Markaudio CHN 110 building and comparisons with Genelec 8340 SAM monitors

[Waxx]

wel, i'm going to post it again so you can compare. This is my 15° off axis measurment in my listening room, without eq or other processing. The box is a MLTL tuned to 30Hz with the CHN110. Room is about 3.5x5.5x2.7m in size and measurement was made to eq the driver in my room (but little eq was needed outside bass frequencies). Mic was the Umik-1 and software is off course REW. It's an average of 10 measurements on the same spot.

 

[Tangband]

It is not difficult to get a rough estimate of the on-axis frequency response (assuming that the half-space manufacturer driver measurement is reliable) and speaker radiation.

For this we take Tangband's specification (31L, BR @36Hz) and simulate the low frequency response of the speaker with baffle step (orange doted line). Then we use the manufacturer half-space FR measurement and simulate the driver on-axis FR on the speaker baffle and adjust the low FR response.
This gives us approximately the following on-axis FR:
View attachment 304446
In reality, it should look a bit more even in the low frequency range below 100Hz, as I suspect that some room resonances are included in the manufacturer's measurement.
Because VCAD does not include the cabinet depth in the simulation, its influence on the on-axis FR is missing.
Due to the slim and high cabinet shape of the loudspeaker, a cabinet resonance occurs around 170Hz, which also will have an influence on the on-axis FR, if this resonance is not damped (the same applies to possible BR port resonances).


Since Tangband will make detailed measurements anyway, I will spare myself an time consuming BEM simulation of the speaker to get the radiation and use the simplified simulation in VCAD (normalized hor and ver radiation):
View attachment 304454
There are a few things to keep in mind when interpreting the radiation simulation.
The influence of the cabinet depth on the radiation is missing, which should slightly change the radiation somewhere around 1kHz.

Above 4-5kHz, the radiation should be less narrow than shown in the simulation, since the driver has a special dust cap that probably takes over the radiation of the high frequencies to a large extent.

I don't know of any off-axis measurements for the CHN-110 driver, but the equally sized, but more expensive, Pluvia 11 has an identical dust cap and shows significantly less beaming than one would expect from a 5.5'' driver.
First manufacturer on-axis measurement, then magazine 0,15,30 deg half-space measurement of the Pluvia 11:
View attachment 304455 View attachment 304456
Source: Klang und Ton 2021-3

Bottom Line: The pronounced on-axis frequency response dip in the range of 2-5 kHz is not likely to be completely flattened off-axis either. This could lead to a rather warm, laid back sound impression. It could also lead to the sound stage being perceived more at height or behind the speakers ("deep sound stage").
The on-axis FR is actually too flat for a full-range driver of this size; one would actually tune to a flat 20°-30° FR. But maybe the special dust cap is sufficient to give a balanced sound impression.

Thanks for the simulation. My chn110 speakers are 24 cm deep.
Now I have the 3 components needed for a passive line level baffle step correction for this speaker , I not sure its needed for best sound though because of the concrete walls in my listening room. The timbre from the Rega dac is also perceived slightly warmer than my Yamaha wxc50. I will try it anyway next week.

Edit: That Pluvia 11 driver you mention seems to be a good alternativ if one needs to have smaller boxes.

 

[ctrl]

Mic was the Umik-1 and software is off course REW. It's an average of 10 measurements on the same spot.

An averaged room measurement with heavy smoothing can give a rough overview of the tonality of the loudspeaker in the room, but is not comparable to a "quasi anechoic" on-axis measurement or a full set of hor and ver measurements. In your measured microphone signal it is no longer possible to distinguish between direct and reflected sound, nor between hor and ver reflections,... but the human ear is able to do this.

Here an on-axis measurement at 1.2m of an 170mm waveguide. Yellow curve 1/6 smoothed room measurement, pink curve the gated "quasi anechoic" measurement. You can see that room resonances can add up 3dB sound preassure level (e.g. at 2kHz) and above 3kHz resonances are clearly attenuated by the 1/6 octave smoothing.


Turn your loudspeaker so that it points into the center of the room (away from reflecting surfaces) and measure the axial frequency response at a distance of 0.3m. For the graph of the measurement choose 1/24 octave smoothing.

This should show us the resonance behavior of the driver above 1kHz quite well (even better would be with gate). If you then set a gate to get the quasi anechoic measurement, the CSD diagram is meaningful about the decay behavior of the driver.

 

[badspeakerdesigner]

I don't find the in room measurements all that useful. I have plenty of speakers that have pretty glaring issues when measured quasi-anechoic, but the rooms smooths it all out. You wouldn't know the problems were there if you didn't take better measurements. It's not that hard to get some good quasi-anechoic info but yet no one with these drivers gets that info :/

This is the dispersion I got on my friends speaker with chn40. It sounds spacious and nice mid range coherency but imaging was nearly impossible for the brain to figure out IME.

 

[kemmler3D]

It is not difficult to get a rough estimate of the on-axis frequency response (assuming that the half-space manufacturer driver measurement is reliable) and speaker radiation.

For this we take Tangband's specification (31L, BR @36Hz) and simulate the low frequency response of the speaker with baffle step (orange doted line). Then we use the manufacturer half-space FR measurement and simulate the driver on-axis FR on the speaker baffle and adjust the low FR response.
This gives us approximately the following on-axis FR:
View attachment 304446
In reality, it should look a bit more even in the low frequency range below 100Hz, as I suspect that some room resonances are included in the manufacturer's measurement.
Because VCAD does not include the cabinet depth in the simulation, its influence on the on-axis FR is missing.
Due to the slim and high cabinet shape of the loudspeaker, a cabinet resonance occurs around 170Hz, which also will have an influence on the on-axis FR, if this resonance is not damped (the same applies to possible BR port resonances).


Since Tangband will make detailed measurements anyway, I will spare myself an time consuming BEM simulation of the speaker to get the radiation and use the simplified simulation in VCAD (normalized hor and ver radiation):
View attachment 304454
There are a few things to keep in mind when interpreting the radiation simulation.
The influence of the cabinet depth on the radiation is missing, which should slightly change the radiation somewhere around 1kHz.

Above 4-5kHz, the radiation should be less narrow than shown in the simulation, since the driver has a special dust cap that probably takes over the radiation of the high frequencies to a large extent.

I don't know of any off-axis measurements for the CHN-110 driver, but the equally sized, but more expensive, Pluvia 11 has an identical dust cap and shows significantly less beaming than one would expect from a 5.5'' driver.
First manufacturer on-axis measurement, then magazine 0,15,30 deg half-space measurement of the Pluvia 11:
View attachment 304455 View attachment 304456
Source: Klang und Ton 2021-3

Bottom Line: The pronounced on-axis frequency response dip in the range of 2-5 kHz is not likely to be completely flattened off-axis either. This could lead to a rather warm, laid back sound impression. It could also lead to the sound stage being perceived more at height or behind the speakers ("deep sound stage").
The on-axis FR is actually too flat for a full-range driver of this size; one would actually tune to a flat 20°-30° FR. But maybe the special dust cap is sufficient to give a balanced sound impression.

Good analysis, thanks. The 3P measurements of the Pluvia vs. manufacturer definitely give me pause. The >5khz response is ragged AF in the 3rd party graph!

 

[Tangband]

Good analysis, thanks. The 3P measurements of the Pluvia vs. manufacturer definitely give me pause. The >5khz response is ragged AF in the 3rd party graph!

… the ragged response must be seen in the light of using no crossover at all in the usual crossover region . There are advantages in the midrange without a crossover. This might be the reason I prefere the sound of this chn110 compared to Kef ls50, a coaxial which measures better but has a crossover .

Theres a ” directness” to the sound with the chn110 and at the same time the sound dont seem to be colored , which was the case with my earlier two DIY speakers that had only one driver ( la petite audiophile och Mupps )

 

[jaakkopetteri]

Some MA measurements at Dibirama:

DIBIRAMA - ITALIAN DIY LOUDSPEAKERS

{loadposition carosello} Vai al libro

www.dibirama.altervista.org

DIBIRAMA - ITALIAN DIY LOUDSPEAKERS

{loadposition carosello} Vai al libro

www.dibirama.altervista.org

Plenty of Dayton, Fostex and TB fullrangers measured too

 

[ctrl]

Good analysis, thanks. The 3P measurements of the Pluvia vs. manufacturer definitely give me pause. The >5khz response is ragged AF in the 3rd party graph!

Yes, the fact that the break-up resonances are virtually non-existent in the manufacturer measurements and figuratively poke your eye out when looking at it in the independent measurements is "normal".
I have independent measurements of the Mark Audio Alpair 10 Gen 2, and the difference between the manufacturer's measurement and that of the magazine is just as dramatic in the range above 5 kHz.

I can hardly believe it myself, but it makes a difference if measurements are presented with 100 or 120 dB scaling plus smoothing or with 50 dB scaling and no or little smoothing
The interesting thing is that I know that the manufacturer's measurements are deliberately misleading and yet I feel that the driver with the smoother frequency response is "better", would therefore opt for the left driver (although both identical) - this is how sales psychology works

By the way, the measurement data of Mark Audio drivers are really not bad, they just can not perform miracles and show break-up resonances with delayed decay in the high frequency range like almost all driver, maybe less severe.

 

[Waxx]

An averaged room measurement with heavy smoothing can give a rough overview of the tonality of the loudspeaker in the room, but is not comparable to a "quasi anechoic" on-axis measurement or a full set of hor and ver measurements. In your measured microphone signal it is no longer possible to distinguish between direct and reflected sound, nor between hor and ver reflections,... but the human ear is able to do this.

Here an on-axis measurement at 1.2m of an 170mm waveguide. Yellow curve 1/6 smoothed room measurement, pink curve the gated "quasi anechoic" measurement. You can see that room resonances can add up 3dB sound preassure level (e.g. at 2kHz) and above 3kHz resonances are clearly attenuated by the 1/6 octave smoothing.
View attachment 304493

Turn your loudspeaker so that it points into the center of the room (away from reflecting surfaces) and measure the axial frequency response at a distance of 0.3m. For the graph of the measurement choose 1/24 octave smoothing.

This should show us the resonance behavior of the driver above 1kHz quite well (even better would be with gate). If you then set a gate to get the quasi anechoic measurement, the CSD diagram is meaningful about the decay behavior of the driver.

I can't do real anchionic measurements here, but i did a non gated measurement (had no time to check the gate config) of that same box. You got a rise in the treble (all cone drivers have that, but not as controlled as here). But i thought we wanted a flat response in room, not a anchionic flat response for a good speaker... A single driver will always beam a bit, and you can make that sound good in a room by having some tilted treble on axis. But nobody listens on axis (they are sitting between 2 speakers for stereo) so...

For speaker design this can be usefull, not for juding how a speaker sound in a room.

anyway, here the measurement at 0.3m 0° on axisn, the file (REW) and a picture of the graph at 1/24th smoothing (like requested). A in my other measurements, below 250Hz my room comes into play. This is all dsp disabled, mono (the right one in my setup) from rew to mindsp flex to amp (NCore) to speaker. I know this is a good speaker and everybody who heared it agrees (a few copies are made already by friends). But it's not everybody's cup of tea, i know.

 

[kemmler3D]

Yes, the fact that the break-up resonances are virtually non-existent in the manufacturer measurements and figuratively poke your eye out when looking at it in the independent measurements is "normal".
I have independent measurements of the Mark Audio Alpair 10 Gen 2, and the difference between the manufacturer's measurement and that of the magazine is just as dramatic in the range above 5 kHz.

I can hardly believe it myself, but it makes a difference if measurements are presented with 100 or 120 dB scaling plus smoothing or with 50 dB scaling and no or little smoothing
The interesting thing is that I know that the manufacturer's measurements are deliberately misleading and yet I feel that the driver with the smoother frequency response is "better", would therefore opt for the left driver (although both identical) - this is how sales psychology works
View attachment 304518

By the way, the measurement data of Mark Audio drivers are really not bad, they just can not perform miracles and show break-up resonances with delayed decay in the high frequency range like almost all driver, maybe less severe.

I guess this makes sense, I just somehow managed to believe someone had actually created a midwoofer with breakup above 10khz. The amount of smoothing on that graph is enough to make John Hurt look like Johnny Depp.

 

[Tangband]

I guess this makes sense, I just somehow managed to believe someone had actually created a midwoofer with breakup above 10khz. The amount of smoothing on that graph is enough to make John Hurt look like Johnny Depp.

The treble sounds fine with the chn110, I cant hear any breakup resonanses ( even if they are there ) . As I stated earlier , this driver sounds much better than it should. Theres more details , more open sounding and an illusion of 3D depth than with Genelec and also Monitor audio rx6 .

 

[ctrl]

But i thought we wanted a flat response in room, not a anchionic flat response for a good speaker...

Actually, no. The in-room response should actually drop slightly at higher frequencies. The slope of the in-room response depends on the speaker concept and your own taste.

For "normal" speakers the rough guideline is as flat as possible direct sound, i.e. flat on-axis FR (+ listening window) and an evenly sloping in-room response to high frequencies - this applies to "normal" speakers that are tuned on axis.

This guideline doesn't applies to broadband loudspeakers, since they usually radiate too narrow at high frequencies (except for some small 2''- 3'' drivers).

These are usually not listened to on axis and mostly tuned to a flat anechoic FR between 10-30°. Therefore, the on-axis FR of these speakers, usually increases to high frequencies.
This results in a more even sloping sound power and in-room response.

This is all dsp disabled, mono (the right one in my setup) from rew to mindsp flex to amp (NCore) to speaker.

Thanks for the measurement.

There seems to be something wrong with your measurement, as it is extremely different from the manufacturer's measurement.
In your measurement, the SPL on axis increases by a whopping 16dB from 1-10kHz. In the manufacturer's half-space measurement, the FR doesn't show this behavior for the same frequency range (it stays nearly even). The dip between 2-5kHz also does not show up in your measurement - a cabinet with 17-30cm width should enhance the on-axis dip further.

My guess is that your miniDSP "bypass mode" is not working properly, in other words your filters are still active.

 

[Waxx]

Actually, no. The in-room response should actually drop slightly at higher frequencies. The slope of the in-room response depends on the speaker concept and your own taste.

For "normal" speakers the rough guideline is as flat as possible direct sound, i.e. flat on-axis FR (+ listening window) and an evenly sloping in-room response to high frequencies - this applies to "normal" speakers that are tuned on axis.

This guideline doesn't applies to broadband loudspeakers, since they usually radiate too narrow at high frequencies (except for some small 2''- 3'' drivers).

These are usually not listened to on axis and mostly tuned to a flat anechoic FR between 10-30°. Therefore, the on-axis FR of these speakers, usually increases to high frequencies.
This results in a more even sloping sound power and in-room response.


Thanks for the measurement.

There seems to be something wrong with your measurement, as it is extremely different from the manufacturer's measurement.
In your measurement, the SPL on axis increases by a whopping 16dB from 1-10kHz. In the manufacturer's half-space measurement, the FR doesn't show this behavior for the same frequency range (it stays nearly even). The dip between 2-5kHz also does not show up in your measurement - a cabinet with 17-30cm width should enhance the dip further.
View attachment 304676 View attachment 304677
My guess is that your miniDSP "bypass mode" is not working properly, in other words your filters are still active.

My filters look very different. It could be something went wrong altough, i can take a look at the measurements later and use an other dac (to avoid dsp processing). But that won't be today or this week. I got many other things to do first. I did those measurements in a hurry, and in my living room with a lot of furniture. I don't really have a good space to do these kind of measurements as i live in a very old building with a lot of small spaces, no big ones. And the belgian summer we have this year does not help (stormy, wet and cold for the last 3 weeks) to do them outside...

 

[ctrl]

My filters look very different. It could be something went wrong altough, i can take a look at the measurements later and use an other dac (to avoid dsp processing). But that won't be today or this week.

No problem, since @Tangband will be taking measurements anyway, so we will soon have a comparison to your measurements.

I assumed that you used an individually calibrated measurement microphone and the calibration file was active when measuring with REW - otherwise this would be another possible source of error.
It would also be possible that the manufacturer's measurement is crap (very low probability) or that the manufacturer has made changes to the driver (very unlikely with such huge impact).

 

[kemmler3D]

If I were to build something with these, it would probably be nearfield desktop-type speakers (I don't want to use the word "monitor" necessarily), where the beaming might be dealt with directly by pointing them directly at my ears. In that case maybe they would work OK... but the ragged HF response then becomes possible more of a problem. People keep saying you can't hear notches, but ... yes you can. If you turn a comb filter on and off, the difference is always stark.

 

[badspeakerdesigner]

I personally don't agree that dispersion matters less nearfield, if anything I feel it matters more. Just take a woofer and get nearfield and you'll quickly see that the placement there doesn't fix the problem.

 

[Tangband]

If I were to build something with these, it would probably be nearfield desktop-type speakers (I don't want to use the word "monitor" necessarily), where the beaming might be dealt with directly by pointing them directly at my ears. In that case maybe they would work OK... but the ragged HF response then becomes possible more of a problem. People keep saying you can't hear notches, but ... yes you can. If you turn a comb filter on and off, the difference is always stark.

Again… with a fullrangedriver you trade the less good frequency response in the upper registers to no need of a crossover of any kind - this (might) sound more open and real in certain cases .

Some loudspeakerconstructors says that one can not hear high Q notches that well, and using 1/6 oct smoothing at measurements, would show about the same result as the human ear hear .

Im not sure about any consensus about this though.

 

[RickS]

Yes, the fact that the break-up resonances are virtually non-existent in the manufacturer measurements and figuratively poke your eye out when looking at it in the independent measurements is "normal".
I have independent measurements of the Mark Audio Alpair 10 Gen 2, and the difference between the manufacturer's measurement and that of the magazine is just as dramatic in the range above 5 kHz.

I can hardly believe it myself, but it makes a difference if measurements are presented with 100 or 120 dB scaling plus smoothing or with 50 dB scaling and no or little smoothing
The interesting thing is that I know that the manufacturer's measurements are deliberately misleading and yet I feel that the driver with the smoother frequency response is "better", would therefore opt for the left driver (although both identical) - this is how sales psychology works
View attachment 304518

By the way, the measurement data of Mark Audio drivers are really not bad, they just can not perform miracles and show break-up resonances with delayed decay in the high frequency range like almost all driver, maybe less severe.

A few years ago, after seeing some rave subjective reviews of Mark Audio drivers, I spent some time pondering them.

I came to many comparable conclusions as you state in this post. The manufacturers posted measurements were often incomplete and/or misleading. As you said, no miracles!

 

[Tangband]

Thanks for the simulation. My chn110 speakers are 24 cm deep.
Now I have the 3 components needed for a passive line level baffle step correction for this speaker , I not sure its needed for best sound though because of the concrete walls in my listening room. The timbre from the Rega dac is also perceived slightly warmer than my Yamaha wxc50. I will try it anyway next week.

Edit: That Pluvia 11 driver you mention seems to be a good alternativ if one needs to have smaller boxes.

Finnished with the passive line level baffle step correction, very similar to the Linkwitz drawings , but the correction is a little lower in frequency , thus another value of the capacitor. The 2 resistors ( vishay ) are 4,7 kOhm ( R1 ) and 12 kOhm ( R2 ) and the polypropylene capacitor C2 is 22 nF.
This filter is put between my dac and my amplifier. Now listening….

Linkwitz drawings:

 

[Tangband]

With the Yamaha as digitalstreamer and Rega dac as source, the sound results with the passive baffle step correction wasnt entirely any better , the dynamics became slightly worse and Im more directly in touch with the music without it. So now Im listening without baffle step correction. Soundwise I dont feel there is to little bass.

The amplifier testing with gear up to 1000 dollars havent payed of yet, - The Aiyima a04 is an encredible bargain soundwise with a very clear basspitch and good dynamics, without sounding bad in the higher registers.

I have instead pulled the trigger for an Audiophonics ncore 250 poweramp - it will be interesting if this will be a sound upgrade or downgrade from the Aiyima tpa3251 with lm4562. I keep everything open.

(If this turns out well, theres a chance I will swap my Yamaha wxc50 as streamer/preamp and buy a WiiM pro+ or a Linn DS instead . Both the WiiM and the ncore amplifier are 14 cm wide…)