"Theoretical aspects of the Groothuis NLFP system" paper is poor. It provides no summary of what Groothuis system does, or how it is implemented.
A single driver diaphragm regardless of number of voice coils has single output function. So I am not sure what you are getting at with the patent links.
The level of harmonic distortion of speaker is non factor for Hi-Fi, but its phase performance is, as you have found out via your numerous investigations using DRC. A good acoustical square wave response up to 4kHz is as good as it gets; this indicates that phase relationships of source signal are being reproduced acoustically. NLFP appears to be hardware EQ that must be custom tailored to drivers. DRC does this for you.
David Griesinger has much better series of papers on phase structure and perception:
http://www.davidgriesinger.com/ica2010/paper1.doc
http://www.davidgriesinger.com/ica2010/paper2.doc
http://www.davidgriesinger.com/ica2010/paper3.doc
I'm certain that from outside your listening room that many would believe live sound is happening.
A single driver diaphragm regardless of number of voice coils has single output function. So I am not sure what you are getting at with the patent links.
The level of harmonic distortion of speaker is non factor for Hi-Fi, but its phase performance is, as you have found out via your numerous investigations using DRC. A good acoustical square wave response up to 4kHz is as good as it gets; this indicates that phase relationships of source signal are being reproduced acoustically. NLFP appears to be hardware EQ that must be custom tailored to drivers. DRC does this for you.
David Griesinger has much better series of papers on phase structure and perception:
http://www.davidgriesinger.com/ica2010/paper1.doc
http://www.davidgriesinger.com/ica2010/paper2.doc
http://www.davidgriesinger.com/ica2010/paper3.doc
I'm certain that from outside your listening room that many would believe live sound is happening.
Last edited:
I just had time to go trough the first paper and I agree with you. To me it seemed it was more about the understanding of the author of that paper's view on the system from Hans Groothuis. There was nothing really useful in there.
I haven't had time to go trough the other papers yet. I do remember his system to sound big. But I don't recall anything about it's imaging performance. Reading what I've found so far on Dutch forums I'm remembering more details about the talk(s) I had with Hans Groothuis.
Even his tiny desktop system hooked up to a laptop sounded way bigger than it actually was. He used a 3 way concept as the base but manipulated the time arrival in an unusual way. It could very well be that I'm romanticising what I heard back then. I'm way more critical about my own sound system than I ever have been listening to other setups
.Getting the time arrival right has brought out a lot of surprises in songs as well so I'm quite confident it is a more right way to approach the problem of audio playback.
I agree with Bob, there are so many variables to play with making it hard to get it all right. That's why I said earlier, If I could combine every good part of all the different things I've tried over the past months I'd be very happy.
My desktop is filled with all kinds of DRC test files that didn't even make it to a listening test
.
Last edited:
Finally had the time to go trough the patent applications and connect that info to the talks I had with Hans.
A bit of background, most demo's I've heard were with speakers that were nothing special to behold. But the sound they had was really something else!
It seems he build his own motor arrangements for his drivers (multiple coil arrangement) and used an amplifier with separate driving modules incorporated to be able to advance the signals below 200 Hz and delay above that 200 Hz mark. Taking advantage of some of the psycho acoustic features of our hearing mechanism.
Basically as a sort of trickery to make relatively small speakers sound bigger. I can vouch for that! That really worked. But in relation to my own project it is of no real value to me.
Judging in hindsight it is a complex way of making small speakers sound more like big sound sources. He was very proud of the fact that in most demo's the subs were not playing and yet you had a convincing low end.
His little desktop/satellite speakers had a full engaging sound for their size.
Going back to my own setup I build speakers that have the capacity to do 20 Hz to about 17-18 KHz. No trickery needed except to overcome the effects of the placement in a room. DRC is helping me there to get the signal back to phase linear behaviour. Giving me the right sounding kick of a big drum etc.
Either baffle size, room placement/size or some other effect gave me trouble in the first arrival around the 130-160 Hz area. One of the reasons to see if the impedance rise that falls within that window played any role in that part of the spectrum.
Increasing the excess phase window in DRC beyond it's original intended length at low frequencies made it possible for me to correct that problem area. Making that part time coherent as well.
I also had resonances left in the ~60-70 Hz that I was able to partially solve by boosting one side while attenuating the other. After the longer phase window this problem isn't as big an issue anymore. To keep distortion down I still cut a bit of the left signal at 70 Hz and apply boost there in the right channel, but way less than I had to do before.
With the remark about "romanticizing" what I heard back in ~2003/2004 I do not mean to downplay Mr. Groothuis's concept. But I've learned it is quite easy to impress on a first (brief) listen to speakers. It is way more difficult to remain impressed in the long run!
The main trouble I have left in my current setup is being unable to use more damping and/or diffusion in my listening space. I would still like to gain more depth in my imaging. Though I've had mixed results with depth proving that in time I should be able to discover how to maximise that.
But I'm convinced damping the front wall would help immensely. Sadly that's not possible for me.
Any tips/tricks that could help me with that? Apart from applying more damping that is...
A bit of background, most demo's I've heard were with speakers that were nothing special to behold. But the sound they had was really something else!
It seems he build his own motor arrangements for his drivers (multiple coil arrangement) and used an amplifier with separate driving modules incorporated to be able to advance the signals below 200 Hz and delay above that 200 Hz mark. Taking advantage of some of the psycho acoustic features of our hearing mechanism.
Basically as a sort of trickery to make relatively small speakers sound bigger. I can vouch for that! That really worked. But in relation to my own project it is of no real value to me.
Judging in hindsight it is a complex way of making small speakers sound more like big sound sources. He was very proud of the fact that in most demo's the subs were not playing and yet you had a convincing low end.
His little desktop/satellite speakers had a full engaging sound for their size.
Going back to my own setup I build speakers that have the capacity to do 20 Hz to about 17-18 KHz. No trickery needed except to overcome the effects of the placement in a room. DRC is helping me there to get the signal back to phase linear behaviour. Giving me the right sounding kick of a big drum etc.
Either baffle size, room placement/size or some other effect gave me trouble in the first arrival around the 130-160 Hz area. One of the reasons to see if the impedance rise that falls within that window played any role in that part of the spectrum.
Increasing the excess phase window in DRC beyond it's original intended length at low frequencies made it possible for me to correct that problem area. Making that part time coherent as well.
I also had resonances left in the ~60-70 Hz that I was able to partially solve by boosting one side while attenuating the other. After the longer phase window this problem isn't as big an issue anymore. To keep distortion down I still cut a bit of the left signal at 70 Hz and apply boost there in the right channel, but way less than I had to do before.
With the remark about "romanticizing" what I heard back in ~2003/2004 I do not mean to downplay Mr. Groothuis's concept. But I've learned it is quite easy to impress on a first (brief) listen to speakers. It is way more difficult to remain impressed in the long run!
The main trouble I have left in my current setup is being unable to use more damping and/or diffusion in my listening space. I would still like to gain more depth in my imaging. Though I've had mixed results with depth proving that in time I should be able to discover how to maximise that.
But I'm convinced damping the front wall would help immensely. Sadly that's not possible for me.
Any tips/tricks that could help me with that? Apart from applying more damping that is...
Thinking more about that "depth" part I'm thinking with the boundaries set by my room I'd probably have the best results with a "Haas Kicker" signal from behind. I'd have to create that artificially because of my room. It would consist of a L-R and R-L signal from behind the listening position, delayed (about 20 ms) and as a band pass signal. It would mean extra channels and speakers and more processing. It's something I had in mind trying all along but haven't got all the needed parts yet to incorporate it. Still trying to figure out how to do it. I'd want it to be a somewhat diffuse source.
The processing isn't that hard, routing channels is a bit harder due to using a single stereo DAC at the moment. I do have a multi channel expansion card for my xonar soundcard though.
It's not that I don't get depth. But I'd like to be able to get a little more out of this. I can steer the side channels by toe in/toe out, they can go from in line with the main voice in the center to about 150 degree coming more from the sides. But the center voice is almost always placed between the speaker planes except for a few songs.
With the back speakers working as Haas Kickers without the mid information present it will probably work to pull the phantom voices closer to the listening position.
So far it's pretty dependant on the recording what amount of depth I get. The (phase) steering of the sides is way stronger than in the phantom center. At least that's how I interpret it.
I've experimented with an allpass (at about 200 Hz) on the mid signal only. That did have an effect to pull out the phantom center. But it also negates the time coherency I've been working on. But it does demonstrate that a non time coherent setup probably gains a little extra depth due to the gradual delay at lower frequencies?
Without any extra speakers balance seems key to maximise the depth. But that is very hard to accomplish while keeping tonality right. As said before on this thread a slight knee somewhere between 1 KHz and 2 KHz goes a long way to gain a depth impression.
Any pointers? Halair, are you watching this? How is your depth, you do have damping panels on the front all behind the speakers at their first reflection points. Please chime in with your findings.
The processing isn't that hard, routing channels is a bit harder due to using a single stereo DAC at the moment. I do have a multi channel expansion card for my xonar soundcard though.
It's not that I don't get depth. But I'd like to be able to get a little more out of this. I can steer the side channels by toe in/toe out, they can go from in line with the main voice in the center to about 150 degree coming more from the sides. But the center voice is almost always placed between the speaker planes except for a few songs.
With the back speakers working as Haas Kickers without the mid information present it will probably work to pull the phantom voices closer to the listening position.
So far it's pretty dependant on the recording what amount of depth I get. The (phase) steering of the sides is way stronger than in the phantom center. At least that's how I interpret it.
I've experimented with an allpass (at about 200 Hz) on the mid signal only. That did have an effect to pull out the phantom center. But it also negates the time coherency I've been working on. But it does demonstrate that a non time coherent setup probably gains a little extra depth due to the gradual delay at lower frequencies?
Without any extra speakers balance seems key to maximise the depth. But that is very hard to accomplish while keeping tonality right. As said before on this thread a slight knee somewhere between 1 KHz and 2 KHz goes a long way to gain a depth impression.
Any pointers? Halair, are you watching this? How is your depth, you do have damping panels on the front all behind the speakers at their first reflection points. Please chime in with your findings
.
Last edited:
Sorry Barley, I just removed both RC networks from the compensation network and left the RCL in. World of difference...
So I just can't disregard that part of the network just yet.
Resulting impedance and electrical phase:
A hair raising difference!
It still pretty much flattens the group delay plot of the impedance measurement:
So I'm thinking, there just might be something in current drive to take this further. The group delay with a 10 Ohm resistor looks like this:
So far I do like what I hear very much! My highs are smoother again. That's without series resistance. Ideally I'd like to try a current source amplifier instead of a series resistor.
There's definitely a difference between the Conjugate Network with and without the high frequency compensation. Hard to tell if the resonance compensation (the LCR) does something positive but it sounds like it!
Highly subjective of coarse. But the raising hairs on my arm do help
.
Just to be able to compare here's the uncorrected GD tab:
Ignore the wiggle at 50 Hz. The new measurements were taken on battery power as an experiment and don't show that wiggle.
This one was taken earlier...
Strangely enough it even gave me back the depth I had been missing. Positioning front to back seems improved in the Phantom center.
Just had someone over for a listen: same comment as I've heard before, close your eyes and you are there...
So on my way again after a bit of doubt. The system didn't sound any like this when I had the Audiophile guest here. No domes will be cut!
Ignore the wiggle at 50 Hz. The new measurements were taken on battery power as an experiment and don't show that wiggle.
This one was taken earlier...
Strangely enough it even gave me back the depth I had been missing. Positioning front to back seems improved in the Phantom center.
Just had someone over for a listen: same comment as I've heard before, close your eyes and you are there...
So on my way again after a bit of doubt. The system didn't sound any like this when I had the Audiophile guest here. No domes will be cut!
Really need to see acoustic response of speaker with and without network(s), with no changes to EQ and microphone position. Comparison of plots for frequency response, phase response and harmonic distortion will then have meaning.
If your amplifier behaves as low impedance voltage source, then these networks only waste power. If amplifier behaves otherwise, an equalization change will happen.
If your amplifier behaves as low impedance voltage source, then these networks only waste power. If amplifier behaves otherwise, an equalization change will happen.
Barleywater, I don't think you can infer from Griessingers papers what the effects of simple phase shifts are; his work is on halls and the effect of different ratio's between direct and reflected sound, and it is very informative. However, his analysis of how the brain processes auditory information, the first paper in your list, is speculative and imo not correct.
I know, I know, you need prove ....
It's going to be a struggle for me to prove it to you though.
I will make the measurements but first I'm going to enjoy this setup and experiment with series resistance.
The reason it's going to be hard is the room I'm in. Distortion is swamped with several things in the room once I measure above 80/85 dB at the listening position. And still it makes a difference if there's traffic outside, especially in the region of interest for this part of the spectrum. Short of making measurements in the middle of the night, annoying the neighbours in the process after first kicking my son and girlfriend out of the house I wouldn't know how to make them clean enough to tell us something.
What I see in the impedance plot and its group delay does tell me something though. A few pages back I showed the plots of different impedance measurements and it's effect at higher frequencies. The wide spread on the Group Delay plot that's still evident on the above graphs. The conjugate network does not help the frequencies above 400 Hz but it does seem more controlled in everything below that number.
When I re-did my baffle, that improved the upper part of the spectrum. It also showed up on the waterfall plot making the mid part way more clear with a tall leading edge and a huge drop before the first bumps etc showing up. I'm not prepared to show those yet simply because I'm not completely satisfied with it. The difference was noticeable though.
Ahhh... what the .... we're all learning here, I'll show the mid part of the plots...
Note these are taken in a life room, so a lot will be dependant on what the room does as well.
Here's (part of) the waterfall prior to the different damping of the baffle:
And after using butyl between the two aluminium baffle plates:
I'd say an improvement, I could have used different settings to make it prettier but it shows what I wanted to show. Making the window smaller would leave out more hash, as this still has early room stuff in it, you clearly see it build up by flicking trough the window length.
But the improvement in the "cliff" is clear. I left the settings in there so you'll know what I used. Not proud of it yet, I think I can do better but that's mainly in the region above the part shown here.
So I'm starting to think that the impedance plot, the Group Delay part to be precise, already shows us something that will turn up in the CSD. I'm not certain about it though.
Nobody is telling me anything (lol). Just trying to connect what I see with what I hear.
This mod was very clear in listening tests. But what I do in FIR processing also shows up in the early waterfall milliseconds. The DRC window was actually quite a bit shorter in the better graph, making it longer would/could show an even better result. My goal is not to use longer windows than strictly necessary.
But coming back to the resonance flattening, I think it will have some effect there. But the area it works in also has huge room problems showing up. So I try to find the advantage in the impedance plot. And I could be very wrong there but for now I believe there is a link.
....It's going to be a struggle for me to prove it to you though.
I will make the measurements but first I'm going to enjoy this setup and experiment with series resistance.
The reason it's going to be hard is the room I'm in. Distortion is swamped with several things in the room once I measure above 80/85 dB at the listening position. And still it makes a difference if there's traffic outside, especially in the region of interest for this part of the spectrum. Short of making measurements in the middle of the night, annoying the neighbours in the process after first kicking my son and girlfriend out of the house I wouldn't know how to make them clean enough to tell us something.
What I see in the impedance plot and its group delay does tell me something though. A few pages back I showed the plots of different impedance measurements and it's effect at higher frequencies. The wide spread on the Group Delay plot that's still evident on the above graphs. The conjugate network does not help the frequencies above 400 Hz but it does seem more controlled in everything below that number.
When I re-did my baffle, that improved the upper part of the spectrum. It also showed up on the waterfall plot making the mid part way more clear with a tall leading edge and a huge drop before the first bumps etc showing up. I'm not prepared to show those yet simply because I'm not completely satisfied with it. The difference was noticeable though.
Ahhh... what the .... we're all learning here, I'll show the mid part of the plots...
Note these are taken in a life room, so a lot will be dependant on what the room does as well.
Here's (part of) the waterfall prior to the different damping of the baffle:
And after using butyl between the two aluminium baffle plates:
I'd say an improvement, I could have used different settings to make it prettier but it shows what I wanted to show. Making the window smaller would leave out more hash, as this still has early room stuff in it, you clearly see it build up by flicking trough the window length.
But the improvement in the "cliff" is clear. I left the settings in there so you'll know what I used. Not proud of it yet, I think I can do better but that's mainly in the region above the part shown here.
So I'm starting to think that the impedance plot, the Group Delay part to be precise, already shows us something that will turn up in the CSD. I'm not certain about it though.
Nobody is telling me anything (lol). Just trying to connect what I see with what I hear.
This mod was very clear in listening tests. But what I do in FIR processing also shows up in the early waterfall milliseconds. The DRC window was actually quite a bit shorter in the better graph, making it longer would/could show an even better result. My goal is not to use longer windows than strictly necessary.
But coming back to the resonance flattening, I think it will have some effect there. But the area it works in also has huge room problems showing up. So I try to find the advantage in the impedance plot. And I could be very wrong there but for now I believe there is a link.
The experiment with series resistance was short lived. To test it I first listened without the series resistance and measured the SPL level with my RadioShack SPL meter. I established about 86 dB average on the meter and rearranged the wiring for series resistance.
After setting the SPL at the 86 dB average again I played a couple of songs.
As soon as I played "Royals" by Lorde(*) I noticed a strange sound in my right speaker. I went over there to look at the cone movement: not excessive.
Looked at the digital Analyzer in JRiver: nothing excessive there.
Backing off the volume helped. The 10 ohm resistor was very hot to the touch on the right side. The left is cold in comparison. No sign of a raised temperature. I guess I'm sending my amp into clipping on the right side. It has the most boost of both channels.
Removed the series resistors and played the song again: clean as a whistle. Even at louder volume. So I guess I'm at the boundaries of my amp with series resistance. It was only 10 Ohm. According to XSim I'd lose about 7.5 dB... The rise at higher frequency due to the impedance curve was evident though with the series resistor.
Not getting any wiser from this experiment
(*) I picked that song for the high levels of bass in that song. It hits very hard at ~40 Hz.
I did think about building a set of Joe Rasmussen's "Trans Amp". But now I'm not too sure if that would be strong enough to power the arrays. Theoretically the array could sustain 240 watt before exceeding x-max with EQ engaged. My amp is about 100 watt. So I'd need a pretty strong amp to experience an advantage.
After setting the SPL at the 86 dB average again I played a couple of songs.
As soon as I played "Royals" by Lorde(*) I noticed a strange sound in my right speaker. I went over there to look at the cone movement: not excessive.
Looked at the digital Analyzer in JRiver: nothing excessive there.
Backing off the volume helped. The 10 ohm resistor was very hot to the touch on the right side. The left is cold in comparison. No sign of a raised temperature. I guess I'm sending my amp into clipping on the right side. It has the most boost of both channels.
Removed the series resistors and played the song again: clean as a whistle. Even at louder volume. So I guess I'm at the boundaries of my amp with series resistance. It was only 10 Ohm. According to XSim I'd lose about 7.5 dB... The rise at higher frequency due to the impedance curve was evident though with the series resistor.
Not getting any wiser from this experiment
(*) I picked that song for the high levels of bass in that song. It hits very hard at ~40 Hz.
I did think about building a set of Joe Rasmussen's "Trans Amp". But now I'm not too sure if that would be strong enough to power the arrays. Theoretically the array could sustain 240 watt before exceeding x-max with EQ engaged. My amp is about 100 watt. So I'd need a pretty strong amp to experience an advantage.
Last edited:
I had seen some measurements from you on a tweeter I believe. But you're right about the mixed results. I actually think it's going to be next to nothing on this Vifa. Pure guess on my part, partly based on the earlier experience where I had a series resistor on my left speaker. It didn't seem to result in a different enough sound to notice, while after treating that same side baffle (other one not done yet) it did make a noticeable impression.
So, maybe it's time to let go of the current drive idea for a while and focus on improving the damping some more to get the waterfall results I crave for. It could very well be they are better than what I see due to room pollution of the measurements. Up to a 1.1 ms window actually is pretty clean up top while there's build up right after that around 6 KHz. But how do I make sure it's the baffle/enclosure and not something else.
I'd also like to know if a good class D amp will sound different. I currently use a class A/B and run it trough the entire range with the EQ needed on the Arrays. So if there is crossover distortion in this kind of amp I am bound to run it right trough that region.
vacuphile, you mentioned a few pages back coming over for a cup of coffee and a listen. You wouldn't by any chance have some Ncore amps we could try?
If there is bad specs your amp maybe run a RightMark Audio Analyzer sweep can help, did in past test my old amps Sansui AU-D9. Use your high spec ZONAR ESSENCE ST as measurements device and compare measurements when sound card is looped with a wire up against when your power amp sits in loop with a resistor divider network. If any bad specs turn up try raise bias current for output devices a bit that often helps. To have tech data the amp suggest either buy service manal somewhere as example via this link Manuals-In-Pdf.com: Service Manuals, Owner's Manuals, Repair Manuals, Operating Instructions or contact shop that did service your amp. ARTA is maybe even better but don't know if the free version will do it.
Finally had the time to go trough the patent applications and connect that info to the talks I had with Hans.
A bit of background, most demo's I've heard were with speakers that were nothing special to behold. But the sound they had was really something else!
It seems he build his own motor arrangements for his drivers (multiple coil arrangement) and used an amplifier with separate driving modules incorporated to be able to advance the signals below 200 Hz and delay above that 200 Hz mark. Taking advantage of some of the psycho acoustic features of our hearing mechanism.
Basically as a sort of trickery to make relatively small speakers sound bigger. I can vouch for that! That really worked. But in relation to my own project it is of no real value to me.
Judging in hindsight it is a complex way of making small speakers sound more like big sound sources. He was very proud of the fact that in most demo's the subs were not playing and yet you had a convincing low end.
His little desktop/satellite speakers had a full engaging sound for their size.
Going back to my own setup I build speakers that have the capacity to do 20 Hz to about 17-18 KHz. No trickery needed except to overcome the effects of the placement in a room. DRC is helping me there to get the signal back to phase linear behaviour. Giving me the right sounding kick of a big drum etc.
Either baffle size, room placement/size or some other effect gave me trouble in the first arrival around the 130-160 Hz area. One of the reasons to see if the impedance rise that falls within that window played any role in that part of the spectrum.
Increasing the excess phase window in DRC beyond it's original intended length at low frequencies made it possible for me to correct that problem area. Making that part time coherent as well.
I also had resonances left in the ~60-70 Hz that I was able to partially solve by boosting one side while attenuating the other. After the longer phase window this problem isn't as big an issue anymore. To keep distortion down I still cut a bit of the left signal at 70 Hz and apply boost there in the right channel, but way less than I had to do before.
With the remark about "romanticizing" what I heard back in ~2003/2004 I do not mean to downplay Mr. Groothuis's concept. But I've learned it is quite easy to impress on a first (brief) listen to speakers. It is way more difficult to remain impressed in the long run!
The main trouble I have left in my current setup is being unable to use more damping and/or diffusion in my listening space. I would still like to gain more depth in my imaging. Though I've had mixed results with depth proving that in time I should be able to discover how to maximise that.
But I'm convinced damping the front wall would help immensely. Sadly that's not possible for me.
Any tips/tricks that could help me with that? Apart from applying more damping that is...
Using the hand clap method, I found that I was able to greatly reduce room ringing by nailing 2 inch cotton rope into most of the corners of my room (2 and 3 surface corners - nothing along the floor). Cotton rope from the fabric store turns out to be very flammable, so I don't recommend that, but foam, preferably soft, should work too. I think Auralex makes 2X2X24 inch sections, for example, which are legally fire retardant (non-flammable). If we could get a place like Auralux to make it look like "crown moulding" (decorative), then our wives and girlfriends might put up with it... It makes perfect sense to me that room ringing would damage the sense of depth in a recording. I also glued 1/2 inch thick felt padding to the entire front of by baffle board, with cutouts for the midrange drivers. I use 1/8 inch felt around the tweeter. The depth on a good recording is stunning to me.
My system is a tri-amp'd open baffle arrangement, with 12 inch woofers in separate closed cabinets. Active EQ makes them somewhat flat 20 - 20kHZ. The midrange is 4 five inch drivers in a vertical array. The Seas Millenium tweeter is at the top, about 42 inches off the floor. I've also got a rear tweeter that only does above 12kHZ. So yes, I'm cheating... generating more depth than is necessarily in the recording. But it doesn't get old.
Last edited:
Thanks for that Bob, I don't mind cheating as long as it sounds good .
I went trough the room clapping like a maniac and it further convinced me my ceiling is the biggest room draw back. Though I could still improve in the corners. 2 of the corners are filled in with big damping panels. One, behind the right array is untreated.
The right side also has the roughest FR response.
I'll PM you with the details, could be fun indeed!
.I went trough the room clapping like a maniac and it further convinced me my ceiling is the biggest room draw back. Though I could still improve in the corners. 2 of the corners are filled in with big damping panels. One, behind the right array is untreated.
The right side also has the roughest FR response.
I'll PM you with the details, could be fun indeed!
It's not so much I don't trust my old amp. It is in excellent condition and I have no reason to doubt it really. Still I'm wondering if there could be more with a bigger amp that doesn't switch operating mode from one class to another. Class A with high power is out, as well as tubes. Class D seems to have evolved enough by now to be taken seriously. It's winning Car SQ competitions, that much I know. In my car I'm running old school Genesis amps with the tweeters running in class A.
Halair build a Pass Labs Class A amp for his Arrays. From what I've heard from him first impressions were good. But it isn't a power monster amp.
I know, I know, you need prove
It's going to be a struggle for me to prove it to you though.
I will make the measurements but first I'm going to enjoy this setup and experiment with series resistance.
The reason it's going to be hard is the room I'm in. Distortion is swamped with several things in the room once I measure above 80/85 dB at the listening position. And still it makes a difference if there's traffic outside, especially in the region of interest for this part of the spectrum. Short of making measurements in the middle of the night, annoying the neighbours in the process after first kicking my son and girlfriend out of the house I wouldn't know how to make them clean enough to tell us something.
What I see in the impedance plot and its group delay does tell me something though. A few pages back I showed the plots of different impedance measurements and it's effect at higher frequencies. The wide spread on the Group Delay plot that's still evident on the above graphs. The conjugate network does not help the frequencies above 400 Hz but it does seem more controlled in everything below that number.
When I re-did my baffle, that improved the upper part of the spectrum. It also showed up on the waterfall plot making the mid part way more clear with a tall leading edge and a huge drop before the first bumps etc showing up. I'm not prepared to show those yet simply because I'm not completely satisfied with it. The difference was noticeable though.
Ahhh... what the .... we're all learning here, I'll show the mid part of the plots...
Note these are taken in a life room, so a lot will be dependant on what the room does as well.
Here's (part of) the waterfall prior to the different damping of the baffle:
And after using butyl between the two aluminium baffle plates:
I'd say an improvement, I could have used different settings to make it prettier but it shows what I wanted to show. Making the window smaller would leave out more hash, as this still has early room stuff in it, you clearly see it build up by flicking trough the window length.
But the improvement in the "cliff" is clear. I left the settings in there so you'll know what I used. Not proud of it yet, I think I can do better but that's mainly in the region above the part shown here.
So I'm starting to think that the impedance plot, the Group Delay part to be precise, already shows us something that will turn up in the CSD. I'm not certain about it though.
Nobody is telling me anything (lol). Just trying to connect what I see with what I hear.
This mod was very clear in listening tests. But what I do in FIR processing also shows up in the early waterfall milliseconds. The DRC window was actually quite a bit shorter in the better graph, making it longer would/could show an even better result. My goal is not to use longer windows than strictly necessary.
But coming back to the resonance flattening, I think it will have some effect there. But the area it works in also has huge room problems showing up. So I try to find the advantage in the impedance plot. And I could be very wrong there but for now I believe there is a link.
Based on my experiences measuring radial array, these results could well be do to changes in microphone and speaker placement. Very small changes completely change summation from driver array.
Measurements that I am interested in seeing don't take long, and need not be loud. Swept sine has excellent noise rejection; longer settings in REW work quite well. I also live along very busy street, and yet have no difficulty in getting good signal to noise ratio. Expectation is that if audible difference exists with your speaker when impedance network is only changing variable then the difference will be change in equalization, and not do to change in distortion level. For change in distortion level to be cause of your perceived change using compensation network, it would require difference of several orders in magnitude; not likely.
At least the results in the waterfall have been consistent after the change in baffle damping. My initial (re)praise is based on the removal of the dual RC network in the high frequencies. Without it, the sound is just way more pleasing.
No measurable change in SPL level though. But a real difference in sound. I completely removed that part of the network and don't plan to test it again.
These measurements are from different days, the purple one is with conjugation in place. Raw measurement with some pre-EQ and 1/48 smoothing. If anything the purple one seems to have deeper valleys. Figuring in the fact that microphone position isn't exactly the same the remaining result is quite close. No DRC here in this example.
It remains hard to tell if the remaining RCL network does something positive but I will do measurements with and without it. It might be the impression of more depth comes from these minute SPL changes. With the RC network in place I had a more gritty high frequency sound. Without it, it sounded more smooth.
With it I never had chills running down my spine. With it removed that "feature" was back. The sound became fuller without the RC network in place. The one thing I cannot check now is a change in phase between the two.
I already use the longest REW settings but still it's hard to get good results when it comes to distortion. I manage to work my way trough it but cannot compare for minute changes (if any). I never measure up close, there the array won't sum properly.
I just went back trough things like group delay (filtering to 20 ms window and smaller) between these two measurements and the changes are small if any exist even at the resonance spike.
So you might be right that it does not do me any good. As stated, the removal of the correction up top clearly made an improvement. It might hint at trying a little shorter window in the Bass frequency for a bit more "live" feeling without any conjugation network.
No measurable change in SPL level though. But a real difference in sound. I completely removed that part of the network and don't plan to test it again.
These measurements are from different days, the purple one is with conjugation in place. Raw measurement with some pre-EQ and 1/48 smoothing. If anything the purple one seems to have deeper valleys. Figuring in the fact that microphone position isn't exactly the same the remaining result is quite close. No DRC here in this example.
It remains hard to tell if the remaining RCL network does something positive but I will do measurements with and without it. It might be the impression of more depth comes from these minute SPL changes. With the RC network in place I had a more gritty high frequency sound. Without it, it sounded more smooth.
With it I never had chills running down my spine. With it removed that "feature" was back. The sound became fuller without the RC network in place. The one thing I cannot check now is a change in phase between the two.
I already use the longest REW settings but still it's hard to get good results when it comes to distortion. I manage to work my way trough it but cannot compare for minute changes (if any). I never measure up close, there the array won't sum properly.
I just went back trough things like group delay (filtering to 20 ms window and smaller) between these two measurements and the changes are small if any exist even at the resonance spike.
So you might be right that it does not do me any good. As stated, the removal of the correction up top clearly made an improvement. It might hint at trying a little shorter window in the Bass frequency for a bit more "live" feeling without any conjugation network.
Last edited: