Waveguie and Cardioid on a Slim Baffle
This design will incorporate known solutions to control directivity over most of the audible range. It's currently only a concept but I'm looking to construct a prototype soon. It will be a slow advancing endeavor with projected completion for the late summer. If someone else likes my plan and decides to pursue something similar at a faster rate I will be happy to exchange observations and ideas. Constructive suggestions are welcome although I already have my mind set on how to achieve most of my goals.
I believe the perfect speaker should be able to play well in most listening environments. I want a solution that reduces the negative effects of the listening room while retaining most of the positive ones. Although there is much disagreement and mystery surrounding the human auditory perception I think there is considerable consensus on the following topics:
Constant Directivity - In the highly reflective environment of the typical home, the quality of sound that speakers radiate off-axis is almost as important as it is on-axis. Differences in the direct and reflected sound are audible and detrimental for stereo imaging. Constant directivity designs are often difficult but reduce the impact of other aspects of concern such as cabinet reflections and the need for room treatment.
Floor and ceiling reflections - Ideally, one would want to eliminate all unwanted reflections but in practice our auditory perception is less sensitive on the vertical axis than the horizontal. For that reason and in order to reduce the complexity of the design, I choose to relax the requirements on the vertical response to what is generally acceptable. The vertical lobe resulting from the tweeter-mid crossover should allow at least a 30 degree listening window to accommodate a seated position. Ceiling reflections will not be addressed, however I aim to eliminate the "floor bounce" problem which is known to negatively affect the perception of midbass.
Reducing reflected sound - Very often home speakers end up in the worst acoustic place - close to a room corner. While subwoofers can happily utilize corner loading for extra gain, the region between 50-500Hz can suffer serious irregularities. Waveguide based designs control dispersion well at the high end but their mids and bass are not immune to this phenomena. Dipole designs reduce lateral radiation but there is still considerable amount of energy at the rear of the speaker which can become a problem. My goal is to extend unipolar dispersion to the lower frequencies. The listener will be able to control the amount of room interaction by toeing in the speaker.
Flat frequency response - I aim for a very linear response on axis (+-1.5dB) while allowing for progressively relaxed limits off-axis to +-3dB at 45deg. I might be setting very strict requirements but I feel that this is doable with the help of DSP filters and quality drivers.
Low frequency extension - Based on personal preference I would like to have an F3 of 30Hz and I'm afraid I will need a dedicated subwoofer to reach that. A well designed 12" monopole subwoofer will be able to get there and multiple subs will be even better. I think it will be best to relieve the speaker from the stress of reproducing low frequencies in controlled directivity at the point where the effects of the room begin to matter less. Ideally this speaker should be able to cross over to a sub at 80Hz or lower. However, I will leave this requirement somewhat open for now because I'm not sure what the optimal crossover point is. Also I would like to run some experiments and get real-life data on the low end extension without a sub.
SPL - I don't play very loud and this is a low priority for me. I'm much more concerned with minimizing audible distortion. I will be satisfied if the system is able to hit peaks of 105dB without running out of excursion or power.
Linear phase - Currently I'm using IIR filters in my main system to arrive at a fairly linear phase response through the critical 300-3000Hz range. It's admittedly hard to hear phase distortion but since I'm spoiled with it now, it will be difficult to justify building a new speaker without it. Linear phase will be easy with FIR filters and they are quickly becoming more available so that's an option to consider as well.
Space considerations - Many households don't have the luxury of a large living room. Unless high efficiency is desired there is no need for refrigerator size monsters. In terms of size I feel that the typical floorstander speaker with a slim baffle is good reference point.
To control the directivity at the highs and mids this speaker follows the design ideas developed by MarkK for the popular ER18DXT. At first I thought that I can easily come up with a more original driver pair but eventually I came to understand just how many considerations went into Mark's decision. There aren't many tweeters that license the DXT technology and I didn't want to use a large waveguide because I prefer to keep the drivers as close as possible. The Seas 27TBCD/GB-DXT is probably the best option for a tweeter because of it's excellent directivity control and a small package.
As for a midwoofer there are a lot of capable drivers at the <$100 price point. The underhung Tangband W6-1721, the Pierless Exclusive HDS 830883, the SB17NRX35 and the Dayton RS180 can all work in this application. But after much consideration I could not pick a replacement for the Seas ER18RNX. Honestly, I don't think I can pass on the looks of that reed/paper cone.
- MarkK's ER18DXT: The Seas ER18DXT ported two way
MarkK's ER18DXT succeeds in keeping excellent directivity control from the high end down to about 800Hz where the woofer begins to radiate in full space. My goal is to extend the half-space radiation pattern as low as possible. Running the ER18RNX in an open baffle configuration can retain a constant directivity pattern because energy from the back of the driver will interfere with the front wave and prevent sound radiation to develop into full space. One way to get a cardioid radiation pattern at low frequencies is to build a lossy box with holes and fill it with absorbing material. The idea is to attenuate the sound energy at the back of the driver but not fully absorb it. There are a few existing designs and experiments with lossy boxes that show very promising results.
- Keyser: http://www.diyaudio.com/forums/multi...ioid-like.html
- Kimmo Saunisto: Cardioid bass
A great amount of displacement is required with the lossy box method in order to retain decent sensitivity at low frequencies. Since I want a slim baffle for aesthetic reasons I decided to have a group of four 8" woofers handle the low end. I have my heart set on the Dayton RS225 but I didn't make a final decision yet because of the price for 8 of these. The lossy box construction will undoubtedly hinder bass performance. If an additional subwoofer is need to cover the lowest octave, it's more sensible to use much cheaper guitar speakers instead. On the other hand it might be possible to get satisfactory bass from four 8" drivers at normal listening levels. I don't expect to get much output below 50Hz even with a Linkwitz Transform but even a little may turn out to be acceptable in a small living room like mine.
The crossover will be implemented in DSP using the KX-Project driver for the prototype speaker. I plan to upgrade to miniDSP or another suitable platform later on. On my current system I implemented a phase linear crossover designed by Jean-Michel LeCleac'h. It uses 3rd order Butterworth slopes with a calculated offset and a digital delay to align the phase response. It produces excellent results but I didn't have to worry about the vertical lobe in my previous speaker. With this new design I'm concerned that the digital time delay may result in an undesirable tilt of the vertical response.
Le Cleac'h crossover: Expanded Soundstaging and 3D-Imaging
There are still many unknowns surrounding this design so building a prototype and taking measurements is of high priority. I hope to get insight in the following areas of concern:
- Will damping the rear wave result in the desired cardioid response? How does the composition, thickness and placement of the absorbing material affect the response?
- What will the vertical polar response look like? Will the mid-tweeter crossover allow a good listening window? Will the bass array integrate well with the midwoofer?
- What are the best crossover points and slopes?
- Will the off-axis response remain linear throughout the entire system bandwidth?
- How can I improve the construction of the box to address vibration issues? I want to experiment with magnet-mounting the woofers. Some dipole designs run woofer arrays in alternate polarities (push-pull) to reduce vibrations and I wonder if that approach will work here?
- What is the low end limit of this speaker in terms of safe driver excursion? Will I be happy without an extra subwoofer?
- Most importantly, does cardioid midbass reduce room interactions enough to be worth the extra work?
That's a really beautiful project, lot of good points that are barely seen on the same speaker, the cardioid bass section gives a nice touch of originality. You can expect either nice results and a lot of fun before finishing it.
But I am not so enthusiastic about the Le Cleac'h filter and it's implementation.
First, it's not a linear phase or whatever but only a "quasi optimal", more conceived by his author with in mind something easy to do and accessible to a majority of DIYers, than challenging the ultimate transient perfect.
Second point, finally it's not at all easy to do. There is a gap between the theory and it's application, the drivers have to be textbook linear, the Fc far from their natural roll offs.
In other words, the magic recipe of the spreadsheet needs a lot of measurements and a parametric EQ to comply the target...all this for at best a quasi optimal. This said for the sake of discussion, the important goal is only listening.
I have also the KX drivers on my PC, pretty amazing for a freeware. They are perfectly able to do the job.
You are right of thinking bracing and magnet mounted drivers, the side panels can be vibration friendly. Will you separate the mid cavity from the boomers ?
Love what you've got in mind.
I've built Mark's design and it easily competes with any 2-way standmount out there. The tweeter doesnt have me yearning for anything more expensive, and ive heard all the exi scan speak domes but as always, your mileage may vary!
Boris, I think this speaker will be great
Are you sure about those vertical reflections? Reading through some threads here at diyaudio left me with the impression that vertical reflections are worse than horisontal and always detrimental to sound quality (mainly in the lower mid bass region). OTOH delayed lateral reflections should be beneficial.. Maybe I should re-read the giant directivity thread again?!? :)
I am planning a very similar speaker build.
My front baffle will be slightly wider: total around 35-40cm.
And there will be no side panels.
I might experiment with damping of the rear wave too.
Driver set-up will be similar: 3 way with 4x8" woofers.
I will add two subwoofers if needed: closed box or H-frame OB.
I am going for vintage alnico Seas drivers.
Woofers are Seas 21TV-EW (I already have 8 of those).
I hate rubber/foam surround, high excursion midrangers.
My choice of midrange is a 8" cloth/paper surround widerange driver with a qts of 0.7
Mids will be either Seas 21TV-G or 21TV-GW (I have both in hand).
For HF I will start with the Seas 9TV-LG and 5TV-HF combination (I know that makes it 4 way).
Seas 21-9-5 combo works very well with first order CO as used in many commercial speakers of 1970s.
I will initially try first order for the woofers too. Might end up going active/dsp for them.
For HF I will later try Heil AMT (small one, I have in hand) and perhaps B&G Neo3.
I am considering not having side and rear panels as well. I'm thinking of building the whole thing around a rigid skeleton that holds the woofers and the front baffle. Absorbing material can be secured to the driver basket and the rear of the speaker can be covered with fabric.
Here is a quick simulation I did with DSP that you might find interesting. It looks at how 1st order filters may crossover in an open baffle. Let me state however, that although the simulation looks good electrically, I did not verify it acoustically and I give no guarantees that it works as shown.
My baffle size causes a natural response roll-off to occur at 500Hz. I'm using a 1st order high-pass filter on all simulated drivers to account for that. The red trace represents the midwoofer. The green trace represents 4 woofers on the same baffle. I discovered that in order to sum all of them flat the 4 woofers require a 1st order filter low pass filter at 125Hz. So in essence, when dealing with 1st order slopes, quadrupling the number of woofers on the low end allows the crossover point for the LP to be moved back 2 octaves.
I've got a 2-way thing developing right now, so this will be a very interesting project to follow. Little if any of what I've done/noticed so far will apply directly to what you've proposed here but I'd still like to offer up a couple of points FWIW.
My bass is handled with a 0.47Qt driver (12") firstly on pure OB with minimal support below which evolved into full length wings @10" of depth, a narrower cap above for tweeter support and a bottom of roughly the same depth as the sides as the speaker is elevated off the floor about 10". Bass response was enhanced with the bigger wings ... wing resonance seemed to be confined to a small frequency area and was only really problematic at pretty high SPL.
The damping was tried as per various www resources (keysers thread, johnk, and many others) The thing, for me, that changed with all of the usual damping attempts .. just from a purely listening perspective now ... is that the speaker begins down the road to "sounding like a box". More damping, more box sound ... dead/lifeless .. especially in the mid range. Even just adding the full length wings changed the sound from pure OB to more like a box ... and it EQ's like that as well.
EGDE, at least, was not at all helpful in modeling an expected response with a "U" frame. My suspicion is that as side wings fold more toward the perpendicular ... the driver(s) sound characteristics, that you'd hear in a box alignment, become more "enhanced". I think the amount of damping might therefore become more room dependent than anything else as far as listening enjoyment is concerned. More of a "valve" for spacial level than anything.
Sadly, I don't have any hard data to share as I'm just getting deeper into this fascinating hobby. Good luck with your project boris81 !
Hey I'll be watching this also as I'm working on a design with cardioid bass (found I don't like it fullrange as the image gets smeared) and a different kind of directivity for the highs as I don't like waveguides...
|All times are GMT. The time now is 11:46 PM.|
vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2016 DragonByte Technologies Ltd.
Copyright ©1999-2016 diyAudio