Auto-Routing Double Sided Board with non Plated Through Holes

On 16/12/2017 7:18 a.m., Antzy Carmasaic wrote:

As a hobbyist, I find it much faster for prototyping make my own boards and recently got a CNC router to speed up prototyping. I usually route the boards myself but sometimes don't have enough time to do it by hand. The problem is that while the auto-router works fine for hand-made single sided boards, for double-sided boards, there is no way to turn off PTH for the pads. It will always assume the top and bottom of all pads to be connected by default and route accordingly. As everyone knows, making PTH at home is not feasible(time and money).

Is there any setting to tell the auto-router not to assume pads' top and bottom to be connected?

One way I see is by using a ULP that places pads' duplicate shapes on bRestrict for parts on bottom side and on tRestrict for parts on the top side. Is there any such ULP available?

 

--

To view any images and attachments in this post, visit:

https://www.element14.com/community/message/232393

 

Hi

I don't remember hearing of such a ULP but one could be built.

Maybe a project over the Christmas break.

As with any coding, one needs to define the problem to be solved so lets

clarify yours.

(1) You wish to use the auto-router to route traces on the top and

bottom of the board. A double sided board.

(2) There are some PTH pads on the top or bottom layer that you wish to

have the auto router not connect to. For those PTH pads, the auto-router

can route on one layer but not the other.

(3) Not all pads on a side (layer) would be barred to the auto-router.

That would be a single sided board which can be handled already.

(4) You accept vias that the auto router places which you will use wire

to connect top and bottom layers

(5) There may be areas that vias should not be placed

(6) For a given device the restrict and keep-out requirements would be

the same for all other identical devices on the board

Let us know if that covers it

Warren

--

... use NNTP://news.cadsoft.de and a functional news reader like

Thunderbird!

---

This email has been checked for viruses by Avast antivirus software.

https://www.avast.com/antivirus

This is what I got till now:

if (!board)
{
    dlgMessageBox("Start this ULP in a Board");
    exit (-1);
}


string result;
    
board(B)
{
    B.elements(E)
    {
        E.package.contacts(C)
        {
            if (C.pad)
            {
                string temp;
                sprintf(temp, "%s (%d) - %d, %d", C.name, C.pad.diameter[LAYER_BOTTOM], C.pad.x, C.pad.y);
                result += temp + "\n";
            }
        }
    }
    
    dlgMessageBox(result,"+OK");
    
    exit(0);
}

This does give me the correct pad names for all pads on board. But the diameter, x and y are way too large. For a board on a 0.05" grid, I'm getting diameter of 422656 and positions like 2844800 and 2032000. Can anyone help me get the correct pad positions and diameter?

I'm planning to draw circles in tRestrict and bRestrict layers at pad positions according to diameter of pad. Not perfect for square or oblong pads but should work as a base code.

If I'm going in the wrong direction, do guide me in any way to copy the exact pad shape to another layer. Thanks.

On 12/17/2017 02:26 PM, Antzy Carmasaic wrote:

I'm getting diameter of 422656 and positions like 2844800 and 2032000. Can anyone help me get the correct pad positions and diameter?

I think you need to add a member identifier to specify units on your

output. ie mils in mm or such. Look at some of the ULP's for pick and place.

Paul

Thanks Paul. It was actually because ULPs get all positions and dimensions in Eagle's internal units. They have to be converted using u2mic, u2mil, u2mm or u2inch. The funny thing is all their examples in their ULP manual just prints without conversion as well. Only by going through other ULPs could I find these functions, given only a small section in the ULP doc:

EAGLE stores all coordinate and size values as int values with a resolution of 1/10000mm (0.1µ). The above unit conversion functions can be used to convert these internal units to the desired measurement units.

Anyways, I got the script made. It works well and does almost everything I had initially planned. Here it is for those looking for a similar solution:

#usage "Allows using auto-router for double-sided boards without PTH\n"
       ""
       "Eagle assumes all pads as PTH which is not true for hand fabricated boards. "
       "This ULP copies pad shape in tRestrict for normal elements' pads, "
       "and in bRestrict for mirrored elements' pads. "
       "This prevents the auto-router from routing between top and bottom layers "
       "using the pads, and so forces it to use vias."
       "The vias can be physically connected to top and bottom using "
       "a wire soldered at both ends."
       "
"
       "Currently, only circular pad shapes are copied in restrict layers. "
       "Other shapes are assumed as the largest circle which can fit inside the shape. "
       "This works well for square, octangon and long pads."
       "
"
       "Author: antzy.pantzy@gmail.com"


// THIS PROGRAM IS PROVIDED AS IS AND WITHOUT WARRANTY OF ANY KIND, EXPRESSED OR IMPLIED


if (!board)
{
    dlgMessageBox("Start this ULP in a Board");
    exit (-1);
}


string result, s;


s += "GRID MM FINEST;\n";
s += "CHANGE WIDTH 0;\n";


board(B)
{
    B.elements(E)
    {   
        E.package.contacts(C)
        {
            if (C.pad)
            {
                string temp;


                if(!E.mirror)
                {
                    s += "CHANGE LAYER tRestrict;\n";
                    sprintf(temp, "CIRCLE (%.4f %.4f) (%.4f %.4f);\n", u2mm(C.pad.x), u2mm(C.pad.y), u2mm(C.pad.x) + (u2mm(C.pad.diameter[LAYER_TOP]) / 2.0), u2mm(C.pad.y));
                }
                else
                {
                    s += "CHANGE LAYER bRestrict;\n";
                    sprintf(temp, "CIRCLE (%.4f %.4f) (%.4f %.4f);\n", u2mm(C.pad.x), u2mm(C.pad.y), u2mm(C.pad.x) + (u2mm(C.pad.diameter[LAYER_BOTTOM]) / 2.0), u2mm(C.pad.y));
                }
                
                s += temp;
            }
        }
    }
    
    s += "GRID LAST;";
    
    exit(s);
}

Here's a screenshot of a quick test I did. 1 of the caps is mirrored, so will be placed on the bottom side. So traces should only go to it from the top side. The autorouter did exactly that. There are also 2 bRestrict areas separating caps, an 8-pin connector and a TO-220 component. The autorouter only took traces to them across the bRestrict using top layer and then assuming pads as disconnected on top layer, used vias to connect to them from the bottom layer. So yeah, works perfectly!!!

Thanks Paul. It was actually because ULPs get all positions and dimensions in Eagle's internal units. They have to be converted using u2mic, u2mil, u2mm or u2inch. The funny thing is all their examples in their ULP manual just prints without conversion as well. Only by going through other ULPs could I find these functions, given only a small section in the ULP doc:

EAGLE stores all coordinate and size values as int values with a resolution of 1/10000mm (0.1µ). The above unit conversion functions can be used to convert these internal units to the desired measurement units.

Anyways, I got the script made. It works well and does almost everything I had initially planned. Here it is for those looking for a similar solution:

#usage "Allows using auto-router for double-sided boards without PTH\n"
       ""
       "Eagle assumes all pads as PTH which is not true for hand fabricated boards. "
       "This ULP copies pad shape in tRestrict for normal elements' pads, "
       "and in bRestrict for mirrored elements' pads. "
       "This prevents the auto-router from routing between top and bottom layers "
       "using the pads, and so forces it to use vias."
       "The vias can be physically connected to top and bottom using "
       "a wire soldered at both ends."
       "
"
       "Currently, only circular pad shapes are copied in restrict layers. "
       "Other shapes are assumed as the largest circle which can fit inside the shape. "
       "This works well for square, octangon and long pads."
       "
"
       "Author: antzy.pantzy@gmail.com"


// THIS PROGRAM IS PROVIDED AS IS AND WITHOUT WARRANTY OF ANY KIND, EXPRESSED OR IMPLIED


if (!board)
{
    dlgMessageBox("Start this ULP in a Board");
    exit (-1);
}


string result, s;


s += "GRID MM FINEST;\n";
s += "CHANGE WIDTH 0;\n";


board(B)
{
    B.elements(E)
    {   
        E.package.contacts(C)
        {
            if (C.pad)
            {
                string temp;


                if(!E.mirror)
                {
                    s += "CHANGE LAYER tRestrict;\n";
                    sprintf(temp, "CIRCLE (%.4f %.4f) (%.4f %.4f);\n", u2mm(C.pad.x), u2mm(C.pad.y), u2mm(C.pad.x) + (u2mm(C.pad.diameter[LAYER_TOP]) / 2.0), u2mm(C.pad.y));
                }
                else
                {
                    s += "CHANGE LAYER bRestrict;\n";
                    sprintf(temp, "CIRCLE (%.4f %.4f) (%.4f %.4f);\n", u2mm(C.pad.x), u2mm(C.pad.y), u2mm(C.pad.x) + (u2mm(C.pad.diameter[LAYER_BOTTOM]) / 2.0), u2mm(C.pad.y));
                }
                
                s += temp;
            }
        }
    }
    
    s += "GRID LAST;";
    
    exit(s);
}

Here's a screenshot of a quick test I did. 1 of the caps is mirrored, so will be placed on the bottom side. So traces should only go to it from the top side. The autorouter did exactly that. There are also 2 bRestrict areas separating caps, an 8-pin connector and a TO-220 component. The autorouter only took traces to them across the bRestrict using top layer and then assuming pads as disconnected on top layer, used vias to connect to them from the bottom layer. So yeah, works perfectly!!!

Way to go!  That was some nice work!

Thank you Gene! Thanks to you, Warren and Paul, I kept at it, else would have given up long ago.