The Square Wave Project

[Jared]

well I meant that the output from a mic when fed a perfect audio square wave would be pretty far from a square wave, wouldn't it? Wouldnt it tend to round off your verticals ??

And I didnt htink you could put a scope on the output snce it seems like you have to use multiple drivers to produce the wave, and each driver is behind crossover components?
If you feed the speaker a perfect square wave, and scope it, and the speaker is ading distortion, would the scope still see a perfect square wave or would the distortion be evident?

[roddyama]

Hi Jared,

If you feed the speaker a perfect square wave, and scope it, and the speaker is ading distortion, would the scope still see a perfect square wave or would the distortion be evident?

That's the idea. We're aimin' to make a system (speaker-xover-amp included) that will reproduce a 20kHz limited square wave input, at the listener.

Bill,

What's the status on the xovers?

Rodd Yamas***a

[JohnG]

How about putting the 6 midranges around the tweeter, i.e a circular array instead of a line array?

John

[jackinnj]

take a 30 Hz sine wave and all it's odd harmonics -- who is to say that a series of pass band filters with the appropriate K's will not perform as you desire, yet sound horrible.

[Bill Fitzpatrick]

The square wave can't be perfectly square for two main reasons; the chosen tweeter is going to have a rolloff somewhere between 20K-30K and the woofer is going to have a low frequency limit. The mic will have limits as well. These can be modeled to obtain a picture of the target. This target is the bandwidth limited square wave previously mentioned.

Distortion products will not be significant for this experiment.

A circular array would be OK if you think in terms of a one dimensional sweet spot but it doesn't have any advantages so I would say it's not worth considering.

jackinnj: That would be one of those black box tricks where you get a flat frequency response output under sine wave conditions but it sound like ****. Fortunately that's not what we would be doing. Anyway, the resulting system has to pass other standard tests as well; impulse, step, waterfall, group delay, etc. Oh, it's got to sound good too!

Right now, my workbench is in disarray due to some remodeling so I haven't begun to breadboard anything. But, there are some interesting possibilites floating around in my mind on the active filter end of things.

It would be great if someone could suggest a tweeter with low resonace and ferrofluid at a reasonable price.

[cocolino]

Originally posted by Rodd Yamas***a
Why the low resonance in the tweeter?

The low resonance is essential because otherwise the tweeters acoustical phase (time if You prefer) response would srew up the desired crossover behavier by interacting with the crossover response at the Mid/Tweeter point. IMHO the resonance should be at least 10 times (>3 octaves) away from the crossing point. The more the better.

Here some 1" dome tweeters that might be suitable
Ferrofluid:
Scan Speak D2905/930000; fres= 650Hz
Scan Speak D2905/950000; fres= 550Hz
Seas Excel T25CF001; fres= 650Hz
Thiel C23; fres= 500Hz (Ceramic diaphragm)
Non-Ferrofluid:
Vifa XT25TG30-04; Ringradiator; fres= 500Hz
Scan Speak D2905/970000; fres= 500Hz
Scan Speak D2905/990000; fres= 500Hz
Scan Speak R2904/700000; Ringradiator; fres= 520Hz (extremly expensive)

Unfortunately all of this drivers unless the Vifa Ringradiator (but it has no ferrofluid) are quite expensive but I don`t know any other which might be suitable also (unless perhaps a few ribbons which are even more expensive).
The only other way to go as I see it would be to electronically equalize (Linkwitz circuit) the resonance of a cheaper driver to lower frequencies. This also would have the benefit that the high Xover point restriction could be weakened somewhat and that it could be changed to maybe 3,5kHz (depending of the final highpass slope of course) or so what again would have positive influence regarding lobing in multidriver midrange configurations (arrays) .

Otherwise, as the upper Xover point has to be around 5kHz very likely due to the mentioned restriction, Ferrofluid may not be absolutely neccessary IMHO.

Originally posted by Bill Fitzpatrick:
2. The mid range speakers should be 6 or 8 Tangband W4-654S in a vertical curved array


Originally posted by planet 10:
This is pretty head-in-a-vise restrictive (and realistically totally impractical). A single mid/full-range in the middle would make it a lot more sense and make the goal of a square wave easier. Something like a Jordan JX92 would be a good choice.

Me too thinks that such a long line array for the midrange might cause troubles since due to the neccessary relatively high Xover point and the required flat slopes there may be serious lobing effects which extremly restrict the vertical listener position. From such an configuration I expect huge frequency deviations (and of course changes of the shape of the "square" as well) only when the vertical position is changed by very few centimeters (You`ll need two different chairs for two listeners of different height).
A Jordan JX92 indeed might be one of the very few choices for a MT configuration and if it sounds as excellent as the other Jordans (I only know the JX53) it would be hard to overcome (quality) soundwise (but I have to admit I don`t know the Tangbands) .
Also an option could be the Jordan JX62 in an MTM configuration or maybe even MMTMM if higher SPL is desired.

I´m aware that we can`t always get everything what we want at one time and that there are to make compromises and that on axis reponse is the number 1 issue here but excessive bad vertical off axis behaviour due to lobing could spoil the "square wave" approuch to some extend though IMHO. Therefore by all means I believe that a MT or in case of multiple midrange driver configurations a MTM or MMTMM arrangement will give more freedom regarding vertical lobing effects in this design.

Originally posted by Bill Fitzpatrick:
Right now, my workbench is in disarray due to some remodeling so I haven't begun to breadboard anything. But, there are some interesting possibilites floating around in my mind on the active filter end of things.

Let them float yet and into here Bill Maybe some of us can contribute something useful.

[roddyama]

Hi Christoph,

The low resonance is essential because otherwise the tweeters acoustical phase (time if You prefer) response would srew up the desired crossover behavier by interacting with the crossover response at the Mid/Tweeter point. IMHO the resonance should be at least 10 times (>3 octaves) away from the crossing point.

I see your point. I usually go by at least 1 octave, but in this case you’re right, “more is better".

Rodd Yamas***a

[phase_accurate]

Square-waves and microphones:

Even though microphones are not perfect (in fact nothing in sound recording and reproduction is), good ones might introduce less temporal deviations than loudspeakers since they are very small FR transducers.
I.e. their step response will never look as bad as the step response of a conventional multiway speaker where the responses of all the different frequency-range's drivers arrive one after each other. So a good microphone will be a valid piece of test equipment to see the improvement in square-wave response over a conventional multiway speaker.

Midrange arrays:

Why not try what has already been done: The outer drivers are taken back in level by an additional lowpass filter in the frequency range where lobing starts to be a problem. The delay, introduced by this lowpass, can help the arrangement to behave like a point source. This has already been done with some ESLs to reduce lobing by arranging the stators in segments that are driven seperately. I was also able to listen to a small Swiss-made active speaker where they did the same (It was a WWTWW arrangement). Where the cone area is needed most (at the lower end) all drivers are working together. The upper response drop might be equalised actively.
I know that this adds complexity so I would first start with two FR drivers in the midrange (MTM) and try such tricks afterwards.

Regards

Charles

[Bill Fitzpatrick]

phase_accurate:

Good point on the array.

I have yet to build and listen to a curved array but it might offer a solution should the lobing be unacceptable.

Right now I'm finishing up a woofer project that is tied to the structure of the house. I've always had excellent results from bolting down woofers but this time, as confirmed by a frequency sweep, a nearby gas fireplace is being exited at several frequencies. I'm going to have to get into it and find the offending areas.

When that's done I'll make a curved array of probably 8 or 12 small Tangbands per channel and see if my ears are happy with it. I have been considering an additional driver with a larger voice coil on the rear of the array to eliminate the need for step response equalization. Anyone have any thoughts on this?

[roddyama]

Hi Bill,

How about a tall diamond shaped baffle. Narrow at the top and bottom and wide in the middle. Conceptually, this can be combined with Phase-Accurate’s approach of using passive equalization on some of the mid drivers. Instead of a lo-pass on the top and bottom drivers, you progressively step the level down (with resistors only) as you move from the top and bottom toward the center. The diamond shaped baffle will spread the baffle step over a much broader range, and the progressive drop in level will provide compensation for the drivers most affected by the baffle step. You also get away from the lo-pass phase shift.

What do ya think?
Rodd Yamas***a