When routing track around a corner:
I have been told that 90 degree routing is not allowed (due to the possibility of acid traps I think).
Idealy it should be 45 degrees (mitered).
I have also been told that 90 degree routing is ok for large tracks
e.g. 50 Thou (thousant of an inch).
Why does acid get trapped in e.g. 20 Thou bends but not in 50 Thou bends ?
(The 90 Degree corner will exist what ever the track width.)
I'm just guessing here, but I presume that if an acid trap causes a
decrease of the track (at the inner corner of the bend) by, say, 2 thou
it will mean that 10% of a 20 thou track is lost while "only" 4% is lost from
a 50 thou track.
For high speed signals, the 90 degree bend looks like a capacitive load attached to the transmission line. Rounding the corner of the bend would leave a constant width, reducing reflection and signal rise-time. It is simpler to chamfering the corner to 45 degree. This would work for frequencies up to 10 GHz. Most layouts does not allow 90 degree bends for this reason.
For high speed signals, the 90 degree bend looks like a capacitive load attached to the transmission line. Rounding the corner of the bend would leave a constant width, reducing reflection and signal rise-time.
The the idea that right angle pcb traces degrade signal integrity is a frequently repeated "rule of thumb" that just isn't correct. See the references below for just a few of the many papers available on the subject:
The acid trap concern is real for some PCB fabricators. For narrow traces, overetching on the inside of the bend eats away a higher percentage of the available copper than for wider traces (i.e. .5mil from a 6mil trace is more significant than .5mil from a 20mil trace). How much is really overetched depends on the equipment, process, and skill of the fabricator.