crossover phase coherence question

[timbley]

I've been reading about crossovers a bit, and I'm confused by a lot of what I'm reading. From what I gather, a properly designed two way crossover will be phase coherent across the audio spectrum. But a 3 way can mess things up. Is this true?

[Al Klappenberger]

Timbley,

Yes, it's true but it means nothing! The only thing that matters is the total phase plus time delay right where the two drivers are generating the same sound. If they happen to be out of phase, you get a frequency response dip. If the crossover area has a good response, forget it. The human ear is totally deaf to the phase of components of a complex waveform, that is, the harmonics in music.

Al K.

[sunnysal]

of course this is a subject that generates some disagreement, not everyone agrees that people are "phase deaf"...tony

[Al Klappenberger]

Tony,

This is true. I only know for a fact the I am phase deaf. I did an experiment using 3 seperate sine wave signal generators to simulate a square wave where none of the harmonics were phase-locked. The phase was random and continuously changing, not just phase locked to the wrong algnes like happend in a crossover. I could hear no difference at all as the phase changed to spite visual clues looking at the waveform on an oscilloscope. It was a continuous raspy tone. What you CAN hear though is interferance between drivers casued by two sound sources making the same sound at the same time.

Al K.

[timbley]

Al K,

I was wondering about this because it was explained to me that my digital crossover and EQ emulate analog designes, with all of their phase shift issues included. More expensive digital crossovers and EQs are "phase coherent." Some people were making a big deal out of this, and I just started to wonder what I might be missing out on.

[Al Klappenberger]

Guys,

I really believe that phase is something your brain simply can't deal with. I am 100% cnvinced that I can't her it. I would like more people to try the 3 signal generator experiment, but I realize not to many people have 3 seperate signal generators lying around! I don't have any reason to beleive that my inability to hear phase distortion is a personal malidy, but the only way to know for sure is for others to try the experiment.

Al K.

[timbley]

I experimented a little with shifting the phase on the tweeter using the Behringer. I could hear a little happen when listening to white noise, but very little if nothing when listening to music. I think the cancellation effects are minamal because the crossover slopes I'm using are pretty steep.

But, that test was when I was using lots of EQ. Last night I tried listening with no EQ, just a pair of 10ohm resistors in series with the tweeters to bring them down to about the same level as the midrange, and hopefully flatten out their response a bit, (although that didn't appear to happen much looking at the RTA, they still roll off fairly sharply above about 7K.) The un-EQ'd balance was suprisingly listenable despite the response not looking all that flat. I moved the crossover point down a bit to a point with a smooth looking transition. The sound was very, very clear and non fatiguing, with a prominent midrange, giving vocals a lot of power. I listened until 1:00a.m., and wanted to crank it way up. Didn't want to go to bed. While listening, I tried to apply a litte EQ here and there, and I got the distinct impression that the clarity was reduced each time, somehow the sound had become a little confused. Switching the polarity on the tweeter wasn't good. Hmm, I'll have to listen some more.

[WMcD]

I've read the same thing: It is possible to get some two-way to add properly, in theoretical perfection. Three-ways have issues which make them more difficult. (Constant power versus constant voltage?)

There are very many big "but"-s in any theoretical analysis of either. It is really just saying that the electrical output of the crossover does this or that. Call that the transfer function between the input to the filters and the split outputs would work into resistors.

The problem is that there is another transfer issue between the electrical input to the driver and its output. Any given driver has its own roll off of output and phase issues. This even if you use active filters. If you use passive filters their behaviour is effected by the varying electrical impedance of the drivers.

Any difference in position of the drivers messes things up more.

People working with only the theories address some of the issues by the use of Zobels (to adjust impedance) and offset of the drivers. Then there is a lot of listening and fooling with values.

The very sophisticated testing and design software can experiment with the value of the components in the filters and get things working much better. LMS and LEAP. This is based on careful measurments and simulations. I expect the new generation crossover in the K-Horn etc, were designed that way.

I recall reading some comments by one of the theoretical people that once they have gotten to use the newer analysis and simulation tools, they realize just how poor their assumptions and solutions were in the real world.

I had built a two-way crossover for a two-way horn. Theory said that the polarity of the HF unit should be inverted to get a good phase match. This is not uncommon. At the crossover freq, one output is 90 degrees minus and the other 90 degrees plus.

Acoustic testing showed that to be basically correct. Without polarity reversal, there is a dip, albeit a narrow one.

I decided to live with the dip. I just couldn't live with the thought that the HF (or the LF for that matter) over its entire working frequency was going to be reverse polarity. Yes, theory says you can't hear polarity except on some very special test signals. But I can do whatever I darn well please. Smile.

My conclusion is that all the worry about theory is for naught if anyone believes it is perfect. There is a lot to be adjusted away from theory to get better real world results.

Best,

Gil

[timbley]

Thanks for all of your responses guys. I'm learning about a lot of stuff.

I must admit, I know little about crossovers. I've heard of a Zobel before, but don't know what it is. I understand caps impede low frequencies and coils high, resistors everything.

I'm starting to think that an emphasis in a certain frequency range can make the sound clearer. When EQ is applied to smooth out the response, the apparent clarity is reduced at first listen. It sounds more confused because you can hear more, not because the phase is messed with. That's what Ethan Whiner says.

Looking at the RF-7 response with active crossovers, it really rolls off quite a bit in the bass and treble. The passive crossover accomplish a lot of EQ.