On 2017-06-29, Dave Hunt wrote:
I think that you'll find that an active speaker with an AVB input will require more current than can be supplied over an ethernet cable.
In practice that is probably true, but in theory it needn't be. You see I've just been delving into some old time speaker design theory, and would like to flaunt... ;)
Quite a number of variants of Power over Ethernet exist. If we only take a look at the IEEE standardized variants, the minimum seems to be something like 13 Watts of continuous power, after cable dissipation, over the permitted cable length (cf. IEEE 802.3af-2003).
By modern standards that ain't a lot. You'd have to factor in amplifier losses, low sensitivity drivers and whathaveyou. It'd seem that an active speaker driven from such a power source would have to be pretty quiet. Probably it will be as well, using modern designs which tend to exchange power for sensitivity and small size.
However, if the only design margin you have to abide by is that 13W total expenditure of already more or less regulated DC coupled power, you can do much more. First, you put in two separately optimized class-D amps, with peak efficiency in excess of 90% at full power. The first one you use to drive a compression horn at MF/HF, at high voltage. The second one you optimize for (relatively) high currents to an LF dynamic woofer set of extremely wide area, again leading to high radiative resistance, and so high sensitivity.
Going that way, you could in theory mimic many classical studio monitor designs, from before the time when power was ample and to be wasted. Those things were sensitive enough to make your head spin by modern standards, simply because they had to be when driven by the time's vastly underpowered amplifiers; at one time you were supposed to make do with 10 Watts of electrical input power for 100 seats, and they did. Wasting half of it using class-A amps, who knows how much in linear power conversion and regulation, and even then not really going with the highest of sensitivity in the speakers.
Engineering for the limited power conveyed by PoE is or should be a return to the earlier times when available power was the relevant design margin. As such the design should take note of how the high general sensitivity of the time was derived. Borrow from the most successful designs of that time, instead of inventing the wheel all over again, square.
So, at least go with something like one of Klipsch's designs. The RF-7 is reputed to go to a bit over 100dB SPL at one metre, using 1W of continuous power. There are also many other, wider band, more well behaved and sometimes even more sensitive speaker designs out there, from the days of yore. Things which were even considered suitable for studio monitoring use, yet which will make an auditorium full of people walk away at sustained 13 Watts.
The only slight nit with those thingies is their size. Typically, nowadays, you'd expect something fed off an Ethernet socket to be pocket-size. Extremely high sensitivity loudspeakers are anything but. They fill walls. In the high end they require compression driver driven horns which just can't be folded in any realistic way; the MF/HF unit will reach back at least tens of centimetres in its normal application. And then, despite the many attempts at folding a bass horn, that doesn't really work too well either; you'd rather go from the high efficiency of a horn loaded compression driver to a conventional cone of large diameter at the low end.
Or rather a set of them. Big honkin' ones, with coherent radiative area to spare. While it might seem counter-intuitive to use such big cones and so many of them, driven by such little power, actually what you're doing is exchanging extra efficiency in the *least* efficient part of the chain (radiative conversion) into something even *less* linearly manageable by current and even past standards (huge forces and wide throws in bass cones, so as to rapidly compress lots of air in another manner). After you tune your bass cones and their drive magnets somewhat lighter while increasing their area, as such increasing their resistive, radiative loading, suddenly you can drive unreal amounts of rather low frequency noise into the environment even at transistor radio like power outputs. (I'm only exaggerating a bit, here.) Certainly at the 13 watts minus the slim efficiency margin of current class D amps, delivered by a single standardized Ethernet wire carrying PoE (of many) to just one (well-tuned) driver (of many), could basically drive you deaf, blind and insane within a typical home.
...if the thing needed to do that wasn't so big in every way as to be unimplementable by modern standards. The best, most sensitive studio monitors I seem to recall seeing the specs for somewhere were, after all, a metre and a half on both sides, closer to a metre thick, and in addition flush mounted to a purpose built monitor room's front wall. They perhaps just took a Watt of power to drive, and were nigh perfect for their time (could be even better now in their particular application, too), but then considering the extra baggage in excess of the driving Ethernet cable... ;)
Still, there are lessons to be learnt from the earlier times, even with regard to the design of more reasonable PoE driven amps and speakers.
Perhaps for very small speakers with a low power digital amplifier, but anything decent would require mains (or DC batteries) to the speaker.
I'm not too sure it would... ;)
The available current may power the DAC though.
Any competent mixed domain engineer of today can engineer the DAC into the amplifier, or vice versa. They are no longer separate things at the low end of the spectrum, and vice versa, at the high end of the switching spectrum, we can also do a lot of things at the same time which don't really seem like either power or signal conversion.
-- Sampo Syreeni, aka decoy - de...@iki.fi, http://decoy.iki.fi/front +358-40-3255353, 025E D175 ABE5 027C 9494 EEB0 E090 8BA9 0509 85C2 _______________________________________________ Sursound mailing list Sursound@music.vt.edu https://mail.music.vt.edu/mailman/listinfo/sursound - unsubscribe here, edit account or options, view archives and so on.