Go Back   Home > Forums > Loudspeakers > Multi-Way

Multi-Way Conventional loudspeakers with crossovers

Please consider donating to help us continue to serve you.

Ads on/off / Custom Title / More PMs / More album space / Advanced printing & mass image saving
Reply
 
Thread Tools Search this Thread
Old 30th May 2007, 10:33 PM   #921
diyAudio Member
 
Lynn Olson's Avatar
 
Join Date: Jul 2004
Location: Northern Colorado
Default Meta-View

Stepping back a few steps and looking at the larger picture, the overall goal of the project is low intrinsic distortion and spacious sound. This was the goal with the Amity, Aurora, and Karna amplifiers, and they were met.

It's not so easy with loudspeakers, which usually have low distortion, lots of headroom, or spacious, wide-open sound. Electrostats have low distortion and spacious, almost distant, sound, but there's no headroom. To my perceptions at least, horns have an up-front sound that is thrilling in one way, but not so good at spatial impression (room size, height, ambience, etc.). As an old-time quadraphonic guy, this is important to me, part of what stereo is all about (not a mono fan at all).

The project as it sits almost splits Magnetar's suggestions down the middle: dipole from 70 Hz to 1.2 kHz, and something like the Radian 745PB (3" diaphragm, 1.4" exit) for the high frequencies. But there's a method to this: if we want constant directivity when transitioning from a 12" driver to a 90-degree horn, we can follow Geddes' example and cross where the diameter of the woofer equals one wavelength. Thus, 1.2 kHz (in practice adjusted for best phase match between LF and HF drivers at the crossover frequency).

I'm still considering a rear horn, as well, on the rear side of the baffle, tucked between the midbass and widerange driver. I don't think it'll need to be identical to the front horn, but made from similar diaphragm materials, at a minimum. If the Radian 745PB is on the front, the Radian 465PB on the back might be a good companion, running at reduced level (-6dB), and with a crossover slope and frequency designed to maintain constant power into a sphere. This can be confirmed by measuring off-axis all the way around the speaker, and more importantly, total power into the room, as measured on a flat surface.

You can see where this is going. Low IM distortion, ample headroom, and constant dipole directivity from 70 Hz to 20 kHz. In the below-70 Hz region, stereo subwoofers are essential for maintaining spatial impression, preferably placed as widely as possible, even at the sides of the listener. It might even be desirable to have four subwoofers, at each corner of the room. This would maximize the left-to-right velocity component, which is responsible for LF spatial impression.
  Reply With Quote
Old 31st May 2007, 04:01 AM   #922
diyAudio Member
 
Join Date: Feb 2004
Location: Texas
Quote:
Originally posted by ScottG

The question I have here is:
Precisely why do you like the sealed over the vented servo? (..for a given freq.)
I have done direct comparison of the same "servo" driver with two different servo feedback circuitries so that they will be flat or either sealed or vented box. I have a toggle switch on these servo boards, which I no longer sell them. My original idea was that people can choose either sealed or vented with a toggle of a switch. Later on I realize that most people already have their mind set when they purchase the kits and adding the toggle switch is just a waste of time. But that method has given me a convenient way to build a vented box tuned to 20hz, and then I can also plug to vent and turn it into aperiodic. In this setup, I can hear how the sealed version (in the case is really an aperiodic) sounded cleaner with a more "coherent" sound vs the vented box which sounded just a bit lacking in definition and less coherent. I have matched the frequency response from 20hz up to 80hz between the two. So frequency response is not an issue. Cone control wise, they have about the same effective driver Qts (which is about 0.15), so cone control is not an issue. The only thing different between the two is one is 2nd order and the other is 4th order and the sealed has completely control of the active radiator vs vented box does not have control on the vent resonance. And phase is another big difference. The sealed version has just a bit less than 90 degrees phase shift at 20hz. The wavelength at 20hz is about 50ft and 90 degrees is 12ft (vs in vented box, it can easily have 180 degrees phase shift) and the group delay is just smaller with the sealed box. My experience has been quite consistent though.
__________________
Servo Subwoofer kits > Rythmik Audio | Articulate bass for the discerning audiophile .:: Brian Ding
  Reply With Quote
Old 31st May 2007, 06:08 AM   #923
diyAudio Member
 
Lynn Olson's Avatar
 
Join Date: Jul 2004
Location: Northern Colorado
Quote:
Originally posted by rythmikaudio

The only thing different between the two is one is 2nd order and the other is 4th order and the sealed has completely control of the active radiator vs vented box does not have control on the vent resonance. And phase is another big difference. The sealed version has just a bit less than 90 degrees phase shift at 20hz. The wavelength at 20hz is about 50ft and 90 degrees is 12ft (vs in vented box, it can easily have 180 degrees phase shift) and the group delay is just smaller with the sealed box. My experience has been quite consistent though.
Brian zeroes in on the key point here. To an old Thiele/Small guy that took a class from Robert Ashley in the Seventies, the whole point of using a vented system was a sharp decrease in excursion in the box-frequency region, and a more compact box for a given value of F3 and efficiency. The downside, and it's a real issue, is more group delay errors from a 4th-order highpass filter, and substantially greater sensitivity to tuning issues (because of the high-order filter).

That class was the first and only place I'd ever seen a physical copy of Richard Small's doctoral thesis - which became the series of AES articles that kicked off the T/S transformation of the industry in 1973. The interesting thing is that Small had a whole chapter on resistive-vent systems that was never published in the AES Journal - they were 3rd-order highpass filters, much to my surprise. At the end of the class, I asked Bob Ashley why he didn't publish that article in the AES Journal, and the answer was unexpected - because the space-efficiency factor (kn) wasn't any better than a conventional closed box, and excursion was similar.

Closed-box systems always suffered from considerably worse distortion (several times) at the bottom of the working range compared to vented systems, which is why T/S theory led to the renaissance of vented systems - they could be designed accurately for the first time, the boom chased out, and most important, had lower distortion. All this with a few nomograms and a HP scientific calculator. That's why closed-box systems, the traditional favorite of the Fifties and Sixties, fell out of favor by the late Seventies.

But with stable, well-designed servo system, the T/S landscape changes once again. Direct sensing allows arbitrary alignments, as well as real-time sensing and compensation of driver nonlinearities. Much of the vented advantage evaporates, since the distortion figures between the two are now comparable, which leaves behind the disadvantages, worsened group delay and issues with vent turbulence and pipe modes.

One thing I never liked about both closed and vented systems was the presence of a response zero around 1 Hz or so; this is caused by box leaks (Ql), and cannot be completely removed, even with really over-the-top sealing-up measures. It's real enough it has to compensated for when building a test box, otherwise the compliance data will be off by a few percent. One minor advantage of a resistive-vent system is that the zero isn't there - actually, it's moved up in frequency, and becomes part of the 3rd-order highpass alignment, so the "leak" works for the design, instead of against it.

I'm hoping the servo system knows how to compensate for the LF zero, and removes it from the alignment. This zero, along with nonlinearities in the spider, is the reason for the nasty in-and-out "breathing" we see in LF drivers.

To see the breathing, all you have to do is apply continuous heavy drive to a mediocre woofer in a vented system, and watch the cone as it gradually creeps forward or back, then cut off the signal and watch the driver spring back to center. I've seen woofers gradually work themselves almost out of the gap if the spider design is bad enough.

A zero in the response, combined with a nonlinearity that is unpredictably excited by it, is obviously undesirable. I'm expecting the servo system senses this condition, particularly the off-center VC, and compensates accordingly.
  Reply With Quote
Old 31st May 2007, 07:09 AM   #924
diyAudio Member
 
Graham Maynard's Avatar
 
Join Date: Dec 2003
Location: UK
Hi Brian,

You write about 90 degree phase shift at 20Hz for closed box, and 180 degree for vented.
Either of these phase shifts will sound 'off' (grotesquely yukky to me, which is why I so dislike ABRs) when compared to the phase linear drive which might be experienced via a low impedance voltage driven large baffle.

It is not possible to electronically equalise a LF characteristic (even via a servo arrangement) and retain both a faithful amplitude response on transients and a linear phase response on steady sines.

I would rather have a naturally falling bass amplitude response (when bench measured via steady sines) than have different (confusing) first and second cycle bass responses, the first transient response being unmodified by driver-system characteristics or EQ as the increasing amplitude response builds up during *waveform reproduction time*, before being followed by a delayed boost which provides the 'flat' response, but which is out of phase (not coherent) with the start of the waveform.

If the boost cannot arise in phase (coherently) with input then the bass driver output timing becomes shifted; compensation for this phase shift wrt a listening position then affects the transient response (coherency). And yes this is quite clearly audible below 50Hz, which is why increasing driver cone area with reducing frequency cannot fail to sound better.


Cheers ......... Graham.
  Reply With Quote
Old 31st May 2007, 08:26 AM   #925
diyAudio Member
 
Lynn Olson's Avatar
 
Join Date: Jul 2004
Location: Northern Colorado
All loudspeakers have at least a 2nd-order highpass characteristic; there's no way around this, no matter what you do. This is what you get with a closed-box monopole (setting aside for now questions about very low-frequency zeroes caused by box leaks).

All other systems, including dipoles, have higher-order highpass-filter characteristics. A dipole has the 2nd-order highpass resulting from the driver's Fs (which is close to, but not quite the same as, the free-air figure), and an 1st-order resulting from the dipole's 1/f rolloff. Net result, a very spread-out 3rd-order highpass.

Resistive-vent monopoles are 3rd-order, if my memory and quick scan of Richard Small's Doctoral thesis was correct. Classical vented systems, servo-ed or not, are 4th-order highpass filters, and I suspect transmission lines are too.

Given that all loudspeakers are intrinsically highpass filters, what can we do to improve step response? High-order highpass filters always exhibit overshoots - it's just a question of "how much" and "how long". Spreading out the filter, either by cascading 1st and 2nd-order characteristic (as with dipoles), is one way to do it - but "undoing" this 1st-order rolloff with corrective EQ gets us right back where we started.

You can also choose a gentler rolloff slope - this is similar to spreading out the highpass filter over a broader frequency range - like a Bessel or Gaussian filter. My first subwoofer at Audionics in 1979 was a 4th-order Bessel alignment, which almost popped out of the "unequalized" portion of a 6th-order vented-equalized Butterworth alignment. No free lunch with the Bessel, though - the distortion is almost as high as a closed box, due to box tuning being quite a bit lower than classical B4 alignments.

My own guess is that when we hear "crisp" or "fast" bass, it's not so much transient-perfect as low distortion, combined with freedom from dynamic mistuning artifacts. That was one of the subtle benefits of the vented-Bessel alignment; it was much less sensitive to dynamic Q and Fs variations than the "standard" vented alignments, and that offset steady-state distortion figures that weren't quite as good.
  Reply With Quote
Old 31st May 2007, 08:38 AM   #926
SunRa is offline SunRa  Romania
diyAudio Member
 
Join Date: Aug 2004
Location: Romania
Hello Lynn,

I know this is a little off-topic, but it's about the subs. You mentioned at the begining of this thread Garry's Pimm subwoofer design. Is this an option to consider anymore?

The specs where quite amsaing if I remember, ruler flat from 20Hz up. I don't know though what the design was (heavy stuffed U-baffle?) so if there are some references on the web, maybe you could point them out...

The design so far looks great. I followed Magnetar's opinions about ultimate dynamics, and I wonder if a double weaveguide on the widerange driver wouldn't enhance this a little (a 120 degrees WG means around 3-4dB I guess), leaving it a dipole in the same time.
Besides that, with a proper weaveguide shape (maybe EnABL treated), I believe edge difractions would be eliminated. Another plus could be less interactions with the room's walls, without having that focused, projector-like image you dislike horns for.

I hope this makes any sense
  Reply With Quote
Old 31st May 2007, 03:40 PM   #927
mige0 is offline mige0  Austria
diyAudio Member
 
mige0's Avatar
 
Join Date: Jan 2005
Location: Austria, at a beautiful place right in the heart of the Alps.
Default sensing the speaker

Hi


Brian, having not played around with speaker feedback amps yet, I wonder why you have chosen a sense coil wound over the voice coil instead of using a separate coil in a separate magnet structure as outlined in your page ( " Even worse, some authors have even suggested using one coil in a dual voice coil driver as the sensing coil. " ) ? Aren't there some speakers that are especially designed for this ?

From any technical point of view to me it seems to be the worst place to get uncontaminated information about the speed of the VC / diaphragm as the magnet field is heavily affected by the current of the VC ? Furthermore you also should copy all the temperature memory effects right to the sense coil unless you don't use constantan or compensate electronically.

Are there any special compensations effects I am not aware of that make the VC as sensing place a preferable decision ?

Greetings
Michael
  Reply With Quote
Old 31st May 2007, 03:53 PM   #928
MBK is offline MBK  Singapore
diyAudio Member
 
Join Date: Apr 2002
Location: Singapore
Quote:
I wonder if a double weaveguide on the widerange driver wouldn't enhance this a little (a 120 degrees WG means around 3-4dB I guess), leaving it a dipole in the same time.
That exact thought has crossed my mind as well. I actually planned to play with this using 3 mm MDF horn side walls of different length and shape, some time far, far, into the future. Here with experimentation one could choose the double horn parameters in such a way that the loading picks up where the dipole rolloff starts (going down in frequency). Another plus is increased separation distance without cavity resonance, due to the horn profile vs. parallel U-walls. Disadvantage is size and the necessary experimentation.
  Reply With Quote
Old 31st May 2007, 06:34 PM   #929
diyAudio Member
 
Join Date: Feb 2004
Location: Texas
Default Re: sensing the speaker

Quote:
Originally posted by mige0
Hi

Brian, having not played around with speaker feedback amps yet, I wonder why you have chosen a sense coil wound over the voice coil instead of using a separate coil in a separate magnet structure as outlined in your page ( " Even worse, some authors have even suggested using one coil in a dual voice coil driver as the sensing coil. " ) ? Aren't there some speakers that are especially designed for this ?

From any technical point of view to me it seems to be the worst place to get uncontaminated information about the speed of the VC / diaphragm as the magnet field is heavily affected by the current of the VC ? Furthermore you also should copy all the temperature memory effects right to the sense coil unless you don't use constantan or compensate electronically.

Are there any special compensations effects I am not aware of that make the VC as sensing place a preferable decision ?

Greetings
Michael
First of all, our patent is not on how the sensing coil should be implemented, rather it is on how the servo feedback and current feedback are used. It is possible to get a sensing coil in a separate gap. However, that introduces an potential failure mechanism. The good thing about sensing coil wound on the same former as the driver coil (and co-center) is so called "self alignement". It is controlling where the force is applied. This is biggest advantage over any other configuration. At high frequency, when there is any distance between the voice coil and sensor, it is going to introduce extra phase shift (because the cone movement can no longer be view as piston motion, any cone break up and such and be a big problem) and it will get worse over time.

The sensing feedback has a input impedance of 1Meg to 51k ohms (frequency dependent and DC coupled, yes our sensing feedback is DC coupled) and the sensing coil resistance is 24ohms. The error introduced by the temperature is -66db (or 1/2000). It is almost neligible.
__________________
Servo Subwoofer kits > Rythmik Audio | Articulate bass for the discerning audiophile .:: Brian Ding
  Reply With Quote
Old 31st May 2007, 06:49 PM   #930
diyAudio Member
 
Join Date: Feb 2004
Location: Texas
Quote:
Originally posted by Graham Maynard
Hi Brian,

You write about 90 degree phase shift at 20Hz for closed box, and 180 degree for vented.
Either of these phase shifts will sound 'off' (grotesquely yukky to me, which is why I so dislike ABRs) when compared to the phase linear drive which might be experienced via a low impedance voltage driven large baffle.

It is not possible to electronically equalise a LF characteristic (even via a servo arrangement) and retain both a faithful amplitude response on transients and a linear phase response on steady sines.

One thing I would like to clarify is that there is no "linear phase" in HP filters. So all the discussion of group delay is just a notion. It is pretty much determined by the conjugate roots of the transfer function and it is always additive. So eventually, two things help. The lower the order, or the lower the Q, the better. One big Q is all it will take to make it worse and there is no way to get it back. I have seen Ilkka's measurement of REL sub on the other forum and it looked like a first order roll-off starting at 60hz. What a sacrifice Our approach is Q=0.5 which is same as two first order group delay added together. That is pretty much the best one can do (and still has -6db at 10hz).

The group delay is merely an indication of how the source will move when the signal gliding on the frequency axis. It is about coherence. It is about how the ambience information will be "correlated" to the source and give us the connection of the sound stage. The absolute phase is really not that important.
__________________
Servo Subwoofer kits > Rythmik Audio | Articulate bass for the discerning audiophile .:: Brian Ding
  Reply With Quote

Reply


Hide this!Advertise here!
Thread Tools Search this Thread
Search this Thread:

Advanced Search

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are Off
Pingbacks are Off
Refbacks are Off



New To Site? Need Help?

All times are GMT. The time now is 04:56 AM.


vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2014 DragonByte Technologies Ltd.
Copyright 1999-2014 diyAudio

Content Relevant URLs by vBSEO 3.3.2