A lot of this comes down to listener acclimation, and eventually, expectation. There's a certain spatial quality to CD speaker systems that is quite different than conventional all-direct-radiator systems or conventional horns - and as Dr. Geddes mentions, getting used to this requires a period of acclimation.
Similarly, high-dynamic-range speakers have a different presentation - and this is independent of CD vs non-CD directivity. Electrostats can sound surprisingly similar to horns and waveguides until the electrostats hit the inevitable wall between 95 to 105 dB, which tells us that the conventional direct-radiators are doing something to the dynamics at levels well below 95 dB.
But there are limits to acclimation, and I suspect this is a matter of individual perception. I haven't warmed to the spatial qualities of CD speakers, despite liking many other things they do. Spatial presentation is as important for me as tonal rendition - this is probably due to the years I put into quadraphonic sound on the Shadow Vector project.
I spent a long time on the Shadow Vector subjectively optimizing the ballistics of the direction-sensing circuits and setting the decoding coefficients of the dynamic matrix so the ambient reverberant field was free of nulls and hotspots. As a result, small deviations in depth perception and any tendency towards "detenting" or "hotspotting" in the reverberant field really jumps out for me. Since very few people have tuned a dynamic-matrix surround decoder, it's not something I'd expect most people to notice, unless they're messing around with Soundfield microphones or similar ambient-capture systems.
Most of the folks I know in Triode-Land go for tonality and dynamics - thus, many varieties of horns are popular. Many of the triode DIYers are trimming the harmonic spectra of their amplifiers to subjectively match the balance of their horns - my guess is that 2nd to 3rd harmonic ratios are used to "warm" or "cool" the sound to taste.
Similarly, high-dynamic-range speakers have a different presentation - and this is independent of CD vs non-CD directivity. Electrostats can sound surprisingly similar to horns and waveguides until the electrostats hit the inevitable wall between 95 to 105 dB, which tells us that the conventional direct-radiators are doing something to the dynamics at levels well below 95 dB.
But there are limits to acclimation, and I suspect this is a matter of individual perception. I haven't warmed to the spatial qualities of CD speakers, despite liking many other things they do. Spatial presentation is as important for me as tonal rendition - this is probably due to the years I put into quadraphonic sound on the Shadow Vector project.
I spent a long time on the Shadow Vector subjectively optimizing the ballistics of the direction-sensing circuits and setting the decoding coefficients of the dynamic matrix so the ambient reverberant field was free of nulls and hotspots. As a result, small deviations in depth perception and any tendency towards "detenting" or "hotspotting" in the reverberant field really jumps out for me. Since very few people have tuned a dynamic-matrix surround decoder, it's not something I'd expect most people to notice, unless they're messing around with Soundfield microphones or similar ambient-capture systems.
Most of the folks I know in Triode-Land go for tonality and dynamics - thus, many varieties of horns are popular. Many of the triode DIYers are trimming the harmonic spectra of their amplifiers to subjectively match the balance of their horns - my guess is that 2nd to 3rd harmonic ratios are used to "warm" or "cool" the sound to taste.
ScottG said:
Correction (..urgg, second time this night):
I don't necessarily disagree that CD is a worthwhile pursuit.
(..should have seen the double negative there..
) 
A few things of "note" with the CD:
1. The dominate (spl) axis never sounds correct, basically you can't listen on the 0 degree axis. This leaves a "toe-in" or "toe-out" configuration. The "toe-in" reduces soundstage width while enhancing center image fill and image specificity. The "toe-out" method does the opposite, BUT it also makes listener movement translate into image shifting.
2. I've not seen a CD design that extends low enough in freq.. To maintain a good freq. balance usually requires a certain distance, otherwise it can be either to "bright" or to "dull" - to a greater extent than your average design.
As far as application - I've not heard a CD design where "tone", image development, and soundstage development were correct (all 3). This isn't to say that it's not possible, but rather that I've not heard it (..and I haven't heard Earl's speakers).
"CSD is not "basically a series of SPL curves at different times"...
this is an erroneous interpretation in my opinion.!"
This is a practical argumantation related to the particular software used only. What counts is the concept. I'm afraid this concept isn't remembered to often.
this is an erroneous interpretation in my opinion.!"
soongsc said:
This is a practical argumantation related to the particular software used only. What counts is the concept. I'm afraid this concept isn't remembered to often.
Lynn Olson said:
Hi,
did You use a level meter to generate that given assumptions? Shadow Vector sounds cool, but what is it? I don't understand what You are after by mentioning Shadow Vector, dynamic matrices etc. that You optimized subjectively. Complete white out metaphorically said.
Regarding 2nd and 3rd HD in triode land You find Yourself in direct contradiction to Mr. Geddes. A discussion on that topic was highly appreciated. I'm absolutele keen on knowing what has to come beyond Your ARIEL.
by
ScottG said:
Hi,
Stereo doesn't privide You with all that 3 items. That is so by first principles. Unfortunately sound engineers mix records using mostly CD designs. May be thats the reason for poor (pure?) sound. They ruin all charms that a non perfect disc may have. In that if You listen to it it is not for the hifi-ish zound but for the music itself.
It would be straight forward to make our own recordings. No CD designs during mix and quality check, no mix due to 2 mic stereo and no quality checks at all. For the sake of ambience we would remember the original sound and in the first place the feel when we were there.
But - as we are just much to laid back for such - we could think stereo over, as an alternative. What does it mean, "tone", image development, and soundstage development. Especially when it is "correct"?
thank You
"1. The dominate (spl) axis never sounds correct, basically you can't listen on the 0 degree axis. This leaves a "toe-in" or "toe-out" configuration. The "toe-in" reduces soundstage width while enhancing center image fill and image specificity. The "toe-out" method does the opposite, BUT it also makes listener movement translate into image shifting."
I find the opposite. It doesn't matter all that much if I am on or off axis. I also find that there seems to be less image shifting and more uniform tonal balance through out the room using a CD type system. You end up with a wider sweet spot.
I also find that the sense of space and depth are driven by the recording. If it's not there you hear it plain as day.
Rob
I find the opposite. It doesn't matter all that much if I am on or off axis. I also find that there seems to be less image shifting and more uniform tonal balance through out the room using a CD type system. You end up with a wider sweet spot.
I also find that the sense of space and depth are driven by the recording. If it's not there you hear it plain as day.
Rob
ScottG said:
Well this is the key. You speak with such absolute authority about things that turn out to be completely wrong. Nothing that you say above about CD is true in a "good" design, so don't generalize them as such. Ask anyone who has heard my speakers if anything that you list above as "characteristics" of CD is true. You will find that none of it is.
gedlee said:
... and this implies that Mr. Geddes' speakers are a reference for the CD paradigma. Which implication turns out to be wishfull thinking.
In some respect the Gedddes approach is quite old fashioned. Example given in using a compression driver and horn like wave guide. Have a glance at elder JBL designs as 4035 etc. As far as I know JBL invented CD but at least commercialized it first about 40 years ago.
Others prefer a wider spacial radiation but still CD, along Genelc, Klein&Hummel Geithain.
The caveat here is not, that Geddes's speakers havent't been listened to first. It is that
"tone", image development, and soundstage development were correct (all 3)
is argueing with categories not yet defined. 'Hope we get the definitions soon.
so long
BudP said:
Sorry Bud, I'm afraid no.
Where my tools were sufficient to do FR measurements down to *very* low levels
http://www.diyaudio.com/forums/showthread.php?postid=1337520#post1337520
– this isn't true for CSD processing with the tools I have available in the range down to –80 / -120dB you might be looking for.
Maybe SoundEasy does the trick – but I don't have it – JohnK might know better if possible at all.
Really would have liked to do you a favour as the trick with "galvanic connecting of the speakers plates" your thread brought up works excellent for my current setup.
Michael
xpert said:
So is the voice coil driven cone. Doesn't make it unworthy of refinement or use.
If you or I were Earl Geddes, I'm not sure we would retain our composure in the face of such superficial and time-wasting 'analysis'.
... "Hope we get the definitions soon." .. Why? So you can suspend creative thought and become an "xpert" in these matters, too?
Russell Dawkins said:
Russell - Agreed
I just ignore it - a common occurance on the web - no point in even considering it.
I would like to appologize for bringing up my designs in the context of a "subjective" discussion. This is quite unlike me and I will try and refrane from it in the future. I think that I had too many Boubons at that point and lost my composure.
Shadow Vector links for non-Google users:
Nutshell page
Shadow Vector PDF
The Shadow Vector quadraphonic decoder was the invention that started my career in high-end audio. It was an early (1973~1975) dynamic-matrix decoder for the popular CBS SQ matrix and a direct predecessor of the Dolby Pro Logic II decoders used in cinemas and home theater systems today. This patent is cited twice by subsequent Yamaha patents and three times by subsequent Dolby Labs patents.
The working prototype was up and running in 1974, and the Audionics crew demo'ed it at the Chicago CES that year. We used Radford electronics and KEF 104 speakers in what would now be called an IEC layout, with the fronts at 60-degree spacing and the rears 80~90 degrees away from the fronts, giving a 220~240 degree arc of sound. CBS and EMI used the then-standard 90-degree spacing for all speakers, to the great detriment of the forward image. (No Center speakers were used back then - this was several years before Dolby Surround.)
We later went on a roadshow to demo the prototype at EMI Labs and BBC Labs - that tour was when I met Laurie Fincham at KEF, and saw how they designed Target Function crossovers on their $150,000 DEC computer, complete with a programmer from a nearby university that wrote the FFT routines. Big difference between the fantastic outlay they had to make then and the simplicity of PC-based programs now.
The dynamic-decoder field was very small and easy to research for patents - there couldn't have been more than 15-20 people working on it worldwide. Scheiber (EV4), CBS Labs (SQ), Willcocks (Tate DES), Sansui (QS Vario-Matrix), and the earliest work at the BBC (Ambisonics and the UHJ matrix). The re-formatting of QS into Dolby Surround was a few years later, and I played a small role in that, providing Wesley Ruggles with the equations to turn a professional QS decoder into what I later discovered was the Dolby Surround decoder. (Wesley paid me $500 and made sure I never heard of Dolby Labs and what this strange new LCRS matrix was going to be used for. This experience was a good introduction to how Hollywood really works.)
It was after returning from our European road-show in 1975 I was unwillingly drafted into speaker design. The "chief designer" for speakers at Audionics had high-tailed it to Seattle with no forwarding address (no fool he), and I got stuck with completing a 4-way monstrosity, based on nothing more than my fading memories of the 2-hour conversation with Laurie Fincham in England. The choice was simple enough: either finish the 4-way thing, or look for another job. Not really my finest hour or my best design. But I did complete it 6 months later, and Audionics sold all of 8 pairs. I went on to design a variety of speakers for Audionics, most of them not all that great in retrospect, but probably decent speakers for the day. The real money-maker was the CC-2 amplifier, designed by Bob Sickler. Audionics sold thousands of those things.
Nutshell page
Shadow Vector PDF
The Shadow Vector quadraphonic decoder was the invention that started my career in high-end audio. It was an early (1973~1975) dynamic-matrix decoder for the popular CBS SQ matrix and a direct predecessor of the Dolby Pro Logic II decoders used in cinemas and home theater systems today. This patent is cited twice by subsequent Yamaha patents and three times by subsequent Dolby Labs patents.
The working prototype was up and running in 1974, and the Audionics crew demo'ed it at the Chicago CES that year. We used Radford electronics and KEF 104 speakers in what would now be called an IEC layout, with the fronts at 60-degree spacing and the rears 80~90 degrees away from the fronts, giving a 220~240 degree arc of sound. CBS and EMI used the then-standard 90-degree spacing for all speakers, to the great detriment of the forward image. (No Center speakers were used back then - this was several years before Dolby Surround.)
We later went on a roadshow to demo the prototype at EMI Labs and BBC Labs - that tour was when I met Laurie Fincham at KEF, and saw how they designed Target Function crossovers on their $150,000 DEC computer, complete with a programmer from a nearby university that wrote the FFT routines. Big difference between the fantastic outlay they had to make then and the simplicity of PC-based programs now.
The dynamic-decoder field was very small and easy to research for patents - there couldn't have been more than 15-20 people working on it worldwide. Scheiber (EV4), CBS Labs (SQ), Willcocks (Tate DES), Sansui (QS Vario-Matrix), and the earliest work at the BBC (Ambisonics and the UHJ matrix). The re-formatting of QS into Dolby Surround was a few years later, and I played a small role in that, providing Wesley Ruggles with the equations to turn a professional QS decoder into what I later discovered was the Dolby Surround decoder. (Wesley paid me $500 and made sure I never heard of Dolby Labs and what this strange new LCRS matrix was going to be used for. This experience was a good introduction to how Hollywood really works.)
It was after returning from our European road-show in 1975 I was unwillingly drafted into speaker design. The "chief designer" for speakers at Audionics had high-tailed it to Seattle with no forwarding address (no fool he), and I got stuck with completing a 4-way monstrosity, based on nothing more than my fading memories of the 2-hour conversation with Laurie Fincham in England. The choice was simple enough: either finish the 4-way thing, or look for another job. Not really my finest hour or my best design. But I did complete it 6 months later, and Audionics sold all of 8 pairs. I went on to design a variety of speakers for Audionics, most of them not all that great in retrospect, but probably decent speakers for the day. The real money-maker was the CC-2 amplifier, designed by Bob Sickler. Audionics sold thousands of those things.
Thank you for your efforts Michael. If you have access to Litz wire you might try the loop of wire scenario on an inexpensive opamp buffered CD player, just for fun. Then you might come join the herd of arm waving , mass hysteria deluded folks here.
http://www.diyaudio.com/forums/showthread.php?s=&postid=1460031#post1460031
And learn how to remove those non linear propagation activities from speakers, that are not frequency response derived, also free of charge. How to alter direct radiators to show the character, lack of compression and clarity of very good CD systems.
Bud
http://www.diyaudio.com/forums/showthread.php?s=&postid=1460031#post1460031
And learn how to remove those non linear propagation activities from speakers, that are not frequency response derived, also free of charge. How to alter direct radiators to show the character, lack of compression and clarity of very good CD systems.
Bud
In some respects, F1 cars are quite old-fashioned. For example, they have four wheels, one engine, and two driven wheels, just like Benz and Ford cars from more than 100 years ago.
Perhaps the more interesting question is how the newer exemplars differ from the old implementations rather than how they are similar?
xpert said:
About what you would expect..
Tone (in the "monitor" sense, not "sound reinforcement sense") - better reproduction of harmonics in amplitude and time.
"Imaging" - more life-like 3D "images" of performers.
"Soundstage" - more life-like 3D "images" of the performance venue.
(real or virtual).
gedlee said:
It *may* be the "key".
What's "completely wrong" - my perception, or my ability to recall it? True, I am fairly convinced of my prior reactions - so yes, there is a *degree* of "absolute authority". As for application to others..
- um No. (..and we have gone over this before, would you like the link?)If you wish give your own subjective response, perhaps those *particulars* that you find are contrary to my own, go ahead - I would welcome it. Additionally, feel free to ask this from others as well - but don't ask me to do it for you.
And one other request - please don't ever demand something from me. Most people (including myself), find this sort of behavior socially unacceptable (..or rude if you prefer).
BTW, if you go back and look at this tangent on CD designs - you might come to the realization that you started it by wanting to expound it's virtues (in an effort to minimize the "less than stellar" CSD you provided - which wasn't a big deal). If you had just managed to stick to the discussion on CSD, or simply have let it "pass", none of this would have occurred.
Hi,
here I disagree. But I come back to this that later.
And this is not true. A CSD is complex frequency response (also known als magnitude and phase response) + a lot of crap due to errors of the underlying DFT. So its even worse than you mentioned.
But why I still think they're useful?
It's because most of us are simple structured, No-PhD human beings, and most of us don't understand group delay. Do you ever tried to teach a non-technician about group delay and he understands the case within one hour?* But show him a CSD, tell him "there is a resonance" and he knows the problem.
But, I have to add, the classical CSD is ********. It lies whenever possible. I saw a few CSDs of a driver with a extreme resonance, where the engineer (yep, a graduate engineer in EE (in german a Dipl.-Ing.)) misinterpreted the graph, because this big, fat resonance was the cause for DFT-errors (you know why you sometimes should use a windows before applying a DFT?), and these DFT-errors apparently lengthened the decay at lower frequencies. I would like to show you this graphs, but I don't have them anymore. Oh, btw, the mentioned engineer is the most-famous DIY speaker builder in Germany. **** happens. I know this, I misinterpreted enough results in my short life.
So, what to use?
Use the burst decay, preferably with the time axis normalized to periods. You should really read S. Linkwitz on this subject, see http://www.linkwitzlab.com/frontiers_2.htm#M
And don't forget to read Citation [4] (the text is by me) Pay me enough money (I also accept beer**) and I translate it to English.
Bye
Baseballbat
* If you ever did, please write down this explanation and mail it me.
** I mean real beer. You know: American beer is like making love in a canoe... F*ing close to water.
Edit: Goddammit, does this forum censor strong language? Since when?
xpert said:
here I disagree. But I come back to this that later.
And this is not true. A CSD is complex frequency response (also known als magnitude and phase response) + a lot of crap due to errors of the underlying DFT. So its even worse than you mentioned.
But why I still think they're useful?
It's because most of us are simple structured, No-PhD human beings, and most of us don't understand group delay. Do you ever tried to teach a non-technician about group delay and he understands the case within one hour?* But show him a CSD, tell him "there is a resonance" and he knows the problem.
But, I have to add, the classical CSD is ********. It lies whenever possible. I saw a few CSDs of a driver with a extreme resonance, where the engineer (yep, a graduate engineer in EE (in german a Dipl.-Ing.)) misinterpreted the graph, because this big, fat resonance was the cause for DFT-errors (you know why you sometimes should use a windows before applying a DFT?), and these DFT-errors apparently lengthened the decay at lower frequencies. I would like to show you this graphs, but I don't have them anymore. Oh, btw, the mentioned engineer is the most-famous DIY speaker builder in Germany. **** happens. I know this, I misinterpreted enough results in my short life.
So, what to use?
Use the burst decay, preferably with the time axis normalized to periods. You should really read S. Linkwitz on this subject, see http://www.linkwitzlab.com/frontiers_2.htm#M
And don't forget to read Citation [4] (the text is by me
) Pay me enough money (I also accept beer**) and I translate it to English. Bye
Baseballbat
* If you ever did, please write down this explanation and mail it me.
** I mean real beer. You know: American beer is like making love in a canoe... F*ing close to water.
Edit: Goddammit, does this forum censor strong language? Since when?