Hi hifijim,
Hi soundbloke,
Completely agree with your last two posts.
Hi markbakk,
Often times a small difference in some aspects of a measurement is very audible. This is certainly true of electronic equipment. Experience now comes in to figure out what is important and what is not, it all agrees with physics. The exact same thing holds true for loudspeakers and systems. The other important thing to remember is that the test setup and conditions can make a difference in what you measure. For that reason the conditions and equipment used should be recorded and conditions noted with any deviations noted as well.
I think that's valid. I don't know you will ever be able to keep a clean thread, just try to keep people on-topic.
Hi soundbloke,
Completely agree with your last two posts.
Hi markbakk,
Often times a small difference in some aspects of a measurement is very audible. This is certainly true of electronic equipment. Experience now comes in to figure out what is important and what is not, it all agrees with physics. The exact same thing holds true for loudspeakers and systems. The other important thing to remember is that the test setup and conditions can make a difference in what you measure. For that reason the conditions and equipment used should be recorded and conditions noted with any deviations noted as well.
Such reasoning has good logic but is much less undebatable than it may look: "Solid statistical research of car sales numbers shows unequivocally what is the best car on the market today. Your preference for Civic R (or Tacoma 4x4) does not count, for obvious reasons."
The last thing I intended was to sound pompous or put anyone off getting started on their hobby. But I do think there are established references that deal with matters much better than most of the content here. IMHE most of the content here is from people seeking answers - hence why I ask so many questions.
Can you explain this
Have you tried this? I've certainly tried this in my quest to find a better 'test signal'. Angelo's log chirp ALWAYS comes out better if you use the same length or lengths of signal. You can either use a slow log chirp or average several log chirps of the same total test time but the important thing is to use Angelo's method to analyse it.
Can you show an example measurement(s) with your different test signals? In my experience, if done properly, you should only see differences at LF.
Angelo's sweep is about the only sensible test signal if you want to investigate LTI of treble unit at high level without killing them.
Sure, have a look at the residual error err=mic-filter(rir,1,spk):
Do you see the dark band after the main sweep line? If rir was unbiased, there should have been noise there, same as before and after. If you see the dark band, that means your model overfitted the observation noise.
Regarding tweeters: let's consider ATM tweeter from ADAM. That's how it measures on a tone, producing 80dBSPL:
You can expect lots of distortions if you cross it at about 2500 Hz, right? on Mozart piano concerto it looks like that
... quite a lot. But if you cross it at 5000Hz, the residual becomes practically inaudible:
You can not expect such a difference from sine sweep test, can you?
(pls ignore fridge noise bands)
If you measure RIR with -90dB precision, your relative error is 3e-5. Temperature coeff for copper is 4e-3, per kelvin. You'll need less than 0.01K change to get TI distortions exceeding your precision, correct?
For other questions: They are very good questions but it would be much easier for you to obtain the answers to all of your questions by yourself if you download and install Matlab, than for me to reply to the growing-in-difficulty never-ending stream of the products of your curiosity. You will need signal processing and audio, home edition. current price is 149+N*49. For a beach bum like you, that's peanuts. Then the quality of your question will rise above clouds for you are a sufficiently mathematically literate man.
Do you see the dark band after the main sweep line? If rir was unbiased, there should have been noise there, same as before and after. If you see the dark band, that means your model overfitted the observation noise.
Regarding tweeters: let's consider ATM tweeter from ADAM. That's how it measures on a tone, producing 80dBSPL:
You can expect lots of distortions if you cross it at about 2500 Hz, right? on Mozart piano concerto it looks like that
... quite a lot. But if you cross it at 5000Hz, the residual becomes practically inaudible:
You can not expect such a difference from sine sweep test, can you?
(pls ignore fridge noise bands)
If you measure RIR with -90dB precision, your relative error is 3e-5. Temperature coeff for copper is 4e-3, per kelvin. You'll need less than 0.01K change to get TI distortions exceeding your precision, correct?
For other questions: They are very good questions but it would be much easier for you to obtain the answers to all of your questions by yourself if you download and install Matlab, than for me to reply to the growing-in-difficulty never-ending stream of the products of your curiosity. You will need signal processing and audio, home edition. current price is 149+N*49. For a beach bum like you, that's peanuts. Then the quality of your question will rise above clouds for you are a sufficiently mathematically literate man.
I don't understand the quantum jump from electronics to speakers. We all agree that correct design and execution and then reliability and mantain in time of the original product is no question that a system of procedures to ensure QC must incorporate some standard scales that are universal.
With loudspeakers your model fails
The extension of the model is pretension.
There's air in between the two membranes, speaker and ear, and waves are elastic.
You wrote: ‘incredibly small changes in amplitude and phase response, supposedly quite ‘negligible’ according to objectivist ideas, can actually have large audible effects.’
I fail to see how post and pre echo’s relate to that. Furthermore, I thought we were discussing nonlinear distortions. If you please, point out the research that lead to the conclusion in your statement.
There is a difference between preference and audibility. DBLT should show people actually can discern. Now an audio enthousiast will probably tell you that, failing it, such a test is compromised. But I just can’t help thinking otherwise.
Hi picowallspeaker,
I'm only giving an example of how sometimes a small measured difference can be more important for subjective sound quality - that's all. This is all I am saying, period.
That matters not. I've always said that we reproduce the same pressure variations near the observer, and that is the job done. This accounts for air not being a solid. It is also how we captured the original acoustic sound. It would matter if I had stated that if we measured the actual cone movement to be a replica, but not once has anyone suggested this.
Firstly, the Wigner Distribution relates to linear system modelling/analysis. Why then is the different audibility of pre- and post-responses in linear systems not related??? Low energy pre-echo can be more audible than the equivalent post-echo. So it is instead a good example of where measurements and analyses fall short of correlating well with subjective impressions.
Secondly, nowhere was this thread limited to non-linear modelling/analysis. It merely provides another example of where measurements and analyses fall short of correlating well with subjective impressions.
Last edited:
I have also provided in this thread an explanation of why blind testing and your thinking can be flawed. I am no defender of self-proclaimed "golden ears", but I am understanding of the limitations of testing and analyses where humans are concerned.
This thread was limited to loudspeaker drivers. I never ever have seen electrodynamic or other traditional transducers suffering from pre-echoing. Sorry for my ignorance.
And for testing, I’m more into the proof of the pudding. If literally no one can taste the difference, there might be one, but is it of any value other than academic?
And for testing, I’m more into the proof of the pudding. If literally no one can taste the difference, there might be one, but is it of any value other than academic?
I never implied there were. I merely supplied an obvious example that supported Michael Gerzon's statement re linear measures.
Hi markbakk,
Well, my loudspeaker designs improved rapidly due to the fact that I started testing myself and using test data from driver manufacturers. So I guess I got the proof very early on. Thank goodness!
I just sparked up a pair of "proof of concept" boxes I designed in about 1986. They were not standard enclosures and extended the bass down a lot. These used cheap drivers (Philips) just to prove the design was worth considering. We built production using Peerless drivers of the time, modified Philips ADF3000/8 (I think) crossover. I was surprised at how good they sounded loaded with cheap speakers. Today's good speakers blow them away (but not all). There is no way these would have sounded as good without measurements. These boxes sat in a room for well over a decade unused. I forgot how good the design was. I should have loaded them up with the production drivers and retuned the boxes - stupid me. The mids and highs are not up to snuff with these ones.
Well, my loudspeaker designs improved rapidly due to the fact that I started testing myself and using test data from driver manufacturers. So I guess I got the proof very early on. Thank goodness!
I just sparked up a pair of "proof of concept" boxes I designed in about 1986. They were not standard enclosures and extended the bass down a lot. These used cheap drivers (Philips) just to prove the design was worth considering. We built production using Peerless drivers of the time, modified Philips ADF3000/8 (I think) crossover. I was surprised at how good they sounded loaded with cheap speakers. Today's good speakers blow them away (but not all). There is no way these would have sounded as good without measurements. These boxes sat in a room for well over a decade unused. I forgot how good the design was. I should have loaded them up with the production drivers and retuned the boxes - stupid me. The mids and highs are not up to snuff with these ones.
I am lost. Due to my fault, my loudspeakers spent half a year in poor conditions. When I brought them back, the midrange turned out to have deteriorated beyond listenable. The speakers are old, there is no direct replacement. How do I find out a nice 3-4" midrange, 250...2k5 Hz? I ordered a bunch, started to test them, and found that I had more questions than answers. Then I ran into this thread and found that I am not alone.
It appears that the thread was started and supported by people saying that they could taste the difference between drivers and that this difference was critical for them, but the existing "standard" tests do not shed any light on these differences. These tests are reasonably good for discriminating against poor drivers but ... orthogonal to sound quality when we aim for perfection. Thus:
Could we interpret the existing tests differently?
Could we test differently?
What is it exactly we need to test for?
What are the test vectors that would focus on the high sound quality?
and so on...
Hi mikets42,
First, they have to be about the same impedance. They also have to physically fit or be close to the same size. The same style (cone, dome) would give similar sound characteristics.
Then you look at efficiency specs, resonant frequency (your crossover should be at least 2 x this so you use the driver within its range). You would choose one that has a smooth frequency response and this would normally have lower distortion. You may have no choice but to have an upgrade in performance, and you must do both speakers.
So, what speakers are they? A picture of front and back would help. If it's something like 5" cone with metal sealed back, you aren't looking at an expensive speaker. Someone here may be familiar with them and can provide guidance.
First, they have to be about the same impedance. They also have to physically fit or be close to the same size. The same style (cone, dome) would give similar sound characteristics.
Then you look at efficiency specs, resonant frequency (your crossover should be at least 2 x this so you use the driver within its range). You would choose one that has a smooth frequency response and this would normally have lower distortion. You may have no choice but to have an upgrade in performance, and you must do both speakers.
So, what speakers are they? A picture of front and back would help. If it's something like 5" cone with metal sealed back, you aren't looking at an expensive speaker. Someone here may be familiar with them and can provide guidance.
Speakers are Focal Spectral 908.1, ca 1994. Someone on this forum had exactly these midranges for sale - but in 2008. Unfortunately, Focal does not make anything suitable as a replacement anymore, I asked a local dealer already. I tried to measure some candidates in his shop (car stereo) ... no success. So... let's look at the other vendors, right? There are so many measurements published, right? But then you read the comments about how badly people disliked the drivers that measured so well. I got some drivers from solen.ca Again, no luck. There are acclaimed Purify drivers, but they are of much lower sensitivity. Little by little, I became very interested in the topic of measurements vs reality, and here we are