I have a requirement to provide 12v AC at around 800mA at the end of
20m or so of CAT5. Ignoring the issue of voltage drop for now what would be the best way to use the two "spare" pairs to provide the power?One pair for +V the other for -V or split the + and - across each pair? I suspect the latter, common mode rejection etc, but can't really "prove" it in my mind.
On Sun, 11 Apr 2004 21:34:44 +0100 (BST), in uk.d-i-y "Dave Liquorice" strung together this:
I'd split the +ve and -ve over the two spare pairs, basically for balancing.
Umm .. I can't get *my* mind around how you talk about + and - in an AC circuit. I thought you had made a typo and meant 12v DC, but you went on to mention "common mode rejection" and just confused this simple country boy completely ... Well whatever, IMHO, ignoring the volt drop, DC on a pair would be ok, but AC will raise the noise floor on your comms circuit. You'll probably get away with it but it's hardly good practice!
The official way is to use 802.3af Power over Ethernet.
If you Google on 802.3af there are tons of references. However it does use a 48v supply, which may not be what you want.
Here's a DIY way for specific applications.....
.andy
To email, substitute .nospam with .gl
On Sun, 11 Apr 2004 21:47:13 GMT, in uk.d-i-y snipped-for-privacy@microsoft.com (Lurch) strung together this:
I've just noticed the AC bit, don't do it. It's not a good idea sending AC down the same bunch as data signals. Although you could possibly get away with it at 12V it will probably have some sort of unwanted effect. Regulate at one end the un-regulate at the other so you get DC down the cat5.
There's a standard for this now, with switches providing power to things like wireless base stations over their ethernet connection, and products are shipping which use this (we just installed 3 wireless basestations working this way, but don't ask me the details as I didn't install it).
Why not? Think balanced and phase. One wire is positive going when the other is negative.
Stuff induced into a twisted pair is induced more or less equally into each wire, as each wire of balanced pair has opposite phase the induction gets cancelled out.
50Hz worrying 100Mbps ethernet? Strikes me as odd don't they have highpass filters... There could be higher frequency noise present though and spikes for my application I doubt it would be a problem.
From a quick google (thanks Andy) sending 48v AC at up to 350mA (max power 13W all but a smidge) is what 802.3af does...
Yeah but all the add on boxes that I've found only deliver 5v @ 1A and they cost around =A330...
It delivers 48v AC 350mA to the far end. Less a bit for volt drop, the higher supply voltage and lower current reduces that effect for a given power though. All the commercial boxes I've seen change the 48v AC to 5v DC @ 1A.
802.3af also contains stuff to prevent the power being applied until the correct negotiations have taken place, ie the power is interlocked.data. They use one pair for + and one for - rather than splitting +/- across the pairs. The link to loss the calculator was useful that gave .6v loss over 20m, I don't think that is worth worrying about.
On Mon, 12 Apr 2004 00:00:48 +0100 (BST), in uk.d-i-y "Dave Liquorice" strung together this:
Think you'll find that 802.3af is DC actually.
Assuming you're deriving the 12 volts from a transformer, that should take care of the worst of the spikes and HF noise.
I sit corrected, having dug through pages and pages of the spec that spends most of it's time wibbling about the firmware side to impliment the interlocking...
There will not be a +V and -V if it is AC ;-)
Definately I would provide the supply and return down a twisted pair so that the two will cancel each other out. If you were thinking of using two twisted pairs to share the current between them, then balancing the load equally might be a problem. An identical low value of resistor at the feed point of each side of the supply would help, but you still need to ensure both feed and return come down a twisted pair to balance out any interference.
The wire resistance will be enough to balance the load effectively.
Not nice to run 12v AC down signal wires, but definitely do-able.
I suspect that at 800mA over any decent lengtht the v drop will be considerable.
Ethernet transceivers have no response at 50 Hz so it shouldn't be that much of an issue. Agree DC would be a better option though.
See previous answer, a lot of people seem to think of AC as unbalanced like the mains. This is AC from each end of the secondary of a transformer two phases 180 degrees apart.
Thanks thats what I was sort of looking for.
Over the distance I specified 20m (60') and using the calculator found at
This fits with the CAT5 spec having maximum resistance of 9.38 ohms/100m per wire which also fits the resistivity of 24AWG copper wire.
That, to me , is considerable. YMMV.
If it's an AC power supply, it's almost certainly not. It'll be a transformer in the wall-plug, and nothing else.
There will certainly be a bridge rectifier, then either a switched mode or linear power supply on the other end.
If the power supply was as I remember 12V nominal, then there should be at least 1.2V of margin on it, almost certainly more if the transformer is rated at 1A.
HomeOwnersHub website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.