Keith_W DSP system

[Keith_W]

Have you ever thought of selling your main speakers system to start again? Yes, with everything that you now know...

I will never sell these speakers. I mean, who is going to buy them? Some of the modifications are irreversible, e.g. the woofer replacement where the actual cabinet has been modified. If I get a new pair of speakers, I will still keep these but put them in storage somewhere.

There are no speakers on the market that fit my list of needs: point source, good directivity, looks good enough to be placed in a living room, and NO PORTS. I hate ports! Sorry, that is a pet peeve of mine. I was considering KEF Blades but it has a couple of drawbacks - firstly the passive crossover is inside the speaker and looks as if it is difficult to access it to bypass it. That by itself is not a dealbreaker. But what IS a dealbreaker are the ports. I was thinking of writing to KEF and asking them if they could deliver a pair of Blades with no passive XO and no ports, but decided that is too much of a bother. Because otherwise, they would be the perfect speaker - looks great, and juuust enough flaws for me to have fun with.

If I had bought Genelecs all those years ago, I would have had better sound than what I am getting now. But the problem with Genelec is - they are too perfect. Having to deal with problems taught me a LOT. I wouldn't have learnt all this stuff if I owned Genelecs, because there is nothing to fix.

Great system.
And I thought that you, the main contributors of ASR, had put the whole set away in the closet for some time now, to deal with and worry only about measurements!!!
then I see your system, I saw in a JBL review the one about " director” and I am amazed that you also use tubes, double amplifications and 60 kilogram giants….

I listen to more sine waves than music actually Last week I spent a few hours doing measurements, and after that 30 minutes listening to music. I can tell how audible an intervention is by how long it takes me to pick it up. If I hear the difference immediately, it is a huge intervention. If it takes me a long time, it is more subtle. Some of these interventions don't show up in conventional measurements, e.g. I changed the slope of the woofer/horn crossover to a gentle one to avoid the sudden change in directivity. The on-axis FR plot looks exactly the same, because I made it look the same. Even the off-axis FR plot looks similar. But then, I know that my measurements are compromised so I have to listen. And there is quite a nice improvement, the woofer and horn seems more "tied in".

 

[3ll3d00d]

The mic was positioned about 5cm away from the subwoofer driver. I know this gives me a nonsense measurement, especially since it is a dual opposed sub. I wondered whether it was even worthwhile doing it, since I would be generating a nonsense correction.

I don't think it will be a nonsense measurement, the native response of each driver should be the same after all so it should capture that. The pair of them will couple in the range you're using it so correcting on the basis on one doesn't seem fundamentally flawed to me. However I would not generally expect the response of such a driver to need, or benefit from, linearisation, it should be basically smooth in that range whereas yours is anything but hence I'd suspect your measurement is not good. 5cm does seem too far away to me, nearfield is <11% of the effective radius of the driver which is about 1cm or so away. I assume you stopped the sweep at ~120Hz btw to cause such a brickwall to appear too? Try measuring wider bandwidth with the mic much closer and see how it looks then. You're also coercing it to a flattish response down to ~10Hz which seems too aggressive to me and will also be counteracted by the room, you have estimated room gain or not?

FWIW I don't bother with this step on subs, I generally find room impact is larger than this so linearising the sub is not important. Instead I apply a correction to the subs just to shape their response to roughly combine with my room gain estimate then apply MSO correction on top then just run normal correction procedure.

 

[Paul_Sound48]

I was thinking of writing to KEF and asking them if they could deliver a pair of Blades with no passive XO and no ports, but decided that is too much of a bother. Because otherwise, they would be the perfect speaker - looks great, and juuust enough flaws for me to have fun with.

Compared to everything else which you have done, a short email to KEF is nothing.
They might well (will probably) say "no", but on the chance that they might just say "yes"............. send KEF a query Keith. You want to. We want you to.
What you want is far less work for them, than building the regular version after all.
Paul.

 

[Keith_W]

5cm does seem too far away to me, nearfield is <11% of the effective radius of the driver which is about 1cm or so away. I assume you stopped the sweep at ~120Hz btw to cause such a brickwall to appear too? Try measuring wider bandwidth with the mic much closer and see how it looks then. You're also coercing it to a flattish response down to ~10Hz which seems too aggressive to me and will also be counteracted by the room, you have estimated room gain or not?

Here is a selection of subwoofer measurements performed over the past few years. One of them is an extremely close measurement (in brown). You are correct, I stopped the sweep half an octave above the low pass filter.

I don't estimate room gain. The subs are corrected twice - first with a nearfield measurement, then again at MLP. If you look at the subwoofer correction baked into the crossover in post #10 you will see that the correction is not that much.

FWIW I don't bother with this step on subs, I generally find room impact is larger than this so linearising the sub is not important. Instead I apply a correction to the subs just to shape their response to roughly combine with my room gain estimate then apply MSO correction on top then just run normal correction procedure.

That would be a rational way of doing it The problem with DSP is that it is a cool toy. You can get beautiful looking results which at best are an audible improvement, but most of the time don't mean anything, and at worst may have a deleterious effect. No matter what the audible result, the correction staring back at me on the computer looks beautiful. If you combine "cool toy" with "OCD" you end up doing pointless corrections like what I do. The "turn up the settings to 11" mentality common with gamers also exists in DSP, and I have had to learn to cut back. I actually get nicer sounding corrections that way.

Toole has said many times that the microphone does not represent reality. I used to appreciate that on some abstract level, but nothing convinces you more of this truth than making a beautiful looking correction and being disappointed by the results. Ever since I took his advice to heart, I have been getting much better sounding results.

 

[Sokel]

Toole has said many times that the microphone does not represent reality. I used to appreciate that on some abstract level, but nothing convinces you more of this truth than making a beautiful looking correction and being disappointed by the results. Ever since I took his advice to heart, I have been getting much better sounding results.

The big difference that Toole points is that any correction above mid-bass has to be done anechoically.
There's no subsitute to that.
There's a reason (or many) that a klippel goes up to 100k$.

Visual clues without solid base can't be taken seriously.

 

[dualazmak]

Hello @Keith_W,
Thank you for your sharing the details of your amazing multichannel audio system.

At least for me, this thread has been a long time coming since you have briefly touched on your system in your post #85 on the thread entitled "Does Your System "Wow" or "Amaze" You? Looking to upgrade? Tell Us About It...."; I was very much impressed by your brief introduction at that time.

I too briefly touched/shared my audio setup here #68 on that thread, and I surprisingly have found that we, yourself and myself, have very similar approach and step-by-step journey towards our common goal of DSP-based multichannel multi-SP-driver multi-amplifier fully active stereo audio setups (having individual/dedicated amplifier for each of the SP drivers), even though your audio gears are much more high-end-oriented than mine and you have measured your setup so intensively.

Thank you again, and of course I shall carefully follow this attractive and interesting thread!

 

[JSmith]

I surprisingly have found that we, yourself and myself, have very similar approach

Well maybe apart from the wall of VU meters, which is excellent by all means.


JSmith

 

[Keith_W]

The big difference that Toole points is that any correction above mid-bass has to be done anechoically.
There's no subsitute to that.
There's a reason (or many) that a klippel goes up to 100k$.

Visual clues without solid base can't be taken seriously.

Yeah, I have seen him say that too. However, I don't have a Klippel. The speakers are too big to move outside and hoist into the air. This is a purist approach, which I think excludes the have-nots. You have seen from the way the system is put together, that it does not have a natural frequency response. So for my speaker, there is no such thing as an anechoic measurement. It simply does not exist, nobody has ever put my speaker on a Klippel, and I have never seen anechoic measurements of the speaker. I have to make do with what I have, and what I am capable of doing.

At least for me, this thread has been a long time coming since you have briefly touched on your system in your post #85 on the thread entitled "Does Your System "Wow" or "Amaze" You? Looking to upgrade? Tell Us About It...."; I was very much impressed by your brief introduction at that time.

Geez you have an incredible memory. I forgot that I made that post ... more than half a year ago!

 

[Mikig]

I will never sell these speakers. I mean, who is going to buy them? Some of the modifications are irreversible, e.g. the woofer replacement where the actual cabinet has been modified. If I get a new pair of speakers, I will still keep these but put them in storage somewhere.

There are no speakers on the market that fit my list of needs: point source, good directivity, looks good enough to be placed in a living room, and NO PORTS. I hate ports! Sorry, that is a pet peeve of mine. I was considering KEF Blades but it has a couple of drawbacks - firstly the passive crossover is inside the speaker and looks as if it is difficult to access it to bypass it. That by itself is not a dealbreaker. But what IS a dealbreaker are the ports. I was thinking of writing to KEF and asking them if they could deliver a pair of Blades with no passive XO and no ports, but decided that is too much of a bother. Because otherwise, they would be the perfect speaker - looks great, and juuust enough flaws for me to have fun with.

If I had bought Genelecs all those years ago, I would have had better sound than what I am getting now. There is only so much I can do. But the problem with Genelec is - they are too perfect. Having to deal with problems taught me a LOT. I wouldn't have learnt all this stuff if I owned Genelecs, because there is nothing to fix.



I listen to more sine waves than music actually Last week I spent a few hours doing measurements, and after that 30 minutes listening to music. I can tell how audible an intervention is by how long it takes me to pick it up. If I hear the difference immediately, it is a huge intervention. If it takes me a long time, it is more subtle. Some of these interventions don't show up in conventional measurements, e.g. I changed the slope of the woofer/horn crossover to a gentle one to avoid the sudden change in directivity. The on-axis FR plot looks exactly the same, because I made it look the same. Even the off-axis FR plot looks similar. But then, I know that my measurements are compromised so I have to listen. And there is quite a nice improvement, the woofer and horn seems more "tied in".

measurements aside, I am not capable and have no equipment, my story, I think like that of many, is similar to yours.... except that at a certain point, I abandoned the project I was following: the speakers I had required too much effort and time to give results that were enjoyable for me, the multiplication of devices, cables, and the countless tests led me to no longer listen to music!!

so one day I put everything aside and started again, with a very simple system, but which allows me to spend a lot of time listening!!! Music!!!

 

[Keith_W]

Anyway I should post some pictures.

From May last year.

Taken tonight. You can see the model ship in the foreground has had some work done on it

From the speaker, looking back at the MLP.

View from the MLP. You can see the plasma tweeters glowing at you. I had to turn on some lights to take the photo, but when it is all dark and the only light is coming from the tube amps, the effect is really

Here is a quick experiment to see if placing absorbent foam in front of me would improve uBACCH's crosstalk cancellation. As you would expect, what you hear is certainly different. The most obvious is more of a downward FR tilt. I don't think the apparent width or depth has changed very much.

 

[Mikig]

Anyway I should post some pictures.

From May last year.

View attachment 345447

Taken tonight. You can see the model ship in the foreground has had some work done on it

View attachment 345449

From the speaker, looking back at the MLP.

View attachment 345452

View from the MLP.





View attachment 345454

Here is a quick experiment to see if placing absorbent foam in front of me would improve uBACCH's crosstalk cancellation. As you would expect, what you hear is certainly different. The most obvious is more of a downward FR tilt. I don't think the apparent width or depth has changed very much.

nice and congratulations!! beautiful naval models!! How did you determine the position of the speakers? what do you think of the various theories going around the world?? 1/3 room, aurea, Cardas…??? Have you ever measured them during construction??

 

[Keith_W]

Anyway, this speaker has some very interesting properties that causes it to behave in a way that you would not expect. The first sign that something strange is amiss can be seen with this measurement, comparing a nearfield MMM with a MLP SPS:

You can see that the nearfield MMM has a much sharper treble fall-off than at MLP. This is the opposite of what one would expect. To study the phenomenon, I did MMM's at various distances from the speaker:

What I expect to see from this measurement is something like this (this is a simulation, not a real measurement). All the frequencies should start dropping in loudness at an equal rate as we move further from the speaker.

Instead, we get this:

Notice how everything > 1kHz is bunched together. At 250cm, there is a 10dB drop in the woofer. But with the horns and tweeters, there is only a 2dB drop! This causes the frequency response to tilt upwards the further away you get from the speaker - the horns/tweeter remain the same volume, but the subs and woofers drop in volume as expected. Toole says speakers are supposed to tilt downwards the further away you go. Erin provides a great explanation in this video.

So what is causing my speaker to behave in the opposite way to a normal speaker? Obviously the horns have something to do with it, which is clear from the above graph where I indicated the crossover points. Anything not horn = normal expected behaviour. Anything horn = bunched together with minimal drop in volume.

My understanding of how horns work is rather basic, perhaps @Duke might like to chime in. But it does seem to spray sound like a thumb over a garden hose. No thumb, and the stream of water spreads out and falls earlier (i.e. behaviour of the subs and woofers). Place the thumb on the outlet, and you get a concentrated jet of water that is narrow and travels further. Are the horns causing the sound to concentrate and behave like a "jet"?

I needed to explore the phenomenon more closely. Using the technique I described in this thread I repeated a set of MMM's around the speaker.

Result:

You can see already the difference in directivity between the woofer and the horns. With an omni, you expect the volume to remain the same as you rotate around the speaker taking measurements. With anything that is directional, the volume drops quickly. Look at the difference between the red (0 degrees) and brown (45 degrees) curves. Almost a 10dB drop moving 45deg off axis. What is interesting is that at 90deg and 180deg (i.e. behind the speaker) the response of the horns is almost the same.

I then processed the raw measurements in REW using arithmetic functions.

@joentell helped me process the measurements into a contour plot:

And a polar plot:

Now obviously these are very rough and low resolution measurements, a bit like viewing a blurry photograph that has gone through the wash a few times. But even with blurry photographs, shapes can be discerned. I wouldn't read too much into each individual peak and dip, given the limitations of the measurement technique. But we can observe broad trends. This measurement is good enough for me to see:

- that the directivity indeed does narrow above 700Hz. The previous measurement showed a much sharper change in directivity. This made me redo the crossover between the woofer and midrange horn. Rather than the sharp steep filter I was using, I used Acourate's UBPol 11 filter, 0.5 order. This blended the woofer with the horn more gradually, and you can indeed see that the change in directivity is not so sudden.
- Notice the notch in the polar response at 4.5kHz. This is the crossover point between the midrange horn and tweeters.
- Notice how the directivity dramatically narrows from 13kHz up (more obvious in the polar plot). The tweeter is starting to beam. Why does it beam? I measured the tweeter diameter. The lower limit is: lambda = 2* diameter of horn. The diameter is 12cm, so wavelength 24cm, so 1.4kHz. The upper limit is 3 octaves above this, so 11.2kHz. At progressively shorter wavelengths above 11.2kHz, sound seems to exit the tweeter like a compressed jet of water. It does not lose much energy as it makes its way to MLP.

Amazing that we can see all this with this crappy measurement, eh!! I go to bed at night dreaming that I have won the lottery and I can go buy a Klippel. And then go into competition with Amir and make scathing Youtube videos But until that happens, I make do with the tools that I have.

CONCLUSION
This speaker behaves the reverse of a normal speaker - it produces an upward tilting slope further away, rather than a downward tilt. The reason it behaves this way is because of the mixed driver types - conventional cabinets which radiate sound like a normal speaker, and horns above 500Hz. Measurements seem to indicate that it concentrates sound energy.

THEREFORE ... any attempt to correct the speaker to be "flat" anechoically will result in an upward tilt at the listening position. This is why I don't do it.

 

[Multicore]

I listen to more sine waves than music actually

It seems like a lot of complexity for that application.

 

[Keith_W]

nice and congratulations!! beautiful naval models!! How did you determine the position of the speakers? what do you think of the various theories going around the world?? 1/3 room, aurea, Cardas…??? Have you ever measured them during construction??

Sigh. I was hoping nobody would ask me this question! I used the following reasoning when I placed the loudspeakers:

1. Since I am high passing the main speakers and removing the bass, I do not have to worry about placing the speakers optimally for bass reproduction. The subwoofers will take care of that.
2. y axis: The position of the left speaker was determined by the door on the left of the room. The right speaker was placed 1.5m from the right wall and toed in to avoid early reflections.
3. x axis: rule of thirds
4. MLP: rule of thirds

I don't have strong opinions about most of those theories because I haven't looked at them in detail.

 

[ernestcarl]

BTW, where is the woofer? Hidden behind a transparent cloth? Facing back towards the wall? It’s just not obvious with your pictures.

 

[Keith_W]

BTW, where is the woofer? Hidden behind a transparent cloth? Facing back towards the wall? It’s just not obvious with your pictures.

 

[Duke]

View attachment 345468

Notice how everything > 1kHz is bunched together. At 250cm, there is a 10dB drop in the woofer. But with the horns and tweeters, there is only a 2dB drop! This causes the frequency response to tilt upwards the further away you get from the speaker - the horns/tweeter remain the same volume, but the subs and woofers drop in volume as expected. Toole says speakers are supposed to tilt downwards the further away you go. Erin provides a great explanation in this video.

So what is causing my speaker to behave in the opposite way to a normal speaker? Obviously the horns have something to do with it, which is clear from the above graph where I indicated the crossover points. Anything not horn = normal expected behaviour. Anything horn = bunched together with minimal drop in volume.

My understanding of how horns work is rather basic, perhaps @Duke might like to chime in. But it does seem to spray sound like a thumb over a garden hose. No thumb, and the stream of water spreads out and falls earlier (i.e. behaviour of the subs and woofers). Place the thumb on the outlet, and you get a concentrated jet of water that is narrow and travels further. Are the horns causing the sound to concentrate and behave like a "jet"?

I think it's a radiation pattern mis-match thing, just as you suspect. I see something similar in hybrid electrostats, wherein a very directional panel is crossed over to a woofer low enough that the woofer's pattern is very wide in the crossover region.

Here are some arguably relevant and fairly well-known data points: Anechoic SPL falls off by 6 dB per doubling of distance from a point source; by 3 dB per doubling of distance from a true line source; and by 0 dB per doubling of distance from a true planar source. At least up fairly close, I think the horns are behaving somewhere in between a point source and a planar-ish-source, in that the change in SPL over distance is less than point-source-theory would predict.

 

[Sokel]

So for my speaker, there is no such thing as an anechoic measurement. It simply does not exist, nobody has ever put my speaker on a Klippel, and I have never seen anechoic measurements of the speaker. I have to make do with what I have, and what I am capable of doing.

It seems that there is,usually JA is not far from truth,at least for the original version and it doesn't look like it would easily (if at all) EQ'd.

Acapella High Violoncello II loudspeaker Measurements

Sidebar 3: Measurements

www.stereophile.com

The great value of the tests Amir (and all the others that do anechoic) is not only that shows some good speakers but also give the data so these speakers can be corrected above midbass or so and exclude the ones that is probably better not touch because fixing one thing messes something else.

Given your great effort I think you should organize a plan to take them out or else will be always a shot in the dark.
There you will see if your version of them is EQ-able or not,etc.

 

[Keith_W]

Here are some arguably relevant and fairly well-known data points: Anechoic SPL falls off by 6 dB per doubling of distance from a point source; by 3 dB per doubling of distance from a true line source; and by 0 dB per doubling of distance from a true planar source. At least up fairly close, I think the horns are behaving somewhere in between a point source and a planar-ish-source, in that the change in SPL over distance is less than point-source-theory would predict.

"Well known" ... well, I didn't know that. Thank you, that is valuable information.

It seems that there is,usually JA is not far from truth,at least for the original version and it doesn't look like it would easily (if at all) EQ'd.

Acapella High Violoncello II loudspeaker Measurements

Sidebar 3: Measurements

www.stereophile.com

Thank you. But unfortunately, that is a different speaker. My speakers are Violons, those are Violoncellos. Slightly different design, bigger horns, have a selectable horn crossover, etc. And don't forget ... my speakers are modified

 

[FrantzM]

Hi Keith...

You are hereby assigned/beamed to my hate-list...
WoW!! What a journey!
What a system!
Yep, I hate you.


THis said, your journey is in some ways, similar to mine. I have learned a lot because of the limitations of my system. Trying to overcome them without breaking the bank taught me a lot... BTW I did hear the larger brother of your speakers sometimes ago and I was floored by these, it could have been sighted bias but ..

Looking at your measurements and your process... I know where I need to go... I will likely continue to play/learn with my present system and at one point go toward speakers from one of these: Neuman, Genelec, Revel, Kef or JBL, I will use DSP/DRC, have a soft spot for Audyssey MutEQ-X but open to potentially better correction, such as those offered by Trinnov.. On the DSP/DRC, I can't even begin to understand how an audiophile can get the best of his components without the use of DSP/DRC. especially in the bass, which in many ways is the enveloppe of music: I surmise that if the bass is wrong at the MLP, the perception of the rest of the spectrum will suffer.

Thanks for the write-up.

Can we make of this thread a sticky?

Peace.