Dry bass

[sigbergaudio]

Your wish is my command! This is 20-200Hz unsmoothed.

View attachment 333317



@Geert all the full range measurements have subs included. It is not possible to remove the subs and still get a decent measurement. As I indicated earlier, the speakers have been modified with a replacement woofer that is linear between 50-800Hz with very little bass output below 50Hz. This was a deliberate decision, because I knew that I would be removing all sub frequencies from the main speaker.

Thank you. But before even commenting that, I see that you're writing that the full range uncorrected measurement also includes your subwoofers. The response of your subs uncorrected is absolutely horrible. You should really try to move them around to improve the response before applying any EQ. Like others here I'm a bit worried what your DSP has done to correct that.

Are the subs set up in stereo pairs? So two subs on the left and two on the right? Have you considered trying mono subs to help even out the response?

EDIT: I see now that you only have two subs with four drivers. Anyway, I would strongly recommend testing to run them in mono and EQing them as a pair.

On the topic of your 20-200hz graph: As replied on your fullrange graph, you have too steep roll-off from 80-200hz or 80-300hz depending on what curve you're looking for. It's also seems like you're both tuning the speakers and the room at the same time? It's a bit problematic since we don't know the natural curve of your speakers in-room. Ideally they should be tuned to be reasonably flat anechoically/nearfield first, and then we can see how that looks in the room, and then the bass should be tuned to more or less match that. Now you're just forcing the speakers to a given curve without really knowing what the room is contributing.

 

[Keith_W]

Thank you. But before even commenting that, I see that you're writing that the full range uncorrected measurement also includes your subwoofers. The response of your subs uncorrected is absolutely horrible. You should really try to move them around to improve the response before applying any EQ. Like others here I'm a bit worried what your DSP has done to correct that.

I think all uncorrected subs look absolutely horrible. What is especially horrible about mine? Honest question.

Are the subs set up in stereo pairs? So two subs on the left and two on the right? Have you considered trying mono subs to help even out the response?

EDIT: I see now that you only have two subs with four drivers. Anyway, I would strongly recommend testing to run them in mono and EQing them as a pair.

That's right. This is what the subs look like:

I designed them myself, I took the Vas recommended by Rythmik for Q of 0.7 and got someone to CAD up the design and CNC the boards out. I then sent the boards to a furniture maker and got him to assemble the subs and match the veneer to my existing furniture. It is quite possible there are flaws with my sub design given that I am not a speaker designer, but then the guy who did the CAD on the design is a speaker designer and he thought it looked like a reasonable design.

On the topic of your 20-200hz graph: As replied on your fullrange graph, you have too steep roll-off from 80-200hz or 80-300hz depending on what curve you're looking for. It's also seems like you're both tuning the speakers and the room at the same time? It's a bit problematic since we don't know the natural curve of your speakers in-room. Ideally they should be tuned to be reasonably flat anechoically/nearfield first, and then we can see how that looks in the room, and then the bass should be tuned to more or less match that. Now you're just forcing the speakers to a given curve without really knowing what the room is contributing.

Yes, sadly the speakers don't have a natural curve any more. Removing the passive crossover and replacing it with DSP took care of that. Not to mention replacing the woofer. I have wanted to measure the speakers under anechoic conditions for some time, but these speakers are massive (110kg each), and requires all the amplifiers and the computer for it to even work. It's not as if I can pop the speakers in my car and rent an anechoic chamber, we are talking a major logistics challenge. And nobody in Melbourne owns a Klippel, so I can not measure them in situ. I did try the quasi-anechoic measurement using beamforming as described by Krol et al in 2015 but I did not find the exercise useful. I can only make the best of what I have. So yes, the speakers and the room are tuned at the same time.

 

[sigbergaudio]

I think all uncorrected subs look absolutely horrible. What is especially horrible about mine? Honest question.

Well, two creatively placed subwoofers typically give far less problems than your "before" picture". The circled area is obviously quite problematic.

For the green subwoofer you could theoretically reduce the peaks on each end and then increase the gain of the entire subwoofer. For the red one, it looks like a dip that can't really be fixed. The 30-50hz is an area where what you are missing may well live.

That same 30-50hz area is completely flat in your "after" graphs, which I find ..interesting. I wonder what the DSP has done here.

Here are two uncorrected subs (mono) in one of my listening rooms. I would say it is (as an alternative example) significantly less horrible, since there are no dips to speak of below 100hz. So you can just reduce a couple of peaks and then you're good to go.

 

[Geert]

I think all uncorrected subs look absolutely horrible. What is especially horrible about mine? Honest question.

Major deficiencies in the largest part of their operational frequency range.

Which points to a room issue.

As a comparison, this is what the response of my single 6"-driver bookshelves in my living room looks like:

Potential issue might be that your listening position is in the middle of the width of the room, and also the sliding door behind you might be of importance. It might be interesting to measure the response at the edge of your couch (or just take a listen there to start with).

What I would like is more impact and slam, and @ppataki's suggestion of the Pultec EQ has gone a long way towards solving it. It is a few hours now since I started tuning, and I have improved the result audibly. It is still not quite where I would like it, at the moment I CAN get impact and some physicality, but my impression is that it is too low in the freq range, and tuning it higher results in bloated, muddy bass. There is a sweet spot somewhere, and I am going to find it

That confirms my idea that you just need more lows, and indeed, it's a matter of fine tuning. The higher you go in frequency, the more muddy it usually sounds.

 

[Sokel]

Forgetting their response for a minute,what are you driving your subs with?

Their uncorrected level even with that response should be at least 5-10db higher than the mains so they don''t drug them down with them during the correction.
Dragging the whole thing down with a curve as-is means that you waste 20-30db of power,let alone the penalty of the levels on DSP as if it's not enough the penalty applied by the filters alone.

The huge sensitivity difference between the drivers is not something that fixes easily as for levels like that the real overall sensitivity will go into seventies.
Even applying KW after that for peaks the drivers will be a limiting (or destroyed) factor.

 

[Tangband]

Your wish is my command! This is 20-200Hz unsmoothed.

View attachment 333317


20-20kHz with 1/12 smoothing has already been posted. Sorry I missed your earlier reply, this thread seems to have generated a lot of interest from a lot of big guns on ASR for some reason. I am really grateful for that, thanks guys!



All the REW responses have 1/12 smoothing unless indicated otherwise.

@Geert all the full range measurements have subs included. It is not possible to remove the subs and still get a decent measurement. As I indicated earlier, the speakers have been modified with a replacement woofer that is linear between 50-800Hz with very little bass output below 50Hz. This was a deliberate decision, because I knew that I would be removing all sub frequencies from the main speaker.

So it is clearer what we are talking about, this is a sketch of the listening room:

View attachment 333321

At the bottom of the pic you can see a staircase and a narrow corridor leading to the dining room. At the back of the room you can see a large sliding door leading to the backyard. The listening sofa is positioned 1.5m away from the rear sliding door. The subwoofers were positioned by placing one sub at the MLP and doing sweeps around the room. I selected the positions with the fewest nulls. The main speakers were positioned by ear, where they produced the best stereo image, that's where they went.

All the drivers have individual driver correction (you can see that in the crossover that I posted earlier). Each driver has a reversed all pass filter convolved into the correction to flatten the phase angle. All drivers are time aligned with respect to the tweeter: horns 0.0208ms, woofers 0.312ms, and subs 1.875ms using Acourate's time alignment procedure. Then overall room correction is applied with target curve applied full range. I am aware of the debate of correcting up to Schroder only or full range @Tangband, in fact I used to only correct up to Schroder but I now do it full range. It sounds better, in fact the clarity of the system is pretty astonishing. Veils were lifted, night and day, wife could hear difference from the kitchen, etc. I am not joking, the difference all these corrections make can really be heard from the kitchen!

There is absolutely nothing wrong with the system from 200Hz up, apart from perhaps the polar response but there is nothing I can do about that. Even the bass is pretty good, there are no audible peaks and dips, it sounds tight, and I am almost happy with it. What I would like is more impact and slam, and @ppataki's suggestion of the Pultec EQ has gone a long way towards solving it. It is a few hours now since I started tuning, and I have improved the result audibly. It is still not quite where I would like it, at the moment I CAN get impact and some physicality, but my impression is that it is too low in the freq range, and tuning it higher results in bloated, muddy bass. There is a sweet spot somewhere, and I am going to find it. This post was made while I am in the middle of tuning, so back to tuning I go. I will report if I achieve a satisfactory result.

Ok !
Small difference, but might be in the right direction - put spikes below both main speakers and all subwoofers. ( its a coloration but might be perceived as better in the ”impact” department ) . You can also try a more shallow crossover topology ( i.e. second or third order )

 

[Bjorn]

I thought the graph earlier was without the subs. If it's with the subs, it's not good. And what's difficult to know here is what is kind of DSP has been done. It's one thing to EQ the speaker anechoic, another entirely to correct the room response. If much of the latter has been done, and which I suspect, the result isn't going to be great.

The listening position doesn't look very ideal with close to proximity to the rear wall/windows. Perhaps you should also try to put a chair closer for criticial listening. Room modes is something we have to move out from unless it can be treated.

 

[Deleted member 48726]

I asked what the distortion graph looks like earlier. You may be excused by the vast amount of curious comments, but I am curious to see what that looks like on the corrected measurement. Especially given the thought that your DSP tries to correct for a deep null and it seems to be getting away with it..

Do you have any way to monitor the signal dB from the DSP to check if it is clipping? Might not be audible in the low frequencies.

-Correct me if my theory is off the rails.

 

[Keith_W]

I asked what the distortion graph looks like earlier. You may be excused by the vast amount of curious comments, but I am curious to see what that looks like on the corrected measurement. Especially given the thought that your DSP tries to correct for a deep null and it seems to be getting away with it..

Do you have any way to monitor the signal dB from the DSP to check if it is clipping? Might not be audible in the low frequencies.

-Correct me if my theory is off the rails.

Well, this is where my REW n00bness prevents me from showing the distortion curve.

... it doesn't show up. I don't know why.

 

[Geert]

... it doesn't show up. I don't know why.

Did you try disabling 'Mask harmonics below noise floor', as doing distortion measurements at the listening position can be difficult because of ambient noise. (Better measure near-field).

 

[Keith_W]

OK gents, this is the input sweep into Acourate, before overall room correction. ONLY driver correction and volume levelling between drivers has been performed. This involves a nearfield measurement of each driver. It is then corrected for frequency response, reverse all pass filter convolved, etc. and then all the drivers are volume matched and a sweep is taken.

This is the nearfield measurement of the sub, prior to driver correction.

Those REW graphs I posted earlier were found when I rummaged through my collection of .mdat files. Those are uncorrected sweeps when I was experimenting with MSO. I can not explain why they look so different to Acourate's sweep. I even labelled the sweeps correctly, "MLP Left Full" to me means sweep taken from MLP, left side, speaker+sub. So I think those sweeps include the sub. Sorry guys, I know that I must be really frustrating all of you right now. My apologies.

Did you try disabling 'Mask harmonics below noise floor', as doing distortion measurements at the listening position can be difficult because of ambient noise. (Better measure near-field).

Yeah, I did. I have a measurement of individual drivers somewhere else in my .mdat collection. Let me look through it again.

(EDIT) I found it. Sadly, no measurement of subwoofer distortion. So to keep everyone slightly amused until my I get my interface back, this is a measurement of woofer distortion. I can't remember what volume it was taken at, so sorry. I should have been more thorough and taken repeated sweeps at different volume settings, but I took these sweeps with another purpose in mind, and it wasn't for looking at distortion.

 

[sigbergaudio]

OK gents, this is the input sweep into Acourate, before overall room correction. ONLY driver correction and volume levelling between drivers has been performed. This involves a nearfield measurement of each driver. It is then corrected for frequency response, reverse all pass filter convolved, etc. and then all the drivers are volume matched and a sweep is taken.

View attachment 333371

Is it possible to adjust graph view in Acourate? The Y-graph is like 130dB span, it should be 50dB. And add smoothing if possible.

 

[Keith_W]

Is that better?

 

[sigbergaudio]

View attachment 333410

Is that better?

I guess. And the same view with room correction applied?

 

[Kvalsvoll]

I think all uncorrected subs look absolutely horrible. What is especially horrible about mine? Honest question.



That's right. This is what the subs look like:

I designed them myself, I took the Vas recommended by Rythmik for Q of 0.7 and got someone to CAD up the design and CNC the boards out. I then sent the boards to a furniture maker and got him to assemble the subs and match the veneer to my existing furniture. It is quite possible there are flaws with my sub design given that I am not a speaker designer, but then the guy who did the CAD on the design is a speaker designer and he thought it looked like a reasonable design.



Yes, sadly the speakers don't have a natural curve any more. Removing the passive crossover and replacing it with DSP took care of that. Not to mention replacing the woofer. I have wanted to measure the speakers under anechoic conditions for some time, but these speakers are massive (110kg each), and requires all the amplifiers and the computer for it to even work. It's not as if I can pop the speakers in my car and rent an anechoic chamber, we are talking a major logistics challenge. And nobody in Melbourne owns a Klippel, so I can not measure them in situ. I did try the quasi-anechoic measurement using beamforming as described by Krol et al in 2015 but I did not find the exercise useful. I can only make the best of what I have. So yes, the speakers and the room are tuned at the same time.

Very nice subwoofers. Just wanted to say, because I commented about subwoofers "may be inadequate", what I meant was that you might need even much more, to achieve what you want, not that there is something wrong with what you have and certainly not lacking compared to other alternative offerings.

 

[Sokel]

Well, this is where my REW n00bness prevents me from showing the distortion curve.

View attachment 333367

... it doesn't show up. I don't know why.

It's because your mic is not calibrated for SPL.
You can try change "SPL" in the upper left corner for "%" but I don't know if it works too.

 

[Keith_W]

I guess. And the same view with room correction applied?

Here it is for the left channel only. "Before" and "after" superimposed on each other. The target curve I used is in red.

As discussed earlier, I am not sure why there is a dip between 200Hz - 400Hz because REW's opinion is different. It might be different smoothing algorithms, I don't know. With Acourate I used FDW 6/6, with REW 1/12 smoothing.

 

[mcdn]

Some more questions:

• It seems like you're not using a calibrated microphone, is that correct? What mic are you using?
• According to your picture, your subs have one driver pushed hard against the front wall and the other facing toward you. Is this deliberate? Are the subs standard push-push/monopole or are they push-pull/dipole?
• An REW room sim based on your stated dimensions of 7x6x3m and some guesses on speaker position doesn't give a dip at 80Hz. Have you considered why?

And some suggestions:

• Move the main speakers toward the front wall a _lot_. All standard advice on this is based on old assumptions of passive speakers without subs. Your current setup is just asking for SBIR issues, and since you have DSP you can EQ out any peaks moving closer to the wall creates.
• More bass - move the 20-150Hz region up by 6dB
• Less upper-mid and treble - drop the whole region from 700Hz-20kHz by 3dB

 

[Keith_W]

Some more questions:

• It seems like you're not using a calibrated microphone, is that correct? What mic are you using?
• According to your picture, your subs have one driver pushed hard against the front wall and the other facing toward you. Is this deliberate? Are the subs standard push-push/monopole or are they push-pull/dipole?
• An REW room sim based on your stated dimensions of 7x6x3m and some guesses on speaker position doesn't give a dip at 80Hz. Have you considered why?

And some suggestions:

• Move the main speakers toward the front wall a _lot_. All standard advice on this is based on old assumptions of passive speakers without subs. Your current setup is just asking for SBIR issues, and since you have DSP you can EQ out any peaks moving closer to the wall creates.
• More bass - move the 20-150Hz region up by 6dB
• Less upper-mid and treble - drop the whole region from 700Hz-20kHz by 3dB

1. I have 2 calibrated mics and 1 uncalibrated mic. Earthworks M30, and two Behringer ECM8000's. One of the ECM8000's is uncalibrated.
2. The driver in the sub is not pushed hard against the wall, there is about 20cm of space behind the driver and the wall. They are dual opposed subs, i.e. push-push.
3. If you are talking about that REW measurement, I found that in my collection of REW .mdat's. I don't remember why or what I did that made it look so bad, but as you can see from other measurements it looks OK. Perhaps it's from non-MLP and I labelled it MLP by mistake. I don't know
4. Why should there be SBIR issues if the speakers are not close to a boundary? (apart from the floor and ceiling of course, but I assume you are not talking about those). The moment I move my speakers I have to redo the entire room correction exercise, so there better be a very good reason for moving the speakers!
5. Yeah, I am currently experimenting with more bass with the Pultec EQ.
6. I am happy with the upper-mid + treble the way it is, unless you are suggesting that dropping these might improve bass impact?

 

[mcdn]

1. I have 2 calibrated mics and 1 uncalibrated mic. Earthworks M30, and two Behringer ECM8000's. One of the ECM8000's is uncalibrated.
2. The driver in the sub is not pushed hard against the wall, there is about 20cm of space behind the driver and the wall. They are dual opposed subs, i.e. push-push.
3. If you are talking about that REW measurement, I found that in my collection of REW .mdat's. I don't remember why or what I did that made it look so bad, but as you can see from other measurements it looks OK. Perhaps it's from non-MLP and I labelled it MLP by mistake. I don't know
4. Why should there be SBIR issues if the speakers are not close to a boundary? (apart from the floor and ceiling of course, but I assume you are not talking about those). The moment I move my speakers I have to redo the entire room correction exercise, so there better be a very good reason for moving the speakers!
5. Yeah, I am currently experimenting with more bass with the Pultec EQ.
6. I am happy with the upper-mid + treble the way it is, unless you are suggesting that dropping these might improve bass impact?

Awesome data! No wall issues with the subs, that's good.

On (4) SBIR, it's always there, but the frequencies at which it is of interest decrease as you move the speakers away from the boundaries. If you are using DSP it's actually best to move the speakers close to the front wall. This moves the first SBIR freq upwards towards the baffle step freq of the speakers. Basically approximating soffit mounting. Appreciate it's not plausible for you right now but something to think about. Also your main speaker woofers are currently almost equidistant from the front wall, side walls and ceiling. That's never going to be great.

On (6) The upper mid and treble, yes I am suggesting if you drop it to a more natural curve without that 6dB discontinuity at 700Hz your brain will hear more bass. You don't have a peak at 700Hz, you have a 6dB shelf!