Thank you for the reply Mr. Olson,
Looking at the photos of the actual build didn't appeared to me that there is any phase plug, or phase plug like arrangement. No room for it either...
Probably you'll elucidate this once you get the plans.. You really intrigued me with this midrange solution, I am very curious how this will sound...
edit: anyway, at 105dB, with your Karna's, seems a perfect fit. You'll probably get Carmina Burana in your room
Looking at the photos of the actual build didn't appeared to me that there is any phase plug, or phase plug like arrangement. No room for it either...
Probably you'll elucidate this once you get the plans.. You really intrigued me with this midrange solution, I am very curious how this will sound...
edit: anyway, at 105dB, with your Karna's, seems a perfect fit. You'll probably get Carmina Burana in your room
The Karna's are good for 16 watts at 0.1% distortion in Class A1, and 30 watts at 1% distortion in Class AB2 (no visible transition or charge-storage effects during the A1 to AB2 transition).
Subjectively, thanks to gradual compression/limiting, no hard clipping, and instantaneous overload recovery (not true of RC-coupled tube amps), the Karna sounds like a 60-watt tube amp, or a 100-watt transistor amp. It also helps that the VV/Vaic/Emission Labs tubes and the 6CJ3 damper diodes are capable of 1-amp peaks, which are reflected as 28-amp peaks at the secondary of the output transformer.
The Karna is quite a different animal than Fifties-vintage PP-pentode amplifiers or typical 2A3 or 300B SETs - it's a lot closer to the Western Electric 86A theater amplifier than the other two types.
Keeping the amplifier in the A1 region, that's 117 dB from each speaker. Pushing the amp a bit further, 120 dB. Good reason to always use the built-in "mute" switch on the amplifier when switching cables.
Although I plan to be a good boy and build the OmniTop 12's as directed - with the recommended Eminence Deltalite-II 2512 drivers - I've noticed the T/S parameters, EBP, and rising frequency response are pretty similar to the Altec/GPA 414's I already have. So that'll be something fun to explore.
As for the more detailed specs of the OT12's, it's a good question how they'll do in the time domain. Probably not as good as a closed-box cabinet with large-radius corners - but on the other hand, the OT12's will be several dB more efficient, and with a passive crossover, the efficiency of the bass unit sets the efficiency of the whole speaker.
P.S. The Lowther DX55 measured a lot better than I expected. No surprise about the 4 kHz mechanical crossover, but truly quite remarkable how much output there is above 10 kHz.
Subjectively, thanks to gradual compression/limiting, no hard clipping, and instantaneous overload recovery (not true of RC-coupled tube amps), the Karna sounds like a 60-watt tube amp, or a 100-watt transistor amp. It also helps that the VV/Vaic/Emission Labs tubes and the 6CJ3 damper diodes are capable of 1-amp peaks, which are reflected as 28-amp peaks at the secondary of the output transformer.
The Karna is quite a different animal than Fifties-vintage PP-pentode amplifiers or typical 2A3 or 300B SETs - it's a lot closer to the Western Electric 86A theater amplifier than the other two types.
Keeping the amplifier in the A1 region, that's 117 dB from each speaker. Pushing the amp a bit further, 120 dB. Good reason to always use the built-in "mute" switch on the amplifier when switching cables.
Although I plan to be a good boy and build the OmniTop 12's as directed - with the recommended Eminence Deltalite-II 2512 drivers - I've noticed the T/S parameters, EBP, and rising frequency response are pretty similar to the Altec/GPA 414's I already have. So that'll be something fun to explore.
As for the more detailed specs of the OT12's, it's a good question how they'll do in the time domain. Probably not as good as a closed-box cabinet with large-radius corners - but on the other hand, the OT12's will be several dB more efficient, and with a passive crossover, the efficiency of the bass unit sets the efficiency of the whole speaker.
P.S. The Lowther DX55 measured a lot better than I expected. No surprise about the 4 kHz mechanical crossover, but truly quite remarkable how much output there is above 10 kHz.
Last edited:
Although a bit late, away on holiday, for reference I would like to raise a point about record and recording quality.
Broadcast companys, at least here in Europe, still exel in good live broadcasts. They can be heard on various platforms. But to me in Sweden the most common is still FM Stereo. Here I can listen to lots of quality live broadcasts or live broadcast performance recordings in the classical and jazz familys. POP etc never is live in Radio or TV now.
There is a widespread distribution of recordings between members of the European Broadcasting Union, which means that you can hear performances from all over Europe, well recorded or live.
Swedish Radio, SR, also distributes a qouple of channels via ground-digital-television that posses 256 kbit/s bandwidth. Which means good quality.
However with regard to the Internet, I have studied the Scandinavian countries, and here Radio Norway excells, it puts out 320 kbits/s of jazz and classical material. And that includes live performances, live recordings and ordinary records. All with very good quality. Addess is: www.nrk.no then go to 'nettradio'
What I really meant to say with this post is that a lot of broadcasting companies still hold the flag high and more often they do than recording companies. However of course you can find 1st class classical recordings also.
So I cannot really understand how someone would listen to speakers that cannot reproduce vocals corretly or for that matter the solid sound of symphony orchestra 1st violin strings in full pitch.
Cheers
/Erling
Broadcast companys, at least here in Europe, still exel in good live broadcasts. They can be heard on various platforms. But to me in Sweden the most common is still FM Stereo. Here I can listen to lots of quality live broadcasts or live broadcast performance recordings in the classical and jazz familys. POP etc never is live in Radio or TV now.
There is a widespread distribution of recordings between members of the European Broadcasting Union, which means that you can hear performances from all over Europe, well recorded or live.
Swedish Radio, SR, also distributes a qouple of channels via ground-digital-television that posses 256 kbit/s bandwidth. Which means good quality.
However with regard to the Internet, I have studied the Scandinavian countries, and here Radio Norway excells, it puts out 320 kbits/s of jazz and classical material. And that includes live performances, live recordings and ordinary records. All with very good quality. Addess is: www.nrk.no then go to 'nettradio'
What I really meant to say with this post is that a lot of broadcasting companies still hold the flag high and more often they do than recording companies. However of course you can find 1st class classical recordings also.
So I cannot really understand how someone would listen to speakers that cannot reproduce vocals corretly or for that matter the solid sound of symphony orchestra 1st violin strings in full pitch.
Cheers
/Erling
Hello John,
You'll find in attached file the response and group delay curve comparison of different crossovers (low-pass + high pass):
1) Le Cléac'h crossover (JMLC)
2) Francis Brooke 3rd order quasi Linkwitz number 1 (FB1)
3) Francis Brooke 3rd order quasi Linkwitz number 1 (FB1)
4) your proposed 3rd order quasi Linkwitz (JK)
(Thanks to Francis Brooke for the verification of the quasi-optimal behaviour of those crossovers).
As you can see your proposed quasi Linkwitz is very similar to Francis Brooke's crossover (which are only a little bit better).
So we can admit that your proposed quasi Linkwitz may be called quasi optimal.
Best regards from Paris, France
Jean-Michel Le Cléac'h
You'll find in attached file the response and group delay curve comparison of different crossovers (low-pass + high pass):
1) Le Cléac'h crossover (JMLC)
2) Francis Brooke 3rd order quasi Linkwitz number 1 (FB1)
3) Francis Brooke 3rd order quasi Linkwitz number 1 (FB1)
4) your proposed 3rd order quasi Linkwitz (JK)
(Thanks to Francis Brooke for the verification of the quasi-optimal behaviour of those crossovers).
As you can see your proposed quasi Linkwitz is very similar to Francis Brooke's crossover (which are only a little bit better).
So we can admit that your proposed quasi Linkwitz may be called quasi optimal.
Best regards from Paris, France
Jean-Michel Le Cléac'h
Attachments
The advantage(???) of the approach I proposed is that the crossover sum exactly in phase at the crossover point and the polar response is theoretically symmetrical. The FB crossover sum differently with a slight tilt to the polar response. In any event, the slight tilt or the slight different in amplitude flatness is pretty insignificant for any real speaker system. But I still would object to calling these anything other than what they are. I think referring to them as quasi optimal with regard to transient response is incorrect and I think qualifying the result by saying that the transient response above 4k Hz isn't important is just sweeping the error under the rug. Like I said before, if such a crossover were used at 100 Hz rather than 1k Hz it can not be said that the error above 400 Hz is insignificant. That is not to say that these crossover aren't better than others, but they just aren't TP. Their use is limited to specific conditions.
Hello,
The expression "quasioptimal crossover" was not introduced only to describe any improvement in the transient.
The characteristics of a quasioptimal crossover are:
- less than 1dB or so in the summed response
- loudspeakers operating in phase in the +/-1 octave frequency interval surrounding the common cut-off
(this means also a "in coincidence curve" the nearest possible of the summed response)
- the flatest group delay curve in the frequency range <4kHz inside which "phase distortion" is considered to be audible (in the case it is...) . Generally it is considered that over 4kHz phase distortion is not audible, that's why we only measure (see my spreadsheets) the group delay variation below 4kHz.
For my own crossovers I add another characteristics:
- no more complexity than classical passive crossovers... (no need of gyrators, delay lines...)
Below 4kHz the so defined quasioptimal crossovers will always give better results upon all others polynomial crossovers when considering the mentionned characteristics.
We will be happy to accept any crossover having such characteristics in the small group of "quasioptimal crossovers".
Musical transient are not what many crossover designers say. Generally the transient (defined as the envelop rise of the signal) is made of frequency smaller than the highest frequency in the signal. This is true even for very rapid instruments like cymbals. A flat group delay curve until 20kHz cannot be obtained (until now) without sacrificing 1 or more characteristics used to define a quasioptimal crossover.
Best regards from Paris, France
Jean-Michel Le Cléac'h
The expression "quasioptimal crossover" was not introduced only to describe any improvement in the transient.
The characteristics of a quasioptimal crossover are:
- less than 1dB or so in the summed response
- loudspeakers operating in phase in the +/-1 octave frequency interval surrounding the common cut-off
(this means also a "in coincidence curve" the nearest possible of the summed response)
- the flatest group delay curve in the frequency range <4kHz inside which "phase distortion" is considered to be audible (in the case it is...) . Generally it is considered that over 4kHz phase distortion is not audible, that's why we only measure (see my spreadsheets) the group delay variation below 4kHz.
For my own crossovers I add another characteristics:
- no more complexity than classical passive crossovers... (no need of gyrators, delay lines...)
Below 4kHz the so defined quasioptimal crossovers will always give better results upon all others polynomial crossovers when considering the mentionned characteristics.
We will be happy to accept any crossover having such characteristics in the small group of "quasioptimal crossovers".
Musical transient are not what many crossover designers say. Generally the transient (defined as the envelop rise of the signal) is made of frequency smaller than the highest frequency in the signal. This is true even for very rapid instruments like cymbals. A flat group delay curve until 20kHz cannot be obtained (until now) without sacrificing 1 or more characteristics used to define a quasioptimal crossover.
Best regards from Paris, France
Jean-Michel Le Cléac'h
Last edited:
Those who physically position the axial alignment of a supertweeter might disagree with the 'general consideration' (assumption) then.
Phase distortion within any range of any system IS clearly audible.
Hello Graham,
The result of the misalignment of a supertweeter is a modification of timbers but we cannot call that phase distortion.
Human ability to detect phase (a single ear being considered her) is deeply related to the way the basilar membrane of our cochlea vibrate after a wave receiption.
For high frequency sounds a too small area of the basilar membrane is active and from this result a very poor phase detection of our hearing at HF.
Bestregards from Paris, France
Jean-Michel Le Cléac'h
The result of the misalignment of a supertweeter is a modification of timbers but we cannot call that phase distortion.
Human ability to detect phase (a single ear being considered her) is deeply related to the way the basilar membrane of our cochlea vibrate after a wave receiption.
For high frequency sounds a too small area of the basilar membrane is active and from this result a very poor phase detection of our hearing at HF.
Bestregards from Paris, France
Jean-Michel Le Cléac'h
Well, the phase argument just isn't true. He is phase for a 1k Hz x-o. This is characteristic of the JK, and FB crossovers. Most likely yous as well. The offset of the tweeter leads to rapid phase rotation above the x-o point. The response is in phase only at the x-o point.
An externally hosted image should be here but it was not working when we last tested it.
As for transients..... you require that I accept your definitions of what is or is not important.
As I has said before, these types of crossover are what they are. They are simple to construct. Another version can be constructed using Bessel HP and LP section. Set the corner frequencies of the HP and LP sections so that they crossover ay -6dB. Tweeter is connected with inverted polarity. Offset the tweeter by 0.236 of a wave length. The response will be +0/-1.2 dB and have similar GD and transient characteristics to all the other x-o's discussed to date. And they all compare to the LR2 crossover with the only real advantage being the 3rd order roll off.
Well I started reading this most interesting thread, but at 600+ pages it's quite a read!!.. The above quoted is the very first post of this thread, and I'm wondering if an OB was indeed created by now as stated in this first post, or have things gone in a different direction?
Well, the phase argument just isn't true. He is phase for a 1k Hz x-o. This is characteristic of the JK, and FB crossovers. Most likely yous as well. The offset of the tweeter leads to rapid phase rotation above the x-o point. The response is in phase only at the x-o point.
As for transients..... you require that I accept your definitions of what is or is not important.
As I has said before, these types of crossover are what they are. They are simple to construct. Another version can be constructed using Bessel HP and LP section. Set the corner frequencies of the HP and LP sections so that they crossover ay -6dB. Tweeter is connected with inverted polarity. Offset the tweeter by 0.236 of a wave length. The response will be +0/-1.2 dB and have similar GD and transient characteristics to all the other x-o's discussed to date. And they all compare to the LR2 crossover with the only real advantage being the 3rd order roll off.
John
Wouldn't it be more useful to plot Group Delay through the crossover (actually the whole bandwidth)? The large phase rotations at HF with just a small delay makes it dificult to identify any real issues. In other words the exact same plot that you show but with GD instead of phase.
"you require that I accept your definitions of what is or is not important" - yes this is a problem when we don't all agree on what is important - like CD versus non-CD or almost any other design criteria. For the most part you and I seem to agree with the one acception being the LF source requirements (basically we disgree on its importance, I don't care what they are as long as the room response criteria is met.) I agree with Toole for the most part, except for the importance of first reflections. But Jean-Michel and I are very far appart.
Last edited:
A couple of AES papers could lead one to conclude that generally the opposite is true. E.g., see "On the Audibility of All-Pass Phase in Electroacoustical Transfer Functions", Henrik Møller, Pauli Minnaar, Søren Krarup Olesen, Flemming Christensen, and Jan Plogsties (Journal of the AES 2007 March - Volume 55 Number 3).
Don't understand that. Can you please explain?
Best, Markus
Hello,
When it comes to crossover if someone want to fool another one the most easy method is to discuss phase curves...
As the vast majority of human beings I know since I am born what is time, what is a finite duration, what is one second. Nobody knows intutively what is a phase rotation of 45°. That's why I prefer to show group delay curves as the one previously comparing several crossovers)
http://www.diyaudio.com/forums/attachment.php?attachmentid=142936&d=1254840547
(well then most people on Diyaudio knows that a difference of phase of e.g. 45° at 4000Hz has not the same meaning that a 45° difference of phase at 400Hz ...)
When comparing group delay curves I don't see any superiority of the 3rd order quasi Linkwitz crossover recommanded by John over one of the 2 versions of the Francis Brooke's crossover...
Best regards from Paris, France
Best regards
When it comes to crossover if someone want to fool another one the most easy method is to discuss phase curves...
As the vast majority of human beings I know since I am born what is time, what is a finite duration, what is one second. Nobody knows intutively what is a phase rotation of 45°. That's why I prefer to show group delay curves as the one previously comparing several crossovers)
http://www.diyaudio.com/forums/attachment.php?attachmentid=142936&d=1254840547
(well then most people on Diyaudio knows that a difference of phase of e.g. 45° at 4000Hz has not the same meaning that a 45° difference of phase at 400Hz ...)
When comparing group delay curves I don't see any superiority of the 3rd order quasi Linkwitz crossover recommanded by John over one of the 2 versions of the Francis Brooke's crossover...
Best regards from Paris, France
Best regards
Hello,
I don't have that paper, please give us an idea of the content of the excerpt inside which phase distortion over 4kHz is discussed.
The transient of a rapid instrument like the cymbals look like this:
http://media.freesound.org/data/22/images/22755__FranciscoPadilla__52_Chinese_Cymbal.png
As you can see the duration of the rise of the envelop of the signal (what we can call "attack") is equivalent to dozens or hundreds of periods...
(cymbals with more rapid attack exist though but still what I said about the relation between attack and frequency content is true...)
It doesn't look as a square wave...
Best regards from Paris, France
Jean-Michel Le Cléac'h
I don't have that paper, please give us an idea of the content of the excerpt inside which phase distortion over 4kHz is discussed.
The transient of a rapid instrument like the cymbals look like this:
http://media.freesound.org/data/22/images/22755__FranciscoPadilla__52_Chinese_Cymbal.png
As you can see the duration of the rise of the envelop of the signal (what we can call "attack") is equivalent to dozens or hundreds of periods...
(cymbals with more rapid attack exist though but still what I said about the relation between attack and frequency content is true...)
It doesn't look as a square wave...
Best regards from Paris, France
Jean-Michel Le Cléac'h
I've posted pictures and quick comments on the 2009 RMAF over in the "Clubs and Events" part of the DIYAudio forum. (Apology to the mods - sorry, didn't know the 2009 RMAF belonged over there.)
Anyway, I'll post more pix and comments once the forum lets me (the fourth pic seems to be stuck in the uploader.)
Anyway, I'll post more pix and comments once the forum lets me (the fourth pic seems to be stuck in the uploader.)
An externally hosted image should be here but it was not working when we last tested it.
Here is the GD plot. But I have added a caveat. I have also shown the contribution to the GD from the high pass nature of the woofer alignment which dominates the GD. Again, considering the crossover alone, rather than the complete system, gives a misleading picture. The complete system must be considered.
Looking at the scales, wouldn't it be normal for a transient response to be shown on a linear time scale, and the group delay a log scale?
Here are impedance characteristics of two inductors. How would this effect group delay when used in an XO?
An externally hosted image should be here but it was not working when we last tested it.
