> > Why not run a higher voltage around (12V?) and regulate (switcher?) at
> > each location. This will result in less IR loss because the current
> > will be less, and the regulators will hide a lot of garbage on the
> > supply from the load.
>=20
> My initial concern was less to do with volt drop but more with spikes, in=
rush
> currents and ways to be generally nice to the PSU and the devices connect=
ed
> along the run..
>=20
....
>=20
> The general wisdom seems to be, keep noise to a minimum but that is a bit
> vague.
>=20
> I understand if sensitive analog measurements are required then noise is
> paramount but for general processor operation what is an acceptable noise
> level or W/Hz.
>=20
> I have repaired a bunch of old arcade PCB's from the 80's and on several
> occasions during repair I examined the power rails at various points.  Th=
e
> hash on these rails is huge and often assumed that was the cause of the
> failure but was never the case.  I was regularly amazed that these PCBs
> continued to run flawlessly considering what I saw on the scope.

Noise sensitivity depends on the logic design. If the logic is designed to =
be asynchronous then noise is a greater problem. I remember some discussion=
 many moons ago by some guys who had designed some gear that used 7490 seri=
es counters which are asynchronous internally and had no end of problems wi=
th erratic counting and so on. They redesigned their circuit to use 74191 s=
eries synchronous counters and all the problems magically vanished. It all =
came down to noise during the pre-clock setup times on the waveforms, when =
using synchronous counters all the signals were stable for the required tim=
e before the clock edge.

For most modern microprocessor stuff all the internal logic is going to be =
synchronous, and with the recommended bypass caps on supply lines although =
you will see noise on supply lines that is in time with the internal clock =
it is low enough in amplitude that it is not a problem. Sensible use of low=
 value ceramic capacitors plus a few large value (5-10uf) for bulk energy s=
upply is sufficient.



--=20
http://www.piclist.com/techref/piclist PIC/SX FAQ & list archive
View/change your membership options at
http://mailman.mit.edu/mailman/listinfo/piclist
.