Beyond the Ariel
 User Name Stay logged in? Password
 Home Forums Rules Articles diyAudio Store Blogs Gallery Wiki Register Donations FAQ Calendar Search Today's Posts Mark Forums Read Search

 Multi-Way Conventional loudspeakers with crossovers

 Please consider donating to help us continue to serve you. Ads on/off / Custom Title / More PMs / More album space / Advanced printing & mass image saving
john k...
diyAudio Member

Join Date: Aug 2004
Location: US
Quote:
 Originally posted by Baseballbat Hi, it does not. Or, in other words, the real - I call it effective - frequency resolution is determined by the length of the unpadded impulse response. The zero-padding (to get a power of 2 for FFT) does only add interpolated samples to the result. Bye Baseballbat
Yes, in the pure mathematical definition zero padding is ideal interpolation. However, the interpolation is not necessarily inaccurate or an untrue representation of increased resolution. The application must be considered. This implies that to increase frequency resolution from X Hz to X/2 Hz requires that the length of the measured sample is doubled regardless of the length of the impulse. That is not correct, though commonly accepted. Here is a quote from an AES paper by Eric Benjamin,
Quote:
 "Although the frequency resolution and the length of the time window are tied together in many measurement schemes, there is no reason why that needs necessarily to be so. A long transformation can be used to get a narrow frequency resolution measurement from a short analysis window, as long as all the spectral information about the transfer functions is included within that window."
In other words, the spectral information defining the transfer function is limited to that contained within the window while the frequency resolution is set by the length of the FFt.

If the impulse of a system goes to zero (without windowing) within Y msec does that mean the lowest frequency we can obtain and corresponding frequency resolution is 1/Y Hz? If the impulse goes to zero in Y msec is there any difference between sampling the signal to 16Y or zero padding a sample of length y msec to 16 y msec? If extending the length of the measured sample does not add any additional spectral information to the impulse because it is zero, then zero padding yields an identical result, with the same increased resolution.

If we measure at 48K Hz it makes no difference if we have a sample length of 32768 or a sample length of 4096 zero padded to 32768 if the window is 5 msec long. The FFt of both will be identical. The window type and length set the spectral content of the impulse. The windowing throws away spectral content and "pads" the measured impulse with zeros beyond the window's end. The length of the FFt sets the resolution. So it you prefer to consider the zero padded result interpolated, that is fine. But the result is identical to the unpadded case because the widowing supersedes this.
__________________
John k.... Music and Design NaO dsp Dipole Loudspeakers.

mige0
diyAudio Member

Join Date: Jan 2005
Location: Austria, at a beautiful place right in the heart of the Alps.
Quote:
 Originally posted by xpert But - I give up not to make that fool out of me ...

Excellent !

- but wrong again!

You already have made a fool out of you by your claim I have proven you wrong.
And not even exactly by that - I do mistakes as well - no need to be perfect - not even in the public (around here) - but I also don't come along with the "Mr. Xpert" attitude - meaning - I accept to not being perfect and let others know - even more so if I *am* wrong occasionally.

By by
Michael

ScottG
diyAudio Member

Join Date: Feb 2003
Location: US
Quote:
 Originally posted by xpert Don't take that xpert attitude to serious. Don't let it bring You into anger. I'm not different from You in that I'm quite convinced that I have to say a thing about the topic. I read Your post.

No anger.

(the emoticon was there to try and help identify that, and that the "expert" comment was mild play on words.)

BTW, I do understand what the CSD is, and how it can be used and misused. IMO it is (generally) significantly more useful than looking at the impulse response for short time duration. An energy-time curve (specific freq.) is another useful tool as well with respect to decay performance, particularly for "dialing-down" further into problem areas that might be detected with a CSD.
__________________
perspective is everything

Baseballbat
diyAudio Member

Join Date: Apr 2007
Hi,

Quote:
 Originally posted by john k... If we measure at 48K Hz it makes no difference if we have a sample length of 32768 or a sample length of 4096 zero padded to 32768 if the window is 5 msec long. The FFt of both will be identical. The window type and length set the spectral content of the impulse. The windowing throws away spectral content and "pads" the measured impulse with zeros beyond the window's end. The length of the FFt sets the resolution. So it you prefer to consider the zero padded result interpolated, that is fine. But the result is identical to the unpadded case because the widowing supersedes this.
this is correct, and I agree that interpolation must not be completely wrong. But your example has nothing to do with the reality of speaker measurements. In speaker measurements, you have information up to (e. g.) 32k samples, cut it down to 5ms and then extend the data by interpolation up to 32k bins. See the difference? Of course, this may work at high frequencies, where the frequency resolution is comparatively small, but at lower frequencies you have big problems. As I said, 5ms gives you reliable results above ~400Hz, not 200Hz as in theory. Is that what you want?

I prefer a long time window around 100ms, and then use strong smoothing. 1/3 or even 1/1, depends on the unsmoothed curve. Another method is to use the peak level in a burst decay. This gives you also phase information, the frequency resolution depends on the bandwidth of the used wavelet (usual 1/6 or 1/3 octave).

Bye

Baseballbatboy

John_E_Janowitz
diyAudio Member

Join Date: Jun 2004
Location: Green Bay, WI
Quote:
 Originally posted by ScottG Note however that shorting rings (emphasis on plural) is considerably less effective than a counter wound coil a' la 18 Sound "Active Impedance".)
The most effective method for reducing and linearizing inductance and lowering distortion is a full sleeve over the pole, or getting rid of electrical conductivity in the pole. The problem with AIC is that you are now adding more heat to the motor. You take away any ability to sink heat into the pole. You've instantly lost about 2/3 of the heatsinking of the coil and at the same time add more heat to the motor. Take a look through their own paper which shows that the shorting ring in the right place is more effective. A full sleeve of adequate thickness is even more so effective.

Quote:
 The "slow sound" is principally NOT due to inductance levels, but rather a lack of efficiency and increased moving mass. There are any number of drivers that bear this out.
Take a look at the impulse responses measured with added mass to a driver. The magnitude of the impulse decreases as you are losing efficiency. However, the rise time and decay time will not change. This shows that added mass in itself is not the issue.

Now look at a driver with a shorting ring and then again with the shorting ring removed. With the shorting ring removed, both rise and decay time are increased.

Now, your comment that high moving mass drivers sound "slow" does have some merit, but it's important to see why. If you take a driver and increase cone mass it has no effect on impulse response time. However if you add more mass to the VC which is at the same time increasing inductance, increasing flux modulation, etc this is where the issue comes from. Most typically higher mass drivers have larger coils and have higher inductance.

Quote:
 "Acceleration Factor" in the context of loudspeakers is simply an industry term specific to loudspeaker drivers. And yes, it does seem to have a rather good correlation with a driver sounding more or less "slow".
If you include current into the acceleration factor, then it does definitely give a good correlation. Current again is determined greatly by inductance. Many people look at Bl/Mms. This means nothing as you aren't taking into account the resistance of the coil. Take a DVC driver with 100gram Mms. Say with coils in parallel
(4ohm)Bl is 10 and with coils in series(16ohm) Bl is 20. Looking at Bl/Mms, the ratio doubles, although there is nothing different about the acceleration factor of the driver. It will sound exactly the same. However, change the inductance and there is a huge difference.

John

gedlee
diyAudio Member

Join Date: Dec 2004
Location: Novi, Michigan
Quote:
 Originally posted by John_E_Janowitz The most effective method for reducing and linearizing inductance and lowering distortion is a full sleeve over the pole, or getting rid of electrical conductivity in the pole.
John - agreed even if we disagree on what extent is necessary.

Quote:
 Originally posted by John_E_Janowitz If you take a driver and increase cone mass it has no effect on impulse response time. John
I had trouble following your comments (maybe I'm "slow" ), but I'm fairly sure that the statement above can't be true. Could you explain what you intended here.
__________________
Earl Geddes Gedlee Website

xpert
Account Disabled

Join Date: Apr 2009
Quote:
 Originally posted by ScottG No anger. ... decay performance, particularly for "dialing-down" further into problem areas that might be detected with a CSD.
Group Delay:

http://www.jobst-audio.de/Entwicklun...mt-1214_gd.jpg

Amplitude:

http://www.jobst-audio.de/Entwicklun...214_fgang2.jpg

CSD:

http://www.jobst-audio.de/Entwicklun...t-1214_csd.jpg

The speaker has a little problem around 7kHz which can be seen from either amplitude or group delay. As You say in American English, CSD is a no brainer. In this case it tells less than the first two. Most probably it is some path difference in the horn/driver combination or an interference from ridges reflections on the baffle. What does the CSD help here?

People simply don't understand that CSD is not dynamic data. It is calculated completely from the STEADY STATE frequency response. Take that! What is it worth?

by

by the way perpetuated: reflection and path difference are NO resonances!

Lynn Olson
diyAudio Member

Join Date: Jul 2004
Quote:
 Originally posted by xpert People simply don't understand that CSD is not dynamic data. It is calculated completely from the STEADY STATE frequency response. Take that! What is it worth?
Huh? When I saw my first FFT's and CSD's at KEF in 1975 they were calculated directly from a single gigantic 200-watt impulse fed to the speaker. The impulse data was captured DIRECTLY from the digitizer as it came into the computer. No steady-state there. To lower the noise floor, they'd then average a series of impulse measurements, but that meant adding a series of time-domain measurements directly on top of each other, not a conversion into the frequency domain.

FFT's and CSD's were calculated from the raw time-domain data, not the other way around. As a long-time user of 500 and 7000-series Tektronix scopes and spectrum analyzers, I know the difference between real-world time measurements and computed frequency-domain measurements. What you see on the screen of a scope is not "computed" in any way - scopes are nothing more than vertical amplifiers that show the waveshape directly on a scanned, non-storage display. If you see the trace wiggle on a scope, you can pretty sure something really happened to the electrical signal.

I'm not saying that frequency domain measurements aren't useful, but they are computed from the real world of the time domain. As computations, you have to intelligently select the appropriate algorithm to transform the domains, and be aware of what artifacts are generated by the choice of algorithm. People are so comfortable with computers and software they forget that brickwall lowpass filters that ring in the time domain, ADC's running at low sampling rates, and choice of algorithm do not have a neutral interpretation of the raw electrical data.

xpert
Account Disabled

Join Date: Apr 2009
Quote:
 Originally posted by Lynn Olson Huh? When I saw my first FFT's and CSD's at KEF in 1975 ... time domain. ... you have to intelligently select the appropriate algorithm to transform the domains ... ADC's running at low sampling rates, and choice of algorithm do not have a neutral interpretation of the raw electrical data.
Today CSD is calculated from a normalised impulse response. The normalised IR is too the basis for amplitude and phase response calculations. Because the said transformations are simply calculations there is no more information in one than in the other. How CSD is of less readibility can be seen in my last post above.

 12th May 2009, 06:28 PM #5540 Lynn Olson   diyAudio Member     Join Date: Jul 2004 Location: Northern Colorado OK, now I'm really confused. In your own words, the normalized impulse response - which, by definition, is in the time domain - is used to calculate the amplitude and phase response. Ipso facto, time domain is the starting point, and further calculations yield frequency, phase, and non-minimum phase information. What confuses me is that you stated in a previous post that the CSD is calculated from STEADY STATE (your emphasis) frequency response. So what is it? Is the CSD calculated from the impulse response, or not? Since loudspeakers are complex electromechanical devices, minimum-phase operation cannot be assumed - they are not amplifiers, after all, and have non-minimum-phase crossovers, regions of driver breakup, diffraction off the cabinet edges, not to mention nonlinear distortion at all power levels. All of these defects are audible and degrade the sound - the point of any measurement is to find correlations between subjective impressions and possible methods to improve the loudspeaker.

 Posting Rules You may not post new threads You may not post replies You may not post attachments You may not edit your posts BB code is On Smilies are On [IMG] code is On HTML code is Off Forum Rules
 Forum Jump User Control Panel Private Messages Subscriptions Who's Online Search Forums Forums Home Site     Site Announcements     Forum Problems Amplifiers     Solid State     Pass Labs     Tubes / Valves     Chip Amps     Class D     Power Supplies     Headphone Systems Source & Line     Analogue Source     Analog Line Level     Digital Source     Digital Line Level     PC Based Loudspeakers     Multi-Way     Full Range     Subwoofers     Planars & Exotics Live Sound     PA Systems     Instruments and Amps Design & Build     Parts     Equipment & Tools     Construction Tips     Software Tools General Interest     Car Audio     diyAudio.com Articles     Music     Everything Else Member Areas     Introductions     The Lounge     Clubs & Events     In Memoriam The Moving Image Commercial Sector     Swap Meet     Group Buys     The diyAudio Store     Vendor Forums         Vendor's Bazaar         Sonic Craft         Apex Jr         Audio Sector         Acoustic Fun         Chipamp         DIY HiFi Supply         Elekit         Elektor         Mains Cables R Us         Parts Connexion         Planet 10 hifi         Quanghao Audio Design         Siliconray Online Electronics Store         Tubelab     Manufacturers         AKSA         Audio Poutine         Musicaltech         Aussie Amplifiers         CSS         exaDevices         Feastrex         GedLee         Head 'n' HiFi - Walter         Heatsink USA         miniDSP         SITO Audio         Twin Audio         Twisted Pear         Wild Burro Audio

 New To Site? Need Help?

All times are GMT. The time now is 12:17 PM.

Search Engine Optimisation provided by DragonByte SEO (Pro) - vBulletin Mods & Addons Copyright © 2017 DragonByte Technologies Ltd.