SoundStage’s review of Dutch&Dutch’s 8C

[Juhazi]

About distortion of loudspeakers, please remember that music has most energy between 50-500Hz and peak levels of 95-105dB(C) aren't at all rare for people who imitate live levels of a club, big band or a symphony orchestra. Environmental noise is typically measured with A-weighting, but you shouldn't compare that to speaker measurements that have no weighting at all.

I find it funny, that many people here are very much worried of dacs and amps having 0,01% distortion, but easily accept 1% for a speaker.

 

[RayDunzl]

I find it funny, that many people here are very much worried of dacs and amps having 0,01% distortion, but easily accept 1% for a speaker.

I pay attention to speaker distortion...

I like my low-distortion panels, and added cheezy subs to experiment with reducing low frequency distortion by speading the load among 3 drivers instead of one (per side).

The only posted measures I've seen here to beat me were @dallasjustice M2 setup.

I might note, in the context of this post, that he was using them with the Benchmark AHB2 amps, vs. my probably less than stellar in the distortion measure semi-antique Krells.

I also have a JBL LSR 308, they distort early, but were cheap, and end up being my casual listening choice.

 

[DDF]

the basic difference to normal speakers is 200-500Hz range which has unusually high and constant directivity. Typical narrow-baffle speakers "leak" much more energy widely sideways and backwards there (60-180deg)

This is why I think they may be boosting the response as frequency progresses downwards through the cardiod's range, because music is mastered typically for monopoles below 1 kHz, room response becomes more audible the lower the frequency and monopoles dump more power into the room at lower frequencies than this cardiod if they were both flat on axis

 

[andreasmaaan]

Those measurements are not applicable to near-wall performance and I am pretty sure that midrange distortion is lower with near-wall settings too.

I can't see how distortion could be reduced with a change of settings. Distortion is almost 100% a result of driver nonlinearities. Unless changing settings changes the crossover frequency or slope (which I don't believe it does in this case), distortion at a given SPL cannot be affected by a change in settings.

 

[andreasmaaan]

Kii Audio Three has non-typical directivity in low midrange too, with a closed box mid and set-back side woofers and dsp with obviously some delay control. My guess is that its midrange directivity is lower but also with lower distortion. Stereophile has measured it, but NRC not.

I agree, it would be very likely to have lower distortion as a result of the active cardioid configuration, which allows the midrange to be placed in a closed box, which - all else equal - will reduce distortion at the lower end of the midrange's passband relative to the 8C's open-back configuration.

 

[Juhazi]

I can't see how distortion could be reduced with a change of settings. Distortion is almost 100% a result of driver nonlinearities. Unless changing settings changes the crossover frequency or slope (which I don't believe it does in this case), distortion at a given SPL cannot be affected by a change in settings.

My guess is that with near-wall setting the mid doesn't need that much gain/eq in low end. This is in controversy with midrange being cardioid, but still, regarding range 100-200Hz. If I am wrong, and that setting lowers only bass drivers' level, distortion actually gets higher!

In my diy-dipoles I cross monopole closed box 10" woofer to a 12" dipole around 170Hz using LR2 slopes. With LR4 and 150Hz Fc distortion is lower. Still we see similar elevated distortion around xo.

Another thing we see here, that this on-axis shows a bit lower relative level for the monopole region. In-room the response is almost a straight line but it doesn't sound thin.

 

[andreasmaaan]

My guess is that with near-wall setting the mid doesn't need that much gain/eq in low end. This is in controversy with midrange being cardioid, but still, regarding range 100-200Hz. If I am wrong, and that setting lowers only bass drivers' level, distortion actually gets higher!

In my diy-dipoles I cross monopole closed box 10" woofer to a 12" dipole around 170Hz using LR2 slopes. With LR4 and 150Hz Fc distortion is lower. Still we see similar elevated distortion around xo.

Another thing we see here, that this on-axis shows a bit lower relative level for the monopole region. In-room the response is almost a straight line but it doesn't sound thin.
View attachment 25134

Ok, it just comes down to differing definitions of distortion then. I’m thinking of distortion relative to the fundamental, either in % terms or decibels. So changing a setting like that is not reducing distortion, it’s just reducing the fundamental. But we’re on basically the same page here obvs

I’m not sure I understand I quite understand what’s happening with your speakers there! You have a monocle midrange and dipole bass, or the other way around?

 

[Juhazi]

The other way - bass is monopole downfire, crossed to a dipole mid around 170Hz (and dipole all the way up).

And yes, I was talking about the system's relative distortion, like it was measured. This is one of the reasons why steep slopes are used in PA and many pro monitors - to relieve stress from the low end of high-passed driver.

 

[dc655321]

his is one of the reasons why steep slopes are used in PA and many pro monitors

Can you define "steep slopes" here? Slopes > -24dB/octave?

Any specific examples of such PA/pro monitors that exhibit this, with measurements?

 

[andreasmaaan]

And yes, I was talking about the system's relative distortion, like it was measured.

I don't follow then... Earlier you seemed to be saying that by reducing the SPL of the low-midrange (and changing nothing else), overall distortion was being lowered. I'm agreeing with that, but pointing out that lowering the SPL of any system under any circumstances will lower distortion in absolute terms, but not in any meaningful way, since distortion relative to the fundamental has not changed at a given SPL.

This is one of the reasons why steep slopes are used in PA and many pro monitors - to relieve stress from the low end of high-passed driver.

IME, LR24 slopes are fairly ubiquitous in PA and professional audio. Of course I agree that misguided ideas like 1st order Butterworth slopes are absent from PA (but I'd argue they should be absent from home audio too)

The other way - bass is monopole downfire, crossed to a dipole mid around 170Hz (and dipole all the way up).

Ok, so a similar approach to D&D. I think your use of a larger midrange and a higher xo frequency makes sense.

 

[Juhazi]

^ My point is that if we change a multiway-speaker's crossover slopes or gain per each way, distortion% at same system spl changes.

This is about driver excursion difference, but it translates to distortion well, by Kreskowsky:

Here Kreskowsky tells what happens when different radiation patterns are combined, like in D&D 8c and my AINOs

Well, acoustic fourth order LR4 (24dB/octave) is a steep slope for a hifi speaker. Revel speakers use mostly that, but symmetric LR2 or something like that is often consired better for hifi, because of less phase rotation and because max. spl/distortion is not a challenge really. Pro speaker systems often use 48 or even 96dB/octave and are capable of much much higher spl.

 

[andreasmaaan]

My point is that if we change a multiway-speaker's crossover slopes or gain per each way, distortion% at same system spl changes.

I agree 100% re: xover slopes, but I thought here we had presumed that changing the 8C's settings had no effect on crossover slopes or frequencies.

Merely changing the gain doesn't reduce distortion in any meaningful way for the reasons I mentioned above. There is less distortion, but also less fundamental. The % distortion at a given SPL remains unchanged, unless we measure the SPL only at a frequency at which we haven't changed the gain - at which frequency distortion will not have changed either.

Well, acoustic fourth order LR4 (24dB/octave) is a steep slope for a hifi speaker. Revel speakers use mostly that, but symmetric LR2 or something like that is often consired better for hifi, because of less phase rotation and because max. spl/distortion is not a challenge really. Pro speaker systems often use 48 or even 96dB/octave and are capable of much much higher spl.

That's interesting. Which PA speakers use 48 or 96 dB/octave xovers that you're aware of? I agree that these kinds of slopes are very uncommon in home audio, but I'm not familiar with many PA speakers that employ them either.

 

[Juhazi]

I guess that we don't really disagree, andreasmaan, but just use different phrasing.

Still, I want to add, that the main effect of near-wall placement and using BCB setting for 8c, is to get the benefit from boundary reinforcement of bass frequencies. When the speaker's response at listening point is equalized to match the response of a free-standing speaker, bass frequencies benefit up to 6dB in efficiency - and has lower distortion - hopefully also a bit for the cardioid midrange.

Here is what the User's manual says about this:
Boundary Coupled Bass (BCB) By placing the 8c’s close to the front wall*, the two rear-firing subwoofers of the 8c can be used in boundary coupling mode. When the 8c and the wall are acoustically coupled, the two will effectively act as a single source. In the 8c Boundary Coupled Bass (BCB) has three advantages: • Boundary coupling avoids the destructive interference between the speaker's direct sound and the boundary reflection. • Low frequency headroom is increased by up to 6 dB. • Increased directivity in the bass, for a good directivity match between the BoundaryCoupled Bass and the cardioid midrange. ... You can further flatten the in-room response with the parametric equalizer if necessary. Dutch & Dutch recommends that you use in-room measurements to properly optimize the equalizer settings. Generally speaking, the closer to the wall, the better boundary coupling works.

(Bolded by me: because of 6dB gain from boundary reinforcement and after the response is equalized, better efficiency gives 6dB more headroom = lower distortion at same spl)

Actually now I start to wonder, if the BCB setting actually helps only the backside woofers. A cardioid midranges will get only minimal benefit from the wall, because of the cardioid pattern. So, only the distortion below xo gets lower. We need measurements to verify this!

ps. I said PA systems use up to 48/96dB slopes, not PA speakers!

 

[andreasmaaan]

Actually now I start to wonder, if the BCB setting actually helps only the backside woofers. A cardioid midranges will get only minimal benefit from the wall, because of the cardioid pattern. So, only the distortion below xo gets lower. We need measurements to verify this!

I think we just see this in fundamentally different terms but that’s ok, there doesn’t seem to be any technical disagreement

Concerning the question re BCB mode, it should have no effect in the cardioid range of the speaker, as rearward output in this range is negligible.

I think your earlier speculation that the slight elevation in the amplitude response in the cardioid region of the speaker is intended to compensate for the lower power response in this region (compared to conventional monopoles) remains the most plausible explanation.

 

[Juhazi]

I link the D&D designer's thread again, this is after 8c was published.
https://www.diyaudio.com/forums/multi-way/192737-2-waveguide-cardioid-18.html#post5422744

The patent https://patents.google.com/patent/EP3018915A1/en

The basic differences between 8c and Kii Three are
1) 8c midrange makes cardioid response acoustically 100Hz>, with the resisitive side vents and enclosure's internal shape and bass below 100Hz is backfire monopole. Bass and mid responses are matched with eq and delay settings of dsp.
2) Three uses sound source separation and electric delay to set and adjust cardioid response from low bass up to where midrange looses backside radiation (around 500Hz). The auxiliary BXT module adds monopole backfire bass to compensate low efficiency and thus get lower distortion and higher clean spl.

We see backside radiation profiles very seldom, this is the best resource , but so far neither of these two have been measured/published
https://www.princeton.edu/3D3A/Directivity.html

Gainphile's diy project combines different patterns - dipole bass and mid, monopole tweeter in waveguide. Notice how sharp transition is.
https://www.princeton.edu/3D3A/Directivity/Gainphile R16/index.html

 

[youngho]

Is it possible that the strange dip around 60 Hz represents a problem with the "free-standing" setting? It seems to me that incorrect delay for the rear woofers so that their output is out of phase around 60 hz with respect to the front woofer's output might explain this. There is no issue with the rear woofer only measurement or with the listening window curve with woofers turned off. Other settings were not tested, but if D&D's own measurements were done with another setting, that could explain the discrepancy.

Young-Ho

 

[Juhazi]

^ Remember wavelength of 60Hz is 5,67m. My guess is that it is a reflection+mode null between chamber's walls. Chamber is sais to be anechoic only from 80Hz up. When they measured the speaker rotated 180¤, direct bass sound hit the microphone first, so reflection was masked. In frontside measurement direct bass sound was weaker and the reflection caused some nulling.

Delay matching is needed to fine-tune phase of bass and mid at crossover frequency 100Hz, and to correct added delay from reflection of bass from the wall. Physical distance between mid and bass drivers is roughly 50cm, so appropriate delay for mid+tweeter to bass is 1,5ms and it stands for 54¤ rotation of phase at 100Hz (if I calculated it right)

https://nrc.canada.ca/en/certificat...ment-calibration-service/acoustical-standards

 

[youngho]

You're probably right. I looked up the Kef Blade Two measurements, since it has side-firing woofers, and it shows a similar null around 60 Hz. (https://www.soundstagenetwork.com/i...&catid=77:loudspeaker-measurements&Itemid=153 )

Young-Ho

^ Remember wavelength of 60Hz is 5,67m. My guess is that it is a reflection+mode null between chamber's walls. Chamber is sais to be anechoic only from 80Hz up. When they measured the speaker rotated 180¤, direct bass sound hit the microphone first, so reflection was masked. In frontside measurement direct bass sound was weaker and the reflection caused some nulling.

Delay matching is needed to fine-tune phase of bass and mid at crossover frequency 100Hz, and to correct added delay from reflection of bass from the wall. Physical distance between mid and bass drivers is roughly 50cm, so appropriate delay for mid+tweeter to bass is 1,5ms and it stands for 54¤ rotation of phase at 100Hz (if I calculated it right)

https://nrc.canada.ca/en/certificat...ment-calibration-service/acoustical-standards

 

[andreasmaaan]

When they measured the speaker rotated 180¤, direct bass sound hit the microphone first, so reflection was masked. In frontside measurement direct bass sound was weaker and the reflection caused some nulling.

The reflection would have to have been very strong - almost as strong as the direct sound - to cause such a null.

I wouldn’t expect the NRC to have so greatly overestimated the low frequency ability of their chamber, but of course it’s possible.

When they measured the speaker rotated 180¤, direct bass sound hit the microphone first, so reflection was masked. In frontside measurement direct bass sound was weaker and the reflection caused some nulling.

Delay matching is needed to fine-tune phase of bass and mid at crossover frequency 100Hz, and to correct added delay from reflection of bass from the wall. Physical distance between mid and bass drivers is roughly 50cm, so appropriate delay for mid+tweeter to bass is 1,5ms and it stands for 54¤ rotation of phase at 100Hz (if I calculated it right)

I think those calculations are correct. But unless the front wall of the anechoic chamber is less than 25cm from the back of the speaker (which surely it couldn’t have been), the direct sound from the woofers would have hit the microphone before the reflection with the speaker oriented in either direction.

In other words, I don’t think your theory can account for the significant difference between the front-facing and rear-facing measurements.

Btw, sorry we seem to keep disagreeing on various minor points here. I actually get the impression we have a pretty similar outlook on most of the central issues here. These discussions are to me just an interesting way to explore what might be happening, so I hope it doesn’t come across as an attempt to score points.

 

[Soniclife]

These discussions are to me just an interesting way to explore what might be happening, so I hope it doesn’t come across as an attempt to score points.

That's what I've been taking from this. It is interesting that even with a chamber things are hard to measure, and translating that to how they are expected to work in real rooms far from easy. I'm more interested in this than how these particular speakers measure.