Capital Audiofest 2022 - November 11th-13th

[Mr. Widget]

Let me know if you have questions about any of this.

A very interesting approach. Off the top of my head I can't think of anyone doing anything quite like this with horns.

How does the sound change when you disconnect the rear firing horn? I imagine both the imaging and the timbre changes. Does having this radiation pattern make room placement even more critical, or is there so much reflected energy that typical first reflections become less significant?

 

[Linearphass]

I just saw this response and hope to clarify the comments I have made. my answered are interspersed therein.
For background I am an EE who worked in test and measurement for 32 years at Tektronix, one the the leading companies in the world for measuring all things electrical. During that time I had direct access to the finest test equipment money can buy including both equipment from my company but also the best of the best of competitors equipment. None of these comments are based on theory alone, they are all measurable and hence numerically quantifiable. As a listener I never trust anything I hear unless I can measure it. If I can not measure it using an existing technique I work until I can develop a technique that correlates both with theory and what I hear.

Active crossovers are great, but let's not get way out over our skis here.


1. Is obviously wrong, or at any rate the result of multiple invalid assumptions. One can of course have active speakers with VERY LONG cables because the equipment is in a different room, on an entirely different floor. The only time they're shorter than usual is in the case of a plate amp speaker. Given that LX-521 was not shown with a plate amp on the "cabinet" and the spec does not contemplate a plate amp. this argument is odd to even include let alone make the starting point of your framing.

2. Well...a growing body of knowledge suggests certain passive filters may offer distortion benefits that equivalent active filters may not. See, e.g. Purifi notch filter white paper. Interestingly, Purifi's paper is also somewhat an admission against interest, given Purifi sells (great!) amp modules and would doubtless prefer to have the story be "use an active crossover with our driver, and double up on amp modules!"

3a. Matters if you're using pathologically garbage amps. If not... (If you reframed this as "amp quality can be markedly worse with no real world performance penalty for the system" then I'd give you the point.)
3b. Um, no. The driver still sheds heat and the cone still moves.

5-6. Irrelevant audiophile gobbledegook

Overall you get 2/8 here: 4 (passive = some loss by definition) and 7 (especially if using modern technology, i.e. DSP) are reasonable points.



1. The only benefit of 2 DAC channels over of 8 is cost of DAC. Give that bespoke analog circuitry is bound to be a lot more expensive than mere DACs (which are cheap - elite performance is >$800 for 2 channels if you want 4V balanced outs, and half that if 2V unbalanced outs will suffice) + reliable off-the-shelf DSP, that is a fairly extreme form of false economy.

2. OK, but possibly at a resolution penalty compared to a good ADC and digital processing. Also, you'll get better real-world sound quality from analog sources by digitizing them because you can then employ room correction, loudness compensation, and so on.

3. An individual LX521 could be processed and EQ'ed by many commercially-available 3-way plate amps, with greater flexibility than analog circuits can provide. (To my knowledge the EQ for the bass units is the same, so there's little reason to use a second amp channel if you can get the required voltage at the drive impedance. For that matter, with modern processing and amplification one could put it all in one box for the stereo pair with an output for additional subs, instead of dividing the electronics in half

So that's 0 for 3, 1 for 3 if grading on a curve and giving you 2 even though it's irrelevant.

The advantages of analog crossover/EQ over DSP are
(a) Better press from the dumber audio influencers, who are easily swayed by frothy buzzwords
(b) Sound quality phobic audiophiles will pay more for it a bespoke analog kit than for an equal or better commercially-available drop-in digital board
(c) Serviceability, maybe

Active crossovers are great, but let's not get way out over our skis here.



1. Is obviously wrong, or at any rate the result of multiple invalid assumptions. One can of course have active speakers with VERY LONG cables because the equipment is in a different room, on an entirely different floor. The only time they're shorter than usual is in the case of a plate amp speaker. Given that LX-521 was not shown with a plate amp on the "cabinet" and the spec does not contemplate a plate amp. this argument is odd to even include let alone make the starting point of your framing.

You have taken this out of context. My response was to a question asked about the 521. In this case the cables are indeed very short, perhaps <1M. The amps were sitting right beside the respective speakers. They are 5 very small but quite powerful amps totaling some 925 W rms, 1 per driver, except the tweeters that share a single amp, wired directly to the driver via very short high quality cables.

2. Well...a growing body of knowledge suggests certain passive filters may offer distortion benefits that equivalent active filters may not. See, e.g. Purifi notch filter white paper. Interestingly, Purifi's paper is also somewhat an admission against interest, given Purifi sells (great!) amp modules and would doubtless prefer to have the story be "use an active crossover with our driver, and double up on amp modules!"

The only case that this would be so is if the amp is crapy and the passive crossover was removing some of the crud it is generating. Moreover since the amps are restricted at their inputs to specific frequency ranges IM distortion is drastically reduced.

3a. Matters if you're using pathologically garbage amps. If not... (If you reframed this as "amp quality can be markedly worse with no real world performance penalty for the system" then I'd give you the point.)

Not at all. Passive crossovers, are very reactive. They fundamentally work by inserting impedance in the signal path. Even with the best possible components they are not linear when faced with a non linear load like a driver. You are right in the point of bad amps. if you have a poorly designed amp with bad phase margins a combination of the crossover, cable reactance and driver reactance can drive it into instability, even to the point of oscillation. See also 3a.

3b. Um, no. The driver still sheds heat and the cone still moves.

3a is not taking about heat but heat changing voicemail resistance. It also talks about back EMF. When the amplifier directly drives the voice it behaves like a voltages source, the voltage does not change regardless of the drivers impedance. If you put a passive crossover in the signal path it introduce a large series impedance. At the crossover frequency this impedance is EQUAL to the (nominal) driver impedance. Since this is an (AC) voltage divider this means the voltage drops by 50% which is the intended goal. Hence if the voicemail impedance change in a passive crossover case the speaker response will change. Don't think the effect is very much? See the following research paper.
It is well done and shows very well how much response can change.

https://www.researchgate.net/publication/323935533_Voice_coil_temperature_in_loudspeaker_performance_Causes_effects_and_prevention_techniques

For the short answer look at figure 6.20
I have observed these same effects 30 years ago in my home lab, they are real, audible to virtually anyone and are dynamic, that is they change not only with overall level but even on music peaks. This forms the basis for frequency dependent compression, a partially nasty form of non-linear distortion.
Eliminating the passive network eliminates these problems as active does not function by changing high level signal path impedance.

5-6. Irrelevant audiophile gobbledegook

Overall you get 2/8 here: 4 (passive = some loss by definition) and 7 (especially if using modern technology, i.e. DSP) are reasonable points.

I don't know what to say here except this absolutely is solid engineering and there is no audiophile gobbledegook involved. Both H field ( magnetic) and E field are real. If you are technical and can experiment it is easy to to prove every one of my points. I have measured every one of the effects stated.

1. The only benefit of 2 DAC channels over of 8 is cost of DAC. Give that bespoke analog circuitry is bound to be a lot more expensive than mere DACs (which are cheap - elite performance is >$800 for 2 channels if you want 4V balanced outs, and half that if 2V unbalanced outs will suffice) + reliable off-the-shelf DSP, that is a fairly extreme form of false economy.

One benefit of 2 channel DAC over 8 is cost. The rest of this point is arguable. What is a the 8 channel DAC you describe that has such wonderful and accurate response for $800? I was not talking about economy at all, the analog variant of the solution costs roughly the same as the digital.

2. OK, but possibly at a resolution penalty compared to a good ADC and digital processing. Also, you'll get better real-world sound quality from analog sources by digitizing them because you can then employ room correction, loudness compensation, and so on.

Where is there resolution loss? Technically analog has infinite resolution. Digital on the other hand is discreet. In practical terms there is no difference between modern digital and analog in terms of "resolution". The dynamic range of this system processor is comparable or better than the best ADC/ DAC. This system readily accommodates any DSP you choose if you are using a digital source as that processing can easily be done in the music server in the digital domain BEFORE it arrives at the analog processor. In this particular case sure DSP signal processing could be used ( and was in the proceeding generation) and it would sound basically the same as long as the transfer functions matched AND the multichannel DAC problem solved. One side reason to use phase compensated precision analog is that DSP processors come and go relatively frequently. To minimize the inconvenience of the user to have to change the DSP system should it fail and no longer be available the analog approach was chosen. It is to avoid planned obsolescence.

3. An individual LX521 could be processed and EQ'ed by many commercially-available 3-way plate amps, with greater flexibility than analog circuits can provide. (To my knowledge the EQ for the bass units is the same, so there's little reason to use a second amp channel if you can get the required voltage at the drive impedance. For that matter, with modern processing and amplification one could put it all in one box for the stereo pair with an output for additional subs, instead of dividing the electronics in half

The LX 521 is a four way system. The amplifiers used are higher quality than you would find in the plate amps I am aware of. The second sub amp, one per driver, is to avoid the issues associated with driving the woofers in parallel since they would then present a 2 ohm load. You most surely are aware of iron induced distortion since you quote Purifi documents above. The sub driver are wired such a way that when the bottom is moving towards the magnet the bottom is driving away for the magnet thus cancelling some of the effects of this distortion. This and the separate amps keep misbehaviors from one driver influencing the other.

So that's 0 for 3, 1 for 3 if grading on a curve and giving you 2 even though it's irrelevant.

The advantages of analog crossover/EQ over DSP are
(a) Better press from the dumber audio influencers, who are easily swayed by frothy buzzwords
(b) Sound quality phobic audiophiles will pay more for it a bespoke analog kit than for an equal or better commercially-available drop-in digital board
(c) Serviceability, maybe

No doubt that analog appeals to a different audience than a digital solution. There are serious benefits for making this a true key system that many audiophiles do not appreciate. The ability to control all system parameters if one of them. I personally am in the camp of DSP and use it in all my designs, at least for the last 15 years. I have designed speakers since the early 1960's and all of my designs before that, there are about 35, used passive high level analog crossovers. I can say with 100% certainty that I do not miss them in any way.

Thanks for your time.

 

[jhaider]

@Linearphass I'm not going to reply to you point by point, but a few things.

First, it is clear from your response you did not comprehend the Purifi paper I linked. I recommend going back to it.

Second, you seem to harp on "quality" of electronics a lot, when that's more often than not an insignificant factor, while simultaneously making assumptions about market offerings that show you don't really know what's out there (e.g. Hypex plate amps that use the same turnkey amp modules as the "official" amp).

Third, a regular reader of this forum should know that digital to analog conversion is a solved problem. I refer you to Amir's DAC reviews.

Fourth, I'm quite surprised that someone who claims to be a measurement person and further claims to have been doing this stuff since many Gen Xer’s parents were in grade school doesn't know the basics of the impedance of drive units operating in and around their Fs. The impedance curve for paralleled LX521's woofers is mostly well, well above 2Ω in their passband, so even if that low impedance were a problem for modern amps (which it generally isn't) the impedance of paralleled “4Ω” woofers for bass is not that low anyway.

Lastly, the paper you cited relates to voicecoil heating causing compression generally. Nobody disputes that is a thing. Here, your false and farcical claim is, active drive somehow magically solves that issue. Let's throw in some actual data here. ErinH takes thermal compression measurements of the loudspeakers he measures. Here's his description of the test:

-------
Dynamic Range (Instantaneous Compression Test)
The below graphic indicates just how much SPL is lost (compression) or gained (enhancement; usually due to distortion) when the speaker is played at higher output volumes instantly via a 2.7 second logarithmic sine sweep referenced to 76dB at 1 meter. The signals are played consecutively without any additional stimulus applied. Then normalized against the 76dB result.

The tests are conducted in this fashion:

76dB at 1 meter (baseline; black)
86dB at 1 meter (red)
96dB at 1 meter (blue)
102dB at 1 meter (purple)
The purpose of this test is to illustrate how much (if at all) the output changes as a speaker’s components temperature increases (i.e., voice coils, crossover components) instantaneously.
-----

So let's compare an active and a passive speaker with as similar configurations as can be found in Erin’s dataset.

For the active let's take Linkwitz LX Mini, which has a relatively expensive Seas 6" midwoofer in a closed “box” actively crossed to a full-range driver with a decent sized voicecoil. An LX Mini kit costs $640 the pair as of this writing.

For the passive let's take Arendal 1961, a mini monitor with a 5.5" woofer in a closed box that costs $800 the pair for complete speakers as of this writing.

How did each fair in the Erin's instantaneous compression test?

LX Mini:

Arendal 1961:

So clearly the active crossover isn't making LX Mini immune from compression, and the passive crossover doesn't seem to be hurting the tiny little Arendal.

 

[Linearphass]

@Linearphass I'm not going to reply to you point by point, but a few things.

First, it is clear from your response you did not comprehend the Purifi paper I linked. I recommend going back to it.

Second, you seem to harp on "quality" of electronics a lot, when that's more often than not an insignificant factor, while simultaneously making assumptions about market offerings that show you don't really know what's out there (e.g. Hypex plate amps that use the same turnkey amp modules as the "official" amp).

Third, a regular reader of this forum should know that digital to analog conversion is a solved problem. I refer you to Amir's DAC reviews.

Fourth, I'm quite surprised that someone who claims to be a measurement person and further claims to have been doing this stuff since many Gen Xer’s parents were in grade school doesn't know the basics of the impedance of drive units operating in and around their Fs. The impedance curve for paralleled LX521's woofers is mostly well, well above 2Ω in their passband, so even if that low impedance were a problem for modern amps (which it generally isn't) the impedance of paralleled “4Ω” woofers for bass is not that low anyway.

Lastly, the paper you cited relates to voicecoil heating causing compression generally. Nobody disputes that is a thing. Here, your false and farcical claim is, active drive somehow magically solves that issue. Let's throw in some actual data here. ErinH takes thermal compression measurements of the loudspeakers he measures. Here's his description of the test:

-------
Dynamic Range (Instantaneous Compression Test)
The below graphic indicates just how much SPL is lost (compression) or gained (enhancement; usually due to distortion) when the speaker is played at higher output volumes instantly via a 2.7 second logarithmic sine sweep referenced to 76dB at 1 meter. The signals are played consecutively without any additional stimulus applied. Then normalized against the 76dB result.

The tests are conducted in this fashion:

76dB at 1 meter (baseline; black)
86dB at 1 meter (red)
96dB at 1 meter (blue)
102dB at 1 meter (purple)
The purpose of this test is to illustrate how much (if at all) the output changes as a speaker’s components temperature increases (i.e., voice coils, crossover components) instantaneously.
-----

So let's compare an active and a passive speaker with as similar configurations as can be found in Erin’s dataset.

For the active let's take Linkwitz LX Mini, which has a relatively expensive Seas 6" midwoofer in a closed “box” actively crossed to a full-range driver with a decent sized voicecoil. An LX Mini kit costs $640 the pair as of this writing.

For the passive let's take Arendal 1961, a mini monitor with a 5.5" woofer in a closed box that costs $800 the pair for complete speakers as of this writing.

How did each fair in the Erin's instantaneous compression test?

LX Mini:

Arendal 1961:

So clearly the active crossover isn't making LX Mini immune from compression, and the passive crossover doesn't seem to be hurting the tiny little Arendal.

Once again my initial response was to a specific question from an individual regarding the 521. You seem to have plenty of issues with my response and I am OK with that. I did not read the paper you referenced because the link was broken. Upon further research I think I found what you were referring to and I agree that a series notch has validity to solve some problems where active does not work like cone breakup. I am not aware of any 5 channel plate amps that could be drop in replacements perhaps you could share those with the group. 5 channels are needed for the 521 for other reasons not just because of impedance.
I am glad to hear that all DAC problems are solved, we can close the patent office. Funny Amir wastes his time testing new ones. I am also glad in 32 years my bosses never figured out that I was incompetent. TEK was not like a university or the government, you actually would be fired for incompetence.
No further response is required, we obviously disagree and further discussion is unlikely to change that. We should not turn this thread into an argument. If you really want to air this out I would be happy to discuss further on the phone. Have a good day.

 

[jhaider]

5 channels are needed for the 521 for other reasons not just because of impedance.

On closer review it seems we were both wrong on fact in different ways here.

From my side, I was thinking back to my experience listening to Mr. Linkwitz demonstrate Orion, which is a 3-way speaker, as well as the original LX521, which had a passive crossover between midrange and midbass. For those a 3-way plate amp would be fine (assuming there was some place to put it!). However, the final version has an active filter between those drivers as well. I am not aware of a native 4-way plate DSP-amp available to hobbyists, though there are some suitable standalone units, e.g. from Powersoft.

From your side, it is clear in the @magicLX demo the amplifier modules are configured as 4 channels, not 5, likely (though I don't know for a fact) with the bass amp module bridged. Why is that clear? Poorly drafted "replace 10 banana plugs" copy on the LX521 sale website aside, the inference comes from physical reality: you can't run cables for 5 amp channels from an 8-pole Neutrik speakON connector, and, as the late Mr. Linkwitz pointed out, "there is no 10-pole Speakon connector available or a 5-conductor-pair speaker cable."

But to circle back, this discussion goes back to the original very silly claim that "VERY SHORT cables" lead to "much better control of drivers." That's just blowing smoke. If you can point to a measured effect, that's great but we can measure so much stuff that we can't hear that the proof required would be not electrical measurements but a controlled subjective same/different listening test.

I am glad to hear that all DAC problems are solved, we can close the patent office. Funny Amir wastes his time testing new ones.

Cool story, bro. FWIW, I did a quick scan of patents related to DACs, and suffice it to say there isn't a lot of recent activity related to converting digitally encoded audible-bandwidth signals back to analog. I'm sure a trademark search would turn up more activity...

The benefit of DAC reviews is really showing the emperor has no clothes, and the DAC is a particularly stupid object of obsession. The good $100 unbalanced ones and $200 balanced (2-channel) ones offer measured performance well beyond the bounds of audibility. IMO @amirm's spotlight facilitated this state of affairs, as the cheap ones were generally audibly fine but not objectively elite until he started measuring DACs. The flip side of that is the focus on measured electrical performance led a race for engineers to chase measurably superior heights in parameters with relatively low audible thresholds, instead of investing in features such as equalization/automated modal correction, bass management, and loudness compensation...add crossfeed to the list for headphone listeners.

 

[Linearphass]

On closer review it seems we were both wrong on fact in different ways here.

From my side, I was thinking back to my experience listening to Mr. Linkwitz demonstrate Orion, which is a 3-way speaker, as well as the original LX521, which had a passive crossover between midrange and midbass. For those a 3-way plate amp would be fine (assuming there was some place to put it!). However, the final version has an active filter between those drivers as well. I am not aware of a native 4-way plate DSP-amp available to hobbyists, though there are some suitable standalone units, e.g. from Powersoft.

From your side, it is clear in the @magicLX demo the amplifier modules are configured as 4 channels, not 5, likely (though I don't know for a fact) with the bass amp module bridged. Why is that clear? Poorly drafted "replace 10 banana plugs" copy on the LX521 sale website aside, the inference comes from physical reality: you can't run cables for 5 amp channels from an 8-pole Neutrik speakON connector, and, as the late Mr. Linkwitz pointed out, "there is no 10-pole Speakon connector available or a 5-conductor-pair speaker cable."

But to circle back, this discussion goes back to the original very silly claim that "VERY SHORT cables" lead to "much better control of drivers." That's just blowing smoke. If you can point to a measured effect, that's great but we can measure so much stuff that we can't hear that the proof required would be not electrical measurements but a controlled subjective same/different listening test.




Cool story, bro. FWIW, I did a quick scan of patents related to DACs, and suffice it to say there isn't a lot of recent activity related to converting digitally encoded audible-bandwidth signals back to analog. I'm sure a trademark search would turn up more activity...

The benefit of DAC reviews is really showing the emperor has no clothes, and the DAC is a particularly stupid object of obsession. The good $100 unbalanced ones and $200 balanced (2-channel) ones offer measured performance well beyond the bounds of audibility. IMO @amirm's spotlight facilitated this state of affairs, as the cheap ones were generally audibly fine but not objectively elite until he started measuring DACs. The flip side of that is the focus on measured electrical performance led a race for engineers to chase measurably superior heights in parameters with relatively low audible thresholds, instead of investing in features such as equalization/automated modal correction, bass management, and loudness compensation...add crossfeed to the list for headphone listeners.

Let me give you a short calibration so I help you to understand my response.
1. The individual I was responding to asked for a comparison of advantages of fully active over passive implementations. My response was made in that context.
2. Since I retired in 2011 I help people all over the world set up, calibrate and optimize their audio systems ( no HT work) , mostly using REW as it is both competent, readily available, relatively easy for most people to learn and it allows for easy though not realtime interactive sharing of results. The work is over 90% electro acoustic, placement orientation, EQ, room treatment etc, NOT component optimization or change. There has never been any charge for this service and it has become very popular since COVID. The last few years it has varied from 25- 50 system setups and calibration/ optimizations per year. These vary in time from 5 to 50 hours or so per system. One friend is WAY over 500 hours. That particular one has been VERY helpful to me. We have explored lots of valuable concepts and validated those that actually work. A lot of this time is spent training the person to competently use REW. If I was say a lawyer I would be charging $50/ 15 minutes for this service based on how I value my time. I have a backlog of customers waiting for help. I turn down 10-20% of those asking for help either becuase they do not have enough tech knowledge and/ or they want me to do all the work. I do this because I like to help people and tweak on things. For me it is as rewarding as listening. The vast majority of the people come back repeatedly for help and refer others.
3. I am nearly 70 so I am not as fast or is my mind as sharp as it used to be.
4. I have NO monetary interest in Linkwitz Audio , DR Frank Brenner, nor am I compensated by them. I do help as in #2 above from time to time and act as a tech liaison at some shows so I do have a T shirt! I am also a long time admirer of Siegfried, first met him around 1982, where he was a chief competitor in test equipment when he worked at hp and I was at Tektronix. However, we had many non competitive conversations on audio over the years and I found his knowledge to be be absolutely top notch. Since his passing I want to see his products go forth and be successful an hence I do what I can to help.

I'm sure you know his old web site which has been preserved/ enhanced by Frank:

https://www.linkwitzlab.com

IMO it is by FAR the most useful website for me by a large factor. Though not super well organized, one can find incredible knowledge and tidbits there if you are willing to work for it.

Less known is:

https://oplug-support.org/viewforum.php?f=28

A great source of up to date info. Many people the are quite knowledgeable. Note that only the GENERAL sections are accessible to the public. The detailed sections are intended for owners only as they contain IP.


With that in mind the current 521 uses 5 channels of amplification with the channel power optimized to each driver. Of course each channel is directly coupled to the driver via a single pair of wires. Proceeding the power amps but in the same box is a high quality analog signal processor that handles crossover and EQ and phase compensation issues. There in no intervening passive component, even a HP filter cap. The power flows through two .7m (standard)
cable bundles which use two speak-on connectors per end, the cables are OF and heavy but are not exotic. The Speakons are different sizes to prevent any possibility of errors when plugging them in. See here:

https://linkwitz.store/product/lx521-cable-from-powerbox-6pro-ncore/

Each woofer is driven by a separate 250 watt amp for several reasons. Each of the two power boxes can consume in excess of 1kW under condition of heavy bass and loud volumes like the 1812 cannons. Transient current demands for class D amps is higher than traditional ABs because they don't have have huge power supply caps to source current. The current need must come for the wall which should have very low impedance. Because of this, high volume listeners ideally would dedicate two 15-20 amp circuits to their listening room to allow for less than idea electrician work. Contact me off list if you wish to discuss further.

Until 2 years ago I would completely agree with your "blowing smoke" comment regarding cables. Today that is not the case. There are things that I can easily measure L, C and R dielectric absorption, EMI ingress and others that correlate to some extent with what I hear. I am continuously trying to correlate measurements because I have proven to myself that my hearing is enormously unreliable in anything other than a differential judgement, that is I can hear of difference but I can not for sure say which is more accurate. While I can provide cable measurements I will not do so in this environment because I don't want to start a cable war. Contact me off list if you wish to discuss further.

I have done semi-controlled tests using my younger sons as subjects because they have good hearing. patience to endure the detailed and lengthy tests, as well as critical listening skills. The tests were double blinded and precisely level verified BUT the switching was done manually so there was a delay of perhaps 20 seconds or so. Both sons showed good correlation 8/10 for one 10/ 10 for the other with measured results and my thesis. The same comment made above applies here.

On my DAC comment that was mostly tongue and cheek but I mostly agree with what you say. The fact remains that the vast majority of the customers that buy speakers in this price class demand more expensive than $100 DAC's. For the same reason the cables were are on elevators at CAF.
I will say though that I can easily hear a difference between my OPPO 205 DAC and the DAC's in my Mini DSP 4x10 even though the processing is totally bypassed, levels are precisely matched and everything else under my control is normalized. The difference is not subtle.