is this possible that GD of a 88 hz Q10 -8 db Filter is -16 ms ?

user44455555

Registered
Thread Starter
Joined
Jan 16, 2021
Messages
86
Hello
please see screesnhot. this large group delays happen with high Q.
anyway i compare my EQ settings and other speaker correction from sonarworks , arc 2.5, mathroom and notice all i find sound worsest(but look better in FR) have large group delay from EQ. I use for test the speaker output XLR cable to input of soundcard..No speaker is measure only the filters.

group delay (2).jpg


phase frequency.jpg


I then try filter settings that are not so good in FR but they did not have more GD as 6 ms and i hear clear they have more clarity. REW also show higher clarity values at 100 hz when use less Q
so it seem dont use filter with higher Q as 3.

Q 3 seem also 1/3 octave graphic EQ. is there a filter setting file from REW for 1/3 octave EQ ?
 

John Mulcahy

REW Author
Joined
Apr 3, 2017
Messages
5,434
A peaking filter at 88 Hz with Q = 10, G = -8 dB has a peak group delay of about -34 ms. That's entirely normal, and a consequence of the rapid change of phase when passing through the filter's centre frequency. Usually such a filter would be used to counter a similarly sharp and narrow modal boost in the room's response, which would have the opposite phase response to the filter and the net effect would be for them to cancel each other out. What matters is the overall effect at the listening position, not the response of the filters alone. A 1/3 octave filter would have a Q of about 4.4 using REW's generic equaliser setting.

51543


51544
 

user44455555

Registered
Thread Starter
Joined
Jan 16, 2021
Messages
86
the cancelation seem help only a little. dirac look and sound even more worse.
dirac.jpg


When i use this correctiion on headphones after several minutes hear my brain can correct this to a good sound and not muddy. but because the error of speakers(see burst results) and eq together is too much errors. this result in a muddy sound and the low bass sound as it sound from somewhere from side and not in front. when i hear longer this give bad ear feeling. when i use no eq this give this problems not but it is a boomy sound. that bass direction can not hear lower as 120 hz is maybe not true for all peoples. maybe when can hear ITD work on lower frequency too.

I also try foam but even if i put 0.5 qm near over speaker or near on side of speaker it reduce nothing in db in room modes levels. maybe this all come from desktop plate. i sit 50cm away from speaker on desktop. linear phase work better but still not so good in sound as less group delay. maybe the pre ring of linear phase too reduce clarity. also linear phase have too much delay for realtime keyboard play.

My main problems are a 8 db peak at 88 hz and 180 hz and a dip at 110 hz. the other FR mistakes i did not hear after longer time as bad. really frustrating that there seem no better speakers.the burst response of 100 hz look very worse even on carbon fibre speakers. I think with a speaker that have burst response as headphones this can sound much better
 

user44455555

Registered
Thread Starter
Joined
Jan 16, 2021
Messages
86
Why would you do that? How does it make sense to listen to an EQ solution for a room response without the room's contribution?
I do this test to see whats the most important problem. So i think because the correction eq sound not so bad on headphones as on speakers it is mostly speaker problem because the speakers are very worse at 100 hz in precission. i test also with a CEA 2010 burst 160 hz and 300 hz and 800 hz . the focal with carbon fibre woofer look at 300 hz and higher better as other. but at 160 hz it is much more unprecise. so a precise speaker at 100 hz help alot i think

160 hz focal CEA 2010 burst.jpg


800 hz focal CEA .jpg
 

user44455555

Registered
Thread Starter
Joined
Jan 16, 2021
Messages
86
Is there a way that it is possible to shift channel 1 or channel 2 in X without steps so it is possible to use mathfunction channel 1-channel 2 to show the diffrence better ?
 

user44455555

Registered
Thread Starter
Joined
Jan 16, 2021
Messages
86
I have no spreadsheet. rew have for Y . osciloscope also have a channel X offset. have such a setting in rew help alot or the osciloscope support the loopback delay parameter . because speakers look better with continues measure signals than uncontinous music signals it usefull to have a more real world test with uncontinues music signals. or maybe rew can the diffrence to a CEA 2010 burst output as a distortion value in % ?
 
Top Bottom