rename the rear trace to have the same name as the ‘ring’ electrode, but then I find that the ring electrode gets reduced to:

http://web.mit.edu/kimt/www/eagleroutin … lygon5.PNG

http://web.mit.edu/kimt/www/eagleroutin … lygon6.PNG

and I also get a small ‘unrouted line’ between the two vias. The gerber output, according to a viewer, looks identical to the copper (as seen by polygons in real mode) as displayed above in Eagle.

Questions:

  1. Practical: I want to keep the ring electrode as it is, but have connections through vias. Is there a way that I can enforce this result?

  2. General: It seems that Eagle is automatically deciding that I don’t need the extra polygons (of the ellipse) on the side, and insisting that I don’t pour copper on it and instead offers this ‘unrouted’ line. What is the logic that Eagle is applying that yields this result?

  3. More background: It’s also pretty clear that Eagle layout is intended to route signals rather than defining physical electrode shapes (I could not find any other primitives other than straight lines and circular arcs, for instance). For my application, I need specific electrode shapes, and tried to hack a solution through polygon wedges. Will this run into problems down the line in the fabrication? For instance, the CAM process warned me of generating large data in gerber files. I tried reading the gerber output, but didn’t really know how to make sense of it. Is the typical process of PCB manufacture a vector drawing technology, that may get overloaded if I define so many polygons – or does the CAM ‘reduce’ my (bloated) vector data into a simpler form of some kind (i.e. reduce to vector over the actual shape outline, or some sort of raster)?

2

Hi,

The issue is that eagle won’t pour any polygon that doesn’t contain another object of the same net-name.

To work around this - have the script create the elliptical electrode as a single polygon with a continuous border, rather than as disparate polygons. To add the center cutaway, have the script plot all the outer points, then jump to the inner points, around the center cutaway the opposite direction, and back out to the first outside vertex. This creates a “C” shaped polygon, but with the tips of the C closed, such that the created copper will be solid. The polygon should then pour.

–David Carne

3

Or use software that doesn’t suck… Altium Designer handles this a lot more gracefully, though there is still room for improvement.

4

busonerd:
Hi,

The issue is that eagle won’t pour any polygon that doesn’t contain another object of the same net-name.

To work around this - have the script create the elliptical electrode as a single polygon with a continuous border, rather than as disparate polygons. To add the center cutaway, have the script plot all the outer points, then jump to the inner points, around the center cutaway the opposite direction, and back out to the first outside vertex. This creates a “C” shaped polygon, but with the tips of the C closed, such that the created copper will be solid. The polygon should then pour.

–David Carne

Thanks for the suggestion of a “C”-shaped polygon, David.

For your reference, what you suggested did not work. Here’s what happened. First, I made sure that I was indeed generating a C-shaped polygon, by purposely not closing the elliptical ring:

http://web.mit.edu/kimt/www/eagleroutin … lygon7.PNG

http://web.mit.edu/kimt/www/eagleroutin … lygon8.PNG

However, when I tried to close the C, Eagle seems to be “one step ahead” and decided that what I really wanted was a completely filled ellipse:

http://web.mit.edu/kimt/www/eagleroutin … lygon9.PNG

http://web.mit.edu/kimt/www/eagleroutin … ygon10.PNG

Instead, your explanation that Eagle was not pouring copper onto my polygons because it did not contain an element of the same net-name, pointed to an alternate solution of implementing the ellipse as two C’s, mirrored such that their union is the final shape that I want. This has the following consequences:

– Because the shapes are independently C’s, Eagle does not get smart on me and fill in the center,

– Since I have two vias with the netname, both polygons are drawn.

Ultimately, I ended up with what I want (checked Gerber outputs as well). Thanks!

http://web.mit.edu/kimt/www/eagleroutin … ygon11.PNG