Why you need 500Wpc for good sound

[fenderbender]

^^^^

LOLOLOLOLOLOLOL[][][][][][][][][][][][][]

[DrWho]

I also agree that bi-amping reduces the strain on the amp.

Could
it also be that bi-amping greatly reduces harmonic distortion, since
the treble amp cannot be getting any harmonics of tones below the
crossover point? If I understand harmonic distortion correctly, a
100Hz tone has a first harmonic of 200Hz, a second harmonic of 400Hz, a
third harmonic of 800Hz, and so on. If an amp's HD is "heavily
third-harmonic in nature", as I've read in a number of reviews, I take
that to mean that spurious harmonics from the low end will be generated
well into the midrange.

If, for example, the crossover point in
a 2-way system is 500Hz, the highest third-harmonic frequency generated
by tones below that would be 4000Hz. If the system is bi-amped, the
treble amp won't have any of those harmonics, since they'd be generated
by the bass amp and its output is cut off above 500Hz. This implies
that there would be no third-harmonics between 500Hz and 4000Hz,
meaning the midrange would sound cleaner.

Am I understanding
this correctly? I do notice that a bi-amped system does sound more
clear, but I'm not sure of all the factors that give it that clarity.

I
think the way distortion is defined, 200Hz is the second harmonic of
100Hz - so 300Hz would be the third harmonic. In fact, let me find a link:

http://en.wikipedia.org/wiki/Harmonic

So in your example, a perfect 500Hz high pass filter would prevent any 3rd harmonics between 500Hz to 2kHz (so two octaves).

However, the problem is a bit more complicated than that...

With
a perfect 500Hz active low pass filter to the bass bin, you have no way
to filter out higher harmonics...so you will end up with 3rd order
distortion from your bass bin in the 500Hz to 2kHz region. So in a
system setting, with two speakers that cover the entire audible range,
you achieve no decrease in THD.

However, one of the nice things
about horns is that they can act like acoustic lowpass filters...so if
your horn's transfer function drops like a rock above 500Hz, then you
will achieve no 3rd order harmonics from 500Hz to 2kHz. It doesn't drop
like a rock in real life, but any attenuation should reduce any
amplifier or speaker caused distortions.

I think this is a good reason why you want your speaker to have a well behaved frequency response, even above the passband. Any distortion
introduced by the amplifier or driver is going to be augmented by the
natural frequency response of the speaker and horn. You can put EQ
before the amp so that the source material comes out with a flat
response, but anything the amp or driver does weird doesn't benefit
from the EQ.

So let's say you boost the LF by 10dB to extend the
F3, you're always going to see an extra 10dB of distortion at all those
frequencies. Let's also assume that your speaker has a rather wide
additional +10dB in its midband that normally gets EQ'd out in the
crossover...you're now going to see +20dB distortion at harmonic
distortions that fall within that passband (yikes). I think the bigger
issue here is actually the speaker distortions and not necessarily
those imparted by the amp, but it's effectively the same problem in
both cases.

I bring this up because I've noticed that a lot of
horns seem to have a round shaped frequency response that is higher in
the middle, and attenuated at the extremes. HF attenuation caused by a
horn is actually going to decrease harmonic distortions (one of the
nice things about CD horns...the downside to CD horns is the increase
in cone excursion, and thus greater IMD). So really, the bigger issue
is the decaying LF response, which we see in most everything Klipsch
(Khorn, Jubilee, P-39f...just to name the flagships).

Btw, I'm not entirely sure that this theory about distortion is
true, but I think it kinda correlates to a lot of the listening I've
done. The nice thing about a good system is that the distortion should
be extremely low in the first place, so hopefully this doesn't come
across as saying horns are bad...I just think that the subjective world
might be interested in ways to make a horn tailor the distortion
response to something more pleasing...

To come full circle, decreasing the number of frequencies that need
to be reproduced by the amp and driver will always reduce the amount of
IMD.
Is it a 3dB or 6dB reduction for every halving of frequencies? This
will be very dependant on the source material though (since the
spectral content is not at all equally distributed). So in the above
example with a perfect 500Hz xover in a perfect system, you will see
some guaranteed decrease in distortion.

[DrWho]

RE: the fantasy power charts...

I think those charts portray
the necessity for the utmost in efficiency from our speakers. Although
there is no scale on those charts, I think it would be interesting to
compare against the acoustic powers of real sounds....even with 110dB
speakers, I'm fairly certain that there are a lot of naturally occuring
sounds that can't be reproduced without compression.

Also, even though speakers are going to be the limiting factor in
dynamic accuracy, it doesn't justify an amp that makes it even
worse...though sometimes I wonder if people aren't just trying to undo
the dynamics their horns offer - which might actually make more sense
in light of a lot of the studio equipment being used. The advantage
will be less IMD for the same levels of THD (and if your amp soft
clips, that THD will be more 2nd order).

[jacksonbart]

One of my favorite songs is Traffic's Dear Mr. Fantasy.

Those used Peaches are selling like cold cakes

[seti]

I have been asking others about SET power and the use of efficient speakers. Another friend and I have been discussing the same things as we are in this thread. It has been fun. Paul Joppa of bottlehead fame had the following response to me. I am still researching both sides of this as there seems to be more to these ideas.

*****
Typical "peak VU-averaged" (a fairly slow-response average level)
listening level is about 82dB +/-2dB for recording engineers.
Instantaneous peak to "peak VU-averaged" is 14dB for well-recorded
material - this is about the most that can be obtained from a master
tape. The only case where a higher peak to average ratio shows up is in
action movies where car crashes and other special effects may be
louder. Therefor you need a minimum of 96dB peak capability.

In
an average home listening room the 1w/1m sensitivity of the speaker is
vary nearly equal to the room-average reverberant field.

Many
audiophiles add 6dB to this requirement. This makes sense for
high-feedback solid state electronics, which become much less linear
near their peak output. The negative feedback reduces the measured
distortion at these levels, but it pumps up the higher harmonics and
the effect is audible to many careful listeners. Some audiophiles
really want to listen at louder levels than the professionals will use
- I don't know why, but then I am not one of them.

Anyhow, even
with this 6DB increase which I maintain is not really needed for
low-feedback tube amps, you need 102dB maximum. With your 104dB
speakers, that is 0.63 watts. Assuming a typical 300B amp, which might
have 5% distortion (mostly 2nd harmonic) at 8 watts, the distortion at
0.63 watts would be about 1.4%. The limit of audibility for
second-harmonic distortion is about 3%, so the distortion is inaudible.

Of
course, if you want to reproduce a rock concert, or are deaf, or want
to become deaf, you may want to listen at much higher levels. At that
point you must determine your own loudness needs.

*******

[ClaudeJ1]

Of
course, if you want to reproduce a rock concert, or are deaf, or want
to become deaf, you may want to listen at much higher levels. At that
point you must determine your own loudness needs.

OSHA limits on sound levels for an 8-hour workday exposure is 85 db, otherwise hearing protection must be used.

Since my setup now give me more micro detail than ever before (see Avatar), I find myself listening ot about 80-82 db in the sweet spot with my RS meter set on C weighting and Slow response setting. With 108 db efficiency (a padded down 112db/W on my EV horns) I think I measured about 0.10 watts a while back at a much higher listening level.

[DrWho]

Instantaneous peak to "peak VU-averaged" is 14dB for well-recorded
material - this is about the most that can be obtained from a master
tape. The only case where a higher peak to average ratio shows up is in
action movies where car crashes and other special effects may be
louder.

I
can agree with his 82dB number (I would have said 85dB +-2dB, so we're
close), but that 14dB is not conservative at all. One of my favorite
songs from Nightwish (track 5 on the Once album) has a 60dB difference
between the rain sticks at the beginning and the tribal drum hit that
comes in after a few seconds. Blue Man Group (another favorite of mine)
is consistently over 40dB on a lot of the transients (and not just the
percussive sounds either). Massive Attack is usually over 35dB in a lot
of the songs I've bothered to measure.

I know that's only a dozen
or so examples from only 3 bands, but I would really like to know where
that 14dB number came from. Heck, spoken word is usually about 14dB
with an exciting speaker...maybe closer to 6-8dB if it's for the radio.
Music is a whole heck of a lot more dynamic than speech []

This makes sense for
high-feedback solid state electronics, which become much less linear
near their peak output. The negative feedback reduces the measured
distortion at these levels, but it pumps up the higher harmonics and
the effect is audible to many careful listeners.

Is
he comparing clipping behavior or non-clipping behavior? Here's a tip
for him....don't clip the amp, sheesh. Talk about misapplication of
theory (or intentionally misleading).

With your 104dB
speakers, that is 0.63 watts. Assuming a typical 300B amp, which might
have 5% distortion (mostly 2nd harmonic) at 8 watts, the distortion at
0.63 watts would be about 1.4%. The limit of audibility for
second-harmonic distortion is about 3%, so the distortion is inaudible.

Don't forget the speaker distortion, and the non-linear frequency response from the mismatched source impedance...

I also don't agree with the 3% distortion claim either. We were
blindly identifying way smaller distortion changes (like on the order of 0.1%) in the lab last
spring, and it was mostly 2nd harmonic (when not clipping of course). Dunno where the speaker distortion was at though...

[seti]

Instantaneous peak to "peak VU-averaged" is 14dB for well-recorded
material - this is about the most that can be obtained from a master
tape. The only case where a higher peak to average ratio shows up is in
action movies where car crashes and other special effects may be
louder.

I
can agree with his 82dB number (I would have said 85dB +-2dB, so we're
close), but that 14dB is not conservative at all. One of my favorite
songs from Nightwish (track 5 on the Once album) has a 60dB difference
between the rain sticks at the beginning and the tribal drum hit that
comes in after a few seconds. Blue Man Group (another favorite of mine)
is consistently over 40dB on a lot of the transients (and not just the
percussive sounds either). Massive Attack is usually over 35dB in a lot
of the songs I've bothered to measure.

I know that's only a dozen
or so examples from only 3 bands, but I would really like to know where
that 14dB number came from. Heck, spoken word is usually about 14dB
with an exciting speaker...maybe closer to 6-8dB if it's for the radio.
Music is a whole heck of a lot more dynamic than speech

This makes sense for
high-feedback solid state electronics, which become much less linear
near their peak output. The negative feedback reduces the measured
distortion at these levels, but it pumps up the higher harmonics and
the effect is audible to many careful listeners.

Is
he comparing clipping behavior or non-clipping behavior? Here's a tip
for him....don't clip the amp, sheesh. Talk about misapplication of
theory (or intentionally misleading).

With your 104dB
speakers, that is 0.63 watts. Assuming a typical 300B amp, which might
have 5% distortion (mostly 2nd harmonic) at 8 watts, the distortion at
0.63 watts would be about 1.4%. The limit of audibility for
second-harmonic distortion is about 3%, so the distortion is inaudible.

Don't forget the speaker distortion, and the non-linear frequency response from the mismatched source impedance...

I also don't agree with the 3% distortion claim either. We were
blindly identifying way smaller distortion changes (like on the order of 0.1%) in the lab last
spring, and it was mostly 2nd harmonic (when not clipping of course). Dunno where the speaker distortion was at though...

Unfortunately I do not have the tech savy to back up what I re-post which is a hazard of posting comments from others. Ultimately I am searching to the answer to why 8 watts kicks the crap out of everything I have heard to date. I can see both sides but once again I gotta go with my ears. Joppa has designed and listened to more SET systems than I have so I defer to his knowledge on this subject. I should quit discussing and trust my ears.

[mikebse2a3]

Hey guys for another perspective on amplifier design look at Nelson Pass's article I posted in technical. Especially as it pertains to IM distortion and feedback use in amplifiers identified in his research.

mike tn

[mikebse2a3]

Unfortunately I do not have the tech savy to back up what I re-post which is a hazard of posting comments from others. Ultimately I am searching to the answer to why 8 watts kicks the crap out of everything I have heard to date. I can see both sides but once again I gotta go with my ears. Joppa has designed and listened to more SET systems than I have so I defer to his knowledge on this subject. I should quit discussing and trust my ears.

Hey seti

Your ears might be on to something![] SET isn't perfect but neither is any other amplifier design. It's just important to know what each one's weaknesses and strengths are when using them>

mike tn[]