Help EQing the subwoofer

[Marcus Aseth]

Maybe this is a dumb question... And... Do you have an NVDIA graphics card/device in your system? And if you do, why aren't you using that instead of the Behringer UMC-22 for high quality stereo output?

I have no idea if that refers to the monitor or my RTX 2060, either way they both don't have an usb port which the umc-22 needs, and I need the umc22 because I want a knob for the volume of the whole setup
EDIT: yeah, 21G1WG4 is the monitor.

 

[ddude003]

If its a laptop the OP may not have many options... If it is a desktop or desk side there may be a motherboard sound system which may be a better option only to get the output off the USB that is being used as mic in... And sure if the motherboard system or a GFX card has a 2.1 or better why not take advantage of that...

Interesting... "The card features two DisplayPort 1.4 outs, one HDMI 2.0 out, one DVI-D out, and one USB-C/VirtualLink, with the last port designed for future VR headsets." Has its own USB-C port??? Crazy...

@Marcus Aseth said: "I need the umc22 because I want a knob for the volume of the whole setup"... Ah... got it...

BTY, you had asked for an alternative to Equalizer APO... I was just starting to look into CamillaDSP as another choice for a convolution engine... I don't know anything about it at this point... Its free and runs on windoz, unix and MacOS, maybe Pi too...

 

[Marcus Aseth]

Has its own USB-C port??? Crazy...

Not that I can see, maybe that's another model.

Now that I've figured out what fooled me the first time, tomorrow I'll attempt again the video-tutorial posted on page 1.
Although I have 2 doubts: once I figure out the time delay, what am I supposed to do with it in EQ APO? Same for the gain compensation.

those are the two control I assume I must use to set those values, the problem is that they are applied to the previously mentioned "CODEC" which affect the whole system...I cannot specify that a modification is intended for the Left speaker and one is intended for the Right...

 

[ddude003]

Sorry @Marcus Aseth, I don't have any experience in the depths of windoz system sound... At this point I can only suggest that you get back to where you can use REW to sample the room and use Equalizer APO to apply EQ... Maybe center that Target Curve, as has been suggested before, and rerun REWs standard filtering methods or try Serkan's inversion method again...

 

[JStewart]

Although I have 2 doubts: once I figure out the time delay, what am I supposed to do with it in EQ APO? Same for the gain compensation.

those are the two control I assume I must use to set those values, the problem is that they are applied to the previously mentioned "CODEC" which affect the whole system..

With your current set-up no delay or gain settings can be used in Eq APO. That’s because they apply to both the mains and sub simultaneously in your set-up. To time align you must have the ability to affect the mains and sub individually. The only timing adjustment that is available in your set-up is the phase control on the sub. You might want to review my post, #5.

I cannot specify that a modification is intended for the Left speaker and one is intended for the Right...

This is absolutely not necessary in your setup as the left and right are equidistant to you. For future reference though, Eq APO does have the functionality to apply Eq, gains, and delay to each channel on the device.

 

[Marcus Aseth]

This is absolutely not necessary in your setup as the left and right are equidistant to you.

I'm not sure about that, for some reason following the tutorial on page 1, during the time alignment inside REW where in the video he got something like -0.4us I had a pretty big number... I'll share screenshots tomorrow

For future reference though, Eq APO does have the functionality to apply Eq, gains, and delay to each channel on the device.

And my device is seen as a single channel, correct?
If so, you guys know of an USB interface which comes with a volume knob that gives me the ability to set each channel from EQ APO?

 

[JStewart]

I'm not sure about that, for some reason following the tutorial on page 1, during the time alignment inside REW where in the video he got something like -0.4us I had a pretty big number... I'll share screenshots tomorrow

It’s not necessary.
He said at the very beginning he was not covering how to take measurements. Then if you look closely at his measurement tabs you can read they were taken with an acoustic timing reference.

You can not do any alignment with REW unless there is a timing reference. This is because the time it takes from when you press start to when it starts playing is not the same each time as your computer is busy with other stuff it thinks is more important than your measurement. A timing reference removes this inconsistency.

Then, in the video, he also says you can move your speakers and accomplish the same thing.

And my device is seen as a single channel, correct?

It has two channels if it’s stereo. The right and left channels are separate. Eq APO has the functionality to use separate filters for each channel on the device.

 

[ddude003]

I was just about to say that I did not see any acoustic timing reference in the latest L and R .mdat file you shared above... In instructions I linked to above @serko70 does suggest using acoustic timing reference... First bullet point under Measurement section...

If your speakers are on the desk in front of you, any head movement would change the perceived sound...

It appears to me that the UMC22 is a 2 channel stereo device... It should work for what you are trying to do...

 

[JStewart]

@Marcus Aseth I thought I would use your measurements on the first page of this thread for an alignment example. The absolute delays are not correct because the measurements were not taken with a timing reference, however the frequency response that can be achieved with correct timing between L+R and the Sub is correct.

For starters, this is REW's alignment tool. The measurements being aligned are the Sub and L+R. The black line is the combined response and the best I could achieve using the tool.

Next is a comparison of the your L+R with Sub and the Aligned L+R with Sub that was created with the alignment tool. Your measurement is the light blue and the black is the after using the alignment tool.

Perhaps also of note is the Group Delay. See how the subwoofer's sound is lagging behind the mains with the original L+R with Subs and not with the aligned version?

If you want to try to do this then a timing reference will be needed and the convenience of the alignment tool can not be used because you don't have a set up to accommodate injecting the delays needed into a piece of equipment at the right point in your signal chain. The sub phase can be tried though. I want to make mention again that the sub phase control should provide enough range to do this but there's no guarantee. Measurements will tell.

Do you primarily listen near field? (Near field = Sitting at the desk with the speakers on the desk?) If so then at some point you'll should at least try setting the XO at 50Hz and the slope to 24. This should better match the rolloff of the Kali's which are at approx 50Hz and 24dB per octave. But for now set the XO to what's worked best for you so far. Disable any active EQ. You will need to re-eq after this is done. Set the phase to 90 degrees. (we'll start in the middle).

For the measurements set the acoustic timing reference to the right channel.

Now lets measure the left channel. Note where the volume control is on on the sub, then turn it all the way down. Set the measurement range as 10Hz to 300Hz. Now take the measurement.
Expected behavior is you will hear a "chirp" from the right speaker and shortly after you will hear the sweep on the left speaker, but there will be no sound coming from the sub. The measurement will reveal if there is.

If everything worked as expected, measure the right channel. The right channel will remain the acoustic timing reference channel as well. We may or may not use this measurement but it will be easiest to get now.

Now let's measure the sub. Turn the power off to the left Kali so it can't play. Unplug the input after power is turned off if you know that you need to keep it 100% off. Turn the sub volume back to where it was before the left measurement. Now take the sub measurement.
Expected behavior is you will hear the chirp from the right speaker and shortly after the measurement sweep being played on the subwoofer only.

Now let's measure the sub again with the phase knob at 135 and a 3rd time at 180. Please label all measurements accordingly.

That's it. Now you have a some measurements with a timing reference that you can work with. Post the mdat if you wish and we can have a look too.

Edit: After all that I just realized I have this backwards. The mains are closer to your listening position than the sub so the mains will likely be the ones that need delay applied and not the sub. The phase control which will delay the sub even more is likely of no value. The measurements can still be done with a timing reference to verify.

 

[Marcus Aseth]

Edit: After all that I just realized I have this backwards. The mains are closer to your listening position than the sub so the mains will likely be the ones that need delay applied and not the sub. The phase control which will delay the sub even more is likely of no value. The measurements can still be done with a timing reference to verify.

Thank you for your time to try and solve it, but since doing this (the measuring and all) usually drains me a lot, if we know in advance it can't be solved that way then I would rather use the "mental energy" for the day attempting Serkan's video tutorial ^_^' (which will take me a long time already, probably hours)
Also yes, the speakers are placed Nearfield to the sides of the computer screen
But a valuable thing I take away from the post is that there is a group delay window in which I can see if my sub is lagging behind, problem is I cannot find it :\
How do I access Group Delay?

 

[serko70]

Apologize to Serkan, is not his method that failed, as I've just realized somewhere inbetween following steps I've completely destroyed Equalizer APO. I've tryied to uninstall it, reinstall it, restart, nothing, the filters are not being applied anymore.
This software is simply not reliable.
There is a better free alternative to Equalizer APO?

No worries. Btw, I have changed the target curve in the method to the top red line in the below chart (original harman curve - green line - was the second best):

@ddude003 this yields slightly better results in my system.

I have already posted this under the video I guess but attaching once again in this post just in case.

 

[Marcus Aseth]

No worries. Btw, I have changed the target curve in the method to the top red line in the below chart (original harman curve - green line - was the second best):

Thank you, perfect timing, I'll be using this then

 

[Marcus Aseth]

@serko70 That curve you've passed me above cannot be used because "is not compressed", I mean it uses some different scale or something, like it has variations of 60db or more - so I've went back to the previous curve from your video.
This was the final measured result after applying the convolution on Equalizer APO:

I've never understood that dip around 600Hz, is that a result of the homemade absorbing panels I've made? Or is it the room? Something else?
Anyway Is it good overall?
I guess I'll find out in the following days by listening to it.
I'm not happy about the dip in the 45dB region but that was there before and won't go away even by moving the subwoofer to the corners of the room
I should point out I did something different from your video, during the Trace Arithmetic step I've used a Regularization% of "4 (8dB max)" since I thought it needed some strong correction in certain points, and then a Gain of -9,10dB inside of Equalizer APO because after applying the convolution file it would warn me (in red) of a massive Peak Gain generated by that, so it would counterbalance it bringing back the Peak Gain value shown from Equalizer APO down to 0.0
Is that ok?

Finally, below I'll attach all the measurement obtained following the video (with notes for the SUB settings) and the final measurement obtained after applying the convolution file, in case someone wants to doublecheck and see if I could have obtained something better (likely I can, if someone passes me a usable version of that better curve )
Thanks everyone for all the help, I'm optimist it will sound great this time

 

[JStewart]

How do I access Group Delay?

In the row of buttons just above the graphs labeled GD or if you wish to compare measurements as I did then use the Overlays button found in the top row and select GD there.

BTW, if you ever have a mind to you could move the sub closer to the desk. Have you ever tried it on the opposite side of the desk from where you sit?

 

[Marcus Aseth]

In the row of buttons just above the graphs labeled GD or if you wish to compare measurements as I did then use the Overlays button found in the top row and select GD there.

BTW, if you ever have a mind to you could move the sub closer to the desk. Have you ever tried it on the opposite side of the desk from where you sit?

Yes, and it's extremely annoying there because the vibrations/air movement makes my clothing/pants/"feet rest platform" vibrate and the desk vibrates more as well, it becomes so annoying I can't enjoy the music even if it sounded good

 

[JStewart]

Yes, and it's extremely annoying there because the vibrations/air movement makes my clothing/pants/"feet rest platform" vibrate and the desk vibrates more as well, it becomes so annoying I can't enjoy the music even if it sounded good

Ok then. That spot is off the list. Lol.

 

[serko70]

@serko70 That curve you've passed me above cannot be used because "is not compressed", I mean it uses some different scale or something, like it has variations of 60db or more - so I've went back to the previous curve from your video.
This was the final measured result after applying the convolution on Equalizer APO:
View attachment 55403
I've never understood that dip around 600Hz, is that a result of the homemade absorbing panels I've made? Or is it the room? Something else?
Anyway Is it good overall?
I guess I'll find out in the following days by listening to it.
I'm not happy about the dip in the 45dB region but that was there before and won't go away even by moving the subwoofer to the corners of the room
I should point out I did something different from your video, during the Trace Arithmetic step I've used a Regularization% of "4 (8dB max)" since I thought it needed some strong correction in certain points, and then a Gain of -9,10dB inside of Equalizer APO because after applying the convolution file it would warn me (in red) of a massive Peak Gain generated by that, so it would counterbalance it bringing back the Peak Gain value shown from Equalizer APO down to 0.0
Is that ok?
View attachment 55404
Finally, below I'll attach all the measurement obtained following the video (with notes for the SUB settings) and the final measurement obtained after applying the convolution file, in case someone wants to doublecheck and see if I could have obtained something better (likely I can, if someone passes me a usable version of that better curve )
Thanks everyone for all the help, I'm optimist it will sound great this time

That same curve works perfectly with my REW, it's a simple txt file (generated in REW actually):

You can remove any house curve from preferences and then simply add this room curve in EQ which will produce the same target curve (You LF figures will be different of course)

I also have the 47 dB dip, it's a room mode and cannot be digitally corrected

Equalizer APO loves to warn for some peaks, I usually ignore them to be honest. I usually find out the same convolution file works in Roon DSP with no headroom warning. We are actually not boosting anything in this method, it's only lowering peaks. Make sure "normalize IR responses" option is "unticked" while exporting the final correction wav file, normalizing causes most of these warnings in EQ APO.

I checked your mdat file. The 500-600 Hz dips are probably caused by reflections from your desk (or whatever there's 20-25 cm away from your woofer drivers). If you don't like your results, you can try dimming down your subwoofer volume by 2-3 dB before measuring speakers and calculating the corrections.

 

[Marcus Aseth]

That same curve works perfectly with my REW, it's a simple txt file (generated in REW actually)

Thanks, I was mistaken about that, I got confused
This below is the final measured result using the curve.
should I do anything about those peaks at 25Hz and 31Hz? Maybe a peak filter? Or should I just ignore it?

 

[ddude003]

@Marcus Aseth we are really talking about your own personal taste in sound here... As Serkan suggested above "If you don't like your results, you can try dimming down your subwoofer volume by 2-3 dB before measuring speakers and calculating the corrections."

You can also create several Target curves with the method Serkan describes above in that EQ Target setting window... Try a few like flat from 20Hz to 20kHz, flat from 20Hz to 500Hz then -3 or -4 to 20kHz, flat from 20Hz to 1kHz and then -3 or -4 to 20kHz and the other 2 you already have, the Harman "hump" and the Harman +3 at 20Hz to -3 at 20kHz...

The 500-600-700 Hz dips you have may be from your computer monitor... Fold up a beach towel or two and hang them over the monitor and see there are any changes in your sweeps...

 

[ddude003]

Thank you @serko70... I appreciate all the work you've done... I have Martin Logan ESLs which, as you know, are a little different... I use a +2 at 20Hz sloping to a -4 at 20kHz target... I'm not a real bass head... I like my bass light and tight...

I have also done some studying with microphone position and found in my case that pointing the microphone dead center between my speakers and at 3/4 the electrostatic panel height, which it turns out to be about ear hight, gives me the best resultant frequency response sweeps... I read somewhere that for stereo you point the microphone at the center point between your speakers and for surround systems you go with the 90 degrees (point up)... As with all things audio, your milage may vary...

 

[Marcus Aseth]

@Marcus Aseth we are really talking about your own personal taste in sound here... As Serkan suggested above "If you don't like your results, you can try dimming down your subwoofer volume by 2-3 dB before measuring speakers and calculating the corrections."

You can also create several Target curves with the method Serkan describes above in that EQ Target setting window... Try a few like flat from 20Hz to 20kHz, flat from 20Hz to 500Hz then -3 or -4 to 20kHz, flat from 20Hz to 1kHz and then -3 or -4 to 20kHz and the other 2 you already have, the Harman "hump" and the Harman +3 at 20Hz to -3 at 20kHz...

The 500-600-700 Hz dips you have may be from your computer monitor... Fold up a beach towel or two and hang them over the monitor and see there are any changes in your sweeps...

I'm afraid that if I do consider doing that, I'll stay slave to the circle of confusion xD
If noone sees anything glaringly wrong in what I did, then I guess I have no reason to touch it anymore until I have a reason

 

[serko70]

Thank you @serko70... I appreciate all the work you've done... I have Martin Logan ESLs which, as you know, are a little different... I use a +2 at 20Hz sloping to a -4 at 20kHz target... I'm not a real bass head... I like my bass light and tight...

I have also done some studying with microphone position and found in my case that pointing the microphone dead center between my speakers and at 3/4 the electrostatic panel height, which it turns out to be about ear hight, gives me the best resultant frequency response sweeps... I read somewhere that for stereo you point the microphone at the center point between your speakers and for surround systems you go with the 90 degrees (point up)... As with all things audio, your milage may vary...

If you could point the mic directly towards the speaker, it's slightly more advantageous then pointing up however you cannot do that in a dead centre measurement. When you try to move the mic towards each speaker for L&R you will lose delicate timing information. SO pointing up and keeping the mic in place for both L & R is more useful for correction purposes.

 

[Marcus Aseth]

Hey guys, I'm back xD
The one above in the reply #43 was my last measurement with the current convolution filter applied, although in certain songs I'm finding the bass to be a bit "boomy" , is that because my room is small? (this is a small bedroom)
How should I approach this?
Should I go for a house curve that at the highest is 2dB lower than the one in reply #43 or maybe this is something that has to do with the low frequecies in a small space so I should have a curve that rolloff sooner? (like, instead of rolloff at 20Hz, maybe 25? 30? )

 

[Marcus Aseth]

ops, double post, I don't know how to delete this... (ignore this message, read the one above )

 

[ddude003]

Seems like you have a few options... Add some bass absorption, adjust the level/gain on the sub, try a few other house curves to flatten out the bass response a bit... Or any combination of the above... EQ/DSP alone may not get you over the room response especially in the sub bass and bass frequencies... Does that sub have its own EQ? If it does try to flatten those peaks that... Might be faster than trying several convolution filters... It might point the way to go with the filters if you choose to run you sub flat and do all the room correction with convolution...