Time coherence - how important and what speakers?

I think we need to remember that active speakers don't necessarily need to be self contained. One could even argue that isolating cross-overs, amps and drivers have advantages. Even in a self contained active speaker, the various components could be modular in design, so that one doesn't have throw the baby out with the bath water.

Cdc/Unsound you are too smart for your own good...

Cdc asks:
- Is time / phase coherence maintained through the whole recording / playback process so the speaker is the only thing messing up phase and time coherence?
The answer is no, but... ALL gear and speakers and mics add time delays at the high and low ends of their operating bandwidths. Those are (fortunately) gradual changes in the phase vs. frequency. ONLY speakers (and certain microphones) will jerk the phase around abruptly in the middle of the audible band.

- What if a driver naturally rolls off greater than 1st order. Does the crossover have to boost the driver? Is this bad?
Any driver is rolling off because it has a "mechanical crossover", which is usually > 2nd-order. It rolls off because it has moving mass and compliance. The compliance is the flexibility of the hinge-points on the cone as it breaks up. The moving mass is varying with frequency- the outer portion of the cone coming to a standstill.

With less effective moving mass for that voice coil to move, the frequency response peaks before it rolls off- you can see how the wiggles in a raw-driver's impedance curve correspond to the breakup modes, and to peaks and dips in the frequency response.

Cone breakup is a "dirty" or "ringing" sound at worst. Breakup modes are always excited by the fundamentals down at 1/2, 1/3, 1/4, 1/5... of each breakup frequency. A driver should not naturally roll off `till well beyond the actual crossover point. Any breakup modes should be well-damped so they don't ring.

Cone breakup can be disguised with very high damping- such as with the woven Kevlar and carbon fiber cones, and the mineral-filled poly cones. It keeps those drivers' frequency responses smooth, but crushes the dynamic response as the cone structure is absorbing the energy, instead of moving the air. So why use those cones?
a) They don't have the "peakiness" of poorly-designed paper cones.
b) They look hi-tech.
c) Making proper wood-fiber cones is a dying art.
d) Plastic cones are good for sales in tropical environments and cars.
So we lose dynamics, and inter-transient silence, and detail- who cares?

- Would the extra power handling of the voice coils in 1st order x-over degrade performance in other ways like higher inductance, mass, and hysteresis?
Inductance- nothing we can't adapt to.
Moving mass- higher means less efficiency (see my post today on the High Efficiency thread) and eventually more power compression.
Hysteresis- depends on what you're talking about. The longer voice coil does overhang into more of the fringe-magnetic field created by a poorly designed magnet structure. There are others, such as inadvertently changing the center of gravity of the cone/voice coil assembly- which will lead to unwanted rocking motions, which puts really strange eddy currents back into the pole piece, which...

Unsound,
Your observation that "active speakers don't necessarily need to be self contained" is true. It would have all the advantages you list.

But if a speaker manufacturer merely supplied or recommended a certain electronic crossover, then the customer must purchase several "matched sound quality" amplifier channels and interconnect cables and speaker wires. And those are very strong points for putting all that stuff back into the speaker cabinets.

Either way, my main problem: whose gear do we choose- and will they let us? I've spent a lifetime working with speakers, not amplifiers- the only amp I could design would be from a cookbook recipe, or I'd have to mimic one already out there. And we'd have to service it! Egads..

Thanks for the insightful comments and questions. I wish the press would talk about this stuff. Why do you suppose that is? (serious question for another thread)

Best regards,
Roy

Revel literature is interesting:
"The crossover networks . . . maintain a 24db per octave, 4th order acoustic response . . . the steep filter slopes ensure good acoustical behavior in the crossover regions, with a minimum of acoustical interference, along with low distortion and wide dynamic range.
The somewhat steep 24dB per octave slopes also provide the benefits of keeping ALL DRIVERS IN PHASE AT THE CROSSOVER POINTS - A BENEFIT SOUGHT IN THE MORE COMMONLY USED 6dB per octave crossover designs from other companies. The steeper fourth order slopes, however, avoid the power handling problems associated with first order crossover networks."

Between Revels claim of maintaining phase coherence with 4th order x-over and comments made above about bad in-room frequency response (some Revel is very good at) with slow roll-off x-over design 4th order sounds very convincing.

I believe SPicas used a 1st order/4th order x-over configuration in their tc-50...with the 4th order on the mid-bass driver...designer John B. claimed through computer simulation tests that the 4th order had virtually no phase anomalies...I do know this...in terms of 3-d imaging...the SPicas are still one of the best...I also believe B&W uses a 4th order design in their higher end models...

Cdc, it is time to dispel the myth hidden inside their statement, as many other firms say exactly the same thing.

You say Revel states, "The crossover networks . . . maintain a 24db per octave, 4th order acoustic response..."

FYI: Acoustic response- this is the actual frequency vs. amplitude response a mic would measure, freefield at one meter on swept sine wave tones (no floor bounce). This is the right way to describe any crossover- by its final acoustic response (assuming you even listen at one meter... which is a whole other problem).

". . . the steep filter slopes ensure good acoustical behavior in the crossover regions, with a minimum of acoustical interference,.."

This is true -no question- when measured by swept sine waves or pink noise, w/o the floor bounce. FYI: "good acoustical behaviour" means there are major no peaks or dips in the frequency vs. amplitude response on swept tones or pink noise.

"along with low distortion and wide dynamic range."

Yes, because 4th-order rolloffs keep the drivers protected really well from low frequencies- the upper-range cones/domes don't even wiggle on a bass drum.

"The somewhat steep 24dB per octave slopes also provide the benefits of keeping ALL DRIVERS IN PHASE AT THE CROSSOVER POINTS.."

Yes they are in phase, which is a benefit, but THEY DID NOT START AT THE SAME MOMENT, NOR WILL THEY STOP AT THE SAME MOMENT. THEY ARE NOT TIME COHERENT.

With a tone generator, feed that woofer and mid a steady sine wave at the frequency where the crossover occurs. Put a mic out front- look at the combined single sine wave from the woofer and mid on a `scope. You cannot see the beginning or end of this steady tone- it's just a sine wave going up and down.

Then unhook the mid- look at just the woofer. Then look at just the mid. THEIR WAVE'S PEAKS AND VALLEYS LINE UP- they are in phase. But if you could see the beginning or the end, one starts A FULL CYCLE LATER THAN THE OTHER. And when they stop, ONE STOPS A FULL CYCLE LATER THAN THE OTHER. You could even say their combined output rings at that crossover frequency.

ALL of that is in any second or third-year electrical engineering "Filter Theory" book in plain English- that is the behaviour of the 4th-order filter. In phase, yes, but 360 degrees out of step.

Can you hear this? Yes. Just listen to a speaker (or headphone) without that time delay. How much time delay was imposed? Exactly one full period of the crossover frequency. If that was 400Hz, then the time delay between the two drivers is 1/400th second, or 2.5 milliseconds, which is ~32 inches for time of travel, acoustically.

You just smeared the guitar spatially by ~32 inches, front-to-rear, and transiently by 2.5ms, across its 400Hz range (just above middle C). It sounds like the upper strings of the guitar are "leading" the lower strings, or that there is more pick on the string. It also changes the wave envelope, which means a loss of clarity. And all that means it changes the musical message.

But how would you know unless you've heard the real thing? All you really know is that there are many "poor recordings" you can't play, can't enjoy. Many performances "you just don't get". Why? Because that phase distortion is distorting everything that comes into it- including any recorded distortions- so you hear distorted distortion- which is multiplicative, not additive.

You are being presented a speaker that warps not only the image of the guitar, and its dynamic attack, but mis-aligns the individual tones that go into shaping its harmonic envelope, which is its timbre- the very nature of why it sounds like a guitar.

"...The steeper fourth order slopes, however, avoid the power handling problems associated with first order crossover networks."

Yes, if you use less than the best drivers.

Cdc, you remark, "and comments made above about bad in-room frequency response (some Revel is very good at) with slow roll-off x-over design 4th order sounds very convincing."

I am not sure what you mean by "(some Revel is very good at)", but all you have to do is listen to end the debate. Music waveforms have little to do with the math Revel and others use, math based on sine-wave arguments, "proven" by sine wave measurements.

The standard response to all that I've stated above about the drivers not being in sync is, "Well, you can't hear the phase errors, anyway. There are tests that prove that."

The tests don't prove that- they are ambiguous at best, because they were based on clicks and other unfamliar sounds.

If a designer has never built a speaker that is minimum phase (the standard term for any time-coherent system), he has not even tried to hear the difference.

A 4th-order acoustic rolloff is used because:
--the designer believes the sine-wave arguments,
--it allows high-tech-looking metal cones to be used, as most all of those have a severe (>10dB) peak at the cone breakup frequency, along with 5-10% distortion caused by that peak. Look at the SEAS and others' metal-cone driver frequency responses- readily available. The 4th-order rolloff chops them off.
--a 4th-order rolloff lets the designer lower costs by using less rugged drivers.
--a 4th-order rolloff is often computer-aided in its design. Sounds high-tech, and good for advertising.
--the designer doesn't have to learn time-domain math, which is quite difficult.

All the math of acoustics and physics and music supports my claims, fully. What's known about how the ear uses the time domain for image formation and timbre retrieval supports my assertations. The actual times and timing of music's transients support them. The method behind how any timbre-containing wave envelope forms as time evolves, supports them.

As far as off-axis interference? 1st-order drivers overlap a lot. So those drivers must be really good- and it's hard to find "the best". However, they are overlapping TIME COHERENTLY, `most everywhere in the horizontal plane, so there is no interference or lobing. Period.

If you stand up- yes, they start to go out of phase by many degrees. But the 4th-order circuits are already HUNDREDS of degrees out of phase, no matter where you stand or sit.

Some would claim this time-coherence stuff is not a question of right or wrong, but a matter of taste. I would remind you that keeping the speaker a minimum-phase design (not possible w/4th-order) is the ONLY way to preserve the actual shape of the musical waveform received by the mic. For inside that "shape", that irregular, non-sine wave envelope, lay all the tones of the music, all the dynamic changes, all the musicality, and all of the musical message. Why disturb it?

I welcome any attempt to refute my points on any grounds.

Most of all- just listen. Non-time coherent speakers sound like a wall of sound -no depth- as the time domain is scrambled. Time is distance. Time delay is depth. Time is transients coming and going, and tones building then decaying. Scramble the timing between bass and treble- you lose the depth. You lose the timbre. You lose the musical intent. You lose access to many, many recordings.

Time coherent behaviour can be heard in any decent headphone- even a $30 Walkman headphone, let alone the Grado or Stax. Just listen, especially to music that has energy near the crossover frequency.

Thanks. I don't want to be the constant "Answerer" here, but I really hope this helps.

Roy