Arfa said:
It can also depend on whether the driving amp is a tube or semiconductor
type. The general rule of thumb is that tube amps don't like open circuits
( or anything in between ) across their outputs whilst semiconductor types
don't like shorts ( or anything in between ). By this, I mean that if a tube
amp is designed to deliver its rated power into say 16 ohms, then it won't
like having 50 ohms across it when you wind the wick up.
That's an interesting set of assertions, considering the fact that
it's not at all unusual for a speaker's impedance to rise to 50 ohms
and above within its operating frequency range.
Likewise, a semiconductor amp ( transistors, ICs or STK hybrids )
won't like 2 ohms across it, if it's designed to run into an 8
ohm load.
While this might be true of some units, as a generalization,
it's not.
So, to run several sets of speakers, each adjustable, from one amp, it is
not a problem to put a wirewound pot in series with each speaker, assuming
that you are running a semiconductor amp. You can calculate the total load
easily by series addition, and ohms law. It's not quite right, as the '
ohmage ' quoted for a speaker, is its impedance at a particular frequency,
not its DC resistance, but near enough.
I would suggest using series resistors as well, to balance up the levels in
each room, and still leave a good adjustment range, as well as making sure
that the whole network cannot drop below the minimum load impedance before
the amp is being overloaded.
Series resistors and series pots may not be the worst way to
accomplish this, but until someone suggests something worse,
this will stand as the likely candidate.
Given that the impednace of the vast majority of speakers is
a frequency dependent function and can easily vary by as much
as a factor of 1 to 10, any substantial series resistance WILL
result in fairly gross frequency response errors. Let's take
your suggestion below and see what happens.
So let's say you are going to feed out to 5 rooms and use some 8 ohm
speakers that you've already got. The quoted minimum load impedance for your
semiconductor amp, is 8 ohms. If you just hook them all up in parallel, the
impedance presented to the amp will be 8/5 or 1.6 ohms - clearly an
overload.
Now change things so that at each speaker, you have a 22 ohm wirewound
resistor, in series with a 22 ohm w/w pot, hooked as a variable resistor, in
series with the 8 ohms of the speaker. With the pot at minimum resistance
( maximum audio ), each speaker will represent 30 ohms. So with all 5 rooms
set like this, the total paralleled load presented to the amp, would be 30/5
or 6 ohms. Unless you're going to play drum and bass at full volume, then by
the time you've added in a bit of cable resistance, your amp is not going to
mind this slight reduction in the minimum presented impedance.
Now turn each speaker to minimum vol ( maximum pot resistance ) Each speaker
will now be 22 ohms + 22 ohms + 8 ohms = 52 ohms. Parallel all these up, and
the impedance presented to the amp will be 52/5 or about 10 ohms.
In such a scenario, each "nominally 8 ohm" speaker will be looking at
44 ohms in series with it. Let's reasonably assume that the speaker's
impedance varies from a low of about 7 ohms in the midband to say 40
ohms at resonance. With 44 ohms in series, the attenuation at the
high point of the impedance will be:
G = 40/(40+44)
= 40/84
= 0.48
which is equal to -6.4 dB. At the minimum impedance of 7 ohms,
the result will be:
G = 7/(7+44)
= 7/51
= 0.14
or about 17.3 dB.
Thus, the scheme suggested will introduce a frequency response error
or nearly 11 dB on such a speaker. Hardly subtle.
Second issue, since all these resistors are in series and thus
the same current must pass through both the speakers and the
series resistors, and since the power dissipated goes as the
resistance times the square of the current, it becomes quickly
apparent that MOST of the amplifier's power will be used to
heat up the series resistors. Let's assume that the amplifier used
can produce 100 watts into a nominal 8 ohm load. Since:
P = E^2/R
thus
E = sqrt(P*R)
then such an amplifier can produce about 28 volts RMS across the
load. By Ohm's law:
E = I * R
thus
I = E/R
so
I = 28/52
or about half an amp. That half amp into the speaker will produce:
P = I^2 R
= 0.5^2 8
= 2 watts
while the series reistor and pot will dissipate:
P = 0.5^2 * 44
= 11 watts
Fully 85% of the amplifier's output will be devoted to heating
up those resistors, only 15% will find its way to the speakers.
now consider the very likely possibility that the speakers used
You would set all this up by setting each pot to about half way, then
setting the amp volume control to get the desired level in the ' loudest '
room.
GIven that the total series impedance can vary from 22 to 44 ohms,
and assume, for the purpose of simplicity, that the speakers DO
represent a resistive 8 ohm load, the minimum and maximum gain
of the proposed arrangement will be:
Gmin = 8/(22+22+8)
= 8/52
= 0.15
= -16.3 dB
Gmax = 8/(22+8)
= 8/30
= 0.27
= -11.5 dB
A range of 4.8 dB, or only +- 2.4 dB from the "midpoint" setting.
That's completely insufficient for compensating for differences
in speakers, differences in room acoustics and positioning,
differences in the amount of level desired or, for that matter,
to make a substantial audible difference.
The best thing is to try it in the garage or wherever first. A few different
values of resistor and pot to play with, bought from your local Radio Shack
store, will be a lot cheaper and easier than getting special pads.
The difference being, of course, that "special pads" have at least
a chance of working.
The
values I've used are just to make the math easy to understand. If you follow
the principle, you should be able to adapt it. 22 ohms is probably a good
starting point though.
Okay, using that math, I think it can be shown that the scheme is
uworkable.
It's hard to calculate a definitive power rating for
the pots and R's because it depends on many factors, but 3 watt wirewound or
cermet pots would probably be ok, with 3 or 5 watt resistors.
Actually, no, as shown above the math is VERY straightforward. To
generalize it even further, the amount of power dissipated in
each serires resistance is directly proportional to the ratio of
that resistance to the total serires resistance. Take that ratio,
multiply it by the total power output of the amplifier, and that's
your power requirement.
Finally, by presenting something other than the design impedance to the
driving amp's output, audio enthusiasts will tell you that you are
compromising the audio quality.
I might suggest that such "audio enthusiast" are unaware of the
fact that almost EVERY speaker on the market presents an impedance
which varies rather substantially from the "design impedance" of
amplifiers. From your assertion above, it follows that almost
every speaker on the market "compromises the audio quality."
In fact, all audio amplifier MUST cope with the fact that the
load impedance is LIKELY to vary widely. And since most amplifiers,
even most tube amplifiers (with some notable pathological
exceptions) behave essentially as low output impedance voltage
sources, this is not an issue.
Whilst this is strictly true if you start
looking at damping factors and other such esoteric quantities, I defy Joe
Average-Listner to hear anything untoward.
I might posit that an 11 dB error on the frequency response is
likely to be detectable by Mr. Average Listener. That's equivalent
to adjusting an equalizer to have an 11 dB boost in the bass, about
a 3 dB boost in the midrange, and a 3-4 dB boost at high frequencies.
Are you asserting that such is NOT audible?
If the system is used for primarily background listeing and not
the utmost in fidelity, then a 70V distribution system with
levels taps on the speaker transformers is the most reasonable
way to achieve moderate quality, adjustability, safety and
reliability with an existing system at moderate cost. However,
the best way, as suggested by another poster, is to distribute
the audio signal at some low level and then use local amplification
at the listening point. The distribution could be low-level analog
signals over appropriately shielded, twisted pair or preferably,
via either multi-drop digital or even via networking. This would
orovide the maximum quality, efficiency and versatility, but
at the highest cost.
