RE: Power over ethernet

[Emmanuel Mayssat <emayssat@outlook.com>]

> From: nick.rees@diamond.ac.uk
> I would be concerned about running this sort of power over Cat5 because of the current limits. Wikipedia puts the current rating of Cat5 at 0.577A per conductor, so if you use all 4 pairs that is 2.3A, which is right on the limit for 120W with a 48V supply.

> If you factor in the fact that most structured cabling installations are far from ideal, and how stupid you look when everything goes up in smoke because of a cable fire, then being a bit conservative makes sense.

That may explain why it is not a IEEE standard... yet.  Indeed, there may be infrastructure requirements to meet (i.e extremely short CAT6 cable, etc.). To me, what is interesting is that it is already available and commercialized... So asking myself and others 'what if' seems appropriate.

At Lyncean, we are about selling the smallest usable light source possible. At each sale, we rebuild the complete infrastructure. As such cable length, legacy hardware are not much of a concern as it is at large synchrotron. 

That being said, power supplies take a lot of spaces in our chassis... So I am just asking, what if we could 'externalize' power supplies to shrink out total rack space by say 20%. Surprisingly, the answer I was given by the electronics guy was not related to spontaneous combustion, rather it was related to noise propagation from one chassis to another through the switch.

The good news is that in theory, low-power electronics (mostly diagnostics) could potentially be powered through the network.

Of course, in practice that is a pipe dream... but what if?

Ok, back to my routine EPICS programming...