Welcome! Need support, you got it. Or share your ideas and experiences.
Return to Planar Speaker Asylum
71.67.188.239
Really good survey of research on transients in audio and their significance to audio reproduction.
The main point is that we process sound in stages where the location and then size of a sound source are determined in the first 1 ms and subsequent 10-30 ms. Only then is the frequency of the sound determined. By then it has been localized and its dimensions determined. Much of this is phase dependent and having phase differences in the 0.5-6 khz range disrupts the perceptual process.
Contrary to common belief in the acoustics professions, localization and I would speculate textural information as well size, are much faster and far more precise than a pitch perception and can perceive events with timing differences of just 0.01 msec between the speakers. that is a single cycle at the 48khz sampling rate.
Follow Ups:
Too bad the highest frequency attainable with 48k sample rate would be 1/2 that, minus a tiny guard band.
Too much is never enough
nt
I am convinced about handling transients. I'm still not convinced about linear phase. I know the designer of my own speakers(not planars) and they are bass reflex with 4th order crossovers, certainly not a phase linear speaker. Yet they have superb transients including the mid-bass where many box speakers are lousy. From what I've learned it has something to do with the transfer function of each driver and its enclosure and it's high pass/low pass filter(consider the bass roll off as a high pass filter even though it's all passive).
I'll add that I have a strong belief that the problem is not the attack(though that is important, of course) so much as the decay that is critical. It's overhang, ringing after the signal has ceased that muddles the sound on many speakers.
I don't believe it's only because they are my speakers, something I do worry about. I often come home listening to my own software on another system and think I've heard something really good. And then I try it again on my own system and am surprised how poorly the other system handled transients.
I looked at the crossover of the Froy MTM 2 way and it is 1st order on the midrange/bass and 3rd order on the tweeter. The circuits in the midrange crossover are dedicated to flattening the FR and the impedance of the midrange. It is a series notch filter and and impedance adjustment parallel to the mid/bass drivers.
That attention to impedance probably gets the speaker to be in phase in the midbass to the top of the midrange at the 2.2 khz LP. That is most of the critical range and should give you what you experience. If you use the subwoofer at a low enough freq then you will have no phase effects in the critical portion. In a 2 way, even an LR4 has no phase well into the pass band.
In a bandpass the LR4 is truly vicious with the phase, running the entire 360 deg. from the LP to the HP.
It's 4th order acoustically. I'm a personal friend of the designer and if he says it's 4th order, I believe him.
Thanks for that. I wasn't saying that it wasn't so acoustically and electrically combined, just that I see the midwoofers at an electrical 1st order on the schematic.
I was just speculating that in the context of the 250-6000 hz range that seems most critical to transient perception the midwoofers are likely to have low/no phase shift for all but the top octave of the range.
What matters is the total crossover according to my friend and that's the combination of electrical and acoustic characteristics. By the way even though it's 4th order an LR4 crossover properly done is very well damped(Q = 0.49) which is a big factor in low overhang.
You're incorrect with you analysis of the componetry of the Froy crossover. The midbass filter has no (specific) components for impedance adjustment. There's a notch filter consisting of the LC and the CR network forms the 4th order acoustic rolloff in conjunction with L1.
It's incorrect to label the LP and HP as 1st-order and 3rd-order respectively. The electrical networks look that way, but the acoustic design of the Froy is an LR4 design at 2.2khz. All LR4 crossovers have a 360 degree phase rotation at the crossover frequency. It's a defined characteristic of this design.
The bass roll-off would follow the typical vented-box characteristic. Your statement should read "If you use the subwoofer at a 'high' enough freq then you will have 'minimal' phase effects."
There's nothing wrong with the Froy design with respect to phase response. It's as expected.....for what it is.
Cheers,
Dave.
I understand the acoustic and electrical rolloffs should add up to 4th order. Otherwise the MTM arrangement does not work well.
I see that the CR and LCR can be both part of the notch filter, but the low steady impedance of the speaker outside the crossover region and driver resonance seems not to have a high order electrical phase shift. I have not run the values so I don't know. The drivers are that reactive that phase would get to 360?
The MTM arrangement does not require any particular crossover. In fact, the original D'Appolito scheme did not use a 4th-order acoustic network. It was designed as a 3rd-order network to facilitate a less abrupt change in power response. I've seen the MTM arrangement implemented as everything from 1st-order to 4th-order. :)
The MTM arrangement has some disadvantages but the primary advantage outweighs many of those.....IMO. The twin woofers add to a 6db sensitivity increase which generally matches that of many stock tweeters. Thus, you can usually design a crossover network with minimal/no attenuation on the tweeter and have system sensitivity approximately 6db higher than an MT version with the same drivers.
On the Froy design, the CR is not part of the notch filter and the system impedance curve follows that of a typical MTM implemented this way. There is no impedance leveling network employed on the Froy design. If there were you would see an LCR series network connected in shunt with the input terminals.
All LR4 crossovers exhibit a 360 degree phase shift. It has nothing to do with the drivers unless the crossover point is fairly close to the driver native acoustic roll-off point. You can see an excellent example of this with the Froy design.
Cheers,
Dave.
I should have said that it is better executed with a steep XO (acoustic and electric combined) slope and has a narrow vertical sweet spot. My friend with the Focal Utopia has to choose chairs with the right height to match the angle of the speakers.
Thanks for correcting me but I still don't see the 4rth order network in the Froy schematic.
It's 4th order acoustic, not electric. You have to take driver characteristics into account. And if it were 4th order electric in this case it wouldn't be really 4th order. It would be higher order. You need a good computer program to get the transfer function you want. If you just take nominal driver values and do a textbook crossover, you get nothing like what you think you are.
Bohdan has been writing on the topic for years now. It's an excellent tutorial on the subject and an outshoot from his work on SoundEasy.However, unfortunately, you can't achieve "linear-phase" operation with large transducered speakers like Magnepan's. Comb filtering and the large transducer surfaces create a complicated lobing pattern that can't be corrected (to linear phase) with crossover techniques and/or other electronic means.
You're going to have to switch to a speaker that's a much smaller acoustic source. :)
Cheers,
Dave.
Edits: 09/16/14
I think the line sources are much better for this than point sources since they can cancel out top and bottom boundary effects and I know that they are the only way to preserve the height dimension in recordings. Presuming there is something left of the original cues after studio processing and mixing, as Tony Lauck pointed out.
Playing the Neo8 midrange array naked or just with a 1st order HP at 100-150 hz pretty much had no phase through the passband - 250 hz to 8 khz (nearly 10 khz when axis is aimed at the listening seat). I could hear what they are talking about in the article. I miss the magic of that setup but it has practical limits that bug me.
But it is fairly easy to set it up again. I think I will try it overnight if I can't sleep.
you can get deqx to work on magnepans. it's kind of tricky. you have to have them aimed just right and align the measurement mic to the listening chair. elevation of the mic should be half way up the driver and 3 feet away. even then it's a little trial and error. the bg rd50 sounds better to my ears though.
If you can get the result only at a single point in space does it really satisfy the linear phase objective??
With a point source transducer in an anechoic environment the phase response you've shown might be achievable at a variety of microphone positions.
The audibility of phase distortion has been discussed in audio circles for years. There's still no consensus on whether it is or not. :)
Regarding the DEQX: It's difficult to say why that unit has not been more successful. It does what it's advertised to do, but I think this aspect of loudspeaker reproduction is not understood by the vast majority of audiophiles.
Cheers,
Dave.
the mic position to make the filters for the deqx was kind of tricky to get that measurement at the listening position. it shows that the phase correction can extend to the listening position. the measurement at the listening position was even taken with different software just to be sure. i picked the best looking part of the graph, but i think the midrange is the most critical anyways.
Yeah, that's my point. The phase response is only achievable at a single point in space. Once you move your microphone it all falls apart.
You would probably counter with......"isn't the listening position the only point that matters?"
I would counter with......"is it?"
This is a tricky topic and I suspect there's also confusion on the difference between "transient perfect" and "transient response." They're two completely different things.
Cheers,
Dave.
One of the problems may be that many recordings are made with multiple microphones and after these get mixed down there can be a total phase jumble.
Tony Lauck
"Diversity is the law of nature; no two entities in this universe are uniform." - P.R. Sarkar
I'm not sure where the "problems" come in?The test tones, used to test the phase linearity of the speakers, won't be a jumble and if one can straighten out the speakers then at least we will hear the jumble on a recordings the way the producer intended. :-)
The jumble on the recording is what it is. I wouldn't want to "fix" it if I could. That would be like repainting the Mona Lisa.
Tre'
Have Fun and Enjoy the Music
"Still Working the Problem"
Edits: 09/17/14
My point is that the audibility of phase distortion in a loudspeaker assumes that there isn't already gross distortion on the recording. If there is, it will likely mask any phase distortion in the speaker, or worse, make a bad recording sound better, possibly leading to a poor choise of system components or system setup.
Tony Lauck
"Diversity is the law of nature; no two entities in this universe are uniform." - P.R. Sarkar
.
Have Fun and Enjoy the Music
"Still Working the Problem"
I can't believe it hasn't caught on more. Poor marketing?
Wickedly good find, Satie! I'll have to read it carefully when I get home but, at first glance, it makes sense. Tks!
Post a Followup:
FAQ |
Post a Message! |
Forgot Password? |
|
||||||||||||||
|
This post is made possible by the generous support of people like you and our sponsors: