Build Thread: Ardor - point source monitor
- Design Goals
Audio research shows that the quality of sound radiated off-axis from the listener has great impact on the perceived realism of the reproduced auditory event. The vast majority of conventional loudspeakers however aim to deliver coherent sound pressure only at a tiny "sweet spot" and largely ignore the quality of sound that radiates in every other direction. There is an obvious problem in typical living spaces where various reflections combine with the original signal. A couple of successful constant directivity loudspeaker implementations are a testament on how important off-axis dispersion control is.
Multi-way designs rely on electronic crossovers to split the signal into separate frequency ranges and hand over output to different size drivers. The main advantage is using individual drivers in their optimal range which in turn delivers the best distortion and sensitivity performance for the entire system. However, integrating multiple drivers to radiate sound at the same level in all directions for all frequencies is not a simple task. The main problem is that drivers radiate low frequencies equally in all directions but at higher frequencies the sound becomes directional. The narrowing of the radiation field across frequencies is proportional to the size of the driver and larger drivers begin to "beam" at lower frequencies than smaller drivers. The typical 2-way speaker radiates bass in all directions, in the midrange radiation gradually narrows down then widens up abruptly at the crossover point and continues to drop toward higher frequencies.
Another problem caused by crossovers is disturbances in the resulting wavefront when two identical waves from different drivers combine at the crossover frequency. The physical offset between the two drivers and the wavelength of the crossover frequency affects the resulting radiation pattern and therefore can be manipulated by the designer. Interestingly, this issue is widely ignored and priority is usually given to other design parameters. Arguably measurement equipment shows such wavefront disturbances as relatively benign signal dips over narrow frequency on the vertical axis. Psychoacoustic research also shows that sound reflections from the vertical axis have lesser importance to the listening brain than ones on the horizontal axis.
This design takes a holistic approach of preserving the integrity of the reproduced wave in all directions. The omnidirectional radiation pattern requires ample space from the room walls and attention should be given to selecting an optimal placement.
The resultant wavefront of two drivers begins to act as a point-source at frequencies where 1/4 of the wavelength is larger than the center-to-center distance between the two drivers. Designing with this rule can cure both problems of crossovers - integrity of the wave is preserved and drivers are user in a range where they radiate uniformly.
- IMG1 simulation of driver lobbing at high frequency on the left and good integration at low frequencies on the right
The requirements for this design will be most demanding on the tweeter. I started by looking at tweeters that can keep a wide radiation pattern as high in frequency as possible. Although this trait is usually accomplished by making the radiating surface smaller, most tweeters have huge magnet structures and even wider flanges grossly overshooting the distance requirement. The 1/4 wavelength at 1400Hz is about 60mm. Furthermore, very few tweeters can handle such low frequencies and only a handful can go an octave lower to allow a clean LR4 crossover. Harmonic distortion is also a major issue for tweeters playing this low.
60mm is not a lot of space to work with so initially only 3" midranges were considered. This size drivers also fulfill the requirement of keeping radiation uniform in the lower frequency range. There are some very well performing fullranges in this category but the challenge was finding a high sensitivity driver. I was also hoping for a low crossover under 300Hz which proved difficult.
I think it goes without saying that the most challenging part of this design was selecting a tweeter and midrange that can fit 60mm or closer. The combination of drivers I picked came on the market about a year ago, again exemplifying the difficulty of constructing drivers that can operate well under such demanding requirements. I chose the Scan-Speak Discovery 10F/4424G at $74ea because of it's excellent performance and the 4ohm version has higher sensitivity. It's technically a 4" midrange but it's closer in actual size to a 3". I only found 2 tweeters that meet the low crossover and small size requirements. The Vifa NE19VTA-04 is priced at $25ea and the ScanSpeak Illuminator D2004/6020-00 3/4" at $118ea. Even with such small drivers it was clear that I can't get them 60mm center-to-center. Eventually I settled for the cheaper Vifa tweeter because I decided to cut a part of the flange which puts me at 70mm center-to-center.
The 1/4 wavelength of 400Hz is about 210mm which allows for a 12" driver on the bottom. Such a driver if used right has enough output to cover the bottom of the auditory range at reasonable levels. While I certainly encourage such pursuit I decided it was more convenient for me to buy a smaller prefabricated box and limit the low extension to what's practical.
The box is a Dayton TWC-0.75BK 20L Curved Cabinet by Parts-Express that sells for $133/each. The best match for it in a sealed configuration is the SB Acoustics SB23NRXS45-8 priced at $82ea. I use a 6dB bass boost to get flat response down to 50Hz.
I am a beginner at woodworking and I did not have the confidence to build the enclosure. Regardless, I was very eager to put a personal touch on this build so I decided to fabricate the baffle out of MDF.
Based on personal preference I decided to implement this project in DSP with active amplification. I am sure that this design can be converted to work well with a passive components. I will provide the transfer functions for those interested in building their own crossover and EQ.
The drivers used are very linear and the crossover is pretty straight forward. I use 4th order Linkwitz-Riley crossovers at 400Hz and 1400Hz. The woofer gets a High-Shelf filter at 370Hz,-2.4dB,1.51Q for Baffle Step Compensation. An optional Low-Shelf filter is used at 51Hz,+6.6dB,1.68Q for bass boost. Due to it's metal dome the Vifa tweeter has a resonance in the last octave. Don't be tempted to go with the worse performing fabric dome version, a Notch filter at 20kHz,-5.5dB,0.4Q tames this pretty well. The mid and tweeter are padded down -2dB and - 5dB respectively.
- IMG2 crossover and EQ values
Singfried Linkwitz suggests that flat frequency response is not optimal for systems that interact a lot with the listening space. He suggests a frequency response that slopes down towards higher frequencies at -2dB for every frequency decade. I accomplish this with a High-Shelf filter at 450Hz,-2dB,0.87 for the mid and recessed level for the tweeter. I still feel that the sound is somewhat bright for my taste and I will likely end up with a steeper curve. I want to make the point however, that this phenomena is dependent on the room and positioning.
I use a few Notch filters here and there to smooth out the response but I won't publish those values because they are under 3dB and I suspect that most of them are unique to my enclosure.
The system radiates uniformly from 0-60 degrees. The response drops down gradually from around 60 degrees outward and is -9 dB at 90. I didn't do extensive measurements beyond 90 degrees because I only measured indoors but it seems the uniform radiation trend is well preserved. Response in the last octave drops off-axis but the excellent dispersion of the Vifa tweeter is quite linear on axis to 8kHz even at 90 degrees!
- IMG3 horizontal frequency response 0-blue, 60-red, 90-green
I didn't hit my target of 60mm distance for the 1400Hz crossover and I was a little worried about that. The vertical measurements did not register a significant dip in response till 90 degrees where it shows up not very significantly at -3dB. That corresponds with the simulation so I declared it acceptable and decided to leave it alone. I set a delay of 0.084milliseconds for the tweeter to ensure the lobe is centered on the vertical axis. A delay for the other crossover was not necessary because the midrange and woofer overlap significantly.
- vertical frequency response 0-blue, 60-red, 90-green
I want to take some time before presenting the customary listening impressions.
- frequency response and phase on axis (stitched up)
- the finished speaker
Great job, very interesting to see you choose a three way design when most similar projects would skip the mid range driver and go straight from 8 inch woofer to tweeter and xover at 1400hz.
Congratulations, very nice job ;)
Can easily do in passive. Go to 1400 with a 3/4" :eek:
Note i have an other solution for you, a coaxial driver ! Seas does them 8",7" and 4". It is easy to find the 7", 4" but i also see the 8" and i know a shop which sell them in my country. These drivers are not perfect but they act as point source. Other brands make coaxial like Tangbang.
THE ART OF SOUND PERFECTION BY SEAS - SEAS PRESTIGE COAXIAL DRIVERS
THE ART OF SOUND PERFECTION BY SEAS - SEAS EXCEL COAXIAL DRIVERS
The L12 could do very well the job ?
Thank you, jerome 69!
I don't see anything wrong in challenging conventional design trends as long as there is a sound purpose behind it. I don't claim that I'm smarter than anyone else. The goal is to explore possible improvements on the conventional multi-way box by selecting different priorities.
I understand that coaxial drivers along with fullranges are ideally the best solution for a point source design. If however, comparable or better parameters can be obtained with a conventional arrangement multi-way system, then cost can be lowered and performance increased.
I forgot to describe a very important construction detail. The drivers don't share the enclosure! The midrange has it's own box that is attached to the removable baffle. That box is sealed and has about 1L internal volume. Making the box bigger won't significantly increase the low end capability for the 4" midrange and will hurt the internal volume of the 8" woofer enclosure.
The enclosure walls of both boxes are damped with sound absorbing insulation to tame internal reflections. I bought a roll of the UltraTouch Denim Insulation from Hope Depot. It's 3" thick so I separated it down the middle and got two 1.5" sheets. They were enough for a pair of speakers. Additionally the woofer enclosure is loosely stuffed with wool which has the effect of artificially increasing the box volume.
That's a very nice looking speaker. I've considered using a cabinet shape very similar to yours in an upcoming build. How does it sound?
Nice speakers! Very smooth response, small drivers, low crossover-points and nicely rounded cabinets. This should sound really good!
They remind me a bit of NHT speakers: http://www.nhthifi.com/Three?sc=12&category=3772
How is it possible that the 90 degree curve is so much lower in level than the 60 degree curve? Was the 90 degree response curves measured in a different way?
Wrong link. I meant to post this one: Classic Three | Bookshelf Speakers | Premium Audio Equipment
I have no explanation for this phenomena. For off-axis measurements I simply turn the speaker on it's base making sure the tweeter position remains the same distance from the microphone. I may be off an inch or two from one plot to the next but this 6dB difference indicates more than doubling the distance which is not the case here. I got similar results with a prototype I built earlier to test the 1/4 wavelength driver distance concept. An exploration in eliminating driver lobing
I won't rule out an error with my measurements because my setup is quite crude. I measure in my living room at low levels. I have an uncalibrated Behringer EMC 8000 and the distance to the speaker is usually 1m. I gate before the first major reflection.
I want to state that I'm no expert in measuring loudspeakers and I welcome any suggestions on improvement. I have no reason to believe that the plots I am showing here are wrong but i understand that mistakes are possible. I am very open to conducting a set of measurements with anybody in the New York area who may be interested in that.
They sound incredible!
I was hesitant to describe the sound so soon because despite most of the work being finished I am still fiddling around with the EQ. I will present this general initial impression and then have some improvement suggestions at a later time. Before I begin I should assert that this speaker is very sensitive of placement and may not be suitable for some rooms. An optimal performance will not be possible without careful attention to interactions by the speaker and the room. I am working on a placement guide which will address some of the issues that I came across in my far from perfect listening room.
My most immediate impression of this speaker was just how much resolution was available in the reproduced sound. Unimaginable details emerge, some of them great like feeling the intensity of each guitar string plucking and other things less pleasant like a vocalist breathing into the mic. I did not intend to designate this speaker a "monitor" until I built it and listened to it. As such Ardor may not be suitable for people looking for "sweet" and relaxing sound. The presentation is balanced but often "in your face" and certain material can be irritable.
This level of clarity is by no doubt helped by the crossover. I have heard quite convincing rendering of piano and vocals on less stellar systems but I think the real test of crossovers is drums. Somehow the fast transients in a drum sound can give a good indication of the quality of the crossover. Drums on Ardor are much more accurate than what I was ever exposed to. Another feature that I haven't experienced before but was able to attribute to the wide dispersion of the tweeter(through experimentation) is likely what audiophiles refer to as "airy". It has the quality of bringing spaciousness and making the sound more present in the room.
The strong side of this speaker is without question the imaging. I will repeat again that placement is critical for the optimal image presentation and uneven distances from the sidewalls will have negative effects. The soundstage doesn't completely collapse off-axis but when moving to the side the image begins to smear towards the closest sidewall. The sweet spot defined by S. Linkwitz for the Pluto is also the best position for this speaker. Closer - the image is too sharp and and disconnected, further back it's more relaxed and compact. When instruments are recorded with spatial information in stereo they layer on top of one another even if they are presented together. You may hear the vocalist and the piano coming both from the center but with different depth and proportions. Sounds can be felt to "hang in space", they now have an extra attribute of dimension to them. I perceive each sound exerting it's own texture to the fabric of the presentation.
The weak side of the system is currently the low end. I get very satisfactory thump from kettle drums but nothing earth-shaking. I didn't think that I will miss my windows rattling but because the rest of the presentation is so convincing the lack of visceral sensation is confusing. A sub is highly recommended.
|All times are GMT. The time now is 11:59 AM.|
vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2016 DragonByte Technologies Ltd.
Copyright ©1999-2016 diyAudio