At 10:51 AM 8/15/2002 -0500, Joel Hammer wrote:

>I have always been under the belief that if you make a part of a trace/track
>thin you might as well make the *entire* trace/track that width. (holding
>even more true in power & ground runs. in which case the thin run could
>almost act as a fuse.?) If anyone can share logic otherwise I would really
>be interested in hearing it. Or am I safe in what I think to be true?

I don't think that the belief is true. There are lots of reasons to thin 
down a track where necessary and leave the remainder of the track wider. It 
will reduce resistance where that is important. It may improve 
manufacturability (over making the entire length thin). Where it is 
important, it will keep the impedance of the track closer to the desired 
value, but see (2) below.

The concept stated would only be true under two conditions, as far as I can 

(1) The current in the track is high enough that fusing becomes an issue. 
Obviously, one would not thin such a track down at all, unless the thinner 
traces can be paralleled, in which case two thin tracks may be better than 
one track of double the width (because the resistance and thus the 
dissipation would be the same, but they would cool more efficiently).

(2) The thinning of the track is in a configuration that will cause 
significant impedance discontinuities. If this is the case, it is not that 
one "might as well make the entire trace that width" but it is much better 
to make the entire trace a single width, but see below.

A short neckdown in a controlled impedance track will not produce a 
significant impedance discontinuity if the neckdown is short enough for the 
risetime/frequency involved. How short? Well, I've simply made such tracks 
as short as possible, just enough to squeeze between pads. Perhaps someone 
with more specific high-speed knowledge can tell us, probably in terms of 
fraction of effective wavelength or fraction of signal risetime expressed 
as length.

* Tracking #: CAF8C4577485DF458334728ED4948FE1263F8CF1

* * * * * * * * * * * * * * * * * * * * * * * * * * * * * *
* To post a message: mailto:[EMAIL PROTECTED]
* To leave this list visit:
* Contact the list manager:
* Forum Guidelines Rules:
* Browse or Search previous postings:
* * * * * * * * * * * * * * * * * * * * * * * * * * * * * *

Reply via email to