Too much snake oil. I do the spots because it 1st off allows me to get closer to the music.
dave
Dave, 30 to 50db down in a home at night is easy to achieve. But what parameter is it you'd want to find? Phase, CSD, frequency response, distortion, etc. It's not HOM...
Based on the (little) data that is out there, I think it's spreading out break up at high frequencies. Which the audible effects can be 30 to 50db down and related to CSD. That's my guess. If you can lower the q of break up peaks and shut down the ringing a little, then it probably sounds better / clearer. You might hear the music better.
Based on the (little) data that is out there, I think it's spreading out break up at high frequencies. Which the audible effects can be 30 to 50db down and related to CSD. That's my guess. If you can lower the q of break up peaks and shut down the ringing a little, then it probably sounds better / clearer. You might hear the music better.
Now how do you extract that information in the presence of the stimulating 0dB signal? And a typical room does not give you the time dimension you really need to capture the information.
All of them. And if you reduce HOMs, by definition you are increasing DDR.
And what do you mean by distortion. Better DDR means lower distortion -- but in its broadest term, not the ususal, and near useless THD often measured & equated with the term distortion.
Clearer is a good word. A visual analogy that i have used in the past, is imagine looking down onto a mountain valley with a mist that prevents you from making out small details... then EnABL would be having the sun come out and clear away the mist allowing you to see in greater detail what is down there.
dave
Well if you need to capture information beyond say 3ms, then a typical room couldn't do that. Still, outside at night can be very silent, might work. The software extracts the information. But you need to know what you're trying to measure.
I'm not sure a direct radiator has a HOM issue??? Might want to check with Geddes on that. My understanding of HOM is that it's a reflection in the horn throat (simplified explanation).
By distortion, I measure non-linear distortion like 2nd, 3rd, 4th, etc. Those can be easily measured. No, I wouldn't just measure THD. You need to see them individually and at various frequencies.
But distortion comes in various linear forms also, but a frequency response can mostly quantify those.
I'm not sure a direct radiator has a HOM issue??? Might want to check with Geddes on that. My understanding of HOM is that it's a reflection in the horn throat (simplified explanation).
By distortion, I measure non-linear distortion like 2nd, 3rd, 4th, etc. Those can be easily measured. No, I wouldn't just measure THD. You need to see them individually and at various frequencies.
But distortion comes in various linear forms also, but a frequency response can mostly quantify those.
You need to be somewhere where you can eliminate the need to window anything.
As a simple example, as a stimulas use a 0dB signal with a -x0 dB riding on top. You want to extract the FR of the -x0 dB signal.
With a CSD, i want to know what is happening -x0 dB under that 1st curve at 0ms and the CSD that that gives.
It may or may not, but in a device that coule produce HOMs, if you can eliminate them you increase DDR
Those are pigeon holes. By distortion i mean anything that is adding or subtracting information to the signal. Most distortion measurements look for stuff that is added and not those that are subtracted.
Linear refers to that 5% of mathematical problems that can be solved simply (and what science has concentrated on up until the disemination of chaos/information theory. Non-linear distortion is the 95% of problems. The very term non-linear means that it is being looked on from the comfortable point of view that comes from solving linear problems. Information theory brings a more holistic POV. Think in terms of what information id lost or corrupted.
dave
Another discussion of DDR starting about here http://www.diyaudio.com/forums/markaudio/224477-future-woofer-production-10.html#post3329649
dave
dave
By nonlinear I mean harmonic distortion, sorry to confuse. If that isn't what needs to be measured that's fine by me.
Hmm, the first part of your reply is interesting, and doable. I'm not sure how many measurement software packages are equipped to do it, but most have an RTA. Then input a say 1000hz signal at 0db with a 5000hz at -10db and a 4200hz at -20db etc. all random. Then snap an RTA. Do it in a very dead space. Large quite room with thick mats of insulation all around (I have lots). Then do another set of ransoms. Do about 30 different randomness for each speaker (enabled an non). If the non enabled speaker doesn't produce the signals below the main 0db signal as strongly as the enabled speaker, perhaps you've found the difference.
Hmm, the first part of your reply is interesting, and doable. I'm not sure how many measurement software packages are equipped to do it, but most have an RTA. Then input a say 1000hz signal at 0db with a 5000hz at -10db and a 4200hz at -20db etc. all random. Then snap an RTA. Do it in a very dead space. Large quite room with thick mats of insulation all around (I have lots). Then do another set of ransoms. Do about 30 different randomness for each speaker (enabled an non). If the non enabled speaker doesn't produce the signals below the main 0db signal as strongly as the enabled speaker, perhaps you've found the difference.
By using a global term for something very specific that is but an infitesamal aspect of the global term is very limiting to one's thinking of the whole picture.
dave
Like I said, sorry. Harmonic distortion is often referred to as nonlinear distortion in speaker testing, although there are other tests and form of nonlinear distortion. I was asking you what you think needs to be measured.
White noise is 20 to 20,but I'm not sure what software can resolve the two signals. That's why separated inputs would be better. I suppose you could do bands of sound rather than a single frequency. 1000 to 2000 at 0db, 2000 to 3000 at -10, 500 to 1000 at -20db, etc.
White noise is 20 to 20,but I'm not sure what software can resolve the two signals. That's why separated inputs would be better. I suppose you could do bands of sound rather than a single frequency. 1000 to 2000 at 0db, 2000 to 3000 at -10, 500 to 1000 at -20db, etc.
Single sine, multiple sine, swept sine, multiple swept sine, band pass MLS, shaped bursts, and various combinations.
As more sine waves are added to signal, potential IMD components would grow geometrically, but each IMD component taps energy from formant harmonics, pulling measured harmonic value down.
Example:
Burst stimuli are used to explore steepness of crossover slope in rejecting harmonic and IM distortion with small full range driver:
Peerless 2" full range is run with two conditions: Linkwitz-Riley 24dB/oct high pass filter and with 1024 point FIR high pass filter. Each condition is also brick wall filtered with 4096 point FIR 1kHz low pass filter. The upper trace shows a sea of output above 1kHZ for LR4 high pass filter consisting of harmonic distortion and IMD. FIR results show defined much lower simple harmonic peaks against the true noise floor.
Signal passed to driver with low slope 24dB/oct filter and no signal above 1kHz results in significant output above 1kHz. Reality is dirty byproducts continue down below 1kHz. Most of this is induced by relatively low stimulus passed in driver resonance region. When this is rejected as with steep FIR filter, bandwidth is narrower, less cluttered, and produces far fewer distortion components, and at significantly lower level.
This translates directly to listening experience with these crossovers. FIR filter allows much more power before driver gets rough to listen to.
Good old battle of bandwidth and dynamic range.
As more sine waves are added to signal, potential IMD components would grow geometrically, but each IMD component taps energy from formant harmonics, pulling measured harmonic value down.
Example:
Burst stimuli are used to explore steepness of crossover slope in rejecting harmonic and IM distortion with small full range driver:
Peerless 2" full range is run with two conditions: Linkwitz-Riley 24dB/oct high pass filter and with 1024 point FIR high pass filter. Each condition is also brick wall filtered with 4096 point FIR 1kHz low pass filter. The upper trace shows a sea of output above 1kHZ for LR4 high pass filter consisting of harmonic distortion and IMD. FIR results show defined much lower simple harmonic peaks against the true noise floor.
Signal passed to driver with low slope 24dB/oct filter and no signal above 1kHz results in significant output above 1kHz. Reality is dirty byproducts continue down below 1kHz. Most of this is induced by relatively low stimulus passed in driver resonance region. When this is rejected as with steep FIR filter, bandwidth is narrower, less cluttered, and produces far fewer distortion components, and at significantly lower level.
This translates directly to listening experience with these crossovers. FIR filter allows much more power before driver gets rough to listen to.
Good old battle of bandwidth and dynamic range.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.