Has anybody listened to the effect of xovers vs no xover?

[makingmoney]

Yes crossovers are needed and I use them. (This ain't the single driver forum)

Wanting to hear what xovers did to the music I came up with this totally subjective test:

I did a limited listening test over 10 years ago to a 7" Focal mid driver, not a great or bad driver.

I choose 2 pieces of music to listen to, a solo woman vocal and and a solo instrument (can't remember which one). Their sounds were mostly in the linear region of the driver. Some fundamentals or upper harmonics were attenuated.

I listened to no xover, 1st order, 2nd order, 3rd order and 4th order high pass xovers all at about the same xover point. Using decent caps, inductors etc.

No xover was the best sounding. 1st order was still good. 2nd order was starting to sound processed.

3rd and 4th were very processed sounding, almost weird sounding and not natural at all. Very hifi sounding.

Has any one tried a similar listening session? What were your results?
What were your solutions?


Does the DEQX let the xovered music, no matter what the slope, come out as if there was no xover? Does the even cheaper DCX?

[doug20]

What was the frequency range of the content?

I have done many listening tests with my DCX and many drivers...setting slopes definitely changes when only listening to one driver. Of course it should start to sound unnatural when listening to just one driver because we are creating a bandpass and cutting off frequencies with different slopes.

IMO, its hard to know how any XO sounds on its own. Its also very misleading to make any conclusions based on single driver testing.

I do like the DCX for doing these tests and other XO tests though, nothing is quicker in terms of prototyping and experimenting.

[Davey]

I assume the OP's original experiment was using two, identical Focal 7" drivers?

For what's it worth: The DCX2496 can be used to generate an all-pass filter function that would have the equivalent phase-distortion of 2nd/4th/8th order crossovers but with a flat frequency response.

You could also accomplish the same thing with an analog active crossover (a constant-voltage type) by summing the hi/lo outputs together.

These would at least isolate the audibility (if there is any) of phase distortion for a particular "crossover."

Most any other scenario has too many variables to conclude anything about the audible effect of xover vs no xover.

Cheers,

Dave.

[makingmoney]

Originally Posted by doug20

What was the frequency range of the content?

I have done many listening tests with my DCX and many drivers...setting slopes definitely changes when only listening to one driver. Of course it should start to sound unnatural when listening to just one driver because we are creating a bandpass and cutting off frequencies with different slopes.

IMO, its hard to know how any XO sounds on its own. Its also very misleading to make any conclusions based on single driver testing.

I do like the DCX for doing these tests and other XO tests though, nothing is quicker in terms of prototyping and experimenting.

I don't recall the FR of the music except to say it was well in the pass band of all xovers. The xovers were somewhere below 100hz. I tried to keep them low so none of the higher solo music would be attenuated.

I am familiar with the sounds of cutting off frequencies with slopes and my little test didn't suffer from that. Just my subjective ears.

[doug20]

Originally Posted by Davey

I assume the OP's original experiment was using two, identical Focal 7" drivers?

For what's it worth: The DCX2496 can be used to generate an all-pass filter function that would have the equivalent phase-distortion of 2nd/4th/8th order crossovers but with a flat frequency response.

What would the DCX settings be for that?

[prickears]

Makingmoney, What your hearing may be the electrical resonance of the XO. The same experiment using active XO will more than likely yield quite different results.

Some time ago, I was trying to get a passive XO right, 4th order HP, 2700hz. Damn thing rang like a bell. Same XO point, same slope, but active, no such problem.

[Jay]

Originally Posted by makingmoney

No xover was the best sounding. 1st order was still good. 2nd order was starting to sound processed. 3rd and 4th were very processed sounding, almost weird sounding and not natural at all.

Has any one tried a similar listening session? What were your results?
What were your solutions?

Does the DEQX let the xovered music, no matter what the slope, come out as if there was no xover? Does the even cheaper DCX?

That's not subjective at all. I think it is well known by all speaker practitioners.

The most expensive speakers (read: Sonus Faber) uses very simple crossovers. They custom order the driver so that it can be used with say first order passive crossover.

Of course you cannot send full-frequency music to ANY driver with first order filter or worse without filter at all. Finding the balance between compromises (and know the compromises), that's what a designer should be able to do well.

The best implementation for true hi-end system is I believe with active analog crossover. The problem with analog crossover is relatively easy and cheap to solve than with digital crossover.

There are two critical links in audio system: D/A conversion and speaker reproduction. With digital crossover (especially the CHEAP one), you add problem to the signal, so even if the speaker itself doesn't have the problem, it will still produce problematic signal. What is the difference then? Why don't we just solve the problem with the driver to an acceptable degree?

FYI, drivers interaction with series/parallel coils/caps is unique. You can create (almost) SIMILAR slopes with different component arrangements (for example a 3rd order high pass, you can use bigger L and smaller C or you can use smaller L and bigger C) and the result is very different!

[norman bates]

Well, that is a complicated question to answer.

Full range drivers are very good, but lack in other areas.
I had a pair of old thiels 2 ways (6db slanted) but they wern't impressive and I feel my dual 4" sounded better (less haze).

I have an active crossover (rolls) that sounded horrible when I compared running a pioneer b20 (tweaked) wide open versus through the crossover rolled 24db @ 200hz. The opamps are questionable, there are loads of junky electrolytics in the signal path, and the high pass comes off the low path (lots of feedback and noise from low pass fed directly into high pass).

The marchand passive crossovers are good but you need a stout preamp to drive them.
The marchand active crossovers should be nice but they arn't cheap.

Jay is very correct. You design a crossover based on what you want the system to do.

I flip flop between a pair of 4" run wide open with bass support, the pair of 4" crossed at 400hz to a pair of 15's, or a pair of 15's crossed @ 750hz to a large horn/compression driver. The 4" run wide open sound best but even Blues traveller track 11 from "four" makes them sound strained. The horn gives awesome dynamics, slam, and detail (the most fun) but replace the horn with the 4" pair and slide the crossover point down to 400hz, this aids intelligibility but you lose detail that the compression driver has.

Norman

[Davey]

Originally Posted by doug20

What would the DCX settings be for that?

Sorry I took so long to answer. It's been years since I did that testing and I had to refresh my memory.
I was mistaken. The DCX will only provide an all-pass function equivalent to a 4th-order crossover (2nd-order all-pass function.) (Not 2nd/8th order crossovers.)

Anyways, to set it up switch all crossover functions on the crossover tab to off. EQ/delay/etc/etc off as well.
Then use the phase control on the "short delay" tab and set to 180 degrees on the channel you'll be monitoring with.
The phase control picks up the LP filter frequency setting (even though the crossover is turned off) and will phase shift the signal 180 degrees at that frequency.

Here's an example where I set the LP filter frequency to 1khz: (Note zero degrees phase shift at low frequency, 180 degrees at 1khz, and 360 degrees at high frequency. Note also the amplitude trace is flat.)

http://orion.quicksytes.com/content/dcxphase.jpg

Hope that helps.

Cheers,

Dave.

[steph_tsf]

Need to read this first : FRD Consortium tools guide