Columns Retired Columns & Blogs |
Design Guidelines for Practical Near Field Line Arrays
James R. Griffin, Ph.D.
https://audioroundtable.com/misc/nflawp.pdf
Was I in a hotel room? Couldn't tell, because the sound coming from a pair of Grandinote Mach 9 speakers with carbon fiber enclosures ($CAD50,000/pair), a Grandinote 37Wpc Shinai integrated amp ($CAD20,000), and a Grandinote Volta music server ($CAD14,000), with cabling by Nordost, completely obliterated the walls around me. I was somewhere else. But I think what surprised me most about the system's sound was the fact there wasn't a single tube in the signal path. The sound was so spacious, with such luscious tone.
Design Guidelines for Practical Near Field Line Arrays
James R. Griffin, Ph.D.
https://audioroundtable.com/misc/nflawp.pdf
https://audioroundtable.com/misc/nflawp.pdf
Note the examples on page 22 showing a variety of tapers for 12 drivers.
This line array has 16 small wide range drivers, and I suspect the taper might be in parallel groups:(2/3/3/4/4), in series connections:
(8,9/5,6,7/10,11,12/1,2,3,4/13,14,15,16)
in the context of the tapers shown on that page 22.
"James R. Griffin, Ph.D. https://audioroundtable.com/misc/nflawp.pdf "
Not one word on/of Doppler distortion in that paper, which would be rife in this sort of design.
Just like the MacIntosh xk2 b pie or whatever.
https://www.mcintoshlabs.com/-/media/Images/mcintoshlabs/Products/ProductImages/XRT21K/McINTOSH_020118_gnzphoto-1690-edit.ashx
smeared was the word one reviewer said.
unless all drivers are matched to within a poopteenth of each other (impossible task to last any distance).
Cheers George
The problem with the line array is interference at high frequencies, resulting in combed frequency response at those high frequencies. It would not be a problem at frequencies below which all radiating drivers are within one quarter wavelength, effectively a point source, and not the case of a line source. For a line source, it is not a significant problem at frequencies below where center to center driver separation is less than quarter wavelength, but is pronounced where that distance exceeds one wavelength, with transition in between those bounds.
Dr. Jim Griffin's simplified white paper discusses that.
Years ago Jim Griffin sold his collection of Speaker Builder magazines to me, three Priority Mail boxes refilling an empty 20_lb box of copier paper. I could probably dig those out and wade through to find some articles on the subjects that I could point to for easier accessibility, but that would be more effort than I am willing to expend on this fine morning.
Doppler distortion is the effect of where two signals are being radiated from one loudspeaker diaphragm, and the signals modulate each other. The large excursion motion of the low frequency signal causes the high frequency signal to launch from a more highly varied excursion position and diaphragm velocity than would be the case in the absence of the low frequency signal. That causes a varied frequency shift in radiated signals, the Doppler distortion.
It is not a problem associated only with diaphragms in line arrays, rather all loudspeaker diaphragms do this, as do all microphone diaphragms.
It all about the bandwith and excursion. Wider bandwidth of frequency spectrum that is radiated by a diaphragm results in more Doppler distortion, and narrowing the bandwidth of a driver reduces Doppler distortion.
Consider excursion in simple case of a driver mounted on an infinite baffle radiating into half-space (2pi sperical). For constant SPL with respect to frequency (aka flat frequency response) there is an inverse square relationship between frequency and excursion, halved frequency requiring quadrupled excursion. So a small increase in bandwidth by extending low frequency response results in a large increase in excursion and similarly large increase in Doppler distortion.
The solution is in division of labor among multiple drivers suited for their task, dividing spectrum. Rather than getting all of the work done by one overworked single driver, use subwoofers for lowest octaves, woofers for bass, midranges for midrange, tweeters for high frequencies, and use crossovers with stop band slopes steeper than -12_dB per octave for monopole, -18_dB per octave for gradient alignments, to reduce out of band excursion below constant.
Dr. John P. Kreskovski had a couple of papers on the subject on his old Geocities website. The first is available on the Internet archive Wayback Machine, but the more simplified second paper was not archived. Below is a link to the first paper.
http://web.archive.org/web/20090809205449/http://www.geocities.com/kreskovs/Doppler1.html
As a side note ...... Recent TAS review of Devialet Expert Pro integrated amp with SAM (speaker active management) technology discusses some of these issues along with low frequency induced problems with speakers :-) ..........
Stereophile published an article 'Red Shift: Doppler Distortion in Loudspeakers' written by Keith Howard in 2004 :-) .......
Good article. Here is the link.
https://www.stereophile.com/reference/1104red/index.html
The TAS review and discussion about distortion in loudspeakers, the article which I mentioned above, is also an interesting read :-) .........
Given the number of bass drivers in these speakers and the fact that they are ported (which drastically lessens the low end excursion around the port frequency), the Doppler distortion should be fairly low.
You do then have to deal with the fact that the phase response in the low bass is very poor.
As to the comb filtering with the tweeter array, well, yeah! Using very long ribbon, planar magnetic or AMT's would mostly solve that problem. The comb filtering would still happen, but the "comb" would have way fewer teeth (the Steinway Lyngdorf LS Studio, for example).
Several years ago, I emailed Don Keele about his Constant-Beamwidth Transducer technology speaker designs. He was very insistent that keeping the distance between the tweeter array drivers was very important. Even using 3/4" tweeters spaced very closely, there was still comb filtering at 16kHz, but it was rather benign.
This design, using what I'm assuming is 1" tweeters spaced at least 2" apart, is going to start combing below 8kHz and do it quite badly.
On the upside, here's hoping that carbon fiber cabinet is both really light and acoustically dead. There really isn't much info on their site (which looks like it was made 20 years ago).
And, another plus, the THD will be very low due to the multiplicity of drivers so it's not all bad. Also, this things will sound really clean at very insane dB's.
(My own current interests tend more toward gradient alignments in vertical pair-wise symmetric arrays, not long monopole line arrays.)
A line comprised of multi-driver segmented horns brings it's own set of problems, but is a viable solution to the CTC separation distance problem in the line array. Ponder the notion of a ceiling to floor line of something like what is depicted below, flanked on each side by ceiling to floor lines of midwoofers, using Dave Dal Farra's relaxed 3rd order crossover (separated poles, first order through the transition region, third order in the stop band), frequency response compromise limited to keep ripple within 2_dB. Use Earl Geddes' waveguide(/horn) solution of oblate spheroid profile stuffed with reticulated open-cell foam to damp high order modal resonances, the resulting acoustic low pass filter equalized to desired response.
Take a look at the line array pro models from Bose and also some from JBL, available at Sweetwater ........ Those line arrays use multiple small diameter single drivers with separate subwoofers :-) ........
At risk of stating the obvious, tall nearfield line arrays using high quality drivers are very expensive because of the large quantity of drivers. So, the increased expense of a line array may not be worthwhile for many for a casual listening setup, but the performance characteristics may be worthwhile for those with deeper pockets.
Two of the advantages listening in the near field of the line array are that the resulting sound does not change much between seated and standing listening positions, and the nominal listening level does not change much with variation in propagation distances to listeners.
In the nearfield of a line array, sound propagates as -3_dB with doubling of distance, -10*log(x)_dB (where x is the proportional change in distance, eg. x=2 for doubled distance), as compared to sound propagating from a point source -6_dB with doubling of distance, -20*log(x).
I would agree with you in what I think you are referring to in this, that phase distortion and associated group delay (the negative rate of change of phase with respect to frequency) might be a significant part of the problem at low frequencies. Though I would also suggest that it is not the dominant problem.
Rather, I think that in domestic applications at low frequencies, the dominant problem is the interference from Eigentones (harmonics as well as fundamentals) associated with the room's modal resonances, around and below the Schroeder frequency.
At the link below, Keith Howard wrote an article for Stereophile, "Wayward Down Deep", 30 July 2006, related to earlier work on subjective perception of low frequency phase distortion.
https://www.stereophile.com/reference/706deep/index.html
John Atkinsen wrote about the subjectively unnatural sound of acoustic high pass response in recorded sound and playback systems in "Is it Live? Or is it...", 24 November 1995.
https://www.stereophile.com/asweseeit/75
...do not suffer unique phase distortion but they do accumulate more group delay as the result of the port lagging the driver. Both are secondary to the much greater extension provided by the resonator - the speaker is so different to its sealed counterpart that it's hard to argue a distinction just via phase and delay.
... 30 KEF LS50 speakers (per channel).
Wonder how that would sound?
Want to give it a try, Johan?
For about $10k less than the price of those Grandinote speakers, we can get a pair of Lexicon SL-1 wireless, active speakers, which don't need any amplifiers ......... See KR's 2017 CES report :-) ..........
Owning LS50's as my mains in my home theater, they are great speakers when kept in their envelope, but that coaxial driver will not work in a line array. The comb filtering would be terrible. But 30 pairs of LS50 would make for a killer home theater and whole house audio system. You would, though, need a really huge house.
I bolster the low end with a couple Rythmik servo subs and will be getting two more so that sub bass is on par with the mid and highs. I'll then be able to play my basses through my audio system, hear every nuance of the instrument's sound and have no worries about blowing anything up even at very high levels.
I read an article earlier this year where LS50's with four Rythmik subs were compared to JBL EVEREST DD67000 speakers. Apparently, the reviewer thought the much cheaper LS50 based system was every bit as good as the JBL's. Obviously, the JBL's horn based system can play a lot louder than LS50's. The JBL's have a sensitivity of 96dB while the KEF's are around 85dB.
Side note: While I was shopping for the new Sons Of Apollo live Blu-ray (which seems to not exist in the US), I stumbled upon the Devon Townsend Project. I've been watching them play on YouTube while writing. Man, both of these bands kill. So I now have two live Blu-rays coming my way that were recorded in the same ancient amphitheater in Bulgaria with the local symphony. Yeah, baby. Seems like I now have a bunch of Devon Townsend stuff to add to my music collection. I love discovering new music.
I very much doubt the Everest suffers because of its drivers.
I'm sure Ortofan would really enjoy listening to Sons Of Appollo and Devin Townsend Project :-) ......
You need to go listen to the all new Pure Audio ONE.2 Integrated amp
Tube Magic without the tubes too !