Multi-Channel, Multi-Amplifier Audio System Using Software Crossover and Multichannel-DAC

[Kal Rubinson]

In JRiver, we can choose DIYNHK ASIO Driver (ASIO) and set it to 2xDSD in native format;

Why?

 

[QMuse]

Why?

Obviously for the sound to be even more "slightly but significantly natural, clear and better in 3D perspectives, as well as S/N".

 

[Kal Rubinson]

Obviously for the sound to be even more "slightly but significantly natural, clear and better in 3D perspectives, as well as S/N".

If you say so.
Are you aware that JRiver upsamples DSF via a PCM intermediary?

 

[dualazmak]

Next step towards the utilization of a mutli-channel apmlifier...

For a while, at least during the coming several weeks, I will continue to try, test and burn-in my DAC8PRO using the E-460+LC Network single amp system.

On the other hand, I am gathering various information on Hypex module featured configurable multi-channel amplifiers. As I wrote in my post #55 and #88, I am very much interested in configurable NORD ONE MP NCXXX 4-8 122-500W Custom Configurable Channel Amplifier.

After visiting Burning Sound's nice thread entitled "Multichannel amps for active systems", I am also interested in Apollon Audio NCMP8200.

In my post #55, I wrote the possible multi-amplifier trials with two of YAMAHA XM4080, so that I may flexibly try 8 independent amps each of them has volume/level controller. I am now suspending, however, to place purchase order for two of XM4080, since a Japanese local small amplifier factory is offering me to try, free of charge, their rather expensive Hypex NC400 (four NC400 in one amp) stereo 2-way 4-channel Digital Amplifier, DENTEC DP-NC400-4.

I am sorry, but their rather naive web pages are only in Japanese, and your web browser would properly translate the pages into English. The inside view of DENTEC DP-NC400-4 is unbalanced RCA input one, but the president of DENTEC (Sound DEN) kindly offering me to try two of DENTEC DP-NC400-4 specially built with balanced XLR inputs; he said that it is quite easy for him since XLR conectors are already built in.

I assume that two of DENTEC DP-NC400-4 (XLR inputs) should be better than PA type YAMAHA XM4080 for trials in my rather Hi-Fi multi-channel project, and it should be a good lead to NORD ONE or Apollon NCMP8200.

It looks that president of DENTEC (Sound DEN) himself, will be happy to load two of DENTEC DP-NC400-4 (XLR inputs) on his car, and bring and install them for free trial at my home hopefully in the end of June, even though his factory is about 1,000 km away in Hiroshima Prefecture. Looks he is also interested in my project, especially DAC8PRO, first one in Japan, I believe.

Well, by the end of this week, I would like to try the 5-way 10-channel EKIO configuration, still using the single-amplifier+LC network system by changing the XLR cable connections;

As wrote in my post #12, the main reason for still keeping super low SL-L and SL-R channels here is that, in some great but rather old music sources ripped from CD or digitized from LP records, unpleasant small volume super low Fq noises around 20 - 35Hz caused by hall air conditioning or by LP records' bending are deteriorating the total sound quality which can be mute off by the Mute buttons of SL-L and SL-R channel panels.

 

[dualazmak]

If you say so.
Are you aware that JRiver upsamples DSF via a PCM intermediary?

JRiver, as well as Roon, can up-sample up to 8xDSD "on the fly"; down sampling also. Furthermore, I already have many 1xDSD, 2xDSD and 4xDSD files (about 200 tracks out of my 20,000 track library), and this is one of the main reasons that I will preserve the OPPO SONICA DAC (OK for 4xDSD native)+single amp+LC Netwark system, even after completion of my current multi-channel system where software crossover EKIO handles in 192 kHz 24 bit.. Please refer to my post #1, and look carefully at my system diagram there.

Also you would please look at my audio sampler tracks consists of 44.1 kHz PCM, 1xDSD, and 2xDSD files.

 

[dualazmak]

This looks ok, but if you really want to check if the delay is set to correct value you should make phase measurement and see if there is a phase mismatch at crossover points.

It would also make sense to check if impulse responses of sub and main are time aligned.

Have you already read my post series on "Phase Issues"?
Please refer to my post #31"Phase issues...-1-", #33 "Phase issues...-2-", #37 "Phase issues...-3-" and #39 "Phase issues...-4-" .

 

[dualazmak]

It was a really nice day to test and check DAC8PRO for the first time. Let me go to bed, the time here now is 24:53....

Please stay safe!

 

[QMuse]

If you say so.
Are you aware that JRiver upsamples DSF via a PCM intermediary?

I was joking.

 

[QMuse]

Have you already read my post series on "Phase Issues"?
Please refer to my post #31"Phase issues...-1-", #33 "Phase issues...-2-", #37 "Phase issues...-3-" and #39 "Phase issues...-4-" .

Yes, I have, and I haven't found a single phase measurement in those 3 posts. I'm guessing you have the abilitly to alter the phase in that DSP crossover you're using (as every DSP crossover should have it), so how did you check that phase is adjusted correctly around XO points?

 

[Deleted member 2944]

Have you already read my post series on "Phase Issues"?
Please refer to my post #31"Phase issues...-1-", #33 "Phase issues...-2-", #37 "Phase issues...-3-" and #39 "Phase issues...-4-" .

EKIO doesn't even have the ability to adjust phase, independently. You're aware of that, yes????

Dave.

 

[QMuse]

EKIO doesn't even have the ability to adjust phase, independently. You're aware of that, yes????

Dave.

Uh, what kind of software crossover is that if it can't adjsut the phase?!?

 

[Kal Rubinson]

JRiver, as well as Roon, can up-sample up to 8xDSD "on the fly"; down sampling also.

Of course, it can do it on the fly but that doesn't mean that is direct and not through a PCM intermediary. Please check with JRiver before presuming otherwise. In general, knowledgable JRiver users upsample DSF to PCM but skip the conversion back to DSF as unnecessary.

Furthermore, I already have many 1xDSD, 2xDSD and 4xDSD files (about 200 tracks out of my 20,000 track library), and this is one of the main reasons that I will preserve the OPPO SONICA DAC (OK for 4xDSD native)+single amp+LC Netwark system, even after completion of my current multi-channel system where software crossover EKIO handles in 192 kHz 24 bit.. Please refer to my post #1, and look carefully at my system diagram there.
Also you would please look at my audio sampler tracks consists of 44.1 kHz PCM, 1xDSD, and 2xDSD files.

OK but not relevant to my comment.

 

[Kal Rubinson]

I was joking.

Good. I didn't pay attention to your comment not being from the OP, dualazmak, who seems not to be aware of the issue.

 

[dualazmak]

Yes, I have, and I haven't found a single phase measurement in those 3 posts. I'm guessing you have the abilitly to alter the phase in that DSP crossover you're using (as every DSP crossover should have it), so how did you check that phase is adjusted correctly around XO points?

Mainly by my ears and brain while listening to the tracks of "Super Audio Check CD", in comparison with my standard reference non-EKIO sound in exactly the same device setting,,, even though I can use REW and measurement microphne ECM8000.

I usually check the sound in overall, and also in each of the crossovered stereo 5-way channels. At lease for me, if the overall sound quality including "the phase features" is same as, or better than, my non-EKIO reference sound, it should be quite OK. This is always my approach and policy throughout in this project.

I believe my ears+brain and my current reference system including SPs and LC-network are very much sensitive enough for these, and most importantly I can enjoy listening to music, not simply to the sound, with my current system, and my multi-channel project should give somewhat better sound quality for my ears+brain and heart, especially by eliminating/avoiding the LC-network from the SP lines.

I am not an objective measurement addict, and I am not much knowledgable of technical details (and scientific bases?). After trying a lot with REW-Wavelet and other measurements, I now fell that these objective tools and methods would give nice complementary data for me, but at the end of the day, I should trust my ears+brain to listen to music, not to sound.

I like the Halfway Tree's comment and the Linkwitz's words he refered;
But no matter how good the measurements are, at the end of the day our ears are the final arbiter as to whether we like the sound, and I do. As Siegfried Linkwitz used to say 'What is important to the eye is not necessarily important to the ear...,'

 

[dualazmak]

Good. I didn't pay attention to your comment not being from the OP, dualazmak, who seems not to be aware of the issue.

Kal, I understand what you are discussing.

I usually listen each of the music tracks, however, in its native sampling rate; I rip CDs in non-compressed 44.1 kHz AIFF, I digitize my old LPs in 196 kHz AIFF, download DSF music in its available DSD format, i.e. 1xDSD, 2xDSD, and 4xDSD. (I like these formats for the flexible and unlimited tag info...)

Of course I know that "theoretically" up sampling of low bit PCM music into DSD format would give no SQ improvement, but sometimes I can feel slightly better SQ, including better S/N, by on-the-fly up sampling; this issue has been discussed a lot in this ASR Forum, I believe. I do not care how JRiver or Roon is doing the up/down sampling internally....

I my case, I often keep JRiver to "always 2xDSD upsampling mode" only because some of my play lists are the mixture of AIFF and 2xDSD, and I find the all tracks sound nice and OK for my ears+brain in 2xDSD.

After the completion of my project here, maybe my daily listening system would be the "Multi-Channel, Multi-Amplifier Audio System Using Software Crossover and Multichannel-DAC" where the signal processing is 192 kHz 24 bit. This should be quite OK, as long as the overall SQ is somewhat better than my current non-EKIO reference sound. That's all....

By the way, in the future, I believe that I eventually still would like to listen to 2xDSD or 4xDSD music in native format, e.g. Mao Fujita's Chopin Album in DSF 11.2 MHz, then I may easily go back to the preserved single-DAC (SONICA DAC capable of 11.2 MHz DSF)+Single-Amp+LC network system by just changing the SP cable connections at the SP cable terminal boards.

I assume I am not responding to your inquiry directly and correctly, but your kind attention and sympathy will be appreciated.

 

[dualazmak]

Why?

Le me add also;

just to test and confirm DAC8PRO properly works in 2xDSD (DSD128)!

 

[QMuse]

Mainly by my ears and brain while listening to the tracks of "Super Audio Check CD", in comparison with my standard reference non-EKIO sound in exactly the same device setting,,, even though I can use REW and measurement microphne ECM8000.

I'm sorry to dissapoint you but it has been proven scientifically that even the frequency response cannot be adjsuted by ears and phase much less so.

Phase and delay are basically the same thing, for any given frequency one can be coverted to the other. I suggest you measure phase response from app 1.5m distance, use some FDW smoothing (say 10 cycles, or less if you will be getting phase roll-overs) and then you'll see how it behaves. Do you have capability to adjust delay below 1ms?

 

[dualazmak]

I'm sorry to dissapoint you but it has been proven scientifically that even the frequency response cannot be adjsuted by ears and phase much less so.

Phase and delay are basically the same thing, for any given frequency one can be coverted to the other. I suggest you measure phase response from app 1.5m distance, use some FDW smoothing (say 10 cycles, or less if you will be getting phase roll-overs) and then you'll see how it behaves. Do you have capability to adjust delay below 1ms?

Hello QMuse,

I am not disappointed at all, no problem... I understand your point, but let me move forward step-by-step in this project .

Yes, I and EKIO has capability of adjusting delay below 1ms, i.e. 0.1ms step. I decided not to use any delay setting, however, in EKIO configurations mainly because of the possible phase complication given by the delay.

You would please carefully look at my SP alignments shown in post #27; the WO (woofer), SQ (squarker) and TW (tweeter) are well rigidly arranged by YAMAHA in the cabinet of NS-1000 (just magical alignment), and the SW (sub-woofer) YAHAMA YST-SW1000 can be flexibly moved back and forth within 40 cm range, if needed. I do believe such physical movement phase tuning for SW should be much better than the software phase tuning.

Consequently, in my "first to try" EKIO configuration of 12db/Oct crossover for stereo 5-way, I will use no delay in filter, no group delay nor channel gain. The relative gain needed can be controlled by DAC8PRO's nice 8 gain controllers after the DA processing.)

Since all the EQ filters are conventional LR 12dB/Oct, I checked phase "invert" only in SL (15 - 50Hz) and MD (600 - 6,000 Hz) regions. Please remember (as written in post #33) that my SW has phase inversion switch (can be controlled by a remote controller), and TW as well as ST are "inverted" at their SP unit terminals.

My current stance, and possible reply to your inquiry, has been already written in my post #39, where I wrote;

As for the fine tuning of possible slight “delay” of the super-low sound by SWs (sub-woofers), I can easily move the SWs (heavy stuffs, though) back and forth within max. 40 cm range, and I believe such “physical 3D tuning” of SWs should be much better than software crossover’s delay setting which brings phase rotation complication to the total sound.
I know that several people are suggesting me to use software "RePhase", but I believe I should not go into that path at this start-up stage of my project. In this type of "audio research", we should move up step-by-step with modifying single or minimum numbers of factors (parameters, elements) in each of the steps...
To finish my post series on “Phase Issues…”, let me go back to the content of my post #31;
I wrote there;
At least in my current project, from "total sound quality" point of view, my solutions or philosophy on these phase issues are;
"The simpler, the better."
and
"The simplest, the best."

In addition, It would be quite obvious that I need to decide and have a multi-8-channel amplifier in my project before going into the path of fine phase tuning which you kindly suggested.

And, let me touch on this again here;
I like the Halfway Tree's comment and the Linkwitz's words he refered;
But no matter how good the measurements are, at the end of the day our ears are the final arbiter as to whether we like the sound, and I do. As Siegfried Linkwitz used to say 'What is important to the eye is not necessarily important to the ear...,'

 

[QMuse]

Hello QMuse,

Yes, I and EKIO has capability of adjusting delay below 1ms, i.e. 0.1ms step. I decided not to use any delay setting, however, in EKIO configurations mainly because of the possible phase complication given by the delay.

Fine, that is a good prerequisite for phase alignement. I really don't understand what do you mean with "phase complications given by the delay". As i already explained phase and delay are the same thing and th eonly complciations you can expect would be coming if your drivers are not time aligned at XO points.

You would please carefully look at my SP alignments shown in post #27; the WO (woofer), SQ (squarker) and TW (tweeter) are well rigidly arranged by YAMAHA in the cabinet of NS-1000 (just magical alignment), and the SW (sub-woofer) YAHAMA YST-SW1000 can be flexibly moved back and forth within 40 cm range, if needed. I do believe such physical movement phase tuning for SW should be much better that the software phase tuning.

While it is true that drivers are mounted flush on the vertical panel you should still realise their cones are not vertically aligned so there would be some delay differences between them.

Since all the EQ filters are conventional LR 12dB/Oct, I checked phase "invert" only in SL (15 - 50Hz) and MD (600 - 6,000 Hz) regions. Please remember (as wrote in post #33) that my SW has phase inversion switch (can be controlled by a remote controller), and TW as well as ST are "inverted" at their SP unit terminals.

I am not an expert in XO design but I really can't understand based on what have you decided to invert phase in some particular regions but not in others. In other words, how can you be sure you would get optimal summation between drivers in any combination of phase if you don't measure?

And, let me touch on this again here;
I like the Halfway Tree's comment and the Linkwitz's words he refered;
But no matter how good the measurements are, at the end of the day our ears are the final arbiter as to whether we like the sound, and I do. As Siegfried Linkwitz used to say 'What is important to the eye is not necessarily important to the ear...,'

While that is true that quote by Linkwitz shouldn't in any way be used as an excuse NOT to measure. Trust me, Linkwitz did a lot of measurements when designing his speakers.

 

[dualazmak]

Hello QMuse,

http://linea-research.co.uk/wp-content/uploads/LR Download Assets/Tech Docs/CrossoverFilters White Paper -C.pdf

and it is well known in many text books that with 12 dB/Oct BW or LR filter, the phase rotation at the cross-point is 180 degree. So if you use series of 12dB/Oct filerts, e.g. series of 12dB/Oct filters to divide the 15 Hz - 25 kHz into five regions (including 15 Hz low-cut), the theoretical phase status in each of the five regions, in general, are;

(SL) reverse - (LO) normal - (MD) reverse - (HI) normal - (SH) reverse

In the User's Manual of the expensive ACCUPHASE DF-65 Digital Crossover, sorry I cannot find the English Manual now, the basics of these phase rotation and filter type/slope are well described. You may easily find the similar description in many places.....please google it.

This is the very common reason for my "invert" check at SL and MD regions. With these sets in the EKIO configuration, I can achieve exactly the same phase (3D) perspectives throughout 15 Hz - 20 kHz just like without digital crossover EKIO, my reference sound.

YAMAHA intentionally inverted to TW, and I can easily understand it by listening to only the TW stereo sound using the phase check track of the "Super Audio Check CD"; YAMAHA selected the inverted connection through their very careful and intensive listening and measurement trials, they said. I intensively tested and measured FQ response curve and phase check for ST in normal or inverted connection, and finally decided to invert to ST...

I contacted several times with YAMAHA people, some of them already retired, for the reasons of their choice of 12 dB/Oct LC-network in NS-1000 as well as for the parts (capacitors and coils) selections, and their common reply was; "just mainly for the simplicity of the possible phase rotation issues and rather smooth-flat Fq responses, and of course also for the costs of LC parts." After my intensive discussion with them, they also suggested and recommended the possible "first-to-try" EKIO crossover configurations, like the one I am showing you.

Yes, there are considerable overlap in sound of WO, SQ and TW with 12 dB/Oct filters, but these are one of the very nice "characteristics" of still amazingly wonderful NS-1000, which I would like to preserve in my multi-channel system. I just trust YAMAHA for the WO, SQ, TW 3D alignment, and have no intention to put relative delay between them, at least at this timing and stage of my project.

If you use steeper slope filters, like 24, 36, 48, 96 dB/Oct, and/or mixture of these, then the phase rotation becomes much complicated. You may find many text books on these, as you may already aware.

BTW, within sub-woofer YAMAHA YST-SW1000, on the other hand, the built-in high-cut filter is rather heavy duty 24 dB/Oct type; it should be OK since YST-SW1000 has phase inversion switch enables "normal" or "reverse" depending on the room acoustics and/or the distance between the listening position and SW.

Yes, I trust you that Linkwitz did a lot of measurements, like YAMAHA people did. I am not, however, an expert and knowledgeable guy for such measurements, and at least in my current project at present, I feel it would be well enough to compare with my reference sound which is still very nice.

Maybe, after fully establish my system with possible multi-8-channel amplifier (still not yet decided!), I will get back to you on phase and delay issues for your further suggestions, if needed. I will of course also "measure" the established system; FQ response, REW-Wavelet, phase rotations, etc.