How important are chamfered routes?

[rickman]

I was once told by a professional layout person that leaving a 90
degree inside corner on intersections of traces and/or pads was not
good. Seems the acid is harder to wash out of the little corners and
can result in over etching. I have seen tons of boards that don't
bother with this and I have found few who have even heard of this. Is
this a real issue? Is it only an issue on boards with very fine
traces, like 6 or 5 mil?

I have not seen much in layout packages that would make this easy to
do. In fact, it can be very hard to do without direct support from the
package.

 

[Gary Crowell / VCP]

I was once told by a professional layout person that leaving a 90
degree inside corner on intersections of traces and/or pads was not
good. Seems the acid is harder to wash out of the little corners and
can result in over etching. I have seen tons of boards that don't
bother with this and I have found few who have even heard of this. Is
this a real issue? Is it only an issue on boards with very fine
traces, like 6 or 5 mil?

I have not seen much in layout packages that would make this easy to
do. In fact, it can be very hard to do without direct support from the
package.

This is called an 'acid trap' for inside 90 degree corners and more
acute intersections. Many CAD packages will DRC for it and flag them
as problem points.

However, fab processes have improved such that I don't think anyone
really cares anymore. If you've got a really low-tech fabricator,
maybe its still a problem.

BTW I think the term usually used is 'mitering'.

Mitering is not necessary for signal integrity, unless you are well
into multiple gigaHertz.

For lower frequency signal integrity, extensive mitering might reduce
trace length, and that's usually always a good idea.

BTW, what is the difference between a champfer, and a bevel? I've
looked and never found good distinguishing definitions.

Gary Crowell CID
Micron Technology

 

[rickman]

However, fab processes have improved such that I don't think anyone
really cares anymore. If you've got a really low-tech fabricator,
maybe its still a problem.

BTW I think the term usually used is 'mitering'.

Mitering is not necessary for signal integrity, unless you are well
into multiple gigaHertz.

For lower frequency signal integrity, extensive mitering might reduce
trace length, and that's usually always a good idea.

BTW, what is the difference between a champfer, and a bevel? I've
looked and never found good distinguishing definitions.

Just to be clear, I am not just referring to using 45 degree corners in
mid trace turns. More specifically the hard part is joining to a pad
with all angles of 45 degrees.


cham·fer (chăm'fər) pronunciation
tr.v., -fered, -fer·ing, -fers.

1. To cut off the edge or corner of; bevel.

Seems like the non-jargon definitions are pretty much the same.

mi·ter (mī'tər) pronunciation
v.tr.

1. To bestow a miter upon.
2.
a. To make (two pieces or surfaces) join with a miter joint.
b. To bevel the edges of for joining with a miter joint.

Even miter is about the same.

For now I am not going to worry about it since the software does not
support it. I tried beveling the inside corners using just traces on a
very small board I did a while ago and it was a real PITA. It was not
so hard to do the first time, but moving anything was a lot more work.

 

[Gary Crowell / VCP]

Just to be clear, I am not just referring to using 45 degree corners in
mid trace turns. More specifically the hard part is joining to a pad
with all angles of 45 degrees.


cham·fer (ch?m'f?r) pronunciation
tr.v., -fered, -fer·ing, -fers.

1. To cut off the edge or corner of; bevel.

Seems like the non-jargon definitions are pretty much the same.

mi·ter (m?'t?r) pronunciation
v.tr.

1. To bestow a miter upon.
2.
a. To make (two pieces or surfaces) join with a miter joint.
b. To bevel the edges of for joining with a miter joint.

Even miter is about the same.

For now I am not going to worry about it since the software does not
support it. I tried beveling the inside corners using just traces on a
very small board I did a while ago and it was a real PITA. It was not
so hard to do the first time, but moving anything was a lot more work.

Ah, at trace-pad intersections, what you are referring to is usually
called 'teardropping', that is, using added traces at the trace-pad
joint to 'feather' the trace into the pad. Done for three reasons
that I can think of.

1. to eliminate the acid trap. (and, as I mentioned, not usually
important anymore.)
2. to relieve the stress concentration and cracking potential at the
joint.
3. to reduce the potential for cutting the trace by a wandering drill.

You're right, this would be extremely tedious to do by hand, though
many CAD/CAM systems will do it automatically. You could even ask the
board fab house to do it for you, and it should be no charge in most
cases.

Two notes: In some pin layouts, interstitial connector pin patterns,
in particular, teardropping can cause DRC errors due to reduced
clearance. Teardropping with arcs can reduce this problem, but I've
not seen a CAD system that does this.

An alternate to teardropping is a 'snowman'; just a smaller round pad,
off-center from the pad, that overlaps the trace-pad joint. Probably
easier to do if you have to do it by hand.

Gary Crowell CID
Micron Technology

 

[Brad Velander]

Guys, guys, guys,
Entering a pad at 90 degrees is not considered an acid trap,
never was. Not even a 90 degree trace corner, it is any acute
angle entry below 90 degrees. For the 90 degree unmitered
corners, it does effect to some degree impedance controlled
traces simply because of the change in trace width area at the 90
degree corner, effectively changing the impedance of your trace
at each 90 degree corner. A 45 degree miter has a much smaller
change in trace width area at the corner and therefore a much
smaller change in impedance at that corner.

There is an old wife's tale in the industry about electrons
flying off the trace when it runs into the 90 degree corner,or
that it radiates wildly (not in normal circuits below several
Ghz). Sort of like a pile of wrecked cars at the bottom of a
cliff where there is a 90 degree corner at the top of a cliff.
Possibly the previous designer offering advice was referring to
this or other variations of this myth over the years.

As Gary said, teardrops are actually for two mechanical
reasons. They decrease the probability that a drill mislocated
towards the trace entering the pad/via will result in a minimum
annular ring violation severely reducing the traces connectivity
to the via/pad. It also reduces the chance of mechanical cracking
of the trace at the point it enters the pad/via due to thermal
shock, solder joint stresses and board loading in high stress
applications. I disagree that teardrops ever had anything to do
with acid traps because angles of 90 degrees were never
considered an acid trap, acute angles less than 90 degrees was an
acid trap.

As for Gary's comments on the state of technological affairs,
I also agree that it is not so much of an issue with most
fabricators these days. However, it has not gone completely the
way of the Dodo bird. Cleaning techniques are much improved these
days and acid traps are nearly a thing of the past with most
fabricators. At least until you get down to acute angles less
than 45 - 30 degrees.

 

[Gary Crowell / VCP]

Guys, guys, guys,
Entering a pad at 90 degrees is not considered an acid trap,
never was. Not even a 90 degree trace corner, it is any acute
angle entry below 90 degrees.

Oh sure, we agree; I mentioned teardropping, meaning for the case of
an acute pad entry. Of course 90 degrees wouldn't be an acid trap or
the DFM check would leave marks all over the place. Come to think of
it, I wonder what it does consider an acid trap, is 89 degrees? I've
never thought to check (CAM350). A pinhole check is a good idea too.

Gary Crowell CID
Micron Technology