Ascilab C8C - BX8C in-room measurements / P1-P2-P3 / Manual EQ / Dirac ART

[markstein]

Dear community,

I offered to share some in-room measurements of Ascilab C8C - BX8C in modes P1 (max SPL) - P2 (standard cardioid) - P3 (extended cardioid), with manual EQ and with Dirac ART. Here they are:

Microsoft OneDrive

1drv.ms

Edit: now also added MMM measurements (32 averages around MLP including MMM-optimized EQ). /Edit

Obviously my room is suboptimal (an acoustic nightmare ) - nevertheless and despite the quite ugly looking MLP measurements the loudspeakers sound very very good to my ears!

Edit2: The bass shelf in the shown measurements is pretty heavy - it is now reduced to +3dB. Measurements still show +6dB./Edit2

Findings:
- The more you extend the bass cardioid mode, the less gain you have in the bass region.
- Investigate the excess group delay graphs for each setting to see how reflections are influenced. The results are not conclusive.
Keep in mind that there are also strong reflections from the undamped back wall behind the microphone (behind the listener). I might add a broad band absorber there in the future.
- Due to the asymmetric placement (corner and window left, open space and table right), phase around 100Hz region does not match and leads to a broad dip in summed response. Cardioid setting does not help a lot with this issue. Manual EQ neither. Only Dirac ART makes the trick.
- Dirac ART obviously manages to improve the region below 150Hz very effectively (SPL smoothness of L, R and L+R response, excess group delay/ reflections). Due to the 20Hz high pass, group delay in lowest bass increases significantly, however.
- Dirac ART correction up to 400Hz looks much better in the graphs than correction up to 200Hz, however, sounds less natural to my ears - voices start sounding really thick. I prefer to leave the 300Hz region dip as-is and limit any correction to below 200Hz only.
- Notice: mid distortion of the left C8C is noticeably worse than the right one - probably due to a defective Hypex amp module. Ascilab will be sending a new one shortly.

How does this all sound?

Honestly, I am very happy with the speakers in setting P2 with manual EQ and do not feel an urgent need to add Dirac ART. The (difference of) measurement graphs look much worse than they sound to my ears.

These speakers sound very effortless, with a lot of punch and bass pressure - even more so without Dirac ART, probably because my manual EQ ist a littler "rougher" than the super smooth ART EQ.

Please share your findings and thoughts as well!

 

[AudioJester]

Thanks for sharing. Dirac ART with the 2 subs really does work magic upto 150Hz! Above that, its no longer ART, I think just Dirac Live.
Have you measured C8C alone without BX8C?

 

[Satanica]

Yes, ART works up to 150 Hz although when viewing on an analyser, supporting speakers/subwoofers will continue some sort of contribution up to about 200Hz.
Seemingly Dirac Live takes over from Dirac ART somewhere in this frequency range.

Regarding correction range, what I find works well is to go up as high is required to maintain a centre image between the speakers.
So, using an online tone generator tool such as https://www.szynalski.com/tone-generator/ one can determine at what frequency centre image is maintained when both left and right speakers are playing together at the same time.
I found I had to set the upper curtain range to 800Hz to achieve this and that makes Dirac Live correct up to about 700Hz.
But here's the thing, I don't use some straight line target curve for the frequencies greater than about 300Hz.
Instead the target curve follows the general trend of the speaker output, this way the tonality of the speaker should be maintained while correcting where one channel is louder and/or softer than the other.
This should help maintain a centre image by getting the left and right speakers closer together in output at the listening position, especially with a non-symmetrical room setup such as this one.
Using an online generator tool such as the one linked to, is handy to determine if the Dirac Live corrections have audibly improved the centre image across all frequencies.

 

[AudioJester]

Yes, ART works up to 150 Hz although when viewing on an analyser, supporting speakers/subwoofers will continue some sort of contribution up to about 200Hz.
Seemingly Dirac Live takes over from Dirac ART somewhere in this frequency range.

Regarding correction range, what I find works well is to go up as high is required to maintain a centre image between the speakers.
So, using an online tone generator tool such as https://www.szynalski.com/tone-generator/ one can determine at what frequency centre image is maintained when both left and right speakers are playing together at the same time.
I found I had to set the upper curtain range to 800Hz to achieve this and that makes Dirac Live correct up to about 700Hz.
But here's the thing, I don't use some straight line target curve for the frequencies greater than about 300Hz.
Instead the target curve follows the general trend of the speaker output, this way the tonality of the speaker should be maintained while correcting where one channel is louder and/or softer than the other.
And this should help maintain a centre image by getting the left and right speakers closer together in output at the listening position, especially with a non-symmetrical room setup such as this one.
Using an online generator tool such as the one linked to, is handy to determine if the Dirac Live corrections have audibly improved the centre image across all frequencies.

Interesting, thanks for sharing your method.

The problem I see with some of these automated tools (above transition zone) in home environments is they correct measurement anomolies/arefacts that are not actual audible issues - and hence make things worse.

 

[AudioJester]

@Keith_W - what are your impressions of this frequency response and what corrections should be applied?

@markstein curious what troubleshooting steps you took to isolate the plate amp as the cause of the distortion issue?

 

[badtimeticket]

I started to do full phase correction with Dirac ART, but I draw the curve to roughly match my speakers above ~500ish hz, and I try to match the natural curve of the speakers in the room, which still applies phase correction. I think it's slightly better, but nothing mind blowing.

The problem I had with my R3 metas is putting a curtain would add a weird discontinuity by using one of the curves online.

 

[Satanica]

Interesting, thanks for sharing your method.

The problem I see with some of these automated tools (above transition zone) in home environments is they correct measurement anomolies/arefacts that are not actual audible issues - and hence make things worse.

Agreed.

 

[Keith_W]

The first thing I do when I see 24 measurements is to cull them. I am not going to compare 24 measurements which each other, there are thousands of permutations of things that can be compared. So I look to see what question has been asked, and then I delete all the measurements which are not necessary to answer the question. You haven't asked a question, so I don't have much to say.

All I can make are general comments, which I am sure you can see for yourself. For e.g. it is obvious that your manual EQ (yellow) is better than no EQ.

And ART 400Hz (blue) is better than your manual EQ.

You did not take the relevant measurement for evaluating bass with MMM. That would be: L+R no EQ, vs. manual EQ, vs. ART. All these individual channel measurements are useless when evaluating bass (and especially to see the effect of ART, since ART relies on multiple speakers working together). I deleted all of them.

As to your comment about ART 400Hz sounding "thick", here is a comparison of ART 400 (blue) vs. ART 200 (yellow). The only difference is that ART 400 does not have that severe dip between 200-400Hz. The caveat with this measurement is that it is a single point measurement, and not an MMM - so you don't know if that pattern holds true over a wider listening area. As mentioned previously, you did not take that measurement.

There is a very subtle difference in the spectrogram between ART 200 (left) and ART 400 (right). You can't really see it here, you have to create an animated GIF to see. Or you could go to REW and use these settings: Normalize peak to every frequency, and limit vertical range -100ms to 600ms, horizontal from 100Hz - 1kHz. Flick between the two spectros in rapid succession and you will see that ART 400 has marginally more ringing at 265Hz. With a bit of motion, your eye sees it instantly.

Now I don't really think that that marginal increase in ringing would make ART 400 sound thick. All I know is that you really shouldn't be attempting to correct that high. If ART 200 sounds better, then go with ART 200.

Anyway let me know if you have any specific questions.

 

[markstein]

@markstein curious what troubleshooting steps you took to isolate the plate amp as the cause of the distortion issue?

HD is increased over the entire mid range for one C8C compared with the other. This was veryfied with near field measurements of mid driver responses.

Most of the time you won't notice that when just listening to music. In sweeps, when measuring, it is more easily noticeable. Sometimes there is also a buzzing, crackling sound which is clearly audible and shows in measurements.

Ascilab suggests to change the amp module as it is very unlikely that both midrange drivers have a failure, and the failure pattern does not fit to a driver root cause. We should know soon!

 

[markstein]

You did not take the relevant measurement for evaluating bass with MMM. That would be: L+R no EQ, vs. manual EQ, vs. ART. All these individual channel measurements are useless when evaluating bass (and especially to see the effect of ART, since ART relies on multiple speakers working together). I deleted all of them.

Hi Keith, I am the OP and posted the many measurements upon request from other users to show some real world in-room differences between the various cardioid modes of this loudspeaker. Therefore the many measurements


I just threw in some MMM in the last minute, as this a good way to show the perceived SPL response which often is much smoother than a single point measurement might suggest. In fact, I do not like single point room measurements a lot, however, some phenomena cannot be evaluated using MMM (group delay, impulse response, reflections, decay...).

No specific question from my side.

 

[badtimeticket]

Playing a single channel to evaluate ART is fine. Sweeping the left channel is not the same as sweeping the left speaker; ART will map that channel to multiple speakers. It's like saying you need to play LFE+LR to measure subwoofer integration. The only way to measure left speaker only when using ART is to turn off all the other speakers.

 

[Marc v E]

Dear community,

I offered to share some in-room measurements of Ascilab C8C - BX8C in modes P1 (max SPL) - P2 (standard cardioid) - P3 (extended cardioid), with manual EQ and with Dirac ART. Here they are:

Microsoft OneDrive

1drv.ms

Edit: now also added MMM measurements (32 averages around MLP including MMM-optimized EQ). /Edit

Obviously my room is suboptimal (an acoustic nightmare ) - nevertheless and despite the quite ugly looking MLP measurements the loudspeakers sound very very good to my ears!

Edit2: The bass shelf in the shown measurements is pretty heavy - it is now reduced to +3dB. Measurements still show +6dB./Edit2

Findings:
- The more you extend the bass cardioid mode, the less gain you have in the bass region.
- Investigate the excess group delay graphs for each setting to see how reflections are influenced. The results are not conclusive.
Keep in mind that there are also strong reflections from the undamped back wall behind the microphone (behind the listener). I might add a broad band absorber there in the future.
- Due to the asymmetric placement (corner and window left, open space and table right), phase around 100Hz region does not match and leads to a broad dip in summed response. Cardioid setting does not help a lot with this issue. Manual EQ neither. Only Dirac ART makes the trick.
- Dirac ART obviously manages to improve the region below 150Hz very effectively (SPL smoothness of L, R and L+R response, excess group delay/ reflections). Due to the 20Hz high pass, group delay in lowest bass increases significantly, however.
- Dirac ART correction up to 400Hz looks much better in the graphs than correction up to 200Hz, however, sounds less natural to my ears - voices start sounding really thick. I prefer to leave the 300Hz region dip as-is and limit any correction to below 200Hz only.
- Notice: mid distortion of the left C8C is noticeably worse than the right one - probably due to a defective Hypex amp module. Ascilab will be sending a new one shortly.

How does this all sound?

Honestly, I am very happy with the speakers in setting P2 with manual EQ and do not feel an urgent need to add Dirac ART. The (difference of) measurement graphs look much worse than they sound to my ears.

These speakers sound very effortless, with a lot of punch and bass pressure - even more so without Dirac ART, probably because my manual EQ ist a littler "rougher" than the super smooth ART EQ.

Please share your findings and thoughts as well!

View attachment 551201View attachment 551202View attachment 551203View attachment 551204

Thank you very much for sharing! It helps me too. Especially the part about dirac ART.

Regarding eq and the bookshelve to the right I may have an idea that could improve sound.
What if you place the books at the higher shelves, at the height of the tweeter and midrange driver?

 

[badtimeticket]

Btw, here are my measurements too. You should ignore below 30hz and probably the general shape because it's not a consistent target curve and the goal is to more show irregularities. The 30hz dropoff is a neighbor curve, because if the UMIK-1 is accurate, these are going to 14hz in-room.

Basically I started to run ART with the subs in the 2 rear corners. The measured bass is very smooth is very consistent seat to seat.

r3m c8c measures - Google Drive

drive.google.com

 

[badtimeticket]

Blue is C8C with ART + 2 subs (back left and back right). Magenta is R3 meta with ART + 2 subs (but they are front back here, which is the best arrangement I found for those speakers). TBH it's hard to see any cardioid benefits here if any, but maybe those are visible in different measurements? You can see from ~150-600 is still very problematic for me. This is probably a good demonstration of ART with effectively 4 subs.

My subjective impression is the soundstage feels more 3D and percussion is much more clear, with much better instrument separation. But treble is not tapering off in my room, which I corrected with my drawn curve.

 

[markstein]

Blue is C8C with ART + 2 subs (back left and back right). Magenta is R3 meta with ART + 2 subs (but they are front back here, which is the best arrangement I found for those speakers).

How did you end in those two different sub placements?
In my setup as measured in the 1st post, one subwoofer is co-located to the front left loudspeaker (front left room corner), the other subwoofer is behind the listening position, on the right side.

 

[badtimeticket]

How did you end in those two different sub placements?
In my setup as measured in the 1st post, one subwoofer is co-located to the front left loudspeaker (front left room corner), the other subwoofer is behind the listening position, on the right side.

R3M was trial and error. They can’t go that low so they can’t do the fronts well. I tried diagonal, two in front, and one front one back and this was the best for my room by looking at the corrected spread.

I read 4 subs in corners is best though. So when I saw how low these were going, I just did it as 4 corners, and there’s basically no variance in my measured positions.

 

[markstein]

I read 4 subs in corners is best though. So when I saw how low these were going, I just did it as 4 corners, and there’s basically no variance in my measured positions.

Correct. Will try this as well.

I guess you set the F-low support of all loudspeakers to 20Hz, so the front loudspeakers are supporting in the same range as the subs? Did you check how bass distortion changed with/ without ART in this setup?

 

[badtimeticket]

Correct. Will try this as well.

I guess you set the F-low support of all loudspeakers to 20Hz, so the front loudspeakers are supporting in the same range as the subs? Did you check how bass distortion changed with/ without ART in this setup?

Correct. There is not significant distortion but keep in mind I’m effectively high passing at 30 so 20hz is not relevant and also I don’t listen very loud so it becomes a non issue for me. My subs are also very small! 2x SVS 3000 micro

 

[829]

@markstein Amazing! Thank you.

I am not familiar with Dirac ART but I find the 20Hz HPF disturbing especially, since I enjoy pipe organ music (or the extra sub-bass physical vibration that some movies provide) a lot. So I researched and found that there should be an infra-bass bypass option, which does not apply the 20Hz HPF! It is probably also best not to use it with the C8C + BX8C, as I remember a conversation with a friend who owns the C8C and got suggested by @AsciLab not to use high order pass filter because it would add too much delay around center frequency; in order to reduce sub-bass THD or peak (instead, a first order HPF @ 35Hz was suggested - if required). That was for the C8C alone but probably applies for C8C + BX8C equally.

Dirac ART seems to work well, though! Generally, Dirac Live (as any other software) offers better results if used with a single measurement at the MLP and with such excellent speakers(' horizontal directivity), the results will be predictably good also around the MLP. I just wish there was an extended assessment including Dirac's ART: https://ethanwiner.com/dirac/dirac.htm

 

[markstein]

@markstein Amazing! Thank you.

I am not familiar with Dirac ART but I find the 20Hz HPF disturbing especially, since I enjoy pipe organ music (or the extra sub-bass physical vibration that some movies provide) a lot. So I researched and found that there should be an infra-bass bypass option, which does not apply the 20Hz HPF! It is probably also best not to use it with the C8C + BX8C, as I remember a conversation with a friend who owns the C8C and got suggested by @AsciLab not to use high order pass filter because it would add too much delay around center frequency; in order to reduce sub-bass THD or peak (instead, a first order HPF @ 35Hz was suggested - if required). That was for the C8C alone but probably applies for C8C + BX8C equally.

Dirac ART seems to work well, though! Generally, Dirac Live (as any other software) offers better results if used with a single measurement at the MLP and with such excellent speakers(' horizontal directivity), the results will be predictably good also around the MLP. I just wish there was an extended assessment including Dirac's ART: https://ethanwiner.com/dirac/dirac.htm

I tested with the Dirac ART PC version and there is as well the possibility of deactivating the 20Hz HPF. When doing so, there is unfortunately a huge peak at around 22Hz followed by a deep null. So I left the HPF switched on. This setting definitely will require more testing in my end - unfortunately the Dirac test phase ends today.