One thing to bear in mind is that low low level signals are most likely limited by the mic capsule noise level.
soongsc said:
I've tried very low levels for FR measurements. The overlays show that to be of little concern for FR in a quiet environment, might easily be a problem for distortion measurements.
Dave
Questions arising on the other thread in response to technical comments are being rejected, so I'm bringing them over here where I expect they are allowed.
How do you know it's IM and that this specifically is worsened?
Similar question, which impedance and how do you know it's being raised or lowered? This implies quantitative knowledge, as in measurements.
Dave
How do you know it's IM and that this specifically is worsened?
Similar question, which impedance and how do you know it's being raised or lowered? This implies quantitative knowledge, as in measurements.
Dave
Well if Doc Geddes scientific research is believed either Bud's ears are much better than most people's or it isn't IM... ie Bud's use of the term is just a conveinient handle to grab onto.
dave
dlr,
Thank you for the LAUD comments. Do you have any comment on close mike positions for low level signal measurements? Will a 9 inch or closer distance help or hinder? At Nestorovic Labs we measured full range systems at one meter with the mic on axis for the leaf tweeter, but measured individual drivers at 9 inches for woofers, six inches for mid range domes and three inches for tweeters.
It appears to be a resonance problem between original and reflected wave fronts, between whizzer and phase plug. There can also be timing issues between these two devices, that show up as an audible dissonance, roughly in the same frequency band as the shrieks arise within. I have resorted to straws, both soft plastic and metal, to attempt to locate the event physically, but without success.
These are actually treatment issues, but I will be happy to discuss test methods for finding them and investigating what the results from data are, when the audible results have been altered. Actual qualitative investigation will have to wait until I have access to the more typical whizzer phase plug drivers, or until I have time to work on my PM6A Lowthers again.
I will not engage in a debate about whether or not I can alter these conditions, but would certainly not mind discussing what measures might be used to isolate and test them.
Bud
Thank you for the LAUD comments. Do you have any comment on close mike positions for low level signal measurements? Will a 9 inch or closer distance help or hinder? At Nestorovic Labs we measured full range systems at one meter with the mic on axis for the leaf tweeter, but measured individual drivers at 9 inches for woofers, six inches for mid range domes and three inches for tweeters.
The only way "I know" is by comparative listening, as described. What I am listening for is a deviation in character, across a range of female voice that is difficult to reproduce. The shriek is a definite added coloration, that alters the sound from being that of an Alto female voice to a much more aggressive tone, with very noticeable and objectionable peaks. This shriek has a very definite onset and usually a bandwidth. The rest of the vocal range has the normal vocal characteristics. I have yet to hear more than one range of shrieking from this whizzer/phase plug area.
It appears to be a resonance problem between original and reflected wave fronts, between whizzer and phase plug. There can also be timing issues between these two devices, that show up as an audible dissonance, roughly in the same frequency band as the shrieks arise within. I have resorted to straws, both soft plastic and metal, to attempt to locate the event physically, but without success.
These are actually treatment issues, but I will be happy to discuss test methods for finding them and investigating what the results from data are, when the audible results have been altered. Actual qualitative investigation will have to wait until I have access to the more typical whizzer phase plug drivers, or until I have time to work on my PM6A Lowthers again.
I will not engage in a debate about whether or not I can alter these conditions, but would certainly not mind discussing what measures might be used to isolate and test them.
Bud
planet10 said:
I would wager the case is "it isn't IMD". In fact, I think that I can say with some assurance that it's not. I would also suggest that terms with very specific and well-defined meanings not be used haphazardly, certainly not erroneously.
Dave
dlr
As planet 10 Dave points to, I have, likely as not, here also misused a term you have very strict definitions for.
When any driver I am working with has a mismatch in information richness, or on axis "beaming" from a very narrow point, on the center dome surface of a voice coil cover, or a seperate phase plug, I am after a method that drops the radiating efficiency of that area, in an overall response synthesis, from out in the sound field.
In my own model I am impeding the force from acting in one path and this changes the pattern of dispersal of force into and between the two main emitter formats. Since one is pistonic and localized, if I can redirect some of that force into the shear wave, by providing an impedence to it's actual emission, then I will do so. The net effect is a more even balance between on axis and off axis sound, and I am speaking of just off axis from the voice coil here. I can actually depress this central on axis event too much, which is just as specific a locater to the speaker as having an on axis problem is.
Please keep in mind while thinking this over, that the sound field I am discussing is not one you are used to. The sound field produced by an EnABL'd driver does not perceptually arise from the plane of the driver or localized to it in any way. It is instead, from behind that plane and not at all tied to the driver as a location for producing the sounds. This makes an on axis enrichment, or beaming, or a shriek, a very noticeable event.
If you are in center position, out a couple of meters, from a pair of speakers, full range system or full range single driver, and move laterally across to on axis for one side, the image remains behind and around the speaker, when these on axis events are corrected.
When they are not corrected, as you move laterally, the image pops up and on to the speaker, as the location of sound, but only for those portions of the frequency band that are enriched or shrieking. Everything else remains behind and around the speaker as far as perceived location of emission is concerned. Also, the perceived locations are very specific, with respect to other sources also being reproduced, even if you move laterally beyond the pair of speakers a fair distance.
Bud
As planet 10 Dave points to, I have, likely as not, here also misused a term you have very strict definitions for.
When any driver I am working with has a mismatch in information richness, or on axis "beaming" from a very narrow point, on the center dome surface of a voice coil cover, or a seperate phase plug, I am after a method that drops the radiating efficiency of that area, in an overall response synthesis, from out in the sound field.
In my own model I am impeding the force from acting in one path and this changes the pattern of dispersal of force into and between the two main emitter formats. Since one is pistonic and localized, if I can redirect some of that force into the shear wave, by providing an impedence to it's actual emission, then I will do so. The net effect is a more even balance between on axis and off axis sound, and I am speaking of just off axis from the voice coil here. I can actually depress this central on axis event too much, which is just as specific a locater to the speaker as having an on axis problem is.
Please keep in mind while thinking this over, that the sound field I am discussing is not one you are used to. The sound field produced by an EnABL'd driver does not perceptually arise from the plane of the driver or localized to it in any way. It is instead, from behind that plane and not at all tied to the driver as a location for producing the sounds. This makes an on axis enrichment, or beaming, or a shriek, a very noticeable event.
If you are in center position, out a couple of meters, from a pair of speakers, full range system or full range single driver, and move laterally across to on axis for one side, the image remains behind and around the speaker, when these on axis events are corrected.
When they are not corrected, as you move laterally, the image pops up and on to the speaker, as the location of sound, but only for those portions of the frequency band that are enriched or shrieking. Everything else remains behind and around the speaker as far as perceived location of emission is concerned. Also, the perceived locations are very specific, with respect to other sources also being reproduced, even if you move laterally beyond the pair of speakers a fair distance.
Bud
BudP said:
Define what you mean by low level signal measurements? What are you specifically trying to ascertain with the measurements?
As for distance, those distances you quote are probably only useful for distortion measurements. For close-mic'ed woofer measurements one might use to splice into a 1m measurement, they won't do. They also won't be representative of the drivers actually FR due to phase issues as frequency rises. I don't see the benefit. The only practical way to see what can be done is to measure first, mod, then measure again and analyze the delta.
That can all be and likely are functions of FR, having nothing to do with IMD.
Well, a whizzer cone is going to cause a lot of reflected waves, setting up resonances in the gap between cone and whizzer. There's little that can be done to this area, the air gap is going to have its impact no matter in one way or another. There will also be linear distortion due to the same frequencies being produced by cone and whizzer, separate from any IMD (Doppler) that will be present, whatever the audibility is/is not of the latter. The former part (reflected waves) will be altered significantly as one move off-axis. This also can't be eliminated.
Considering that the manufacturers make their business from trying to optimize them, I have serious doubts about making significant improvements. It's also probably impractical to separate the cone from the whizzer insofar as FR contributions and such. They are closely coupled to the point that one affects the other. Even if they were mounted and measured separately the result would not be the two individual parts that can sum to the response when mounted and fixed together as is the case for multi-way systems.
Dave
dlr,
I want to have a look at impulse and distortion on the same driver, across a fairly wide range of power, before and after treatment. Just a simple cone full range to begin with and probably the units that Al at RAW sends back to me for treatment. I intend to do at least 15 and 30 degree off axis comparison runs, along with on axis.
Since I have a very quiet environment I intend to look at a range from -60dB up to one watt drive levels, and as the drivers will likely be nude on stands I will not be interested in signals much below 500 Hz.
I am not looking for absolute performance data here, just a same location, same driver set of tests, with barometric pressure, temperature and humidity data provided.
Bud
I want to have a look at impulse and distortion on the same driver, across a fairly wide range of power, before and after treatment. Just a simple cone full range to begin with and probably the units that Al at RAW sends back to me for treatment. I intend to do at least 15 and 30 degree off axis comparison runs, along with on axis.
Since I have a very quiet environment I intend to look at a range from -60dB up to one watt drive levels, and as the drivers will likely be nude on stands I will not be interested in signals much below 500 Hz.
I am not looking for absolute performance data here, just a same location, same driver set of tests, with barometric pressure, temperature and humidity data provided.
Bud
BudP said:
I'll try again.
Just a word of warning/cattion. What I woudl do here is first set up your system with a 1k Hz sine input to set the gain to that you have nomimally 1 watt (or 2.83V for 8 ohms reference) across the speaker. Then that an impulse and convert it to frequency response to see the level. The, mute you system and take a "silent" impulse and convert to frequency response. The relative difference in level between the first ans second measurments will tell you what you back ground noise is. I think you may find it difficult to get meaningful result 60 dB below the 1 watt level. Given a driver with 90dB/watt you are asking for data at 30dB SPL.
I think this is an interesting issue. THD and IMD seems to be different data generated from different test methods. Unless there is a relation showing a relationship between what is physically happening, how can one say whether what is heard "is" or "is not" IMD?dlr said:
John K,
Thanks John, those comments help with how to think about the process. I will follow your suggestions. The speaker is 87 db sensitive and only 3 inches in diameter across the frame so getting off axis information could also be tricky.
Bud
Thanks John, those comments help with how to think about the process. I will follow your suggestions. The speaker is 87 db sensitive and only 3 inches in diameter across the frame so getting off axis information could also be tricky.
Bud
Just to give everyone a heads up on what we are facing. Analysis is showing that at the higher frequencies, the cone vibration mode consists of concentric and radial modes, which is somthing I did not expect under ideal analysis conditions. Very interesting. I would encourage FEA if people really want to try some stuff out and gain more understanding.
source?
Any data or references? Whose analysis? I thought this was the place for actual data, not "teaser" statements.
John L.
Any data or references? Whose analysis? I thought this was the place for actual data, not "teaser" statements.
John L.
Re: source?
This is something that anyone with the tools can do. If there are difficulties coming up with results, I don't mind providing some tips, but I certainly cannot give away other peoples data. Work is done using Ansys. But other tools such as Comsol can work as well. I have asked before whether anyone had access to Comsol or not, but there was no reply here. If others are doing work, and are willing to share "what works" or any trend that might point to better understanding, then we can see what can be done to make data available.auplater said:
Just for example, this is an example at 16450Hz
At lower frequencies there are only concentric modes.
An externally hosted image should be here but it was not working when we last tested it.
At lower frequencies there are only concentric modes.
Here is a cool illustration of soundwaves at work.
http://www.youtube.com/watch?v=Zkox6niJ1Wc
Cheers
http://www.youtube.com/watch?v=Zkox6niJ1Wc
Cheers
soongsc said:
It would be great if the job could be done in CAELinux... No need to have access to really expensive HW & SW
Code_Aster, Code-Saturne, OpenFOAM & Elmer are the FE & CFD solvers...
as a reference, if you can get those "in the know" interested, to see if there is geometric model compatibility between the packages. That, plus the models constants and coefficients are big offenders.
http://www.caelinux.com/CMS/
If you could get the geometric model, perhaps we could start playing with this. Like adding a ring of blocks
The geometric modeling for something more or less realistic is extremely time consuming...
Gastón