Multiple Small Subs - Geddes Approach - Page 56 - diyAudio
Go Back   Home > Forums > Loudspeakers > Subwoofers

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
Reply
 
Thread Tools Search this Thread
Old 2nd February 2009, 07:45 PM   #551
diyAudio Member
 
john k...'s Avatar
 
Join Date: Aug 2004
Location: US
Quote:
Originally posted by cap'n todd


OK, your s/n looks good. Just thought I'd check since bad s/n can do wierd things to phase.

The method you use for inversion may have an impact here as well (pole/zero approximation in system modeling). Further there are a number of reasons why dereverberation in a real room does not alway work as expected. John Mourjopoulos has done some research on this ("Errors in Real-Time Room Acoustics Dereverberation"J. Audio Eng. Soc., Vol. 52, No. 9, 2004 September), and concluded:


Specifically it was found that under practical real-time
conditions such “high-Q” ringing poles are extremely sensitive
to small variations in the system (room) properties,
which can occur between response measurements and dereverberation tests even if the source/receiver positions and
equipment settings remain identical, so that any such test
will suffer from significant mismatch error. Hence from
this point of view the room may be considered a “weakly
nonstationary system.”


I'm sure we could argue this one for years to come, but my main point is...


Try this on for size: Look at your impulse response, and tell me at what frequency it is strongly ringing after MP EQ?? Looks like around 40 Hz. now look at your mag and phase plots. Where is the biggest cancellation dip (and phase wrap)? 40 Hz. My theory is that when you do this type of eq, namely trying to boost dips as well as redcing peaks, your end up boosting the response at frequencies where it IS non-min phase. Yes, you are also reducing the level at frequencies where (I contend) it tends to be minimum phase, i.e. the peaks. So, you could argue that the initial response was more non MP than MP, since when you flattened it, boosting the dips exacerbated the ringing more than flattening the peaks reduced it. Maybe - more analaysis of different locations and rooms would be needed perhaps to answer that. BUT, my main point is that if you EQ in a sensible way, i.e. just bring the largest peaks down to a reasonable level, you should reduce the overall ringing.

This is why I content that the responses are mostly non MP. Technically they are not MP, but effectively are with respect to (sensible) MP equalization.

I should like to see the result of a more "gentle" type of EQ in either your simulation or measurements.

Respectfully,
Todd Welti
Hi Todd,

I totally agree with you that trying to eq the dips is why the eq'ed impulse resoponse is worse that the raw response. And just cutting the peaks can make things better but still not correct. Cutting the peak will usually make things better, even if not MP(but not always if not MP), because you are reducing energy in the system. Boosting the dip is a double edged soword. If the response were MP then it would help because it would be removing a resonance (a MP response dip rings just as much as a MP boost but with inverted phase).

I'm still of the belief that getting the response to behave as MP

That is why I think the approach has to be to make the response MP, and then worry about eq.
__________________
John k.... Music and Design NaO Dipole Loudspeakers.
  Reply With Quote
Old 2nd February 2009, 07:47 PM   #552
diyAudio Member
 
Join Date: Jan 2009
Quote:
Originally posted by markus76


You can attach a file with every message you compose or upload the image to a provider like http://imageshack.us/ and paste the code you get after uploading the image in your message. Alternatively you can link a file that is already online by pasting its URL in your message.

Best, Markus


ok, here's one. I have found another I can upload if you are interested.
Attached Images
File Type: jpg waterfalldecay.jpg (77.5 KB, 392 views)
  Reply With Quote
Old 2nd February 2009, 08:01 PM   #553
Banned
 
Join Date: May 2008
Location: Switzerland
Thanks Todd, I was hoping that you will post that particular CSD because it's the same Toole used in his book What we see is that the ringing is attenuated (which is good) but at the same time the decay rate has become slower (which is bad). In a minimum phase system the decay rate should be faster after applying an EQ, right?

Best, Markus
  Reply With Quote
Old 2nd February 2009, 08:10 PM   #554
diyAudio Member
 
john k...'s Avatar
 
Join Date: Aug 2004
Location: US
Quote:
Originally posted by gedlee



John

But X(r) must have sources (or sinks) on the boundary otherwise when it is added to g(r) it will not yield the boundary condition since g(r) is not zero on the boundary. See I can also refute you without using that term either. We agree that there can be such an X(r), but I don't think that we see it as the same function and I don't think that it would be easy to derive. I know that I 've never seen it.

Earl you missinterpreted what I posted, due to my poor gramar.

Quote:
It would seem we agree that, in simple terms, X must satisfy the wave equation with no sources in the interior and boundary conditions as required so that when added to g, it forms G, the solution for the bounded region.
How about if I rewrote that as, "In simple terms, X must satisfy the wave equation with no sources in the interior. Boundary conditions as required are applied so that when added to g, g + X = G, the solution for the bounded region."

I would agree that solving the wave equation, with appropriate boundary conditions, for X might not be an easy task, but someone must have done it to come up with the modal expansion for G. So, X = G-g must at least be a reasonable representation of X, at least as good as G is a representation of the bounded region Green's function.
__________________
John k.... Music and Design NaO Dipole Loudspeakers.
  Reply With Quote
Old 2nd February 2009, 08:21 PM   #555
diyAudio Member
 
Join Date: Jan 2009
Quote:
Originally posted by markus76
Thanks Todd, I was hoping that you will post that particular CSD because it's the same Toole used in his book What we see is that the ringing is attenuated (which is good) but at the same time the decay rate has become slower (which is bad). In a minimum phase system the decay rate should be faster after applying an EQ, right?

Best, Markus

I dont think the decay rate is slower. It is certainly much faster in the early part. So overall if you avoid noise floor, I think the decay is faster. Here's more from Karjalainen et al, which shows the "slice" of the waterfall for the resonance in question. Similar to the Toole plot, the decay is much faster at the start and slower later. One reason for this is the noise floor, which I'm sure is right about -25 dB. So, the effect on the decay would start around -15 dB. The other thing, and perhaps this is what you are referring to, is that this type of decay may happen when the Q of the filter does not match the resonance exactly. i may have more on this, if I can find it... . keep in mind that in this example ther're not even trying to remove the resonance, just bring it down to the level of nearby peaks.
Attached Images
File Type: jpg karjalainen.jpg (44.9 KB, 362 views)
  Reply With Quote
Old 2nd February 2009, 08:54 PM   #556
diyAudio Member
 
Join Date: Jan 2009
Quote:
Originally posted by john k...


Hi Todd,

I totally agree with you that trying to eq the dips is why the eq'ed impulse resoponse is worse that the raw response. And just cutting the peaks can make things better but still not correct. Cutting the peak will usually make things better, even if not MP(but not always if not MP), because you are reducing energy in the system. Boosting the dip is a double edged soword. If the response were MP then it would help because it would be removing a resonance (a MP response dip rings just as much as a MP boost but with inverted phase).

I'm still of the belief that getting the response to behave as MP

That is why I think the approach has to be to make the response MP, and then worry about eq.

Hi John, so are you trying to remove all ringing then? Then add some back in to make it sound more natural? I'm dont really think its necessary, but sounds interesting.
  Reply With Quote
Old 2nd February 2009, 08:57 PM   #557
Banned
 
Join Date: May 2008
Location: Switzerland
Quote:
Originally posted by cap'n todd
So overall if you avoid noise floor, I think the decay is faster.
At least it's not clear and that's all I was talking about.
Up to what frequency would you recommend the use of an EQ?

Best, Markus
  Reply With Quote
Old 2nd February 2009, 09:43 PM   #558
youngho is offline youngho  United States
diyAudio Member
 
Join Date: Nov 2008
Quote:
Originally posted by markus76
Thanks Todd, I was hoping that you will post that particular CSD because it's the same Toole used in his book What we see is that the ringing is attenuated (which is good) but at the same time the decay rate has become slower (which is bad). In a minimum phase system the decay rate should be faster after applying an EQ, right?
Markus, I think you mean that the decay RATE should be similar in an approximately minimum phase system, but the decay TIME should be faster because the starting amplitude is lower. You had previously substituted one term for the other many pages ago, and it makes discussion a little confusing. It's hard to make out the slope here, but you can see that the resonant frequency hump goes to zero earlier AFTER equalization, hence the decay TIME is in fact faster.
  Reply With Quote
Old 2nd February 2009, 10:02 PM   #559
diyAudio Member
 
Join Date: Jan 2009
Quote:
Originally posted by markus76


At least it's not clear and that's all I was talking about.
Up to what frequency would you recommend the use of an EQ?

Best, Markus
IMHO
A few hundred Hz for sure, higher than that is another question.
  Reply With Quote
Old 2nd February 2009, 10:12 PM   #560
diyAudio Member
 
Join Date: Jan 2009
Quote:
Originally posted by youngho


Markus, I think you mean that the decay RATE should be similar in an approximately minimum phase system, but the decay TIME should be faster because the starting amplitude is lower. You had previously substituted one term for the other many pages ago, and it makes discussion a little confusing. It's hard to make out the slope here, but you can see that the resonant frequency hump goes to zero earlier AFTER equalization, hence the decay TIME is in fact faster.
Isn't this elementary linear systems theory? If you reduce the Q of a MP resonance, the time constant (i.e. decay rate) is reduced.

time const = Q/(pi*Fo) or somehting like that.

Again, I think that at least part of the complication here is that the Q of the correction filter seldom exactly matches the resonance. In fact it may not always be easy to determine the Q of the resonance. So you end up with a corrected response that has two slopes, from the Q of the resonance and the correction filter not cancelling correctly. Should it be surprising that the decay in time has two slopes then?
  Reply With Quote

Reply


Hide this!Advertise here!
Thread Tools Search this Thread
Search this Thread:

Advanced Search

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
Trackbacks are Off
Pingbacks are Off
Refbacks are Off



New To Site? Need Help?

All times are GMT. The time now is 08:03 AM.


vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2014 DragonByte Technologies Ltd.
Copyright ©1999-2014 diyAudio

Content Relevant URLs by vBSEO 3.3.2