DIY Axially symmetric oblate spheroid CD waveguides, in solid Oak

[jack_bouska]

Thats totally friggin amazing
now i got a new project
gonna take me awhile to save up for 18" drivers
but i really wanna feel that bass

Morbo
Friggin amazing is just the start! After one listening session round my place, an audiophile friend remarked that the deep bass actually cured his constipation!

If you are saving up for drivers, I recommend trying the JBL 2245. The Fs of 20Hz would make a good drop in replacement for the Altec 18" I am using.

But be forewarned, these enclosures do require significant bass boost below 60-80Hz to work properly. Consider using 2nd order low pass, centred on 15-20Hz, as in the circuit attached below.
Good luck, and post the results of your project when you get it going

Jack

[yggdrasil]

Hi Jack.

I believe the mod's was the removing the opamps from the analog section, leaving no gain.

As an aside, considering scratch built crossover systems, I am slowly being convinced by Ed Wildgoose

< http://www.duffroomcorrection.com/wiki/Main_Page >

that it would be beneficial to replace the DCX's with a dedicated rack mount PC

Have not bumped into this site before, but: A PC with everything digital inside has to be interesting. And no need for A/D - D/A of the digital source.

[yggdrasil]

BTW: This could be interesting http://audioheritage.org/vbulletin/s...ossover+theory

[jack_bouska]

BTW: This could be interesting http://audioheritage.org/vbulletin/s...ossover+theory

Johnny - again, thanks for the link to the thread with the LeCleach spreadsheet and PowerPoint, very interesting! - I have briefly reviewed the information, and my interpretation is that J.M. LeCleach is proposing a specific set of time-shift, polarity-flip, and Fs adjustments, to compensate for the inevitable discontinuities, or slow phase rotations inherent in high order crossover filters (3rd or higher order). If a designer is forced to use high order filters because of two or three way topology, or employment of low power handling, or narrow band transducers, then I could see value in testing the LeCleach approach.

My system, and design approach, however, is a very different animal.

I use Dirac-Delta single sample pulse excitation, with a soundcard running at 96kHz A/D to estimate time-of-flight between each driver in my system, and the microphone at the listening position. I dub this "the gold standard" in time delay measurement, and can then adjust the intra-driver mis-alignment to sub-sample accuracy (limited by the time steps on the DCX).

I maintain this phase coherent accuracy by employing only 1st and 2nd order crossovers between frequency bands, which reduces (or eliminates) any of the objectionable phase rotations, and cross band polarity discontinuities which are described by LeCleach. I predict his method would do little for my system, because I have chosen a topology which does not suffer from the same high order xover problems. (no disease, therefore no need for a cure).

Apart from prodigious SPL and ultra-low IM distortion, this is one of the major benefits in five way design. Each driver is only required to cover a 2 octave power bandwidth, which is generally much narrower than the devices broader capability for bandwidth, power handling, and low-distortion frequency range. This means that 1st order crossovers, which generally place higher demands on out of band fidelity and power handling, are entirely permissible in a high-power five way design, employing ruggedpro-driver transducers.

I plan to take this concept to the technical limit (later this month) and test an xover implementation using all 1st order xovers, top to bottom. As extra LF power handling protection for the compression drivers, I will be using the method proposed by T. Sandrik which combines 1st order phase response, with 2nd order (ultimate) roll off. Sandriks method consists of using 1st order crossover slopes at the xover point, and adding an additional 1st order pole at 1 octave spacing on either side of the xover frequency. The phase anomalies associated with the additional poles are attenuated by the primary filters, and the summed phase response shows phase anomalies which are only a couple of degrees away from perfect.

This is the method that I had used 5 years ago when I was forced (by circumstance) to build a two way system with passive filters, although I don't advocate the use of passive filters for HiFi speakers, the results of the pseudo 2nd order implementation worked extremely well. (Good power handling with no audible or measurable phase distortion, have a look at the accompanying slides)

Jack Bouska

[Ian Mackenzie]

So are your drivers going to be in phase (polarity or 180 degrees out of phase)?

The earlier Dynaudios worked like your proposed idea and possibly the Duntech's, but they were true 1st order designs.


Yet another way, opted by Peter Garde (JBL 4430) was to shift the driver crossover co ordinates by rotating the L/C constrants to enable both drivers to be in phase at the crossover point and sum flat. This approach offers minimal group delay. I read about it in an AES article while sitting in the loo in the State Library once. Neville Thiele also has a new crossover design worth considering if you can understand his paper.

[jack_bouska]

So are your drivers going to be in phase (polarity or 180 degrees out of phase)?

All drivers should be wired such that an applied impulse generates a positive acoustic output pressure.

On most transducers, this means connecting the red terminal to the amplifier positive output, but on JBL's it means connecting the black terminal to the amplifier positive output. To avoid confusion, in the following discussion, I will refer to both the black post on JBL's, and the red terminal on other drivers (such as the TAD), as the device positive terminal

Referring to the above schematics, the JBL 2123, and TAD 2002 drivers are represented respectively by the 4 ohm R1 and 8 ohm R2 resistors. (I have omitted all driver reactance for purpose of clarity).

You can see that in both schematics shown, the positive terminal of the transducers should be connected to L1 and C1, which are on the positive side of the signal source (amp).

Polarity flipping of a driver isa "trick" used to fix a phase-induced response anomaly in 2nd order Butterworth filters, but is not needed for the Sandrik method, because the design is not a conventional 2nd order, but rather more like a Linkwitz Reily (cascaded 1st order), with the additional twist of frequency displaced (rather than co-incident) poles.

The basic Sandrik circuit is wired exactly as a first order xover, and I guarantee that when the additional poles are added one octave away (on one side, or the other, or both sides of the xover frequency), you will not be able to hear any difference, other than beneficial effects due to an increase in power handling, or improved out-of-band distortion suppression.

Jack Bouska

[jack_bouska]

So are your drivers going to be in phase (polarity or 180 degrees out of phase)? The earlier Dynaudios worked like your proposed idea and possibly the Duntech's, but they were true 1st order designs.

Thanks again to Ian for raising the question on polarity, and 1st order slopes, It got me to re-check my system, and try out a few more ideas.

Following Ian's posting. I took the opportunity a couple of evenings ago this week, to re-visit the polarity, crossover slopes, and gain settings on my crossovers. In particular, I compared an implementation of true 1st order (6dB/octave) slopes at all four crossover points, against a high order (48dB/octave) Linkwitz-Riely crossover alignment, at the upper two crossover points, between midrange, and my two compression drivers. Both the 1st order, and the 8th order LR require all drivers to be wired in-phase. The first order xover has no phase distortion, but requires high power handling, low distortion, well behaved transducers, and relatively high crossover points, which, fortunately, my system employs. (but I still dare not turn it up anywhere near max volume). The second implementation uses 8th order LR crossovers between the midrange and the big compression driver, and also between the big, and small compression drivers. This has the best power handling performance (I can take the system up to near clipping if I so desire), and also helps to attenuate out of band problems, like cone break-up on the midrange, and diaphragm break-up on the 4" compression driver.

I have not had much chance to listen to the two systems (time for bed when I got done). But rapid switching (using the compare button on the DCX), indicated that the majority of sonic difference is related to the variation in frequency response, rather than phase variation. In other words, the differences were not subtle, and my preference went toward the 8th order slopes, because the 48dB/oct rolloff restricted the interaction/overlap between drivers better than the 1st order. Maybe I can tweak the xover point and response tailoring a little more to restrict the overlap zone in the 1st order xovers from producing a response bump around xover. (which I couldn't cure with simple inter-driver gain settings.)

The first image in this post shows a graphical comparison of the frequency response for the 1st order xover (top graph, both channels shown), and the 8th order slopes (bottom graph). Xover points are indicated with vertical arrows. The biggest effect/problem is at the 1khz xover point, where the 2123 and 2441 seem to sum constructively on the 1st order, and destructively on the 8th order.

Additionally, I could not (easily) implement the Sandrik method, as the Behringer only allows cascaded slopes using the rather cumbersome EQ modules, and the HP and LP shelving options only allow a maximum cut of -15dB. When I applied one of these on either side of the 1st order xover points, I started running into headroom problems in attempting to set output gains between drivers. When I build my next set of dedicated amps, for the compression drivers, I will include a 1st order pole in each amp, around 150-300Hz, so that the low frequency protection is "built in" to the amplifier, and any "fat finger" maladjustments on my part will not run the risk of driving the tweeters with wide band, or low frequency signal.

One other area that I revisited (while I had the laptop/mic test & measurement kit out) was the low frequency compensation for the Altec. After a simple change (2nd order boost instead of 1st order), I managed to get the low frequency -3 dB bandwidth down to an astonishingly low 5 Hz. This is also the approximate low end of the power bandwidth of the system, meaning that the speakers *could* produce around 110dB spl in the octave between 5-10Hz, with the amplifiers near clipping. Of course, there really is no music content down that low, but just in case musicians start producing/recording subsonic tones, I'm ready.

The 2nd image in this post shows the mandatory "eye candy" waterfall plot illustrating the usual room modes (under damped below 35 Hz), and the prodigious extent of amplitude down near the 2Hz end of the scale.

The final image shows a low frequency linear scale amplitude plot (generated using a 5s MLS signal, and 400ms analysis window over the impulse response), for both channels. I'm still amazed that the speakers, (and Panasonic mic), have response down that low.

Jack Bouska

[John W]

Jack,
I was wondering if you could share some thoughts on mounting your 2 in horn flush with the cabinet face, like on the Zingali loudspeakers.
Thanks,
John

[jack_bouska]

Jack,
I was wondering if you could share some thoughts on mounting your 2 in horn flush with the cabinet face, like on the Zingali loudspeakers.
Thanks,
John

If done properly, i.e.: designed with the specific mounting arrangement in mind, then either flush with a cabinet face, or hanging in open air, will work equally well. However, the design of an open air horn, or conversely, the design of both the cabinet and horn in a flush mount arrangement, is far from trivial. I will post a more comprehensive answer later this weekend. If you would like to elaborate with some specific questions, then I can compose a better set of answers, rather than trying to cover too broad a topic. - Jack

[John W]

What I am considering is building the top end to a 4-way system. Basically a box with
a 12in 2202H midrange driver running from 290hz to 1.2khz,
a 2in 2445 using the same profile horn described in your initial post from 1.2khz to 10khz,
and a 2403 cats eye from 10khz up.

There would be a separate box housing the woofers.

I want to mount the horn flush with the face of the cabinet, the tweeter would be flush too, and the 2202H protruding the typical 1/4 in or so. One reason I want the horn flush is I may want to experiment with adding a slant plate lens over the top.

[jack_bouska]

What I am considering is building the top end to a 4-way system. I want to mount the horn flush with the face of the cabinet, the tweeter would be flush too, and the 2202H protruding the typical 1/4 in or so. One reason I want the horn flush is I may want to experiment with adding a slant plate lens over the top.

I understand the question better now, and can answer it quickly, while also taking the liberty to make a few comments specific to your design (in a subsequent post).

Your first question is related to the mounting arrangement for the waveguide on the front baffle. The quick answer is simple; with a slight modification, the contour I posted will work just fine when flush mounted on a front baffle.

The 2” throat waveguide contour is a composite of three curves;
1) Oblate spheroid (3-40deg), plus
2) tractrix (40-90deg), plus
3) quarter round radius (90-180deg).

To construct a flush mount version, simply omit the 3rd section, stopping the waveguide flair at the 90 degree point. You need to keep only the heavy colored lines, and omit the “rams horn” circles on either end. This will make the width of the waveguide mouth exactly 20cm (my diagram is erroneously labeled 10cm, and should read 20cm, sorry). You will need to work out the rebate/recess-flush mounting arrangement cabinetry yourself, depending on your carpentry skills and tools.

Experimenting with a slant plate lens sounds interesting, but I anticipate the results to be disappointing. These lens devices were designed to be placed in front of traditional (non-constant directivity) exponential horns, which exhibit strong narrowing of directivity beam width with increasing frequency.

G. Ausburger described the JBL lens’s as having a non-frequency dependant directivity increase, which means that they are best placed in front of narrow beam width devices (i.e. + - 20deg). The +/- 40deg oblate spheroid waveguide that I use behaves as a reasonable CD device across the full 80deg, (axially symmetric) coverage pattern. Placing a slant plate acoustic lens in front of this horn would expand the coverage in the horizontal plane (really wide), and interfere with the coverage in the vertical plane.

Various other aspects, such as internal reflections in the slant plate, and acoustic impedance contrasts at the front and back of the lens may also impart unwelcome tone to an otherwise simple, and relatively reflection free device. If you need a coverage pattern which wider, or asymmetric, I would advise designing a different shaped waveguide (e.g.: elliptical oblate spheroid), rather than using a lens.

A good analogy would be a trip to the optometrists, where a well meaning individual gets a new prescription for glasses. Based on the observation that these new lens’s improve his vision, he decides to buy two pair, and wears one over top of the other. While it’s true that this will alter the viewer’s perspective, it’s unlikely to be for the better.

However, there may be some benefit to employing a “home made” lens in front of the 2403 cat’s eye, but I’ll cover that topic in my next post.

Jack

[John W]

Thanks Jack. Sounds like I need to dust off the lathe and spin up a couple to give it a try.
I appreciate your explanation for the slant plate lens. I wasn’t sure on the exact function of these, but like the way they sound on some other horns I have. Leaving them off would be the logical first step.

[jack_bouska]

Thanks Jack. Sounds like I need to dust off the lathe and spin up a couple to give it a try.
I appreciate your explanation for the slant plate lens. I wasn’t sure on the exact function of these, but like the way they sound on some other horns I have. Leaving them off would be the logical first step.

I have uploaded a ppt file with the waveguide profile, this should make it easier to print off a 1:1 version for your lathe turning.

As promised, more thoughts on your design as a whole:

Generically, your design looks workable, and is similar to the top end of the 1st system described by D. Daniels: http://www.audioheritage.org/html/perspectives/drews-clues/audiophile.htm

First, bear in mind that I have not worked with any of the three drivers that you mention, so I will base my analysis purely on specifications, my own use of similar drivers, and some measurements made by Mr. Widget.

Second, my comments are my own opinion, and recommendations based on my experience and personal bias towards audio system design. You are cautioned to apply your own critical judgment before implementing any well intentioned suggestion, even my own. It’s your system after all, and you will be living with the sonic results.

You may benefit from some additional tweaking of the design specific to the topics of: Driver Sensitivity matching, directivity matching, distortion and optimum crossover points, as well as driver layout and cabinet shape for lowest diffraction and best imaging.

Starting with the 2202, this driver appears to have good sensitivity (47dB@1mW@30ft = ~ 97dB@1W@3ft) however the 2445 and 2403 may need to be padded slightly to balance overall frequency response. Do not calculate driver padding until after applying all required response tailoring in the crossover, as the additional filtering will have significant frequency dependant attenuation, which will reduce the overall need for driver padding.

The 12” driver has approximately a 10” diameter cone, and will start to exhibit directivity narrowing at about 500Hz, narrowing to a beam width of +/- 50deg at 1.2 kHz, where you selected the crossover point. My oblate spheroid horn exhibits fairly wide coverage between 1 to 1.5 kHz, rapidly narrowing to +/- 40 above 1.8 kHz, which implies that the coverage patterns of the two devices through the crossover zone will be reasonably smooth, and both direct and reverberant sound fields can both be balanced.

I don’t know where the 2202 starts cone breakup, but crossing over at 1.2 kHz should provide a reasonable margin, using 2nd order, or steeper crossover.

The 2445 is an industrial strength device. Use of the waveguide crossed over at 1.2 kHz will relieve low frequency related stress on the 2445, which allows the use of a conical or oblate spheroid horn contour. If you wanted to use the horn in the octave below (i.e.: down to 500 or 600 Hz), then I would suggest using a more traditional exponential, or tractrix contour (with a much bigger mouth), to provide better low frequency throat impedance, in order to keep the diaphragm from bottoming. For home use, with the 2202 running up to 1.2 kHz, and the 2445 on an oblate spheroid waveguide crossed over at 1.2 kHz, you should have a wide margin of safety.


The JBL specification sheet for the 2445 shows a detailed graph of the raw driver mounted to a 2” plane wave tube. Inspection of the graph reveals diaphragm break up commencing just under 10 kHz, and continuing for at least a half octave. I have not auditioned the 2445 in a home setting (but have heard lots of them in concert PA’s), however based solely on the graph, I believe you are wise in choosing to employ a dedicated high frequency device for the top octave.

My first suggestion would be to move the crossover point down in frequency to around 7-8 kHz, so that the distortion associated with the diaphragm breakup will be further down-slope from the xover point for better attenuation.

Above 10 kHz, the 2” exit on the 2445 will start to exhibit directivity which is narrower than +/- 40 deg. The oblate spheroid contour on the throat of the waveguide helps to expand the useful CD frequency range, but the diaphragm breakup at 10 kHz and above makes this effect unpredictable. Best to avoid problems (as I do) and move the xover slightly lower in frequency.

I would also suggest using active crossovers, (dsp based, preferably). If you must use passive crossovers, bear in mind that the response of the 2445 on the waveguide will require significant response tailoring to get the frequency flat in your listening room. If you have a pc, soundcard, and Panasonic based microphone, you will need to allocate a day or two to get the response sorted using a dsp based xover. If you are using analogue active xover, then allocate a week to the task. If you are boiling a passive crossover, it might take several weeks to get everything right using soundcard measurements, and tweaking by substituting components. If you plan to adjust the system by ear, then you probably need to allocate a lifetime to the job.

I am unable to locate a specification sheet for the 2403 cat’s eye ring radiator, however Mr. Widget has kindly posted a selection of directivity graphs for this, (and other ring radiator) devices. See: http://www.audioheritage.org/vbulletin/showpost.php?p=60871&postcount=4

The device appears to have reasonable bandwidth down to as low as 2.5 kHz, with probable power bandwidth starting at around 5 kHz and above. A 7 kHz crossover would probably be quite safe from both power handling, and distortion standpoint; however you should apply caution and increase the volume slowly during the initial testing phase if you choose a lower crossover point than your original 10 kHz.

The use of a 7 kHz crossover is also viable from the standpoint of directivity control, as this frequency appears to be the breakpoint from aperture dominated directivity, below 7K) transitioning into horn controlled directivity (above 7k). This would be consistent with a device having approximately a 3” diameter mouth. (I’m curious; can you measure the 2403 mouth and let me know if I’m close?). A 7 kHz crossover would match the directivity of the +/- 40deg oblate spheroid waveguide below, and provide (according to Mr. Widgets measurement), narrowing dispersion above 7k.

I’m not all that keen on the +/- 15 degree coverage pattern above 10k for the cat’s eye, but there are a couple of things you could do to help alleviate this issue:
1) buy a couple of the quadratic diffraction phase grating devices, (as seen in the first picture in Don’s post: http://www.audioheritage.org/vbulletin/showpost.php?p=125026&postcount=1 ) and place them on the rear wall (behind you) directly in line with the path of the sound from the cats eyes. This will help increase the reverberant energy in the room by randomizing the 1st reflection from the HF devices.
2) Design and build your own version of a slant plate lens, and put this in front of the cat’s eyes, instead of the waveguide/2445. A sheet of aluminum, and a pair of tin-snips, plus a vice (for bending the crinkle shape) is all you need to build your own. One long skinny through bolt on either side, with nuts to space the plates will hold the shape and a can of black spray-paint will make the unit presentable. see Ausburgers article on this forum for operation and design information http://www.lansingheritage.org/images/jbl/reference/technical/lens/page01.jpg

My preference for best imaging and frequency response smoothness is to place the drivers in vertical alignment, minimizing the front baffle area around each driver. For an example of vertical alignment, see Mr Widgets design in the following post:
http://www.audioheritage.org/vbulletin/showpost.php?p=61015&postcount=19

All baffle mounted transducers which have wide (180+ deg) directivity will suffer from diffraction related frequency response anomalies due to acoustic impedance changes at the baffle edges. The easiest way to minimize this in your design is to build a trapezoidal shaped front baffle, where the sides tapers such that the excess baffle area is minimized around each individual driver, and each driver has a variation in distance to all four edges of the front baffle. For examples of this design, see the following Forum web pages:
http://www.lansingheritage.org/images/jbl/specs/home-speakers/1990-250ti/page1.jpg
and
http://www.lansingheritage.org/html/jbl/specs/home-speakers/1999-tik.htm

That’s all for now, best of luck and please start a new thread in the DIY section detailing your progress on this design, with pictures and text.

Jack

[John W]

Wow! Thanks for thorough examination. It appears you’ve spent more time researching this than I did, and it is very generous of you to offer these suggestions.

I am still trying to digest all the information and arrive at a final plan. I’ll break things out into a separate post as things progress.
I recall seeing a post about a year ago on some acoustic lenses for the 075 bullet tweeters, but couldn’t find it. Honestly, I will probably try your first suggestion with the diffraction treatment and see where that leads. I will definitely bring the crossover point down to around 7 kHz.
Anyway, here is a link to a detailed drawing of the 2403 that I made some time ago that should answer your questions about the mouth size.
http://audioheritage.org/vbulletin/s...ad.php?p=62755

[Amok]

Jack,

Very interesting design indeed. I have not been here for a while due to a combination of workload, plus bored to death of the same ol' on many speaker forums. Sure looks like I have been missing out . I use a waveguide in my system (DCX controlled also), built a few years back, but nothing quite this extravagant. Mainly to match the directivity of the dipole mids. I can only dream of woodworking equipment like you have, much less skills.
You have taken things quite a bit further. Your output/bandwidth requirements are rather drastically higher .
Have you considered adding a few more (smaller, lower bandwidth) subs around the room http://www.harman.com/wp/pdf/multsubs.pdf to help with modal issues?
We share a great deal philosophically, but not with neighbor proximity LOL. I'm totally with you and Linkwitz with using cylindrical enclosures. I must also agree that the days of the big passive boxes should have ended decades ago...but I digress.
Excellent work with showing via measurements your design work. Bravo

- Amok