Measureing and correcting Phase

[freedom]

Dear all.

I am wondering how I get around this challenge:

I have a BMS 4590 Dual Concentric Compression Driver, bolted on a BMS 2236 60 x 40 horn. In a trapezoid-shaped box this horn is mounted in the top, and 2 pcs. ATC 12" PA75-314 BLCR low/mid range speakers are placed in the same baffle, just underneath the horn. All three units in a straight vertical line.

The system is controlled by a DCX 2496, in which I have the abillity to correct the phase and delay. I have made a calculation that says the 12" drivers should be delayed close to 6 milliseconds, in order for the compression driver and the 12" drivers to be coherent in time. - This corresponds to about 300 millimeters - which fits well to the physical distance between the drivers. The DCX 2496 has the abillity to automatically test this, and has come up with the same result, within a couple of millimeters. So far so good.

Now, since the phase at the crossover frequency might still not be right I would like to correct this as well.

So here goes my Q: How to do this properly?

At my disposal I have an oscilloscope, ECM-8000 microphone, Behringer DEQ 2496 with RTA, MIC-preamp, FireWire soundcard in decent quality, Smaart Audio Analyser Software and hardware tone generator (not super quality but ok Philips model).

Eventhough the ECM-8000 not has perfectly flat frequency response, I find it adequate for time and phase measurements.

Crossover is at 1210 Hz. 24 db/oct LR.

I've been wondering where to put the microphone - in the vertical plane this must influence the phase between the drivers significantly since 1210 Hz is about 28 centimeters.

Also the distance from the speaker to the microphone is tickeling my mind. My triangular "intuition" says that the further away the less the vertical position affects the phase, and the more the speaker can be interpreted as a single point soure, and not multiple drivers. On the other hand, to far away could perhaps result in a wrong measurement due to reflections - since I do not have an acoustically dead room. Whats your opinion on this issue?

Perhaps I should take a completely other approach?

Looking forward to read your recommandations and suggestions about how to get the best result.

Thanks in advance

Hans-Henning, Denmark

[Ron E]

You are right that there is only one place perpendicular to the baffle where this will work for noncoincident drivers, so you will need to define that point before you continue.

What is the goal of this exercise? You want perfect time alignment, or ? The ear is rather insensitive to time alignment/phase so your effort may be wasted. If your crossover is not a perfect 1st order your time alignment will not produce a transient perfect blending anyway, so...

I haven't used Smaart, so there may be an easier way using that.

My feeling is that you can figure this out with a really simple setup:
Make some tonebursts at the crossover frequency (or impulses) using Audacity. Play them through the high and low frequency drivers and record the result - the amplifier output in the left channel and the microphone output to the right. Measure the difference between signal and sound and you have your time of flight, including phase difference at crossover frequency if you use toneburst.

IMO, you can avoid this entire process - all you need to do is make the system sum flat and have a good reverse phase null to have the speakers as "in phase" as they need to be at the crossover frequency. This you should be able to measure easily with Smaart.

Good luck.

[myhrrhleine]

Originally posted by Ron E The ear is rather insensitive to time alignment/phase so your effort may be wasted. [/B]

old myths sure die slowly!

[Juergen Knoop]

Originally posted by Ron E

What is the goal of this exercise? You want perfect time alignment, or ? The ear is rather insensitive to time alignment/phase so your effort may be wasted.

I have listened to some nice time aligned speakers. Of course speaker building is always about making compromises. So time alignment may not your main issue.
But if you can achieve it with reasonable effort - why not?
Regards

[Ron E]

Originally posted by myhrrhleine
old myths sure die slowly!

http://www.silcom.com/~aludwig/Phase_audibility.htm
There are several schools of thought on the issue of phase audibility, and I am a naysayer until proven otherwise. That someone can concoct a signal that is audible does not say that such a signal is found in music, or that it is representative of phase audibility in general.

Originally posted by Juergen Knoop

I have listened to some nice time aligned speakers. Of course speaker building is always about making compromises. So time alignment may not your main issue.
But if you can achieve it with reasonable effort - why not?

Flat frequency response and smooth off-axis response is orders of magnitude more important than flat phase response or time alignment, IMO. That your time aligned speakers sounded nice does not mean that they did so _because_ of time alignment.

I don't mean to discourage someone from time aligning if they desire to do so, but I think they should have a clear picture of what is to be gained.

[thoriated]

6 ms seems to be quite a bit, considering that sound travels about a foot a millisecond (or a meter every 3 ms) under audiophile compatible atmospheric conditions.

[Iain McNeill]

You have the phase response due to the bandpass nature of the speakers themselves and the phase response of the crossover. There may also be a phase error due to high pass filter in the amplifiers.

These cannot be corrected by a simple delay.

Is there an arbitrary phase function on the DCX2496 or are you looking to add a bit more delay to fix up? In the case of the latter, I wouldn't bother. In the case of the former....well there's a lot of opinion.

My experience is that the measured phase responses are usually so noisy that they are unusable for correction purposes. You have to do a lot of postprocessing to get something that sounds as good as uncorrected.

[myhrrhleine]

Originally posted by Ron E
[B]


http://www.silcom.com/~aludwig/Phase_audibility.htm
There are several schools of thought on the issue of phase audibility, and I am a naysayer until proven otherwise. That someone can concoct a signal that is audible does not say that such a signal is found in music, or that it is representative of phase audibility in general.

phase/time alignment is easily audible by many of us
in the temporal domain, not frequency/steady state tones

[Iain McNeill]

Actually, I agree with Ron E.
Given a flat frequency response, it's the off-axis lobing that is the bigger problem. Use the lowest order crossover to get the job done and set delays for good off axis response at XO frequency.

If your amplifier has too much LF group delay then get bigger and better input caps.

[tthorsten]

take a look at this software

http://www.take-sat.de/english/main.htm