TSSOP 38 routing problems

It's a good idea to read section in the manual on autorouting. The manual

is found in the .docs folder. There it describes the impact of the routing

grid. That is a different grid to the display grid.

When a board design was simple and had no fine pitched devices you could

start the autorouter with no consideration of its settings or those of the

design rules (DRC) and get a result many would accept. As pin pitch, trace

width and clearances get smaller then you have to prepare for these

requirements. If you don't then the autorouter cannot fully route because

all of the numerous rules cannot be met. The goal is to make it easy for

the autorouter otherwise it will appear to do odd things and not even get a

result when manual routing would.

One approach is to identify the smallest pitch device, your TSSOP, and

determine the maximum trace width and spacing it will support. Looks like

5mil for the TSSOP. Next you need to choose the routing grid pitch. Smaller

will ensure routing completes but it increases the routing time

exponentially so you are after the largest you can get away with.

For your simple example I would set the routing grid to match the pitch of

the TSSOP pins. I would temporarily set the display grid to this size and

move the TSSOP so that the pin origins match the grid, then restore the

desired display grid.

Keep in mind that the autorouter routes to the intersections of the routing

grid.

HTH

Warren   

--

Web access to CadSoft support forums at www.eaglecentral.ca.  Where the CadSoft EAGLE community meets.

On Wed, 28 Nov 2012 10:24:54 +1300, Warren Brayshaw wrote:

It's a good idea to read section in the manual on autorouting. The

manual is found in the .docs folder. There it describes the impact of

the routing grid. That is a different grid to the display grid.

 

When a board design was simple and had no fine pitched devices you could

start the autorouter with no consideration of its settings or those of

the design rules (DRC) and get a result many would accept. As pin pitch,

trace width and clearances get smaller then you have to prepare for

these requirements. If you don't then the autorouter cannot fully route

because all of the numerous rules cannot be met. The goal is to make it

easy for the autorouter otherwise it will appear to do odd things and

not even get a result when manual routing would.

 

One approach is to identify the smallest pitch device, your TSSOP, and

determine the maximum trace width and spacing it will support. Looks

like 5mil for the TSSOP. Next you need to choose the routing grid pitch.

Smaller will ensure routing completes but it increases the routing time

exponentially so you are after the largest you can get away with.

 

For your simple example I would set the routing grid to match the pitch

of the TSSOP pins. I would temporarily set the display grid to this size

and move the TSSOP so that the pin origins match the grid, then restore

the desired display grid.

 

Keep in mind that the autorouter routes to the intersections of the

routing grid.

What Warren said.

What happens is that the autorouter is only going to put traces on the

grid, and it'll only put a trace to a pin if it can do so without

violating clearances.  Where I've run afoul of this is with an inch-pitch

board and a mm-pitch part: even with a fine autoroute pitch, there are

inevitable mismatches between the board grid and the centers of the pins;

the autorouter will try to keep the traces on its grid, which means that

it comes off of the part with a little jog.  If that jog puts the trace

into another pin -- it doesn't get routed.

Setting the autoroute grid to match the dimensions of the finest-pitch

parts (which are usually mm these days), and making sure to put all of

those parts on with the correct grid (be sure to pick them up at the

center with a control-left click), should make it all work.

The autorouter works by first doing a "boneheaded" route, with traces

going strictly up and down on one layer and side to side on the other. 

Then it optimizes a bunch of vias away.

If you're not getting 100% routing in that first step it'll go ahead and

optimize anyway (at least 5.x does).  So while you're making your board

fit for autorouting, it's a good idea to turn off the optimization

steps.  That way you can inspect what the autorouter has done and get an

idea of what you need to do to make autorouting possible.

Densely packed parts with lots of pins will spawn a lot of vias in close

proximity, and there needs to be enough space for all of those vias in

that first step.  So once you get the autorouter so it can connect the

pins at all, then you may need to fine tune the board so that it can get

past that first step.

Make sure to define the smallest vias you or your board house are

comfortable with. 

If that's not enough, and if the autorouter seems to be dying because

of not enough space between vias under or around the chip then decrease

the autoroute grid further, to let the autorouter squeeze in more vias.

If that doesn't do it, then make sure that your dense parts have lots

of space around them -- pack things too tightly around that 100 pin LQFP,

and you're autoroute will never happen.

And, if all else fails, hand route.  I find it's generally better to

spend time jiggering the board around so that it'll route itself, because

I inevitably get the board all routed then find out some circuit change

that Must Be Made; it generally doesn't take much to make a minor change

and hit autoroute again if you've already made it work once.

--

My liberal friends think I'm a conservative kook.

My conservative friends think I'm a liberal kook.

Why am I not happy that they have found common ground?

Tim Wescott, Communications, Control, Circuits & Software

http://www.wescottdesign.com

On Wed, 28 Nov 2012 10:24:54 +1300, Warren Brayshaw wrote:

It's a good idea to read section in the manual on autorouting. The

manual is found in the .docs folder. There it describes the impact of

the routing grid. That is a different grid to the display grid.

 

When a board design was simple and had no fine pitched devices you could

start the autorouter with no consideration of its settings or those of

the design rules (DRC) and get a result many would accept. As pin pitch,

trace width and clearances get smaller then you have to prepare for

these requirements. If you don't then the autorouter cannot fully route

because all of the numerous rules cannot be met. The goal is to make it

easy for the autorouter otherwise it will appear to do odd things and

not even get a result when manual routing would.

 

One approach is to identify the smallest pitch device, your TSSOP, and

determine the maximum trace width and spacing it will support. Looks

like 5mil for the TSSOP. Next you need to choose the routing grid pitch.

Smaller will ensure routing completes but it increases the routing time

exponentially so you are after the largest you can get away with.

 

For your simple example I would set the routing grid to match the pitch

of the TSSOP pins. I would temporarily set the display grid to this size

and move the TSSOP so that the pin origins match the grid, then restore

the desired display grid.

 

Keep in mind that the autorouter routes to the intersections of the

routing grid.

What Warren said.

What happens is that the autorouter is only going to put traces on the

grid, and it'll only put a trace to a pin if it can do so without

violating clearances.  Where I've run afoul of this is with an inch-pitch

board and a mm-pitch part: even with a fine autoroute pitch, there are

inevitable mismatches between the board grid and the centers of the pins;

the autorouter will try to keep the traces on its grid, which means that

it comes off of the part with a little jog.  If that jog puts the trace

into another pin -- it doesn't get routed.

Setting the autoroute grid to match the dimensions of the finest-pitch

parts (which are usually mm these days), and making sure to put all of

those parts on with the correct grid (be sure to pick them up at the

center with a control-left click), should make it all work.

The autorouter works by first doing a "boneheaded" route, with traces

going strictly up and down on one layer and side to side on the other. 

Then it optimizes a bunch of vias away.

If you're not getting 100% routing in that first step it'll go ahead and

optimize anyway (at least 5.x does).  So while you're making your board

fit for autorouting, it's a good idea to turn off the optimization

steps.  That way you can inspect what the autorouter has done and get an

idea of what you need to do to make autorouting possible.

Densely packed parts with lots of pins will spawn a lot of vias in close

proximity, and there needs to be enough space for all of those vias in

that first step.  So once you get the autorouter so it can connect the

pins at all, then you may need to fine tune the board so that it can get

past that first step.

Make sure to define the smallest vias you or your board house are

comfortable with. 

If that's not enough, and if the autorouter seems to be dying because

of not enough space between vias under or around the chip then decrease

the autoroute grid further, to let the autorouter squeeze in more vias.

If that doesn't do it, then make sure that your dense parts have lots

of space around them -- pack things too tightly around that 100 pin LQFP,

and you're autoroute will never happen.

And, if all else fails, hand route.  I find it's generally better to

spend time jiggering the board around so that it'll route itself, because

I inevitably get the board all routed then find out some circuit change

that Must Be Made; it generally doesn't take much to make a minor change

and hit autoroute again if you've already made it work once.

--

My liberal friends think I'm a conservative kook.

My conservative friends think I'm a liberal kook.

Why am I not happy that they have found common ground?

Tim Wescott, Communications, Control, Circuits & Software

http://www.wescottdesign.com