Not really ... it's just that the reality of equalization finally sank in. I spent a fair amount of quality time auditioning the Linkwitz Orions at the last several shows, and the amount of equalization the Orion uses is far outside what I'm willing to do with a passive or active crossover. When you're conserving those very expensive triode watts, throwing any of them away as a result of EQ is not an attractive prospect.
For folks who are in the "watts are watts" camp and are also OK with 8 to 12 op-amps and/or digital equalization, OB speakers have a lot of appeal, removing all of those obnoxious box colorations at a stroke. When a 30-watt amplifier represents the outer limits of triodeville, then 6 to 20 dB of equalization (as used in the Orion) becomes a show-stopper. In power terms, of course, that's 4 to 100 times the power required by a flat-response, non-equalized speaker.
Just squeaking another 60% (2 dB) out of an amplifier like the Karna - an all-triode DHT Class A PP amplifier - would practically double the cost, size, and weight. The 300B's would have to go, replaced by 845's running at a plate voltage of 800 to 1000 volts. The driver section would have to be scaled up proportionately - and so on. Once you leave the land of "radio receiver" tubes and enter the world of transmitter tubes, it's a different set of rules.
Sure, I could give into the Dark Side and come to terms with 200 to 400-watt Class AB transistor amplifiers with sliding-bias schemes marketed as "Class A". But I don't like the way these things sound - OK for home theater, not so OK for listening to music. But that's just me.
I've spent enough time designing amplifiers to become prejudiced against conventional practice in the solid-state world - Class A/B switching transitions in large-geometry power devices with charge-storage "sticking" effects, noisy rectification with very large current pulses flowing through the diode bridge, and high-speed transient overshoots at the summing node of the feedback network when the output transistors switch on and off and also when they saturate.
These problems grow worse when large numbers of devices are paralleled, since they all find different A/B transition points and all saturate at slightly different current and voltage levels. What is a moderate problem with a pair of NPN/PNP transistors (as found in a low-powered amplifier) is much more difficult when there are 16 or 24 output devices on a common heat-sink, all operating at different die temperatures and having different quiescent currents. What was a simple transition between B to A to B and back again becomes a forest of transition points, all drifting thermally as power levels go up and down. Smoothing out the A/B transition with large numbers of mismatched and thermally drifting devices remains a very difficult problem.
Class D is an interesting way to bypass the tedious business of Class AB operation - I expect interesting things in the future, as problems with reactive load stability and RF emission are resolved.
That leaves me with 8 to 30-watt Class A DHT amplifiers, and 20 to 60-watt Class AB pentode amplifiers. I'd like a speaker that operates efficiently from 100 Hz on up. Full-range efficiency (from 40 Hz to 15 kHz) would be even nicer, but there's a size and weight tradeoff there.
The large-format horn goes down to 700 Hz without any trouble, but a straight (non-folded) horn that's flat down to 100 Hz is truly gigantic, so that's not an option. An OB that has a baffle peak around 100 Hz is also very large - these are 12-foot wavelengths, after all. If the baffle is a domestically acceptable size and provides a 500 Hz baffle peak, there's a lot of EQ needed between 100 and 500 Hz - and more importantly, 100 to 500 Hz falls in the power region of the musical spectrum, so the drivers and amplifiers need a lot of headroom. That's the blight of the mini-monitor - not enough headroom in this critical part of the spectrum, so the music sounds scaled-down and "miniaturized".
The only way around the challenge of high efficiency and generous headroom in the 100~500 Hz range, as far as I could tell, was plain old boxes, or exploring a transitional approach of an open-backed lossy box, with perforated sides, and filled with damping material. The tough part of the latter is avoiding mass coupling of the box filling to the driver cone, which would once again drag down efficiency.
Frankly, I was glad when Gary Dahl suggested his much simpler 2.5 two-box solution. I'd already spent nearly two years on the horn side of the project, and didn't want to get bogged down in researching lossy boxes.
I still see lossy boxes as an interesting solution in the 100 to 700 Hz range, and would like to explore them after I get the main system up and running.
Last edited:
Hello,
Read my original question: I don't bother if the double impulse response is from the theorical point of view is min phase or not and that I question if the a "simple" min phase equalization was possible in the case of the dipole".
That's different, and I think that in most cases such min phase equalization is very difficult. Only when the wavelength is large it is possible. (well at higher frequency due to the piston directivity, the rear wave doesn't turn around the baffle edge and in an infinite space it should not interfer with the front wave, BUT in a finite space the rear wave will reflect on walls before to interfer with the front wave and we may question a possiblesimple min phase equalization (even at low frequency).
Your graph with the double pulse says quite the same as what mine sent before:
http://www.diyaudio.com/forums/attachments/multi-way/179198d1278347141-beyond-ariel-dipole_arta.gif
Best regards from Paris, France
Jean-Michel Le CLéac'h
Read my original question: I don't bother if the double impulse response is from the theorical point of view is min phase or not and that I question if the a "simple" min phase equalization was possible in the case of the dipole".
That's different, and I think that in most cases such min phase equalization is very difficult. Only when the wavelength is large it is possible. (well at higher frequency due to the piston directivity, the rear wave doesn't turn around the baffle edge and in an infinite space it should not interfer with the front wave, BUT in a finite space the rear wave will reflect on walls before to interfer with the front wave and we may question a possiblesimple min phase equalization (even at low frequency).
Your graph with the double pulse says quite the same as what mine sent before:
http://www.diyaudio.com/forums/attachments/multi-way/179198d1278347141-beyond-ariel-dipole_arta.gif
Best regards from Paris, France
Jean-Michel Le CLéac'h
Last edited:
It is minimum phase in theroy, in practice (measurements) and is easy to equalize assuming you limit the response to it's useful range, as should be apparent by the exsistance of diple speakers like my NaO II, Nao Mini, NaO Note and the Linkwitz Orion. None of which will show a douple impulse.
I suggest you build a simple dipole and eq the response to an acceptable band pass response and look at it youself. I not interested in pointlessly arguing about what is easily verified with measurement.
A well design dipole system will place the high frequency cut off around the dipole peak, ideally lower. If you want to extend the response higher then the baffle width has to be carefully considered in conjunction with driver directionality to avoid significant nulls which are not easily equalized. But I am repeating myself.
-----
1 msec would place the dipole peak at 500 Hz and the first null at 1k Hz. My rule of thumb is that this would be useful to about 250 to 300 Hz. But, depending on the driver directionality, you could maybe push it further. However, I do not like wide baffle dipoles for other reasons. In a very large room, away from the wall behind the speakers, they might be ok, but when placed 3 or 4 feet from the wall and similarly from the side wall the large baffle blocks the reflected rear wave and and leads to severe coloration. The large baffle forms a resoance cavity with the near walls. Obviusly 2 msec reduced the useful range by an octave. 2 msec would be a baffle nominally 4' in diameter.
Hello,
Let's investigate the unequalized dipole subwoofer.
Imagine a loudspeaker having a perfectly flat response and used in a subwoofer dipole intended to be used with the less possible equalization (because passive equalization at frequency less than 100Hz is not very practicable).
In a first example, let's define at 80Hz the upper limit at -3dB of the unequalized pass-band (the null will be at 107Hz). Now the lower cut-off (at -3dB) will be at 27Hz.
This needs a 3.2 meters long difference of path and the required width of the OB will be (roughly) in the same range. Better to use IMHO an infinite baffle.
With a 2 meters long difference of path low F-3dB = 43Hz high and F-3dB = 127Hz.
With 1 meter long difference of path low F-3dB = 86Hz and high F-3dB = 257Hz.
Now let's imagine a 1ms difference of path (as indicated by John), this is equivalent to 0.344 meter.
the low cut-off at -3dB will be at 250Hz and the high cut-off will be at 749Hz. 40Hz will be at -18dB and 27Hz at -21dB it will be not practicable to equalize passively (this requires a +6dB/octave high pass ).
Best regards from Paris, France
Jean-Michel Le Cléac'h
Let's investigate the unequalized dipole subwoofer.
Imagine a loudspeaker having a perfectly flat response and used in a subwoofer dipole intended to be used with the less possible equalization (because passive equalization at frequency less than 100Hz is not very practicable).
In a first example, let's define at 80Hz the upper limit at -3dB of the unequalized pass-band (the null will be at 107Hz). Now the lower cut-off (at -3dB) will be at 27Hz.
This needs a 3.2 meters long difference of path and the required width of the OB will be (roughly) in the same range. Better to use IMHO an infinite baffle.
With a 2 meters long difference of path low F-3dB = 43Hz high and F-3dB = 127Hz.
With 1 meter long difference of path low F-3dB = 86Hz and high F-3dB = 257Hz.
Now let's imagine a 1ms difference of path (as indicated by John), this is equivalent to 0.344 meter.
the low cut-off at -3dB will be at 250Hz and the high cut-off will be at 749Hz. 40Hz will be at -18dB and 27Hz at -21dB it will be not practicable to equalize passively (this requires a +6dB/octave high pass ).
Best regards from Paris, France
Jean-Michel Le Cléac'h
I've been using 140cm wide baffles for bass for almost 10 years now.
2 normal 15" speakers in each, no eq.
Active crossover.
I have never measured them tho.
Sounds fine to me on electric and acoustic bass and so on.
Not much really deep bass.
I don't know how this fits in with the rest of the discussion either.
2 normal 15" speakers in each, no eq.
Active crossover.
I have never measured them tho.
Sounds fine to me on electric and acoustic bass and so on.
Not much really deep bass.
I don't know how this fits in with the rest of the discussion either.
Last edited:
Member
Joined 2009
Paid Member
Hi Gary,
It's great to see that all this theory is resulting in a build.
It looks as if you are really going to town on this, that 'lifetime' speaker project
(I'm not ready for this myself, I was hoping there might be something simpler coming out of this thread that I, a newbie, could build. I'm still trying to decide between an OB with bi-amped supporting woofer, or a BLH - but that's OT for this thread)
John, as far as everyone who loves dipols wold agree from a practical point of view - I really believe you are wrong when we go into some hairsplitting discussion about the more theoretical part = correctability of delayed signals.
The difference is in the quality not in the quantity - meaning if we "can't measure" does not necessarily mean that the dual impulse (original and delayed / reflected / diffracted) does not happen.
Sure - this discussion would possibly be worth an dedicated thread - as there is some confusion with proponets claiming spakers (including dipoles, transmission iines, horns - posslbly backloaded even, etc.) being "true" min phase - and on the other hand there are proponents that claim that this is not the case - me being in between with the concept of "consecutively min phase behaviour" along the time line for any delayed doublet (echo, reflections, multipath - whatever you naim it).
I find this especially interesting as virtually everyone is relaying on the min phase behaviour be for EQing or "phase shaping" via XO.
So the question "min phase or not" - to me - seems to be quite fundamental - worth to do some hairsplitting here or elewhere to be sure.
Michael
Last edited:
Since the intelligent and educated loudspeaker folks hang around these parts and since there isn't a project...how many of you have fooled around with isobaric over the years to get a smaller box with reasonable extension? And...what were your experiences?
I have been thinking about it recently but before had always discarded the idea because of the added expense. Now I'm thinking that it's a great way to get away with a smaller box from drivers that require huge enclosures. My application and interest strictly applies to sealed enclosures as vented enclosures allow you to go smaller for a given voltage sensitivity than their sealed counterparts anyway.
Thank you,
I have been thinking about it recently but before had always discarded the idea because of the added expense. Now I'm thinking that it's a great way to get away with a smaller box from drivers that require huge enclosures. My application and interest strictly applies to sealed enclosures as vented enclosures allow you to go smaller for a given voltage sensitivity than their sealed counterparts anyway.
Thank you,
I am also interested in this question. I don't quite understand how a driver, with nulls and positive and negative going response to a single tone can be described as minimum phase. I do note that a grouping of non minimum phase behavior can be useful in a gross sense, for simplified models, but I don't see how any driver can be called minimum phase, since there is only one frequency at which it actually acts as a piston.
I have attached a series of questions and answers on the subject from a posting I ran into a number of years ago. So long ago that I no longer know it's provenience. It does however address some of my questions but doesn't answer them to my satisfaction. John, I really would appreciate some teaching on the subject.
Bud
A minimum phase system has all the poles and zeros in side the unit circle in the complex z plane. The minimum phase of a driver can be determined by measuring the response. The the minimum phase can be computed through application of the Hilbert-Bode transformation. This reconstructed minimum phase can then be compared to the measured phase. In general the measure phase is the driver's phase relative to the driver's acoustic center plus an excess linear phase component resulting form the time delay from the driver to the measurement point. If the driver's measured phase can be reduced to the reconstructed minimum phase by removing the excess linear phase associated wit the delay, then the driver behaves as a minimum phase devise. In general this is the case, even well into breakup. At some point the driver may deviate from minimum phase but this is generally well at higher frequency where the driver response has rolled off significantly and is of little consequence.
The recognition that driver behavior is minimum phase is paramount in speaker design. When we shape the acoustic amplitude response to a desired target using a passive or active crossover we know the response will sum correctly to its counter part (HP LR4 summed to LP LR4 = flat response) (if the acoustic centers are correctly aligned). And we know that for another driver (with different raw amplitude response) when the acoustic amplitude response is shaped to the same LP target (using a different filter) it will also sum correctly to the HP response. This is because, given that the response shaping filters are minimum phase, they represents a minimum phase correction to the driver's response to match the target. If the driver were not minimum phase we could not have the expectation that once shaped to the acoustic amplitude target it would correctly sum to the HP section. If not minimum phase, then matching the acoustic amplitude target would not imply anything about phase. With out the phase being related to the amplitude response through the Hilbert-Bode transformation speaker design would be a completely trial and error process.
I really don't understand why this is even an issure for discussion. It's a pretty common level of understanding amoung the DIY community.
The minimum phase behavior can also be gleamed from the last 3 plots at the bottom of my stored energy web page. The two plots above show the amplitude response of two very different driver. The 3rd plot from the bottom shows the CSD of an ideal target response. The last two plots show the CSD of the two different drivers when the amplitude is shaped to the target. It is evident that all three plots show basically the same CSD indicating that both the amplitude and phase of the filtered drivers is the same as the ideal target. (They would have to be the same to have the same impulse, and there fore the same CSD.) The slight differences are because the filtered driver responses do not match the ideal target exactly.
Hi Gary, some of us were "beyond the Ariel" before the Ariel.
Good luck with your speakers.
Unfortunately it doesn't get you there. There is no free lunch. Even with line level EQ, if you boost a frequency range you will be needing what? More power. And if you do a cut-only EQ then you have a lower overall level. So you have to turn up the volume to reach the spl you want. Resulting in - more power.
The line level EQ does avoid some overall passive network losses, but they won't be huge anyway.
Nothing wrong with the line level EQ, but it's not a free lunch.
more thoughts to min phase behaviour .
Regarding OB dipole the analogy of two closely spaced (out of phase) point sources - one behind the other - is not that bad a picture for further thoughts.
Now thinking some time about the sonic experience with the latest trend towards nude dipole speakers - there might be more on than then the mere lack of additional resonating baffles.
With two point sources - set some distance apart - there is a sharp "consecutively min phase behaviour" in that the doublet arrives at a discrete time delay
The time before overlay of the two impulses - the first interval of min phase behaviour - is most sharply seperated form the then following time - which will also be min phase behaving, but in a different manner, making it impossible to equalize both time intervals correctly (seen in the nitpicking super zoom).
More or less the same behaviour will occure from a baffled OB speaker with a relatively wide baffle.
Omitting the baffle completely will - for the most outer parts of the membrane (and surround) leave almost no path length to travel before overlaying.
This actually could be seen to be a good thing as the time interval for "consecutively min phase behaviour" is practically no longer existent. Meaning for those nude dipole speakers we get "some improvement" here with respect to min phase behaviour.
Just some thoughts for the passionate thinkers around here
Michael
Regarding OB dipole the analogy of two closely spaced (out of phase) point sources - one behind the other - is not that bad a picture for further thoughts.
Now thinking some time about the sonic experience with the latest trend towards nude dipole speakers - there might be more on than then the mere lack of additional resonating baffles.
With two point sources - set some distance apart - there is a sharp "consecutively min phase behaviour" in that the doublet arrives at a discrete time delay
The time before overlay of the two impulses - the first interval of min phase behaviour - is most sharply seperated form the then following time - which will also be min phase behaving, but in a different manner, making it impossible to equalize both time intervals correctly (seen in the nitpicking super zoom).
More or less the same behaviour will occure from a baffled OB speaker with a relatively wide baffle.
Omitting the baffle completely will - for the most outer parts of the membrane (and surround) leave almost no path length to travel before overlaying.
This actually could be seen to be a good thing as the time interval for "consecutively min phase behaviour" is practically no longer existent. Meaning for those nude dipole speakers we get "some improvement" here with respect to min phase behaviour.
Just some thoughts for the passionate thinkers around here
Michael
Last edited:
Do the Bottlehead folks still do any shows? I went to one of their shows in Silverdale nine years ago.
I brought my Summas out to a show on Mercer Island last summer, but Silverdale is closer to the home I own in Washington.
The most serious problme with double-chamber woofers in acoustic series is the cavity resonance between the woofers, which creates a very large dip-and-peak at a frequency close to the crossover. If the cavity is filled with damping material, it mass-loads the woofers, depressing efficiency, and also throws off the ideal alignment. (Real-world acoustic absorbers are not easy to characterize in T/S equations, and the closer the filling material gets to the cone of the driver, the more difficult the characterization becomes.)
In the commercial sector, the dip-and-peak are simply ignored, and the reviewers are too clueless to hear and comment about it. This might sound like a nasty crack, but audio reviewers almost never know anything about the technology of what they're reviewing.
True story: About ten years ago, I was at the CES and I overheard the editor of nationally famous audio-reviewing magazine describe the technical ignorance of his staff as "objectivity". He was actually quite proud of it, and the $5,000 suit and the top-of-the-line BMW in the parking lot underscored the point. I might be right, but he was rich, and his magazine was one of the few profit-making ventures in a notoriously unprofitable industry.
The laziness and ignorance of the audio press means that serious design issues - like gross response aberrations - almost never get mentioned in reviews. Although double-chamber woofers are a clever technique to synthesize a new, virtual driver, the problem with cavity resonance is unfortunately insoluble, except by sweeping it under the rug with a very steep crossover or more cost-effectively, ignoring the response deviation and staying friends with the high-profile magazines. From a business perspective, the latter is the smarter choice.
With any luck, you can train the reviewers to start praising the defects, saving money and building a distinctive brand identity at the same time. It's worked for a lot of high-profile manufacturers in the past, and appears to be a successful business model for the industry. If you wonder why $10,000 to $100,000 speakers sound so bizarre and unreal, that's why.
Last edited:
Lynn,
Thank you for the thoughtful response. And from the post that you are responding to, I didn't man to accidentally bag on Gary as he does have a project, it's just not the focus of the thread. I like the exchanges I have had with him over the past year or two and did not mean any offense.
Best,
Chris
Thank you for the thoughtful response. And from the post that you are responding to, I didn't man to accidentally bag on Gary as he does have a project, it's just not the focus of the thread. I like the exchanges I have had with him over the past year or two and did not mean any offense.
Best,
Chris