Keith_W DSP system

[3ll3d00d]

And this is FDW 15/1 (thick line above) compared with FDW 15/15 (lighter green line below). Notice that with the FDW 15/1 there is tight bass correction but it smoothens out at the top end, i.e. Toole's "broad tone control".

It looks exactly as you'd expect from the description of how the window narrows, which bit is unexpected to you?

 

[Keith_W]

It looks exactly as you'd expect from the description of how the window narrows, which bit is unexpected to you?

Maybe my thinking is upside down. I will think about it a bit more.

 

[Keith_W]

I had a bit of a heart stopping moment tonight. I was chatting to @joentell and he suggested that I measure my left and right sub, and both of them together in REW. Now I have absolutely no concerns about the subs being different. I use Acourate's ICPA (Interchannel Phase Alignment) feature, and I routinely do a mono (L=R) verification sweep to make sure there is no cancellation.

So I loaded up the filter into the convolver and sent the REW sweep through it. Result (Red/Green = Left and Right sub, Blue = both together).:

I am normally a polite person and I swore when I saw this result. Could Acourate be lying to me? So I started Acourate and did the same sweep:

Subs + Tweeter, with the other drivers turned off. Red/Green = left/right sub, brown = both together. Looks perfect to me.

The most obvious place to look for the problem is the fact that REW and Acourate measure through different signal paths. As mentioned earlier in this thread, REW has no built-in convolver. So the signal path for REW goes through JRiver, then Acourate Convolver, then Merging NADAC, then to the speakers. The incoming mic signal goes through the RME. There is considerable latency when measuring with REW because of the complex signal path, so I always use a timing chirp. The output from Acourate is much simpler, it has a built-in convolver so Acourate outputs directly to the RME.

The more complex the signal chain, the more points of failure there are. Eventually we isolated it down to this:

Inverted polarity in Channel 2 (right subwoofer) on the Merging. I had to fiddle with the control panel on the Merging tonight when I was taking measurements, I must have accidentally inverted the polarity.

So yeah, nice demonstration of bass comb filtering in my room.

 

[Deleted member 48726]

Midrange Horns

The horns are 45cm diameter spherical horns. In theory, the lowest frequency supported by the horn is lambda = 2x diameter, so 90cm wavelength = 381Hz. The upper frequency is 3 octaves above this, so 3000Hz. Below this limit, the horn loses efficiency because the wavelength is not "horn coupled". Above the upper limit, the sound starts to beam.

The horns are built into a separate cabinet which sits on top of the woofer cabinet. Access to the driver is obtained by unscrewing several inconveniently located Allen screws. It then comes off like this:

The enclosure is filled with sand. The driver is protected by a rubber boot. Remove the rubber boot, and we see:

A Dynaudio D-52H driver with Acapella branding.

A friend of mine noticed the construction of the cabinet and said it was very German. Because only Germans would construct a cabinet out of thick plywood, line it with lead, and then think to themselves "let's fill it up with sand".

At the bottom of the cabinet, we see the passive crossover. It was bypassed at this stage:

To reassemble the cabinet, the horns are screwed back into place. There is a small access port at the rear of the cabinet, where sand can be slowly filled with a funnel:

This is a very old measurement of the horns with the native crossover in situ:

And this is the measurement of the horns with the crossover bypassed, with no linearization and no crossover applied:

View attachment 345185

You can see that the lower limit of the horn (440Hz) is fairly close to our calculated theoretical lower limit of 381Hz.

View attachment 345186
That impulse response looks pretty good to me.

View attachment 345187
And this is what the horns look like once I have completed driver linearization.

Oh, the Dynaudio D-52 is a precious relic! Very well built with various configurations. I have the D-52 AF in my Kappas instead of the infamous polydomes of which I couldn't find a proper dome for anywhere in the world.

Below is a snippet from the data-sheet from Dynaudio with measurements done by Bruel & Kjaer-->

 

[dualazmak]

Hello again, Keith;

You wrote:

I prefer Acourate's 15/15 FDW when displaying graphs, it is detailed enough not to hide anything that is going on, but smooth enough for you to discern the trend.

This point and your recent posts made yesterday and today reminded me our (my) wonderful discussion with Dr. Toole last year in February on this thread entitled "What is your favorite house curve".

Dr. Toole kindly wrote at the end of his amazing post here that "Don't worry about little ripples. When I see exceptionally smooth high-resolution room curves I strongly suspect that something wrong has been done. The measurement microphone is no substitute for two ears and a human brain." (Boldface italic color underline are given by myself.)

Then I cordially asked him (ref. here) about Fq response measurement by "short-time sine sweep" vs. "cumulative white noise averaging".
Dr. Toole very kindly responded (ref. here) saying "If properly done both swept tone and noise analysis should give identical answers. It is a choice. The principal difference is in the heating of the drivers in sustained tests at high sound levels - power compression. Low frequencies require longer averaging times."

I am just curious about have you ever tested primitive but reliable and reproducible "cumulative white noise averaging" (ref. here and here) for Fq response measurement at your listening position while all the L+R SP drivers are all singing together, or not?

Just for your possible reference and interest, I measured again the total Fq response of my DSP system (ref. here) by "cumulative white noise averaging" the day before yesterday as shown in this diagram;

(As for the calibration data for my specially selected BEHRINGER ECM8000 microphone, please refer here on my project thread.)

 

[Keith_W]

Hello again, Keith;

You wrote:

This point and your recent posts made yesterday and today reminded me our (my) wonderful discussion with Dr. Toole last year in February on this thread entitled "What is your favorite house curve".

Dr. Toole kindly wrote at the end of his amazing post here that "Don't worry about little ripples. When I see exceptionally smooth high-resolution room curves I strongly suspect that something wrong has been done. The measurement microphone is no substitute for two ears and a human brain." (Boldface italic color are given by me.)

Yeah, I know what Toole said. I think I made it pretty clear in my post that I don't use FDW 15/15 for correction. I only use it to show you the curves on ASR. So you have to ask yourself, why did Toole say that? What is wrong with an "exceptionally smooth high resolution" correction? This is because an omnidirectional microphone does not distinguish between direct and reflected sound, whereas your ears are not omnidirectional, they were specifically evolved to capture directional sound, and our brains evolved to process it. Therefore, omnidirectional mics have the potential to inappropriately correct for reflected sounds as well - in other words, perform corrections for problems that do not exist. Hence Toole's concern.

I have spent a lot of time pondering Toole's point and how to perform proper correction. Toole's solution is to perform "broad tone controls" to adjust for discrepancies in recordings or room induced tilts, but this relies on the speaker being well tuned in the first place (flat on axis response under anechoic conditions, and constant directivity). As I have demonstrated in other posts, the mixed driver types in my speakers do not behave in a typical way - so a flat nearfield anechoic response results in an upward tilt. Then there is the logistical issue of how to do that measurement in the first place, these are not speakers that you can casually carry outside and lift on a table.

Toole is essentially arguing against a finely granular correction. After all, a "broad tone control" is the same as a loosely applied target curve. It is trivially easy for me to choose what kind of correction to make. I can make it really hug the target curve, or I can make it "sort of" follow the target curve. It is not as if all these DSP corrections are set in stone, I can make a couple of filters with different settings and compare. I can immediately switch between them by pushing a button on the convolver, and the music does not even have to stop. Some people on this thread have accused me of "ignoring Toole", but if I have 10 filters loaded into my convolver, some of which "ignore Toole", so what? One push of a button and I am no longer ignoring Toole.

In the absence of a proper anechoic measurement, the far easier solution is to make a lot of target curves and choose one that I like. Even forcing a Harman-like curve on the speakers if necessary. It won't be too far off the actual performance of the speaker if it behaved like a normal speaker.

And one last point: I am a hobbyist, and I perform these corrections for myself. I am not a speaker manufacturer making erroneous or misleading marketing claims about my speakers. I share because I thought some people on ASR might find it interesting, even though I know I am opening myself up for criticism. I am not going around telling you that this is how it should be done, and this is what you need to do in your own system. I have no choice but to compromise, and this is my compromise.

Then I cordiary asked him (ref. here) about Fq response measurement by "short-time sine sweep" vs. "cumulative white noise averaging".
Dr. Toole very kindly responded (ref. here) as "If properly done both swept tone and noise analysis should give identical answers. It is a choice. The principal difference is in the heating of the drivers in sustained tests at high sound levels - power compression. Low frequencies require longer averaging times."

I am just curious about have you ever tested primitive but reliable "cumulative white noise averaging" (ref. here and here) for Fq response measurement at your listening position while all the L+R SP drivers are all singing together, or not?

I haven't. There is another good reason for not doing it besides the one Toole mentioned - I do not want to subject my ears to prolonged high volume white noise. It does not give me any information that I do not already obtain through my existing methods, so I haven't bothered. I do perform MMM's with pink noise though. Pink noise sounds slightly lower pitch than white noise, so I hate it slightly less, but I still hate it.

And BTW @OCA I have bought all the cables I need to rotate the system 90 degrees towards the long wall. I might have some time this weekend to do it.

 

[AudioJester]

For your subs, just run REW directly to your RME dac and turn off amps to horns and tweeters. Then compare with Acourate.

 

[OCA]

And BTW @OCA I have bought all the cables I need to rotate the system 90 degrees towards the long wall. I might have some time this weekend to do it.

 

[dualazmak]

And one last point: I am a hobbyist, and I perform these corrections for myself. I am not a speaker manufacturer making erroneous or misleading marketing claims about my speakers. I share because I thought some people on ASR might find it interesting, even though I know I am opening myself up for criticism. I am not going around telling you that this is how it should be done, and this is what you need to do in your own system. I have no choice but to compromise, and this is my compromise.

Yes, I fully understood your points; neither am I, I too am amateaur hobbyist. I was sharing my measurement approach as one of the possible choices for you and many other people visiting this wonderful and informative thread; you would please go and proceed with your own approach!

Even though the physical setup (DSP multichannel multi-SP-driver multi-amplifier) looks similar in your setup and mine, the major difference is that you are using quite unique big high-end midrange horn and tweeter horn. On the other hand, I use cone subwoofer, cone-woofer, Beryllium-dome-midrange, Beryllium-dome-tweeter, and only one metal horn for my super-tweeter.

I well understand your difficulties (and compromise) in optimization of woofer to midrange horn continuity. And, your various efforts are actually giving us (me) much interesting insights for our system tuning. Yes, we have our own approach and "compromisation" depending on each of our audio setups!

Toole is essentially arguing against a finely granular correction. After all, a "broad tone control" is the same as a loosely applied target curve. It is trivially easy for me to choose what kind of correction to make. I can make it really hug the target curve, or I can make it "sort of" follow the target curve. It is not as if all these DSP corrections are set in stone, I can make a couple of filters with different settings and compare.

This suggestion by Dr. Toole would be rather feasible and suitable at least in my system with five SP drivers, and this is one of the many reasons for that I intentionally use four HiFi integrated amplifiers and active sub-woofers having remote gain/volume for flexible on-the-fly broad relative gain (tone) control in analog domain. Of course I can do such on-the-fly relative gain control with DSP EKIO and/or DAC8PRO in upstream digital domain, but usually I do not like to change DSP parameters on-the-fly in order to avoid possible mis-setting which would have possibility of damaging SP drivers (ref. here and here).

Spoiler

...
Fourth, the digital numeric keyboard value input/change of gains on-the-fly (while listening to music) should be always avoided since we may easily have mis-typing the value, e.g. we may type as dangerous "+35 dB" instead of actually intending "+3.5 dB". In this context, I highly recommend you using mouse-wheel-rotation up-and-down on gain controllers in EKIO for on-the-fly gain/volume control; in EKIO the granularity of mouse-wheel rotation is 0.1 dB, which is really nice. (I myself actually requested the mouse wheel operation, and Guillaume of LUPISOFT very quickly incorporated it into updated EKIO.) Furthermore, EKIO's "Mute" and "Solo" buttons in each of the output chanel panels are really nice and useful for specific-channel(SP-driver)-only listening and/or any combination of all the SP drivers, even L-only, R-only, L+R, etc., during your precise DSP-parameter tuning procedures.

There is another good reason for not doing it besides the one Toole mentioned - I do not want to subject my ears to prolonged high volume white noise. It does not give me any information that I do not already obtain through my existing methods, so I haven't bothered. I do perform MMM's with pink noise though. Pink noise sounds slightly lower pitch than white noise, so I hate it slightly less, but I still hate it.

Well understood, I too do not like to expose my ears and brain to white noise for long!

I have two audio(-visual) dedicated PCs in my listening room, and whenever I perform rather intensive Fq measurement sessions with white noise (not so loudly though), after setting-up microphone at listening position for recording in my second PC, I go up to my office upstairs where I have powerful Xeon CPU workstation PC and large dual PC monitor (total 5120 x 1440 pixel desktop) with which I can control and perform all the recording sessions at downstairs through remote desktop (via wired GB LAN, actually I use AnyDesk) while my wife would be out of home! The recorded white noise tracks can be analyzed anytime afterwards by using Adobe Audition 3.0.1.

 

[Keith_W]

Hi @Mikig, I am sorry I forgot to reply to your post earlier. I am glad I checked, I don't want you to think I am ignoring you

My trivial question was asked because I have the same problem with the room, and I'm looking for points of reference on the fundamental themes, so I need to start from the most "trivial" definitions to at least compare on the basis.

I find myself struggling with a fairly large room, which complicated everything. and, finding an acceptable position for me without measuring instruments was, and is, a challenge. I've already turned it three times, now it seems like the best solution.

The others have already made the point that all multi-driver speakers "converge" a certain distance from the speaker. I would like to tell you that as long as your listening distance is beyond the convergence point, the speaker would behave as a point source and you are good to go. However, this has not been my subjective experience. Maybe it is expectation bias or something else, but I have found that large multi-driver speakers never sound like point source speakers. One good explanation is that the vertical directivity might be very different. Another is that the "size of the wavefront" (something I have speculated on but never been proven because I don't know how to measure it) is much bigger in a multi-driver speaker.

Either way, I think that large multi-driver speakers sound different to point source speakers. I do not have a good explanation why. Point source speakers have their own compromises, which is different from speaker to speaker. In the end I think you have to decide for yourself if you like that type of sound or not.

I have a Mac and if I also want a microphone, what type of program can be used to start giving a more scientific sense to this type of research that I have always had to do "by ear"??

The best place to start would be REW. It is free, and it runs on a Mac. Although I do use REW, I don't think it is a great tool for an active multi-driver setup like mine because it lacks a built-in convolver. I have mentioned this several times in this thread. I am also not as proficient in REW as other ASR members like @OCA (or nearly everyone really) so I don't know what it is really capable of. The downside of REW for new users is that it is not as automated as other software packages like Audiolense or Dirac, and it lacks wizards or macros. Everything is a manual process. This is absolutely not an issue if you know what steps you need to take. But nothing beats "free", and there is no doubting its ability and versatility, so IMO everyone in this hobby NEEDS REW and a microphone. If would go so far as to say that anybody who doesn't have a microphone and REW, or planning to obtain both, should have their ASR membership withdrawn

Beyond REW, IMO the two best software packages are Audiolense and Acourate. Neither of these run on a Mac, unless you use a Windows emulator. They have different strengths and weaknesses, and whether one or the other would be suitable for you depends on your needs. If you do not plan on separating your crossover and going fully active, I would say Audiolense. If you do plan on separating your crossover, I would say Acourate.

I know that Dirac runs on a Mac, but Dirac is astoundingly expensive compared to the above two, especially if you want the crossover option. I hear it is easier to use than both Audiolense/Acourate, but I do not have direct experience with it so I won't say more. I have been told that Focus Fidelity is even easier to use, and it runs on a Mac, but it has no crossover function. It is a relatively new product and I have seen posts shared by David (the developer) that more features are upcoming. So it is a product to watch, I am sure its capabilities will expand.

By the way, the no. 1 reason for poor sounding DSP correction is user error. I am still making mistakes despite doing this for years. See above post where I accidentally inverted the polarity of the sub Therefore, I place a high value on ease of use, and especially automation for new users. I myself like manual control, but climbing that learning curve can be pretty intimidating. I am not going to tell you to do what I have done (except get REW!) because I think there is a different DSP solution for everybody, depending on what your needs are.

I'll ask you the second question!! I noticed that you, and also many other users, use a piece of furniture between the speakers and the wall: have you ever measured whether this piece of furniture becomes a "fake speaker" that creates a sound box?? Thanks good evening!

If you are talking about the entertainment cabinet, I haven't measured its effect. It's not exactly easy to move, I have to remove all the equipment inside it, remove the TV, and it is heavy. It is also irrelevant for me, because I need it there.

 

[Mikig]

Hi @Mikig, I am sorry I forgot to reply to your post earlier. I am glad I checked, I don't want you to think I am ignoring you



The others have already made the point that all multi-driver speakers "converge" a certain distance from the speaker. I would like to tell you that as long as your listening distance is beyond the convergence point, the speaker would behave as a point source and you are good to go. However, this has not been my subjective experience. Maybe it is expectation bias or something else, but I have found that large multi-driver speakers never sound like point source speakers. One good explanation is that the vertical directivity might be very different. Another is that the "size of the wavefront" (something I have speculated on but never been proven because I don't know how to measure it) is much bigger in a multi-driver speaker.

Either way, I think that large multi-driver speakers sound different to point source speakers. I do not have a good explanation why. Point source speakers have their own compromises, which is different from speaker to speaker. In the end I think you have to decide for yourself if you like that type of sound or not.



The best place to start would be REW. It is free, and it runs on a Mac. Although I do use REW, I don't think it is a great tool for an active multi-driver setup like mine because it lacks a built-in convolver. I have mentioned this several times in this thread. I am also not as proficient in REW as other ASR members like @OCA (or nearly everyone really) so I don't know what it is really capable of. The downside of REW for new users is that it is not as automated as other software packages like Audiolense or Dirac, and it lacks wizards or macros. Everything is a manual process. This is absolutely not an issue if you know what steps you need to take. But nothing beats "free", and there is no doubting its ability and versatility, so IMO everyone in this hobby NEEDS REW and a microphone. If would go so far as to say that anybody who doesn't have a microphone and REW, or planning to obtain both, should have their ASR membership withdrawn

Beyond REW, IMO the two best software packages are Audiolense and Acourate. Neither of these run on a Mac, unless you use a Windows emulator. They have different strengths and weaknesses, and whether one or the other would be suitable for you depends on your needs. If you do not plan on separating your crossover and going fully active, I would say Audiolense. If you do plan on separating your crossover, I would say Acourate.

I know that Dirac runs on a Mac, but Dirac is astoundingly expensive compared to the above two, especially if you want the crossover option. I hear it is easier to use than both Audiolense/Acourate, but I do not have direct experience with it so I won't say more. I have been told that Focus Fidelity is even easier to use, and it runs on a Mac, but it has no crossover function. It is a relatively new product and I have seen posts shared by David (the developer) that more features are upcoming. So it is a product to watch, I am sure its capabilities will expand.

By the way, the no. 1 reason for poor sounding DSP correction is user error. I am still making mistakes despite doing this for years. See above post where I accidentally inverted the polarity of the sub Therefore, I place a high value on ease of use, and especially automation for new users. I myself like manual control, but climbing that learning curve can be pretty intimidating. I am not going to tell you to do what I have done (except get REW!) because I think there is a different DSP solution for everybody, depending on what your needs are.



If you are talking about the entertainment cabinet, I haven't measured its effect. It's not exactly easy to move, I have to remove all the equipment inside it, remove the TV, and it is heavy. It is also irrelevant for me, because I need it there.

Hi Keith, no worries!!
my questions are often more philosophical than pointed!

thanks for the advice on REW, and the other programs you mentioned: I think I'll start from this one, which I understand isn't too complicated.
As you say, I also give a lot of importance to ease of use, and this is one of the assumptions that has so far kept me away from DSP or detection and processing systems to be installed in the system: configuring and understanding something has made me always discouraged for fear of obtaining worse results. However, I will download it and try to understand something about it, at most if I manage to obtain some results I will bother you to understand the contents.
In my experience I noticed a fact: when I used the Infinity Epsilons, large speakers, with multiple drivers, I obtained a more "uniform" result with great impact but much less precise. The music was spread out like a giant poster, but the outlines ended up not being clear. With the much smaller Piega speakers and the system I use now, the answer is the opposite. Less “largeness” but more detail. I took inspiration from your interesting thread, and your great availability, precisely because I understood that I wanted a middle ground. Increase the space, therefore increasing the pleasure even when listening to orchestras or great rock, without losing the details that I can enjoy now with small ensembles.
I fear that this answer may be given by a speaker different from mine.
I'm assuming that the rest of the chain is above suspicion.
It's not easy, I know, you will always have to make some compromises, but you can get much closer to good satisfaction. But I don't want to leave this research to chance. I don't want to start buying and selling, trusting only in sensations. Of course those are important because they stimulate research, but I no longer want to play at being a "shop". Thanks as always!! Good day!!

 

[Keith_W]

It has taken 2 days to get to this point. I had to stop because I was exhausted and the guy who was helping me move the heavy equipment around had to pick his kids up from school (if you are reading this, thanks for your help Jayden!). 1 more day to move the amps and (hopefully) recable. Then the measuring starts. Not that I am not already measuring, the tape measure has been getting a bit of a workout. I also realized that this setup actually gives me more distance between MLP and sofa, because the subs are no longer preventing me from placing the speakers on either side of the entertainment cabinet. When I asked (begged) my wife for permission, she rolled her eyes. I think she would be happy with this when she sees it.

 

[OCA]

View attachment 347456

View attachment 347457

View attachment 347458

It has taken 2 days to get to this point. I had to stop because I was exhausted and the guy who was helping me move the heavy equipment around had to pick his kids up from school (if you are reading this, thanks for your help Jayden!). 1 more day to move the amps and (hopefully) recable. Then the measuring starts. Not that I am not already measuring, the tape measure has been getting a bit of a workout. I also realized that this setup actually gives me more distance between MLP and sofa, because the subs are no longer preventing me from placing the speakers on either side of the entertainment cabinet. When I asked (begged) my wife for permission, she rolled her eyes. I think she would be happy with this when she sees it.

These subs exciting all room modes very efficiently from those corners will take a lot of Acourate taps to tame

 

[Keith_W]

@OCA sticking the subs right into the corners is intentional. I want it to excite all the room modes

I hope I like the result, because otherwise it would take another week to move the system back to where it was!!!

 

[OCA]

It seems, you've done most of the hard work. Now, it's only about carrying the amps and making about a thousand cable connections

 

[Keith_W]

The system has been rotated. The subs were deliberately positioned in the corners of the room. I spent a heck of a lot of time routing and hiding cables,

I did a very rough DSP on it. All I did was take the old DSP and do a quick and dirty room correction to make it comply to the target curve, no thought involved. But even with this, IT SOUNDS AMAZING!! I thought that the previous setup already sounded good, but this has an incredible spaciousness that was missing in the old configuration. I am not sure what is creating the spaciousness, maybe it is the widely separated subs, maybe it is the delayed sidewall reflections, I don't know yet. I will create a set of DSP filters with the mains going full range with the subs turned off, that will allow me to A-B them.

What is really exciting was that the DSP was done in a really rough fashion. I know I will be able to improve on this. I can see all sorts of problems in the FR and I will have to look into these in detail. Time alignment has not yet been done. The drivers may have gotten mixed up when I dismantled the speaker to move it so I will repeat the driver linearization process.

When @OCA suggested that I rotate the system my initial reaction was "f*** off!!". But I thought about it a bit more and realized he was right. There are a lot of theoretical advantages to moving the system, besides all the sonic benefits. The room looks nicer, and my wife is really pleased. There is more open space and it feels less cluttered. I have done a better job of hiding the cables (not an easy job when you have 16 channels of cables to hide!).

Anyway, I will post some measurements when I get around to doing them. Plenty of stuff to do.

 

[ernestcarl]

The system has been rotated. The subs were deliberately positioned in the corners of the room. I spent a heck of a lot of time routing and hiding cables,

View attachment 348853

I did a very rough DSP on it. All I did was take the old DSP and do a quick and dirty room correction to make it comply to the target curve, no thought involved. But even with this, IT SOUNDS AMAZING!! I thought that the previous setup already sounded good, but this has an incredible spaciousness that was missing in the old configuration. I am not sure what is creating the spaciousness, maybe it is the widely separated subs, maybe it is the delayed sidewall reflections, I don't know yet. I will create a set of DSP filters with the mains going full range with the subs turned off, that will allow me to A-B them.

What is really exciting was that the DSP was done in a really rough fashion. I know I will be able to improve on this. I can see all sorts of problems in the FR and I will have to look into these in detail. Time alignment has not yet been done. The drivers may have gotten mixed up when I dismantled the speaker to move it so I will repeat the driver linearization process.

When @OCA suggested that I rotate the system my initial reaction was "f*** off!!". But I thought about it a bit more and realized he was right. There are a lot of theoretical advantages to moving the system, besides all the sonic benefits. The room looks nicer, and my wife is really pleased. There is more open space and it feels less cluttered. I have done a better job of hiding the cables (not an easy job when you have 16 channels of cables to hide!).

Anyway, I will post some measurements when I get around to doing them. Plenty of stuff to do.

A number of things could be at play here... some guesses: you are closer to your speakers so you get more of the direct sound overall and less strong early reflections (adding coloration) which improves sense of clarity. Could also be just your "ears breaking-in".

 

[OCA]

The system has been rotated. The subs were deliberately positioned in the corners of the room. I spent a heck of a lot of time routing and hiding cables,

View attachment 348853

I did a very rough DSP on it. All I did was take the old DSP and do a quick and dirty room correction to make it comply to the target curve, no thought involved. But even with this, IT SOUNDS AMAZING!! I thought that the previous setup already sounded good, but this has an incredible spaciousness that was missing in the old configuration. I am not sure what is creating the spaciousness, maybe it is the widely separated subs, maybe it is the delayed sidewall reflections, I don't know yet. I will create a set of DSP filters with the mains going full range with the subs turned off, that will allow me to A-B them.

What is really exciting was that the DSP was done in a really rough fashion. I know I will be able to improve on this. I can see all sorts of problems in the FR and I will have to look into these in detail. Time alignment has not yet been done. The drivers may have gotten mixed up when I dismantled the speaker to move it so I will repeat the driver linearization process.

When @OCA suggested that I rotate the system my initial reaction was "f*** off!!". But I thought about it a bit more and realized he was right. There are a lot of theoretical advantages to moving the system, besides all the sonic benefits. The room looks nicer, and my wife is really pleased. There is more open space and it feels less cluttered. I have done a better job of hiding the cables (not an easy job when you have 16 channels of cables to hide!).

Anyway, I will post some measurements when I get around to doing them. Plenty of stuff to do.

I don't know how it sounds but it sure looks the part!

 

[Keith_W]

A number of things could be at play here... some guesses: you are closer to your speakers so you get more of the direct sound overall and less strong early reflections (adding coloration) which improves sense of clarity. Could also be just your "ears breaking-in".

Actually, I am not sitting closer to the speakers. The triangle is the same size, I did mention that I need to sit far away from the speakers because of the type of drivers. Take a look at this article on "far-field criteria", in particular points 2 and 3:

"More accurate estimates of the far field are found to be:

1. The point of observation where the path length differences to all points on the surface of the loudspeaker perpendicular to the point of observation are the same. Unfortunately this is at an infinite distance and the pressure is zero.

2. The distance at which the loudspeaker’s three-dimensional radiation balloon no longer changes with increasing distance from the source with regard to frequency.

3. The distance from the source where the radiated level begins to follow the inverse-square law for all radiated frequencies. And, a practical definition useful for determining the required measurement distance:

4. The distance from the source where the path length difference for wave arrivals from points on the device on the surface plane perpendicular to the point of observation are within one-quarter wavelength at the highest frequency of interest (Figure 2)."

In fact that is an excellent article and I recommend reading it and chasing up all the little things mentioned. It was certainly eye-opening for me. I only read that article today, but even before this I did experiments here to determine what would be considered "far field" for my speakers.

Whilst I do appreciate your point that my ears and expectation bias are fooling me into thinking it sounds better, particularly after I spent a whole week of moving stuff around (and my extreme reluctance to move it back), I do believe that the differences are real. I wouldn't say that the wife can hear it from the kitchen, but in the sweet spot the soundstage is really absurdly wide.

However, the central image is not solid - I think i unmasked some issues with the crossover. If you recall, I said that the crossover points are 80, 500, and 4500, using extremely steep slopes. If I listen to a solo tenor singing, the high notes are slightly to the right. But when he sings a low note, the image moves to the center. I have noticed this with a couple of recordings already. My theory is that the extreme directivity of the horns and tweeters do a better job of localizing the tenor, but the wider directivity <500Hz causes the sound to "mono out" and shift the image to the center. It is just a theory, I do not know.

Anyway, I am redoing the DSP so that the 500Hz crossover region will be more gentle to hopefully blend the difference in directivity between bass cabinet and horn.

I have quite a bit of work to do! So I am off into the listening room now for more measuring. I will check back on ASR on my other computer in between sweeps

 

[Keith_W]

First step prior to room correction is to confirm the ideal position of the MLP and reposition the speakers. I have freedom to move stuff around my room. My strategy goes like this:

1. I decided that the subs would be fixed in the corner, so the MLP has to move for the most even bass response. The reason I want the subs in the corner is to provide the most corner loading and the most volume/efficiency. I am not too worried about room modes because I can DSP it out of the MLP. Having said that, I still do not want big dips in the MLP. @joentell suggested that I use an RTA to find the best position for the MLP. I was all set to do this, and then I remembered ... Acourate does not have an RTA. REW has an RTA, but no built-in convolver. I can send REW's pink noise through a convolver, but that introduces considerable latency and it won't be an "RTA" any more. In the end I decided it would be easier to simply do a whole bunch of sweeps.

2. Once the MLP is fixed, the speakers will be repositioned within limits of SBIR and listening distance (explained in above post) to obtain an equilateral triangle.

3. Once the position of subs, speakers, and MLP is finalized, I will do an overall room correction.

This is the first step of the process (I am typing this as I go along). A tape measure was laid from the back wall in a straight line towards the speakers. I used Acourate's mic centering tool to ensure correct mic position. I then took sweeps from 10Hz - 400Hz along various distances of the tape measure:

(the sweeps have been staggered for legibility)

It appears that 210-220cm is the least compromised position for the MLP. This happens to obey the rule of thirds (room length is 6m). So there you go, Toole is confirmed once again.

I then placed a marker at the MLP (masking tape to the spot on the floor with an "X" drawn on it). I used the tape measure to measure the distance to the speaker, and repositioned the speakers to form an equilateral triangle with the MLP. Final position: 320cm on each edge of the equilateral triangle. This is now much closer to the speakers than in my previous configuration (400cm distance to speaker).

Now to go on to room correction. I always do this in steps. I make a quick and dirty one and then remeasure and listen, and then go over it again and again every time I pick up an issue. It can take me weeks to do because there is a subjective component involved. As Toole says, don't believe everything the microphone tells you. But then again, I don't believe in "if it sounds good, it is good" either, because it could always sound better, just that I don't know it yet