/looks for fainting couch
-
WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!
Polar response is not the key
- Thread starter Audiojim
- Start date
/looks for fainting couch
- Thread Starter
- #62
The low order bits are getting lost within the high order bits because the resolution of the measurements is too low.
Time coherence is simply the ratio of the time it takes for the tweeter's sound to get to your ears to the time taken by the woofer. Now if this ratio is 1, it's perfect. The problem is, it varies depending on where you sit. With flat bafle speakers, the highs arrive too soon, causing sibilants to sound unnatural.
No they are not. Laser interferometry and other techniques exist for determining issues with drivers and such:
There is a tendency for audiophiles to try to become speaker designers by worrying about stuff you are talking about. That is not your job. Leave that to the speaker designer. What you need to understand is significance of blind preference testing of speakers and measurements that correlate with that. Get that right and the rest becomes easy since there is a small subset of companies who produce speakers this way. The moment you start to worrying about tweeter material this, and time aligned that, you have fallen in the ditch and there is no helping you.
- Joined
- Nov 6, 2018
- Messages
- 1,449
- Likes
- 4,818
You know, when I came here there were some areas where my knowldedge was good to fair, some where it was poor to non-existent. The greatest benefit you can derive from being here is learning from others. There are real experts on this board and they are generous with both accurate information and insight. You may want to do a bit of reading before hitting all the wrong buttons.
If by "problem" you mean what you hear when listening to music, the answer is no, it is NOT a problem. Room reflections come at your ears with many delays and completely mask and annihilate any sense of "time alignment."
The best peer reviewed research here is from journal of Audio Engineering society by our dynamic duo professors, STANLEY P. LIPSHITZ, MARK POCOCK, AND JOHN VANDERKOOY
On the Audibility of Midrange Phase Distortion in Audio Systems
University of Waterloo, Waterloo, Ontario, Canada N2L 3Gl
So no, don't go for the sales pitch. Make sure nothing else is sacrificed in the process of getting you "time aligned speakers." Often there is a step below tweeters to accomplish that which causes its own reflections and diffraction.
Do you have links? Is the review done blind under controlled conditions? If it's not done blind I wouldn't call it a "good review".
How do you know they're good? Sure you can cite reviews, but unless these reviews are done blind (and preferably with a group of listeners, not just one), I would not accept them as evidence that the speaker is "good".
Measurements are done by measuring the speaker at various angles, yes. This gives you many frequency responses at many different angles. The resulting data set is complex because it is a 3D data set (horizontal angle, vertical angle, frequency → amplitude). Polar response (also called "radiation pattern") at various frequencies is just one way of presenting the resulting data, but it's not the most convenient one, far from it. Polar maps and especially spinoramas present the same data but in a way that's easier to read. There also more exotic ones, such as 3D balloons. You can experiment with various ways of visualizing such data sets using software such as CLF viewer or the GedLee Polar Map app.
Up until now I've just assumed that by "polar response" you just meant "off-axis response" (because it's the same data presented in different ways), but now I suspect that you are confused even about that basic concept. If you read that book I mentioned many times, that would clear things up for you.
This is acoustics 101, and your statement is quite frankly borderline insulting. This is such a basic fact that I would expect literally everyone on this thread to understand such concepts.
And no, you will not get a flat response if there are resonances inside the box that the speaker designer did not compensate for, precisely because of interference as you mention. Therefore it will be seen on frequency response measurements. I mean, think about it: you are literally describing a phenomenon that "cancel some frequencies and add at others". The entire point of a frequency response measurement is to show precisely this! So yes, once again, measurements will show the problem. What's your point?
Typical frequency response measurements have much higher resolution than our ears. Even a raw, non-smoothed measurement with only -40 dB noise floor (which is pretty bad) will have an amplitude error of less than 0.1 dB, which is better than a frequency response deviation the ear can detect even on its best day.
I am aware of no evidence that the timing issue that you describe (which is basically a phase response issue) is actually audible in real-world scenarios. I agree with amirm above that the delay seems far too small.
However, what can happen with such delays is interference between the woofer and tweeter at crossover frequency, which is a real problem that speaker designers need to deal with. But, once again, this problem will be visible on measurements, especially off-axis measurements. If the measurements look good, it means the speaker designer was successful at taking care of the problem (or that were was no such problem in the first place for that particular design).
Please realize that an Internet forum is not exactly the most efficient way to deliver a crash course on psychoacoustics and loudspeaker perception research. We can do this all day, but really, you would be much better off doing your own research first. For example you could start by reading this book - it's quite accessible and does not require any specialized prior knowledge. However I am starting to suspect, as others already have, that you don't seem to be interested in learning more on this subject - instead you seem more interested in shooting down any answers that do not match your existing preconceptions. This might feel satisfying, but it will not get you any closer to understanding what makes a speaker sound good.
Last edited:
- Joined
- Feb 23, 2016
- Messages
- 20,793
- Likes
- 37,702
Years ago j-j gave me some good advice though he doesn't suffer fools gladly. I had asked questions, and he said more or less, you'll have to do the work. I'm not doing your homework for you. You'll have to learn some of this or take someone's word for it. He then gave a few references worth reading. I think this is good advice for @Audiojim in this case. If he isn't trolling he looks like he is. If he'd read the book being suggested he'd be up to speed to ask useful questions (or more than likely wouldn't have these questions anymore).
Thanks for the reference. As it turns out, I am planning a visit to Magico next.
Reminds me of arguments on forums and email exchanges I have with conspiracists and the "global warming is a hoax" people. Logic, reason and verifiable evidence will not change their minds. Debunk a couple of statements, and your response might well be ignored, and then followed by a whole new somewhat related gish-gallop of bad logic and known misinformation.
For me, the only benefit for me of engaging with such people or following such exchanges is that it can encourage me to update my knowledge, lead to discovering new research, and even add to my understanding of some fundamental aspects of the science. As I watch the initiator of this thread flail about, I have learned a few new things about speakers from some of the knowledgeable regulars - contributors whom I have learned to appreciate here at ASR.
Also, the ability of respondents to remain civil is commendable, because it is often difficult to distinguish true trolls from those who have dug themselves into a hole, and are desperately trying to save face. The male ego can be a fragile and easily bruised entity. I know that from personal experience, because a number of times in my 70+ years on this planet I have unwittingly - or foolishly - dug myself into similar holes, and sometimes carried on well past the point where I should have stopped digging and crawled out and moved on.
Whether or not there are audible differences under your abnormal hypothetical conditions says nothing about whether or not the polar response is relevant under normal conditions.
I get the impression from your subsequent posts that you are somewhat averse to reading books and averse to data and averse to measurements, so here's one possible alternative way to look at the situation which only requires a little bit of reading:
Question 1: Do reflections make an audibly significant contribution to what we hear?
Answer 1: Apparently they do, otherwise people wouldn't bother designing concert halls and recital halls and cathedrals, nor live rooms in recording studios, nor would we use room treatments at home, nor would we sing in the shower.
Q2: How can we tell whether or not a reflection will be beneficial or detrimental?
A2: In general if a reflection sounds like the direct sound, it will be beneficial. If not, it will be detrimental.
Q3: How do we get reflections that "sound like" the direct sound?
A3: Most reflections start out as off-axis energy, so it makes sense to start out with off-axis energy that is spectrally correct or nearly so. The spectral content of off-axis energy is shown by the polar response.
(The above is somewhat oversimplified for the sake of brevity.)
As for whether what's happening off-axis is "the key" or "a key" or not, it depends on how much of the energy present at the listening position is reflected energy. The more reflected energy at the listening position, the more it matters... which is why your hypothetical equalized outdoor listening test does not tell us anything useful about the role of off-axis sound under more realistic circumstances. If you accept the premise that getting the frequency response you hear at the listening position correct is part of "the basics", then to the extent the off-axis sound contributes to said perceived frequency response at the listening position, getting it right is likewise part of "the basics" (regardless of how many or how few manufacturers seem to concur).
That being said, so much of acoustics and psychoacoustics is counter-intuitive that we cannot rely on our intuition and reasoning. If this sort of thing matters to you, then you really do need to read Toole's book even if you disagree with people who have done so. I suggest the Third Edition as the most accessible if you want to learn about a topic without having read all the background information in preceding chapters.
https://www.amazon.com/Sound-Reprod...coding=UTF8&psc=1&refRID=WW31NPM7359BHZFGYCXG
(Ime Toole, Geddes and Griesinger are all excellent sources of relevant information even if they do not always seem to agree, but imo Toole has put the most effort into making an enormous amount of research contiguous, accessible, and engaging to a non-technical audience. His book can literally save you years of stumbling through research papers written by engineers and acousticians for engineers and acousticians.)
Last edited:
Lost where?
For 16bit, if the lowest bit is lost, there's a 6dB error.
If the 2 lowest bits are lost, 12dB.
3 lowest bits, 18dB.
We don't see that level of error in measurements of competent DACs.
In speakers, perhaps, but I can create a 1-bit tone, play it in the room, measure it, with a microphone and not find that it has gotten lost someplace dark and spooky.
1.025kHz 1-bit 16/44.1 undithered sine wave generated in Audacity recorded via UMIK-1 and displayed in REW:
You said the little bits "getting lost within the high order bits because the resolution of the measurements is too low ".
Add another tone at 525Hz using some of those higher order bits.
The single-bit tone at 1.025kHz is not lost.
So, that, to me, a "Practical Scientist" sort of a guy, says, the little bits probably aren't getting lost when the big boys come out to play.
That verification includes the source data, conversion to a USB stream, conversion to an optical stream, switching by a 4:1 switch, a minDSP doing ASRC to 48kHz, optical to the DAC, the DAC, the interconnects, the preamp, more interconnects, the Amplifiers, speaker cables, the speakers, and anything else I forgot to mention.
Oh yeah.
The microphone, the ADC, the conversion back to USB, reception by the PC, the analysis of the data by REW and the display.
Knew I'd forget something, and I did.
"Bits getting lost"
"resolution of the measurements too low"
Fake news.
Last edited:
Any high frequencies within a 7ms window get treated as one note. Also, you can deal with this within crossover design, it doesn’t need a physical offset to be taken care of (that said, my speaker’s, Buggtussel Amygdala, do utilize a physical offset).
Last edited:
DKT88
Active Member
"One must be careful using Einstein quotes from the Internet, many of them are false attributions." -Abraham Lincoln-
Last edited:
I'm sorry. I apologize profusely.
Pretty cool how I overcame the effects of cabinet resonances, driver materials, even crossover parts and (possibly) time coherence on the first try, though.
Carry on!
Last edited:
One thing I thought I'd add about polar response. I usually see it come to the forefront of discussions on here when people are talking about well-engineered speakers.
When you're talking about poorly engineered speakers compared to good ones, you don't really need to talk about polar response, because a simple on-axis frequency response graph will tell you where the garbage is.
But when you start comparing the good stuff against other good stuff, the on-axis response is often so ridiculously flat that you need to look at more in-depth measurements to understand where there may be flaws. So people will shift from focusing on frequency response to talking about the things that are most difficult to engineer perfectly, and polar response/directivity is one of those.
When you're talking about poorly engineered speakers compared to good ones, you don't really need to talk about polar response, because a simple on-axis frequency response graph will tell you where the garbage is.
But when you start comparing the good stuff against other good stuff, the on-axis response is often so ridiculously flat that you need to look at more in-depth measurements to understand where there may be flaws. So people will shift from focusing on frequency response to talking about the things that are most difficult to engineer perfectly, and polar response/directivity is one of those.
About this "time coherence" - assuming the meaning is phase shift and group delay must be within certain limits across the full frequency bandwidth.
Here is the phase response of a speaker, measured in free space:
And the step response:
This speaker is designed as "controlled phase", meaning the phase and group delay is kept within certain limits to assure there are no audible negative consequences of any phase or time-related errors. Across the full bw, the phase shifts in total around 360 degrees, but this has no impact on sound quality, because this phase shift is spread along a very wide bandwidth. The step response is also quite close to perfect. Compared to most speakers, this is quite excellent performance.
Phase is important, but not because phase in itself is audible. It is important because the low frequency driver and the high frequency driver must sum up correctly around the crossover frequency, where both drivers emit sound simultaneously. Any time offset between the drivers makes it more difficult to get this summation correct. In an active speaker this is easy to achieve, in a passive speaker this is done by choosing the right crossover roll-off for each driver, so that the time offset due to physical misalignment is eliminated. The crossover filter in itself induces delay, and this delay depends on the chosen roll-off characteristics. The speaker is not phase-linear, but more than close enough to be well within audible limits.
Now, if we place the speaker in a typical room, and measure at the listening position, the phase looks like this:
Phase shift is so large, it does not even fit inside the scale. We wrap it, so that for each time it shifts around a full 360 degrees, is just starts over again, and we can fit it on the paper:
What happens here, is that the reflected sound from the surfaces inside the room is added to the direct sound from the speaker, and since this reflected sound has a longer path, the resulting sound will be delayed. This causes a phase shift, which shifts the phase at each reflection point. (Above 2K the wrapping causes the phase to look very erratic because the phase is very close to -180 degrees).
Step response still looks nice, though it has changed visibly:
If we compare the phase in-room to free space, it is obvious that the contribution from reflections inside the room is far more important than the phase response of the speaker itself.
But this also relates to radiation pattern, or polar response. Because frequency distribution and level of all those reflections that causes this phase shift, depends on the radiation pattern of the speaker.
Here is the phase response of a speaker, measured in free space:
And the step response:
This speaker is designed as "controlled phase", meaning the phase and group delay is kept within certain limits to assure there are no audible negative consequences of any phase or time-related errors. Across the full bw, the phase shifts in total around 360 degrees, but this has no impact on sound quality, because this phase shift is spread along a very wide bandwidth. The step response is also quite close to perfect. Compared to most speakers, this is quite excellent performance.
Phase is important, but not because phase in itself is audible. It is important because the low frequency driver and the high frequency driver must sum up correctly around the crossover frequency, where both drivers emit sound simultaneously. Any time offset between the drivers makes it more difficult to get this summation correct. In an active speaker this is easy to achieve, in a passive speaker this is done by choosing the right crossover roll-off for each driver, so that the time offset due to physical misalignment is eliminated. The crossover filter in itself induces delay, and this delay depends on the chosen roll-off characteristics. The speaker is not phase-linear, but more than close enough to be well within audible limits.
Now, if we place the speaker in a typical room, and measure at the listening position, the phase looks like this:
Phase shift is so large, it does not even fit inside the scale. We wrap it, so that for each time it shifts around a full 360 degrees, is just starts over again, and we can fit it on the paper:
What happens here, is that the reflected sound from the surfaces inside the room is added to the direct sound from the speaker, and since this reflected sound has a longer path, the resulting sound will be delayed. This causes a phase shift, which shifts the phase at each reflection point. (Above 2K the wrapping causes the phase to look very erratic because the phase is very close to -180 degrees).
Step response still looks nice, though it has changed visibly:
If we compare the phase in-room to free space, it is obvious that the contribution from reflections inside the room is far more important than the phase response of the speaker itself.
But this also relates to radiation pattern, or polar response. Because frequency distribution and level of all those reflections that causes this phase shift, depends on the radiation pattern of the speaker.
My main speakers have di-polar response above 200Hz or so.
Is that a flaw, or a difference?
Oh boy! Phase Response! I got a post for that, too...
My MartinLogans are in my typical room, and this is measured at the listening position (10 feet):
180Hz crossover point disrupts things a bit.
Is that a flaw, or a difference?
Now, if we place the speaker in a typical room, and measure at the listening position, the phase looks like this:
Phase shift is so large, it does not even fit inside the scale. We wrap it, so that for each time it shifts around a full 360 degrees, is just starts over again, and we can fit it on the paper:
What happens here, is that the reflected sound from the surfaces inside the room is added to the direct sound from the speaker, and since this reflected sound has a longer path, the resulting sound will be delayed. This causes a phase shift, which shifts the phase at each reflection point. (Above 2K the wrapping causes the phase to look very erratic because the phase is very close to -180 degrees).
Oh boy! Phase Response! I got a post for that, too...
My MartinLogans are in my typical room, and this is measured at the listening position (10 feet):
180Hz crossover point disrupts things a bit.
Hi. Would you say that for a particular combination of driver sizes place on a baffle of a certain size, there are going to be 'no surprises' whoever is doing the designing? i.e. if you have a 6" woofer and a standard 1" tweeter on a baffle 8" wide, there's only so much difference in polar response you can achieve, whether it's $10 drivers, or $1000 drivers?
Presumably the more expensive drivers may allow you to use shallower crossover slopes, or bring the crossover frequency down in comparison to the cheaper drivers, which will have some influence on it? But basically the possible range of dispersion patterns is baked in through the choice of driver dimensions and baffle size..?
I ask, because it sometimes seems that the same speakers are churned out over and over, but people somehow expect them to sound miraculously different because of the genius who's putting the drivers in the box. I'm not convinced.
Similar threads
