Beyond the Ariel

I believe the phenomenon that Michael calls CMP is one way of describing the mechanism that causes the measured frequency response of a dipole.

Michael, try repeating your simulation but feed it with an impulse rather than a sine wave. Now perform a FT on the output. I suspect the result will be a classic dipole frequency response curve.

Am I right or wrong?
 
Whilst I'm in a bubble-bursting mood I suppose I'd better deal with what I think is one of the other "CMP" claims, which (without having the energy to trawl through the hyperbole) I think boils down to "I have managed to equalise a null".

I suppose we could talk for a while about minimum phase systems and invertibility, but I suspect there has already been a fair amount of that. The problem of nulls boils down to multiplying by zero. When systems are connected in series, their overall transfer function (frequency response) is the product of their individual transfer functions. If one of the transfer functions is zero somewhere there isn't anything the other transfer function can do about it, so a claim to have equalised a null amounts to having proved that 0 x 1 = 1. The world would be a much stranger place if that were true, so as I think I posted previously if the world doesn't agree with your model, it won't be the world that's wrong.

It is more likely that what has been proven is that 0 + 1 = 1, which while less exciting is at least true. What that means is that somewhere in the miracle equaliser model the system supposedly being equalised and the equaliser are actually connected in parallel rather than in series, with their outputs being added. In the context of a setup where one of those systems represents the transfer function from a loudspeaker to a listener that is rather difficult to achieve, since it would require a path that doesn't go through the loudspeaker - such as another loudspeaker, but I don't think many people would be happy using a loudspeaker to try and equalise a loudspeaker, not least because it wouldn't work. So I'm forced to conclude that the miracle null equaliser doesn't actually exist and the claim is based on a flawed model.
 
Is there any textbook explanation of CMP? Perhaps this would help the discussion significantly. I am interested in knowing about the phenomenon of CMP.

CMP is just something Michael make up. I have said this before and will once more then I'm out of here. What Michael calls a CMP system is really the a system composed of the summation of several systems. Taken separately, each of these individual system would have a steady state response of its own. But taken as a sum, the system response has a steady state response only when both (or all) the components are at steady state.

Now, and Michale perhaps you will like and appreciate this, if the delay between the start up of the additional components of a system is sufficiently long that the first component reaches steady state before the additional components even start, then what we will see in the time response is the approach to steady state of the first component followed by a transition to the steady state of the complete system. But these are not multiple steady states of the complete system.

An example of this is the simple room measurement and quasi anechoic component. When we measure a loud speaker in a reverberant room, if the delay before the first reflection is 5 msec we window the impulse to a length of 5 msec and we perform and FFt of that portion of the total impulse. We say that represents the SS FR of the speaker. It does, at least down to 200 Hz. But this is not the SS FR of the room/speaker system. The SS FR of the room/speaker system is the FFT of the entire impulse response, including all the reflections. The point is that by looking at the initial part of the impulse we are able to get an idea of what the SS FR of one component (the speaker) of the room/speaker system is. This is not one of multiple SS FRs for the system. It is the SS FR of one component of the system, and one which we happen to be interested in.

The same thing applies to gated sine wave measurements. If we can gate the system response so that the speaker component of the system reaches its steady state before the other components of the system enter the picture (reflections etc) then we can see what the speaker is contributing to the system.

What I object to is the implication that the complete system has multiple steady states. It does not. The system has only one SS. However, when divided into individual components each components of the system has its own SS.

Michael is concerned with systems with delays, but this applies to any system composed of multiple sources. A simple two way speaker, for example, has a woofer and a tweeter. The woofer and tweeter each have their own SS response. But neither represents the SS response of the speaker which is the SS response of the sum of the woofer and tweeter response. And certainly there is a delay between the tweeter and woofer response that is typically seen in the system's impulse. If the delay between woofer and tweeter were extremely long, like over 1 msec, then we might be able to window the speaker's impulse and see what the tweeter was doing, at least down to 1 k Hz, and that might represent the SS response of the tweeter, but not the SS response of the speakers.
 
Your analysis is too simplisctic.

Instead if you write
(0 + |e|)*(u + |e|) = 1
where limes(e) -> 0 and limes(u) -> oo
it can be seen that CMP system can be equalised with an equaliser having impulse response h where limes(h) -> 0 and limes(length(h)) -> oo.


- Elias

Whilst I'm in a bubble-bursting mood I suppose I'd better deal with what I think is one of the other "CMP" claims, which (without having the energy to trawl through the hyperbole) I think boils down to "I have managed to equalise a null".

I suppose we could talk for a while about minimum phase systems and invertibility, but I suspect there has already been a fair amount of that. The problem of nulls boils down to multiplying by zero. When systems are connected in series, their overall transfer function (frequency response) is the product of their individual transfer functions. If one of the transfer functions is zero somewhere there isn't anything the other transfer function can do about it, so a claim to have equalised a null amounts to having proved that 0 x 1 = 1. The world would be a much stranger place if that were true, so as I think I posted previously if the world doesn't agree with your model, it won't be the world that's wrong.

It is more likely that what has been proven is that 0 + 1 = 1, which while less exciting is at least true. What that means is that somewhere in the miracle equaliser model the system supposedly being equalised and the equaliser are actually connected in parallel rather than in series, with their outputs being added. In the context of a setup where one of those systems represents the transfer function from a loudspeaker to a listener that is rather difficult to achieve, since it would require a path that doesn't go through the loudspeaker - such as another loudspeaker, but I don't think many people would be happy using a loudspeaker to try and equalise a loudspeaker, not least because it wouldn't work. So I'm forced to conclude that the miracle null equaliser doesn't actually exist and the claim is based on a flawed model.
 
Now, and Michale perhaps you will like and appreciate this,
....


John, as you most certainly know, I basically always appreciate your comments !
:)

Though there is a lot I'd like to hold against your line of thinking and your arguments thereof, I'll better keep my mouth shut for a while, at least regarding CMP - acute oxygen deficit in outer space - need to economize, you know :D :D


Whilst I'm in a bubble-bursting mood ...

I love your mood - much like Don Quixote
;)

Michael


Michael
 
Last edited:
Hi,

Actually I remembered I've dealed with thie earlier in another thread. My ideal row of impulses is a "CMP" system.

That time I pondered what is the "correct" "frequency response" to present.

First picture showing wavelet transform of impulse row using very short wavelet. At time 0 the "frequency response" is flat.
184041d1281793047-beyond-ariel-ideal_reflections_mrwl_1.png



Second picture showing same system but now with a longer wavelet. At time 0 the "frequency response" has comb filtering effects.
184042d1281793066-beyond-ariel-ideal_reflections_mrwl_3.png




- Elias
 

Attachments

  • ideal_reflections_mrwl_1.PNG
    ideal_reflections_mrwl_1.PNG
    44.9 KB · Views: 590
  • ideal_reflections_mrwl_3.PNG
    ideal_reflections_mrwl_3.PNG
    93.1 KB · Views: 636
I believe the phenomenon that Michael calls CMP is one way of describing the mechanism that causes the measured frequency response of a dipole.



Yes, this all started many pages ago when the double impulse of a naked, unequalized dipole was brought up. I pointed out that it was not relevant because the dipole response has to be eq'ed to some band pass response suitable for use as a loudspeaker and once eq'ed to the appropriate band pass response the impulse of the eq'ed dipoles system will be identical to that of the target. Since then the discussion has been a colossal waste of time, mine included.
 
Thanks for the explanation JohnK.
I think a series of systems combined together is in fact one system. Of course any system can be structurally analyzed to determine what part has what effect, but it's generally not necessary to wait for a certain part to reach steady state to do this. It seems inventing a new term like CMP just clouds the issue, things that normally marketing persons do well.

Michael, it seems you have invented a few terms, one of which I seemingly recall is was already trademarked by another company. Are you doing marketing as a proffession?
 
Michael, it seems you have invented a few terms, one of which I seemingly recall is was already trademarked by another company. Are you doing marketing as a proffession?

Great joke, George !!! you opt to buy ???
And I thought, I'm rather in the entertaining biz here. :spin:
 
Last edited:
As it just fits so well what I have lots of fun to read at the moment :
Quite so. There is one gotcha which sometimes can fool you, like tweeter tests with higher power levels involved. There I find power compression visibly influence FR's obtained from logsweep+convolution, depending on the sweep direction. One might precondition the driver to factor out power compression.

- Klaus

consider that preconditioning would need to determine the main thermal behaiour of the speaker at first - as there is no other way to precisely hit thermal steady state (equilibrium) during measurement otherwise.

This is of course of a nitpicking point of view - from a more practical point of view - as you most possibly meant it - quite *any* pre-heating in the same order as the measurement signal will improve the situation.

Michael
Michael, my somewhat more hidden point (not especially addressed to you) was that the impulse from a top-down sweep + convolution on a tweeter with power compression still leads to a valid capture of the stimulus-dependent behaviour. If you play back the same logsweep convolved with that impulse you will get exactly what the microphone saw in the time domain. In other words, the impulse is just the very one needed for a perfect LTI system to then (and only then) faithfully reproduce the original time response when fed with the original stimulus although the original system was sligthly and slowly non-LTI. Faster nonlinear waveform distortion can also be reproduced (with a bit of math involved) when the impulse response contains extractable harmonic impulses in the pre-echos.

Besides preconditioning, compression from DCR rise can be factored out when using current drive (needs preequalizing when you hit impedance preaks).

- Klaus
 
Your analysis is too simplisctic.

Instead if you write
(0 + |e|)*(u + |e|) = 1
where limes(e) -> 0 and limes(u) -> oo
it can be seen that CMP system can be equalised with an equaliser having impulse response h where limes(h) -> 0 and limes(length(h)) -> oo.
Elias, now you're posting gibberish, which is in danger of becoming your trademark. No matter how much sophistry, wishful thinking or self-delusion you multiply that zero by or convolve it with, it will stubbornly remain zero. You're still making the effort though, despite an increasing resumé of gaffs, so I'll be charitable and tell you what you should have said.

If the null in the response was the result of a single delayed copy of the input after time T, an equaliser could be constructed with tap values of (-1)^n, n=0..infinity at intervals of T. The earth-bound among us would look at that infinity and conclude such an equaliser was not going to be much help, given our finite resources. It's success would be rather transient ;). More serious though, especially for our unfortunate listener, is that we're going to have to nail his head to the floor because the slightest change in the delay of that echo would ruin the whole thing. I'm not sure who is worse off, the listener or those of us posting in this thread.
 
Once again Lynn - thanks for the great platform you allow here

Michael

When the cat's away the mice will play (old English proverb). I've been buried in the arcana of 5 to 7-channel home theater systems with the arrival of the Panasonic 58" 3D plasma TV, 3D-capable Panasonic Blu-Ray player, and the latest Comcast/Motorola HDMI-interface cable box (3D-capable also).

Have discovered some truly remarkable HDMI bugs, like the ability to lock out all remote controls until the power is cycled on one of the devices. This is a really weird protocol - DVI with an additional layer of unwanted and bizarre inter-device communication. I understand Hollywood and Congress are the bad guys here, with legally-mandated inter-device authorization protocols that still have a lot of rough edges. HDMI does work most of the time, fortunately.

Will spend today with Karna out of the house - gone shopping with our daughter-in-law and grandson - doing noise-burst auto-measurements of the HT setup. The ancient 2805 Denon receiver uses a crude form of digital parametric EQ, but it should still be interesting seeing just how mismatched the new Center speaker will be. I'm hoping the $3200 Dynaudio SC X center speaker will not need a lot of equalization, but the (admittedly crude) Denon measurements will tell the tale. Over at the AVS Forum, the latest super-duper versions of Audyssey are purportedly capable of sonically acceptable results. Maybe.

The HT world seems to have very different expectations of what music should sound like than I do. Finding an even halfway decent Center speaker turned out to be a major project, and the sound of nearly all HT receivers is truly dreadful. The only one I've warmed up to has been the Marantz AV8003/MM8003 combo - at a mere $5000 price. The Anthem pre/pro and amps, nope. Any Denon, nope. Onkyo, maybe OK, but on the murky side. Pioneer Elite, tipped-up and edgy. Audiophile exotica - no thanks.

It has slowly - very slowly - dawned on me that the HT reviewers have little or no experience listening to unamplified acoustic music, while that's the only thing I'd consider using as a sonic yardstick. HT dreck does explosions, car crashes, and similar adolescent movie fare just fine - but music? Not so much. An iPod Touch & Sennheiser HD 580 phones sounds better than most of this stuff. Very, very thin pickings if you're the kind of person that actually prefers music to the latest Hollywood action movie. The HDMI fiasco has driver most of the specialist manufacturers out of the HT sector, but then again, I don't care for audiophile transistor products anyway, so no loss from my perspective.

As for the now-famous measurements of the TD15M, I did briefly consider an square or diamond array of 9 Vifa 5.5" drivers to get a similar efficiency. But I don't like non-coincident arrivals from large driver arrays - it was a bad enough problem in the Ariel with its compact MTM, and I'd expect a larger array to be worse.

So it came down to subjective and measured preferences with 12 and 15-inch drivers - thus, the TD15M (or Altec/GPA 416) combined with a large-format compression driver and a low-diffraction horn with a smooth falloff on the edge of the dispersion pattern. In some ways, a modern version of the Lansing Iconic studio monitor or the Altec Model 19.
 
Last edited:
Administrator
Joined 2004
Paid Member
Hard to beat a classic. :D

As for HDMI, the cable is cool, the protocol is not. Systems installers and manufacturers are like Extron are already running into huge problems with systems that are only slightly complex. Too many restrictions and bugs. It's a nightmare.
 
I guess I was foolish in expecting high-dollar HT to actually sound good, while mainstream high-end audio doesn't. Gritty generic transistor-amp sound everywhere, audibly and measurably inferior to the 33-year-old Audionics CC-2 amplifier designed by the late Bob Sickler.

This is just inexcusable; good transistor-amplifier design principles have been known since the early Eighties. It's not complicated: adequate and linear current delivery in the driver stage, adequate phase margin with any likely speaker load, a full-power bandwidth at least 3 to 4 times the maximum audio bandwidth, and some attempt to lower rectifier noise in the power supply. I don't see any of that done in the HT field, while we took all of this for granted back at Audionics a third of a century ago.

Either the HT design guys don't know or don't care. I'm appalled to see the art actually go backward over the last 30-plus years. Transistor amps don't have to sound gross; they can actually sound halfway decent if a modicum of care is applied to the design. But it just doesn't seem to be there, with the exception of a handful of products. I'm trying to track down the schematics of the Marantz and Onkyo products, to see what DACs, opamps, and power-amp circuits they are using, but this appears to be Super Secret Eyes Only information, despite the fact that most of these multi-thousand-dollar products are made in a handful of factories in the Shenzen region of China. The Chinese are certainly aware of the schematics, since they're the ones that build them; why the secrecy from the rest of us?
 
Last edited:
That artifact

From John J's comments I take it that this is a suspension or cone issue. It is not as noticeable with the fabric surrounds, although it is still there. It is also most noticeable on the 15" drivers. My guess is it is cone/surround interface issue. Similar to the dip you see around 800-1000hz, on smaller drivers, especially 8" woofers for whatever reason. Or cone flex issue that is mitigated by fabric surrounds but not helped by foam surrounds.

If so it might be interesting to add (temporary) additional to the cone or surround to identify the source.

VAF Research have been doing this on their DC series speakers for more than a decade - the DC-X was, in some ways, solving a similar problem to the question the Ariel answered.


Anyway I haven't seen any big woofers with better IR than the AE drivers so not sure what the fuss is about.

Or one could leave it well enough alone.
 
..and the latest Comcast/Motorola HDMI-interface cable box (3D-capable also).


..Over at the AVS Forum, the latest super-duper versions of Audyssey are purportedly capable of sonically acceptable results. Maybe.
The only one I've warmed up to has been the Marantz AV8003/MM8003 combo - at a mere $5000 price. The Anthem pre/pro and amps, nope. Any Denon, nope. Onkyo, maybe OK, but on the murky side. Pioneer Elite, tipped-up and edgy. Audiophile exotica - no thanks.


Is is 1.3 or 1.4? 1.3 will pass through 3D at 1080i, 1.4 will do 1080P. (..something to look at with the various receivers/processors.)


This one will "only" pass through 3D at 1080i.. (currently).

Emotiva Audio UMC-1 Audio-Video Processor

That in combination with amplifiers of your choice may be a better option. It isn't an Audyssey solution, but I've read opinions where the very best integrated Audyssey stuff wasn't that great. :eek: (and Emotiva's pricing isn't exactly consumer gouging. :) )

There policies aren't bad either:

http://emotiva.com/whybuy.shtm
 
Last edited:
...
Either the HT design guys don't know or don't care. I'm appalled to see the art actually go backward over the last 30-plus years. Transistor amps don't have to sound gross; they can actually sound halfway decent if a modicum of care is applied to the design. But it just doesn't seem to be there, with the exception of a handful of products. I'm trying to track down the schematics of the Marantz and Onkyo products, to see what DACs, opamps, and power-amp circuits they are using, but this appears to be Super Secret Eyes Only information, despite the fact that most of these multi-thousand-dollar products are made in a handful of factories in the Shenzen region of China. The Chinese are certainly aware of the schematics, since they're the ones that build them; why the secrecy from the rest of us?
I think what we are experiencing is a product that meets spec with the lowest recurring and non-recurring cost possible. Even in the software world, we see this happening. Does this mean the days of having schematics in the manual are over? This is the way it seems because everyone is headed towards commonized modules for the consumer market. Some here opened up a Goldmund CD player and found the main circuit board that was pretty much the same as a popular consumer brand. The schematics are probably in the hands of some unkown smaller OEM factory.
 
Member
Joined 2006
Paid Member
This one will "only" pass through 3D at 1080i.. (currently).

Emotiva Audio UMC-1 Audio-Video Processor

(and Emotiva's pricing isn't exactly consumer gouging. :) )

There policies aren't bad either:
Emotiva HQ is just a couple of hills away from my place and I thought I would go over and check 'em out at their upcoming "Emofest". If anyone else is local it would be a good chance to hook up at the show. ;)
Emotiva Audio: Audiophile Quality Multi Channel Amplifiers, Stereo Preamplifiers, Audio/Video Processors, and Award Winning Speaker Systems At Direct Prices

I have not had a chance to see or listen to their gear so it should be interesting. They will have some live musicians at the show - http://compassrecords.com/alison-brown - (it's pretty easy to find musicians around the Nashville area) so comparing "live sound" to their equipment will be interesting. :magnify:

I'm interested if they actually make their gear there or sub it out to you-know-where.
 
Last edited: