How to achieve coherence

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
frugal-phile™
Joined 2001
Paid Member
Small speaker, cross very steep 1kHz, linear phase, and drivers 1/4 wavelength apart at crossover.

Andrew,

When i 1st saw that my thot was that that looks scary. The graphs themselves make me apprehensive, and only suggest an awful lot of processing to get a real flat on-axis curve, which is meaningless without a set of off axis curves. And there is a whole lot more to a speaker than flat-frequency response. I would expect lots of digital artifacts, but i'd have to hear it to really judge.

dave
 
So what do you mean by that? Do you mean the tweeter has a different dispersion pattern to the woofer and that gives it away?

Yes.

But you can hear a lack of coherence even on axis.

That doesn't matter.

Your brain hears timbre as a weighted combination of what it believes to be a direct sound and its reflections with accommodation for less high frequency energy in later reflections as one gets in nature with increasing absorption and diffusion.

Where the reflections' spectra don't have monotonically decreasing energy with increasing frequency it does not sound natural.

This is why both the ubiquitous 2-way with a 6-8" midrange, 1" dome tweeter with a 4" face plate, and 2.5 KHz + cross-over and 6-8" full-range speakers don't work well.
 
Andrew,

When i 1st saw that my thot was that that looks scary. The graphs themselves make me apprehensive, and only suggest an awful lot of processing to get a real flat on-axis curve, which is meaningless without a set of off axis curves. And there is a whole lot more to a speaker than flat-frequency response. I would expect lots of digital artifacts, but i'd have to hear it to really judge.

dave

Measurement at 42", no smoothing:

An externally hosted image should be here but it was not working when we last tested it.


Same measurement, gated to 3ms:

An externally hosted image should be here but it was not working when we last tested it.


45 degrees off axis horizontal: Gated and un-gated results are shown. Rest of results are all gated. (Click on picks to enlarge)

An externally hosted image should be here but it was not working when we last tested it.


90 degrees off axis horizontal:

An externally hosted image should be here but it was not working when we last tested it.


30 degrees horizontal and 30 degrees vertical above:

An externally hosted image should be here but it was not working when we last tested it.


15 degrees below:

An externally hosted image should be here but it was not working when we last tested it.


Speakers do square waves within bandwidth constraints, waveform fidelity to what is on CD/normal listening material is extremely good, and they sound fantastic, no audible digital artifacts.

They are easy to build. I can get this performance using multichannel soundcard ranging from $30 upwards, and four channels of amplification.

Filters are small enough that this will run on antique 500MHz P3 laptop.

Regards,

Andrew
 
Last edited:
With such a large amount spent on drivers & XOs it seems a shame to put them in such a mediocre box.

dave

The box is not mediocre. It's been mistaken a number of times by audiophiles for Sonus Farbers and my wife loves 'em. xo $$? don't think so. One resistor, and two coils don't add up to much compared to many high order xo's I've seen with various cr*p trap circuits included to address typical anomalies associated with mediocre drivers. Get good drivers and the xo doesn't need to be complicated. The simpler the xo, the more coherent the sound - most of the time.

I think I explained the reasoning behind "The Intimates" fairly well in my TT write up.
 
Overlays of axis/off axis measurements, ported to REW:

off axis overlays 30h30v 45 90.png

Plots referenced to 100dB at 500Hz, 20ms Blackman-Harris windowing, and 1/12th octave smoothing.

Speaker stationary, microphone moved.

Workhorse software is Cool Edit Pro2.1 with Aurora plugins, and Sourceforge Convolver plugin.

Microphone is Earthworks OM-1, older, but very similar to M30. Extremely flat response.

Sound Card is E-MU 0404 USB used with Cakewalk Sonar for measurements.

Playback on Roland Octacapture USB using Windows Media Player with Convolver plugin. Octacapture clock is synchronized to E-MU clock.

Also works perfectly fine with ancient Creative SoundBlaster Extigy USB.

Also Octacapture works with Console hosting Convolver, allowing standalone operation with selection of analog or digital input.

Phase overlays:

phase overlays.png

Regards,

Andrew
 
frugal-phile™
Joined 2001
Paid Member
The box is not mediocre.

I've had a number of those actual boxes thru here, they are OK for what they are, but they are still medicore quality -- higher quality than other the 6 layer MDF plywood on the sides.

As to the XOs, even simple ones can cost. 3 parts per in the XOs i am working on now ran close to $100 each.

dave
 
I use arrays of 16 Foster 402S01 drivers as tweeters, and don't hear any "incoherece" with arrays of 8 x 4" drivers with poly cones and pair of 12" woofers in concrete boxes. I don't hear speakers at all, I hear stereo image. Why? Because if speakers "give away" themselves they "give away" because of their distortions. In arrays displacements are tiny, so distortions are much lower than from a single driver that produces the same pressure level.


PIPEDREAMS by Nearfield Acoustics

what about these speakers? These are very tall yet they did not sound like a multi way with 100 drive units. Why?
 
PIPEDREAMS by Nearfield Acoustics

what about these speakers? These are very tall yet they did not sound like a multi way with 100 drive units. Why?

Why should they?
They are much taller than mine that create nice uniform stereo image, and I don't hear them. However, their tweeters are different. My Forters have flat surfaces, like some kind of light metal(?) foam, on thin rubber suspension, 1 inch high.

ES4246.jpg



This picture was taken in my walk in closet using Behringer mike, the speaker was hanging on wires:

666Hz_tweeters.gif
 
PIPEDREAMS by Nearfield Acoustics

what about these speakers? These are very tall yet they did not sound like a multi way with 100 drive units. Why?

Dominant factor is wavelength v cone diameter. And as source spacing decreases, frequency increases at which two sources may be resolved. Somewhere 1/4 or 1/3 of a wavelength source spacing it becomes very difficult to resolve system as 1 or 2 sources.

This is underlying principle of driver coupling with line sources. In case of effective floor to ceiling line source a cylindrical wavefront. If each element of array has flat response on axis and falls as it must, then sum of responses at a listening point is not flat; thus line sources have specific equalization needs.

Keele has multiple papers with details on line sources.


In two way line source crossover lobing is in horizontal plane and will show up in off axis measurements, with all the usual dependencies of sealed, OB, angular separation v frequency etc.

Looking at Pipedreams: How it works indicates crossovers at 60Hz and 6kHz. With pairing of 4" and 1" drivers on flat face baffle, estimating their centers at 3" apart, 1.125" as half wavelength for 6kHz guarantees several horizontal lobes that are modified by off axis directive behavior of drivers.

Pipedreams require dedicated active crossover that also does some form of EQ.

All that driver area does improve linearity and accompanying reductions in IMD.

My Pluto style speaker has 5.24" and 2" drivers with about 3.75" of vertical separation; <1/3 a wavelength for 1kHz crossover. The off axis response follow directive behavior of tweeter both horizontally and vertically:

309145d1351602795-how-achieve-coherence-off-axis-overlays-30h30v-45-90.png


Bang for the buck isn't Pipedream.

Andrew
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.