Anyone know if REW implements Prof. Angelo Farina log sweep method which measures response & each harmonic simultaneously? Clio and the latest AP do this.
https://www.aes.org/e-lib/browse.cfm?elib=10211
Easy with this method to show how NON LTI speakers are
Though we speaker designers find it convenient to pretend that they are Best of all, you can set it so the sweep sounds like a B&K 2010 driven by a B&K 2307 chart recorder, measurements as God intended.
Last edited:
anatech, this only applies to a LTI system. ie It sorta demands the current or voltage has NO THD if the other also has NO THD. Alas, speakers are NOT LTI though its convenient to pretend that they are
Last edited:
The comparison of voltage vs current drive is done by adding 6/12/24 Ohms, and adjusting the volume to match the loudness of 80dB SPL at 1m. The drivers are Focal ps130 and Sounderlink AMT-920.
For reference - ps130 on sine sweep:
as you can see, it's ... somewhat complicated.
For reference - ps130 on sine sweep:
as you can see, it's ... somewhat complicated.
Thanks for the data! It's nice to hear and see how different each driver is from each other.
Little critique though, the video is compliled so that it is easy to compare drivers, but not easy to compare how the resistor changes sound on any single driver because the samples are scattered all over the video. It's obvious that drivers differ from each other so not very interesting, I think it would have been more interesting to compare sound of same driver with added series resistance which would have been lot easier if sequence of clips was one driver 0/6/12/24.another driver 0/6/12/24... I don't want to burden you, I try if I can chop it up and rearrange.
Another, I cannot understand what the data in the images are? The legend of images differ only with sweep number, whats the difference with the sweeps?
Little critique though, the video is compliled so that it is easy to compare drivers, but not easy to compare how the resistor changes sound on any single driver because the samples are scattered all over the video. It's obvious that drivers differ from each other so not very interesting, I think it would have been more interesting to compare sound of same driver with added series resistance which would have been lot easier if sequence of clips was one driver 0/6/12/24.another driver 0/6/12/24... I don't want to burden you, I try if I can chop it up and rearrange.
Another, I cannot understand what the data in the images are? The legend of images differ only with sweep number, whats the difference with the sweeps?
The images are spectrograms of the tracks played. Yes, it would be better to organize the video as 2x3 cell table with arbitrary access - but I do not know how to do that. If you know video editing, a good way to build UI, and wish to help, I could send you the imagery and .mp3's. I am horrible with routine work:-(
mikets, I don't know video editing either what I would want to do is try to facilitate AB test somehow, if you could post one mp3 with 0resistor, and then same driver with 24ohm resistor the difference should be audible, and AB testable if anyone wishes to. Thanks!
For matrix, it's possible to implement with web audio api. You could ask chatgpt something like this. "make simple html page, where there is path to three mp3 files defined in array, all loaded into web audio api nodes. All the tracks should start playing simultaneously, and in loop mode. There should be adjustable master volume node, where one track is connected at a time. The webpage should have button for each audio track, upon click the track would connect to master volume and any previously connected track should disconnect. Put everything into single html file to keep everything as simple as possible, so I can open it in browser without any dependencies."
what I would want to do is try to facilitate AB test somehow, if you could post one mp3 with 0resistor, and then same driver with 24ohm resistor the difference should be audible, and AB testable if anyone wishes to. Thanks! For matrix, it's possible to implement with web audio api. You could ask chatgpt something like this. "make simple html page, where there is path to three mp3 files defined in array, all loaded into web audio api nodes. All the tracks should start playing simultaneously, and in loop mode. There should be adjustable master volume node, where one track is connected at a time. The webpage should have button for each audio track, upon click the track would connect to master volume and any previously connected track should disconnect. Put everything into single html file to keep everything as simple as possible, so I can open it in browser without any dependencies."
OK. I give up. Was is AEC measurement & linearization?
I can't seem to download the paper or join Mathworks.
My MatLab account expired circa 2007 and my present position as a beach bum isn't conducive to $$$ spend.
In an attempt to move back to the OP's question, I'll put the question back to him.
Does he want to make a speaker (system) that measures well?
Or does he want to make a speaker system that sounds good?
I spent much of my previous life dreaming up 'advanced' measurements and trying to correlate them with 'good sound'. eg
https://secure.aes.org/forum/pubs/conventions/?elib=2915
&
https://www.aes.org/e-lib/browse.cfm?elib=3798
up to an including Prof Angelo Farina's log sweep which I worked out 10 yrs before him. He was surprised when I showed him my old code which did exactly what he was doing.
The important paper is
https://secure.aes.org/forum/pubs/conventions/?elib=2476
The title is deceiving cos the Appendix sums up nearly all our work on the Audibility of various Distortions. It's often quoted by false prophets including Sean & Olive.
If the OP is interested, I'll pontificate on what to do for a 'good sounding' speaker.
But if he just want's a speaker that measures good, he just (??) has to use units which measure good
Does he want to make a speaker (system) that measures well?
Or does he want to make a speaker system that sounds good?
I spent much of my previous life dreaming up 'advanced' measurements and trying to correlate them with 'good sound'. eg
https://secure.aes.org/forum/pubs/conventions/?elib=2915
&
https://www.aes.org/e-lib/browse.cfm?elib=3798
up to an including Prof Angelo Farina's log sweep which I worked out 10 yrs before him. He was surprised when I showed him my old code which did exactly what he was doing.
The important paper is
https://secure.aes.org/forum/pubs/conventions/?elib=2476
The title is deceiving cos the Appendix sums up nearly all our work on the Audibility of various Distortions. It's often quoted by false prophets including Sean & Olive.
If the OP is interested, I'll pontificate on what to do for a 'good sounding' speaker.
But if he just want's a speaker that measures good, he just (??) has to use units which measure good
There is nothing surprising about results from research into non-linearities. That is exactly why engineering - and often simple engineering - based on well-understood principles will likely remain the path forward for a while yet.I have talked to Japanese academic researchers (on conferences, for many years) about their efforts to model loudspeakers in the non-linear domain, to predistort adaptively, and the dismal results (obnoxiously huge MIPS, < 15dB gain best case). AFAIK, all papers I am aware of share a defect - they do not agree with experimental data
IMHO, the optical "sub-nanometer" optical sensing has nothing to do with the sub-nanometer precision of Lorentz actuators, which actually work pretty well. Regarding the opposite results of your work - I do not know enough details.
See my post #66 for an explanation as to why the accuracy of optical sensing is irrelevant in this case. Sensing accuracy or methodology is not the limiting issue.
I am not prone to vanity. My attitude is not something you have the ability to know. I am not offensive, and my posts are generally to the point.
I seek only to answer the questions posed and to provide justification for the statements I write. That I have done over and over on this thread. Despite my objections, you still have failed to provide a single technical or scientific reason for your continued objections.
By contrast, your continual objections appear only to reflect those of a closed mind. That is your business. Just please avoid trying to impose it on others.
I little bit harsh I think, no? They produced a valid body of work according to the remit they started with. Maybe the problem they encountered was the non-linearity of us, the listeners, too?
I still think that adding 24 Ohms in series and calling it "current drive" is a misnomer. As it is so close a value to the coil resistance, "power drive" might be a more appropriate name.
So I guess reading what's in that appendix should be worth it for this thread? You've made me curious
.
yeah, but regardless of naming convention 24 series resistor would still slash some motor distortion from acoustic domain, quite a lot actually, enough to demonstrate what "current drive" is all about.
Hi kgrlee,
Distortion in the system doesn't come into it, and we were talking about dynamic correction. The instantaneous results will be the same for a current through the load or voltage applied across the load. You cannot separate those things. We have always been considering dynamic systems.
soundbloke, by commenting and "only to answer the questions posed and to provide justification for the statements I write." gives you license to discuss anything along any line of discussion. I think you'll agree you are giving yourself permission to discuss things way off topic. You clearly dislike it when someone disagrees with you, and you aren't always polite - really.
I suggested you begin a new thread. Some of your discussion is valuable. That material will be lost in this thread. I'm sorry you can't see that.
This was a discussion on test and measurement on loudspeaker drivers. People are talking about how you drive those speakers, and that is not on topic. Anyway, I'll leave you folks to meander.
Distortion in the system doesn't come into it, and we were talking about dynamic correction. The instantaneous results will be the same for a current through the load or voltage applied across the load. You cannot separate those things. We have always been considering dynamic systems.
soundbloke, by commenting and "only to answer the questions posed and to provide justification for the statements I write." gives you license to discuss anything along any line of discussion. I think you'll agree you are giving yourself permission to discuss things way off topic. You clearly dislike it when someone disagrees with you, and you aren't always polite - really.
I suggested you begin a new thread. Some of your discussion is valuable. That material will be lost in this thread. I'm sorry you can't see that.
This was a discussion on test and measurement on loudspeaker drivers. People are talking about how you drive those speakers, and that is not on topic. Anyway, I'll leave you folks to meander.