• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

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

OP
dualazmak

dualazmak

Major Contributor
Forum Donor
Joined
Feb 29, 2020
Messages
2,820
Likes
2,950
Location
Ichihara City, Chiba Prefecture, Japan
Hello friends,

As for the UHF noises which we are intensively discussing here now...

Just for your reference,,, (I shared this also here.)

If you have enough relaxed time, you would please watch all of the four amirm's YouTube clips in his threads;
https://www.audiosciencereview.com/...-high-res-music-antonio-forcione-video.23083/

https://www.audiosciencereview.com/...-live-from-minster-from-the-lake-poets.23095/

https://www.audiosciencereview.com/...pcm-vs-mqa-vs-cd-2l-sampler-comparison.23172/

https://www.audiosciencereview.com/...son-linn-records-free-high-res-samples.23366/

Then, you would please read carefully my above posts here to here.

The issue has been also discussed around 2015 in Japan, like in this page even though in Japanese;
https://sandalaudio.blogspot.com/2015/09/blog-post_17.html
I hope your web browser would properly translate it into English.
Where, "Niserezo (偽レゾ、ニセレゾ)" means "sham HiRes".

The article is really nice and delivers almost the same messages as amirm just gave to us.
 
Last edited:

gene_stl

Addicted to Fun and Learning
Forum Donor
Joined
Jun 14, 2019
Messages
867
Likes
1,198
Location
St.Louis , Missouri , U.S.A.
The Japanese article is quite interesting. I guess if we have concernes about our tweeters, tweeter amps and ears we shall have to examine our sources if DSD or very hi resolution formats are involved. Perhaps we have reached to point of too much of a good thing.
 
OP
dualazmak

dualazmak

Major Contributor
Forum Donor
Joined
Feb 29, 2020
Messages
2,820
Likes
2,950
Location
Ichihara City, Chiba Prefecture, Japan
The Japanese article is quite interesting. I guess if we have concernes about our tweeters, tweeter amps and ears we shall have to examine our sources if DSD or very hi resolution formats are involved. Perhaps we have reached to point of too much of a good thing.

Yes, I fully agree with you.

Consequently, I would like to always use the high cut digital filters at ca. 22 kHz in EKIO's configuration.

I am also discussing with LUPISOFT of EKIO for the possibilities of setting-saving the cut-off Fq at anywhere between 20 - 96 kHz since I always run EKIO in 192 kHz operation. If this would become possible, I would like to set the cut-off point around 23 kHz to 35 kHz, based on my knowledge of 22 kHz - 50 kHz music sound (not noises) cannot be heard with simple sine-wave-tone tests, but when mixed with high quality music sound of 10 Hz - 20 kHz, the 22 kHz - 50 kHz sound may have unproved "audible preferable effects".
 
Last edited:
OP
dualazmak

dualazmak

Major Contributor
Forum Donor
Joined
Feb 29, 2020
Messages
2,820
Likes
2,950
Location
Ichihara City, Chiba Prefecture, Japan
Hello friends,

Early this morning, Guillaume of LUPISOFT (developing and distributing EKIO) kindly responded to me informing he quickly fixed the issue (max. Fc=22.049 kHz even in 192 kHz sampling operation) and he updated EKIO to version 1.0.7.3 which corrects the problem.

This means now I can prepare-save-reload EKIO configurations with High-Cut (Low-Pass) filters of Fc at anywhere up to 96 kHz when EKIO is running in sampling rate of 192 kHz ; very nice, since I always operate EKIO in 192 kHz 24 bit given by JRiver MC (or Roon).

Today, therefore, I prepared EKIO configurations with LR High-Cut (Low-Pass) filters at 25.000 kHz, with slopes of -12 dB/Oct, -24 dB/Oct and -48dB/Oct.
WS001629.JPG


I still would like to allow 20 kHz to 25 kHz signals (hopefully music, but sometimes noises) to get into DAC8PRO and thereafter. And I definitely do not like the bunch of (rather high level) noises over 40 kHz would get into DAC8PRO.

We, myself and my wife, will continue our quasi-blinded ABCx comparative listening sessions with these configurations during the coming one week or two before fully deciding which one out of the three to be the standard EKIO configuration in my multichannel multi-driver multi-amplifier system.

You would please note that, if needed, I can easily and flexibly further fine-tune the relative levels for sub-woofers, woofers, Be-Mid-squawkers, Be-tweeters and horn super-tweeters, at amplifiers in my system.
 
OP
dualazmak

dualazmak

Major Contributor
Forum Donor
Joined
Feb 29, 2020
Messages
2,820
Likes
2,950
Location
Ichihara City, Chiba Prefecture, Japan
OP
dualazmak

dualazmak

Major Contributor
Forum Donor
Joined
Feb 29, 2020
Messages
2,820
Likes
2,950
Location
Ichihara City, Chiba Prefecture, Japan
New Standard EKIO Digital Crossover Configuration with High-Cut (Low-Pass) -48 dB/Oct Filters at 25.000 kHz

Hello friends,

After our intensive quasi-blinded ABCx listening sessions shared and discussed in my above posts #384 and #385, I decided to use this new configuration as our standard one in the multichannel multi-driver multi-amplifier system;
WS001643.JPG


As usual, the master volume is controlled by the volume of JRiver MC, and the rerative gains in downstream can be summarized like this;
WS001644.JPG


Of course, the I/O routing remains unchanged;
WS001649.JPG


The physical configurations also remain unchanged;
WS001637.JPG


In my post #311, I wrote;
Furthermore, throughout my amplifier exploration, I well experienced and learnt that we should never exclude high quality Hi-Fi "integrated amplifiers" to be possibly implemented in this type of multichannel multi-amplifier project. In my case, one of the important "must" conditions (specifications) is that the amplifier should be capable of XRL balanced input from OKTO DAC8PRO.

I daily feel and confirm that it is really nice that I can use the excellent volume controllers of A-S3000, E-460 and TA-A1ES for flexible fine tuning of relative gains, if needed depending on the genre and/or nature of the music tracks.

I usually fine tune the relative gains within plus-minus 3.0 dB range (0.2/0.5 dB step) by the volume dial of E-460 (-18 dB, driving Be-Squawkers) and TA-A1ES (-17 dB, driving Be-Tweeters and Super-Tweeters) keeping the A-S3000 (driving Woofers) at -18 dB (12 O'clock position), as shown in this diagram;
WS001650.JPG
 
Last edited:
OP
dualazmak

dualazmak

Major Contributor
Forum Donor
Joined
Feb 29, 2020
Messages
2,820
Likes
2,950
Location
Ichihara City, Chiba Prefecture, Japan
Hello friends,

This post is a follow-up of my post #344 sharing the possible, soon to be available, amirm's review on Trinnov Altitude.

Just today, amirm finally posted his review of Trinnov Altitude 16 which costs US $17,000! It is very interesting for me that, this time, amirm only objectively measured DAC capabilities of Trinnov Altitude 16, then he intentionally (or not?) abandoned further evaluation of Trinnov Altitude 16 as room correction EQ-DSP instrument.

The reported DAC performance of Trinnov Altitude 16 was really disappointing as you find in amirm's review; much inferior to OKTO DAC8PRO (which I am using) and DAC8 STEREO. Furthermore, Trinnov Altitude 16 is not able to play anything above 96 kHz.

As expected, many people pointed and emphasized that the main value (fit for US $17,000??) of Trinnov Altitude 16 would be its room EQ-DSP software and capabilities. But, in his post #39 there, Amirm clearly, simply and confidently replied to those people by saying "Stuff like xover I could test but i suspect there is nothing exciting there. If you mean room EQ, then measurements are not that useful. Listening tests are paramount." I fully agree with amirm's stance and confidences.

I find the above rather disappointing results and comments on Trrinov Altitude 16 is in good conformity with my worries shared here on Trrinov and Scott Borduin's nice response to my thoughts; he wrote there "For a stereo multi-driver application like you mention, there is probably little advantage in the Trinnov."

In any way, at least in my case, now I feel much relieved about my decision made two years ago that I would not go into the extraordinary expensive Trinnov-path for my multichannel multi-driver multi-amplifier project.
 
Last edited:
OP
dualazmak

dualazmak

Major Contributor
Forum Donor
Joined
Feb 29, 2020
Messages
2,820
Likes
2,950
Location
Ichihara City, Chiba Prefecture, Japan
Frequency Response Measurements by Using Flat Pinck Noise: Another Primitive but Reliable, Reproducible, Flexibly Re-Analyzable Method

First of all, abbreviations in this post;

SW: Sub-Woofers Yamaha YST-SW1000 (L & R)
WO: Woofers of Yamaha NS-1000 (L & R)
Be-SQ: Beryllium Mid-Range Squawkers of Yamaha NS-1000 (L & R)
Be-TW: Beryllium High-Range Tweeters of Yamaha NS-1000 (L & R)
ST: Metal Horn Super-Tweeter Fostex T925A (L & R)

Hello friends,

In my post #318 and #321, I shared Frequency (Fq) Responses (relative SPLs) of my multichannel multi-driver multi-amplifier system measured by REW's quick sine-sweep method and graphically represented after its "psychoacoustic smoothing".

According to the manual of REW;
"Psychoacoustic smoothing uses 1/3 octave below 100Hz, 1/6 octave above 1 kHz and varies from 1/3 octave to 1/6 octave between 100 Hz and 1 kHz. It also applies more weighting to peaks by using a cubic mean (cube root of the average of the cubed values) to produce a plot that more closely corresponds to the perceived frequency response."
I feel that "psychoacoustic smoothing" seems to give the frequency response curve that best corresponds (matches) to my auditory sensation.

On the other hand, I also would like to establish "my own routine procedure/method" of Fq response measurements which should be reliable, reproducible as well as flexibly re-analyzable, and also should be well understandable regarding the internal processing and validity, as I will share in this post as follows.

Before going into the details, let me share again my present digital crossover configurations, master volume and relative gain setting;
WS001708.JPG


And,
WS001707.JPG


EKIO's flexible I/O routing with nice "Mute" and "Solo" buttons on each of the output panels;
WS001706.JPG


The whole physical configuration remains unchanged;
WS001705.JPG


OK, this time I used my second "Completely Silent PC" in the listening room for just recording of the flat pink noise using the omnidirectrional EC measurement microphone Behringer ECM8000;
WS00001007.JPG


I carefully set ECM8000 (together with CEntrance Micport Pro for phantom 48 V supply and USB 96 kHz digital out) at my head position in usual listening sessions;
WS001704.JPG


As for the high quality reliable pink noise, I use track-12 "Pink Noise 20 Hz - 20 kHz" of CBS/Sony's "Super Audio Check CD" 48 DG 3 (1983), which I shared in my post #26.

This paragraph was edited for correction:
While repeatedly playing this pink noise track, I continuously recorded the actual sound of the flat pink noise by using Audacity 3.0.1 into monaural 96 kHz WAV format utilizing EKIO's flexible mute/solo button on each of its output panels. (Of course I can do the recording by using Adobe Audition 3.0.1, but as of today I have not yet installed Adobe Audition in the specific PC; this is the only reason that I used Audacity 3.0.1 for recording.)

Since the amplifier Sony TA-A1ES drives both of Be-TW and ST, I paused the recording while disconnecting and re-connecting the SP line cables for Be-TW and ST so that I could record the sound from only each of them in the individual SP driver recording portions.

The master volume (in JRiver MC27) and other gain values as well as any other variable factors were strictly kept unchanged during the entire recording session.

After the continuous recording session, the recorded 96 kHz WAV file was transferred to the rather powerful Intel Xeon PC Workstation at my office upstairs for further processing and analyses.

Since ECM8000 can hear only up to 20 kHz in flat response, I first carefully converted the recorded 96 kHz WAV into 48 kHz AIFF, by the nice and reliable "Music Converter" of dBpoweramp 17.3, so that I can intensively analyze Fq response up to 24.0 kHz.

The recorded 48 kHz monaural track consists of 6 portions, ca. 30 sec each, as shown here;
WS001703.JPG


Then, I checked/analyzed the track first by MusicScope 2.1.0;
WS001702.JPG


As you can see, MusicScope is really nice for "at a glance" check and confirmation of the Fq response over 10 Hz - 20 kHz; I can see quite nice, almost flat, Fq response of the whole system (with all the SP drivers singing together) represented by the yellow dots spreading 10 Hz to 20 kHz.

You would please note that I kept the windows of rear connecting room behind the ECM8000 opened during the recording session for better room acoustics; this means that we have some amount of low Fq noises of comfortable May weather breeze from outside of my house.

Next, I loaded the sound track into Adobe Audition 3.01 for further intensive analyses;
WS001701.JPG


The Spectral Frequency Display is also very nice to see the system Fq response "at a glance" in one image.
Now, we can analyze the Fq response of each of the 6 parts in this single monaural track.

I first rather intensively analyzed the "all together" portion which is the sound given by all the SP drivers singing together.
In order to closely check the super-low Fq zone of 10 Hz - 100 Hz, we need to set the FFT Size 8192 for Fq Analysis;
WS001700.JPG


I could well confirm again that my SWs are working very nicely having sound peak at around 33 Hz; by just judging from this Fq curve, however, my volume (gain) setting in the L & R active SWs, L & R Yamaha YST-SW1000, would be a little too high, while it can be easily adjustable by IR remote any time while I sit on the sofa at listening position.

The 40 Hz - 600 Hz zone can be better seen simulating "psychoacoustics" by applying FFT Size of 2048;
WS001699.JPG


And, the 800 Hz - 20 kHz zone can be seen "psychoacousticly" by applying FFT Size of 256;
WS001698.JPG


Then I moved onto analyses of Fq response in each of the SP drivers; first for SWs only;
WS001697.JPG


As expected from the shape of "all-singing-together Fq curve", SWs have peak at 35 Hz; please note that, in EKIO's digital crossover, I set Low-Pass (High-Cut) at 50 Hz with -12 dB/Oct slope, and furthermore within SW Yamaha YST-SW1000, I set -24 dB/Oct filter at 55 Hz so that SWs would not widely overlap with WOs.

Next, I analyzed "WOs only" area by applying FFT Size 4096 to nicely see the Fq response in 30 Hz - 100 Hz zone;
WS001696.JPG


Thanks to the full elimination of LC network inductors and capacitors, and also thanks to the dedicated nice amplifier Yamaha A-S3000, now I can confirm the WOs work very nicely/efficiently as low as 40 Hz, and up to 600 Hz where I set the Low-Pass (High-Cut) filter of - 12dB/Oct in EKIO.

The FFT Size of 2048 would give better "psychoacoustic" Fq shape for WOs in 100 Hz to 600 Hz zone;
WS001695.JPG


Now, let's move on to my treasure mid-range Be-SQ with FFT Size 512;
WS001694.JPG


I could confirm again here that Be-SQs of Yamaha NS-1000 is still very efficiently responding to the dedicated amplifier Accuphase E-460 and also to the EKIO's digital crossover target curve.

The Be-TWs were also analyzed with FFT Size 512;
WS001693.JPG


As @mikessi kindly mentioned in his post #350, the Fq response of Be-TWs of NS-1000 is nice up to around 14 kHz, and beyond which it dropped as seen above.

I have been also well aware of these characteristics of Be-TWs of NS-1000, and, therefore, this was one of the main reasons and rationale for my decision on adding STs, Fostex T925A, in my system which was also analyzed here with FFT Size 512;
WS001692.JPG


You can see the slight drop over 18 kHz, but I assume over 18 kHz would be the limitation zone of the microphone ECM8000 and the small USB ADC CEntrance MicPort Pro attached to ECM8000. (So far, I could not find detailed specifications for CEntrance MicPort Pro).

In any way, over 17 kHz, the reasonable/affordable measurement microphone like ECM8000 would not be fully reliable, and I believe I need to have much more expensive measurement microphone and excellent "digital audio microphone interface" of guaranteed flat response up to at least 50 kHz.

I am rather reluctant, however, to have such expensive measurement gears, since my "audible capabilities" has already slightly declined over 10 kHz with my aging, even though I still have fairly better audibility than the "average audibility" of my age segment.

Well, we can very easily and flexibly measure any combination of the SP drivers, just like I analyzed Be-TWs + STs with FFT Size 512 as shown below;
WS001691.JPG



In summary and conclusion, I found that the rather primitive Fq response measurement method of "simple pink noise recording and analyses" shared here is really reliable, accurate, reproducible, flexible (with various FFT sizes) and re-analyzable in any way after having the recorded track/file.

Of course, I understand quite well that Fq response is just one of the various factors determining the total sound quality of our audio systems.
 
Last edited:
OP
dualazmak

dualazmak

Major Contributor
Forum Donor
Joined
Feb 29, 2020
Messages
2,820
Likes
2,950
Location
Ichihara City, Chiba Prefecture, Japan
MusicScope 2.1.0's default Spectral Frequency Display has nice "color gamma scale" to see the low level background noises.
Just for my curiosity and interest, I could identify the sources of major low-level background noises by using Adobe Audition 3.01 and my headphone;
WS001713.JPG
 
Last edited:
OP
dualazmak

dualazmak

Major Contributor
Forum Donor
Joined
Feb 29, 2020
Messages
2,820
Likes
2,950
Location
Ichihara City, Chiba Prefecture, Japan
Hello friends,

I further evaluated/validated the above rather primitive Fq response anaysis of "Cumulative Pink Noise Averaging".

In this post I share the results of Fq response measurement in digital level signals within the audio dedicated silent PC given by software crossover EKIO.
EKIO's crossover and I/O routing configuration remained unchanged;
WS001730.JPG


Please note that I set -6 dB gain in CH5 and CH6 (to drive highly efficient Be-SQs), and -4 dB gain in CH9 and CH10 (to dirve highly efficient STs).

EKIO feeds the crossover 10-channel digital signal into DAC8PRO via the DIYINHK ASIO driver; fortunately this DIYINHK ASIO driver is bidirectional so that Audacity or Adobe Audition can hear the looped back digital signal through the DIYINHK ASIO receiver routes.

This time, I used Adobe Audition 3.0.1 in the same PC to hear and record the EKIO's crossover digital signals (L plus R mixed channel into monaural track) in 192 kHz (my default bit rate), and then the recorded track was carefully converted into 48 kHz by Adobe Audition since the Pink Noise applied was 20 Hz - 20 kHz;
WS001729.JPG


From this recorded track, I could also easily extract each of the 8 portions into short tracks for quick analyses by MusicScope 2.1.0;
WS001728.JPG


Then I loaded the recorded track into Adobe Audition 3.0.1 for further detailed Fq response analyses;
WS001727.JPG


Since the recorded track is the digital level rather pure raw signal without room effects, I may use the FFT Size of 8192 all the way in Fq response analyses. Firstly, the Fq response of "all the channels singing together";
WS001726.JPG


The high-range Fq response can be seen like this;
WS001725.JPG


And the entire Fq response curves can be seen like this;
WS001724.JPG


In my post #321, I shared the similar Fq response measurements in digital level analyzed by REW's "rapid sine wave sweeping" and its psychoacoustic smoothing. According to the manual of REW;
"Psychoacoustic smoothing uses 1/3 octave below 100Hz, 1/6 octave above 1 kHz and varies from 1/3 octave to 1/6 octave between 100 Hz and 1 kHz. It also applies more weighting to peaks by using a cubic mean (cube root of the average of the cubed values) to produce a plot that more closely corresponds to the perceived frequency response."

On the other hand, my present primitive Fq analysis by "Cumulative Pink Noise Averaging" does not include such psychoacoustic smoothing and/or psychoacoustic attenuation.

It is of interest and important (I assume) to see and understand the differences in Fq response curves between the three methods, even though the results and differences were rather well expected/predicted;
WS001731.JPG


In any way, we should be always careful enough that Fq response analysis is based on statistical data manipulation by FFT, and hence the results are greatly depending on the total data size (total energy of samples), sine sweep or pink noise, FFT size (and/or other smoothing algorism), and the X-Y logarithmic (or linear) scales for curve representation.

Edit: A little bit of additions on REW's psychoacoustic smoothing, and typo corrections throughout the post.
 
Last edited:
OP
dualazmak

dualazmak

Major Contributor
Forum Donor
Joined
Feb 29, 2020
Messages
2,820
Likes
2,950
Location
Ichihara City, Chiba Prefecture, Japan
OP
dualazmak

dualazmak

Major Contributor
Forum Donor
Joined
Feb 29, 2020
Messages
2,820
Likes
2,950
Location
Ichihara City, Chiba Prefecture, Japan
Frequency Response Measurements by "Cumulative White Noise Averaging"

Hello friends,

It is well known that nicely prepared pink noise would be rather "gentle" for tweeters and super-tweeters for their burn-in and Fq response (relative SLP) measurements. Typical pink noise, however, has gradual elevation of response in dB scale from high towards low frequency, while typical white noise has actually flat response through 20 Hz - 20 kHz, as shown here analyzed by MusicScope 2.1.0;
WS001808.JPG


Last weekend, therefore, I tested digital level FQ response measurements by "Cumulative White Noise Averaging" since I would like to have similar/identical shapes of curves for EKIO's target crossover configurations with the measured digital level Fq response curves.

This time, I prepared a tentative EKIO configuration which gives essentially flat total target curve throughout 20 Hz to 20 kHz for confirmation and validation of the method of "Cumulative White Noise Averaging".
WS001807.JPG


EKIO feeds the crossover 10-channel digital signal into DAC8PRO via the DIYINHK ASIO driver; fortunately this DIYINHK ASIO driver is bidirectional so that Audacity or Adobe Audition can hear the looped back digital signal through the DIYINHK ASIO receiver routes.

I used Adobe Audition 3.0.1 in the same PC to hear and record the EKIO's crossover digital signals (L&R stereo) in 192 kHz 32 bit, and then the recorded track was carefully converted into 48 kHz L+R monaural track by Adobe Audition as the White Noise applied was 20 Hz - 20 kHz;
WS001806.JPG


And the whole track was first analyzed by MusicScope 2.1.0;
WS001805.JPG


From this recorded track, I could also easily extract each of the 8 portions into short tracks for quick analyses by MusicScope 2.1.0;
WS001804.JPG


Then I loaded the recorded track into Adobe Audition 3.0.1 for further detailed Fq response analyses;
WS001803.JPG


Since the recorded track is the digital level rather pure raw signal without room effects, I may use the FFT Size of 8192 all the way in Fq response analyses.

Firstly, the Fq response of "all the channels singing together" and the high-range Fq responses can be seen like this;
WS001802.JPG


And the entire Fq response curves can be seen like this;
WS001801.JPG


I could nicely confirm and validate this "Cumulative White Noise Averaging Method" gives almost identical "shapes" of the total and individual measured response curves in digital level to the EKIO's crossover target response curves.
WS001810.JPG



I am further planning to apply this method for the Fq response measurements in DAC8PRO's analog line output level, and also the amplifiers' SP output levels. For these purposes, I will soon get a reasonable (but reliable) ADC audio interface (analog RCA input and mic pre-amp with phantom power supply, both into ADC for 192 kHz 24 bit USB digital recording), and also a reasonable (but reliable) SP level to RCA line level High-to-Low converter.
 
Last edited:
OP
dualazmak

dualazmak

Major Contributor
Forum Donor
Joined
Feb 29, 2020
Messages
2,820
Likes
2,950
Location
Ichihara City, Chiba Prefecture, Japan
Where in the Multichannel Multi-Driver (Multi-Way) Multi-Amplifier System should I measure/check Frequency (Fq) Responses?

Hello friends,

About one year has passed since I introduced OKTO DAC8PRO in the system, and six months of burn-in period has passed since I decided to use new Yamaha AS-3000 and Sony TA-A1ES together with the existing Accuphase E460, as shared in my post #311. Also I could recently fully establish the crossover EKIO's new standard configuration as shared in my post #386.
WS001836.JPG


And I could confirm and validate my rather primitive but reliable "Cumulative White Noise Averaging" method for frequency response measurements as shared in my post #392.

I and my wife are now fully satisfying with the wonderful total sound quality of the system, and we really enjoy listening music with this completed system.

Just for my future references, therefore, now I would like to intensively check, confirm and record the Fq responses of the system under the present crossover configuration and relative gains, and I am carefully considering and planning about "Where in the multichannel multi-driver (multi-way) multi-amplifier system should I measure/check frequency (Fq) responses?". In this line, I hope and believe the below diagram is self-explanatory for your understandings;
WS001837.JPG


1. Digital level before DAC8PRO
Just like as I shared in my post #392 with EKIO's tentative flat crossover configuration, I will measure the digital level Fq responses within the PC by looping back the DIYINHK ASIO signal into Adobe Audition 3.0.1 for recording and analyses.

2. DAC8PRO's analog output level
I will soon get a reliable ADC audio interface (analog stereo line-level input and also mic preamp with 48 V phantom power supply) which can convert DAC8PRO's analog level output signals into 192 kHz digital to be recorded and analyzed by Adobe Audition 3.0.1 via dedicated ASIO driver.

3. Amplifiers' SP output level before protection capacitors
I will soon get a reliable SP-level to RCA line-level (High-to-Low) stereo converter which would feed the converted line-level signals into the new ADC audio interface for 192 kHz digital recording and analysis by Adobe Audition 3.0.1.

4. Amplifiers' SP output level after protection capacitors
I will apply the operation as in above 3.

5. Actual SP sound at listening position with measurement microphone (ECM8000)
As I will have the new reliable ADC audio interface with 48 V phantom power supply, I will record and analyze the sound by Adobe Audition 3.0.1 in lower noise floor than using the PC's motherboard integrated Realtek ADC.

I will also use MusicScope 2.1.0 for supplemental purposes together with Adobe Audiotion 3.0.1, just like as shared in my post #392.

Since I still use my DIY SP cabling board with the protection capacitors and fine-tuning resistors, I can perform the Fq response measurements of above 3. and 4. (as well as 1. and 2.) in complete silence by disconnecting the wires to the SP drivers. Please note that I will insert in parallel the SP-level to RCA line-level (High-to-Low) stereo converter before or after the protection capacitors, and I can use the 8 Ohm 100W dummy SP resistor on the board.
WS001838.JPG


You would please find the details and results of these Fq response measurements in my coming posts here within a month or two.
 
Last edited:
OP
dualazmak

dualazmak

Major Contributor
Forum Donor
Joined
Feb 29, 2020
Messages
2,820
Likes
2,950
Location
Ichihara City, Chiba Prefecture, Japan
Frequency (Fq) Responses in the Completed System Measured by using “Cumulative White Noise Averaging Method” under the Present Standard Crossover Configurations and Relative Gains:

Part-1: Fq Responses in EKIO’s Digital Output Level

Hello friends,

Yesterday evening, I did it as Part-1.

The present total system and configurations can be seen like this;
WS001841.JPG


This post shares the results in 1. Fq Responses in EKIO’s Digital Output Level;
WS001842.JPG


The recorded white noise, ca. 60 sec each, by digital loop back into Adobe Audtion 3.0.1 within the PC;
WS001843.JPG


Firstly, analyzed by MusicScope 2.1.0;
WS001844.JPG

And;
WS001845.JPG


Then loaded into Adobe Audition 3.0.1;
WS001846.JPG


The total response curve and the High Fq channels;
WS001847.JPG


The entire Fq responses;
WS001848.JPG


Consequently, the EKIO's target configurations are quite nicely reflected in the Fq responses in digital level;
WS001849.JPG


I will continue analyzing Fq responses by this method in 2, 3, 4 and 5 in this diagram using ADC audio interface, SP-level to RCA line-level (High-to-Low) converter (for 3 and 4), and measurment microphone (ECM8000) (for 5);
WS001837.JPG
 
Last edited:
OP
dualazmak

dualazmak

Major Contributor
Forum Donor
Joined
Feb 29, 2020
Messages
2,820
Likes
2,950
Location
Ichihara City, Chiba Prefecture, Japan
Frequency (Fq) Responses in the Completed System Measured by using “Cumulative White Noise Averaging Method” under the Present Standard Crossover Configurations and Relative Gains:

Part-2: Fq Responses in DAC8PRO’s Analog Output Level

Hello friends,

Yesterday evening, I did it, "Fq Responses in DAC8PRO’s Analog Output Level" , as Part-2.

The present total system and configurations, of course, remain unchanged as can be seen like this;
WS001957.JPG


This post shares the results in "2. Responses in DAC8PRO’s Analog Output Level";
WS001956.JPG


For this measurement and also for the coming posts on 3. through 5., the reasonable but reliable ADC audio interface TASCAM US-1x2HR arrived yesterday;
WS001955.JPG


TASCAM US-1x2HR is an entry class audio interface which TACAM released rather recently;
https://tascam.jp/us/product/us-1x2hr/top
https://tascam.jp/us/product/us-1x2hr/spec
and the spec is just nice enough for my current purpose of Fq response measurements in line-level audio signals.

You would please find a nice review article on US-1x2HR;
https://higherhz.com/tascam-us-1x2hr-review/

Since I do not like to use XLR(balanced)-to-RCA(unbalanced) adaptor or cable, I configure EKIO to feed all the channels to DAC8PRO's CH1 (L) and CH2 (R) which have nice analog unbalanced RCA line-level output for headphone on the front panel;
WS001960.JPG


The RCA unbalanced line-level stereo signals from DAC8PRO's CH1 and CH2 were given into USB audio interface TASCAM US-1x2HR's RCA stereo input ports (in back panel) for digital sound recording by Adobe Audition 3.0.1 via the dedicated ASIO driver.

The recorded white noise, ca. 60 sec each, by digital loop back into Adobe Audtion 3.0.1 within the PC;
WS001954.JPG


Firstly, analyzed by MusicScope 2.1.0;
WS001953.JPG

And,
WS001952.JPG


Then loaded into Adobe Audition 3.0.1;
WS001951.JPG


The total response curve and the High Fq channels;
WS001950.JPG


The entire Fq responses;
WS001949.JPG


As well expected, the EKIO's target configurations are quite nicely reflected in the Fq responses in OKTO DAC8PRO's analog output line-level;
WS001947.JPG



I will continue analyzing Fq responses by this method in 3, 4 and 5 in the below diagram using TASCAM US-1x2HR plus a SP-level to RCA line-level (High-to-Low) stereo converter (for 3 and 4), and measurement microphone (ECM8000) (for 5);
WS001946.JPG


As for the SP-level to RCA line-level (High-to-Low) stereo converter, I have already purchased and received fairly nice and reliable one which I am now carefully testing and validating for the coming Fq measurements in amplifiers' SP output level by using the common "Cumulative White Noise Averaging Method".
 
Last edited:

phoenixdogfan

Major Contributor
Forum Donor
Joined
Nov 6, 2018
Messages
3,294
Likes
5,072
Location
Nashville
Hi Dualazmak I'm enjoying this thread immensely. I plan to purchase Ekio probably next month and I have a couple of questions:

1) I see you are altering everything to 24/192. Is this necessary? I have an I3 machine and when I use 24/192 it just stalls. I also see under Ekio's general setting the sampling rate can be set anywhere from 44.1 to 384. I don't know if Ekio internally resamples everything to 24/192 and you merely want to avoid that resampling by feeding it 24/192 or if there's some other reason. If it's a reluctance to resample material, obviously that won't be the case, b/c very little if anything is natively recorded in 24/192 (and most of it the result of upsampled masters as well) so upsampling in the chain will be inevitable and it's just a question of where it occurs.

2) The system I'm bui;lding is much simpler than yours, it consists of 4 Kef LS 50s (2 Meta in fromt, 2 OGs in back) and a single (for the present) SVS SB 2000 powered sub, I have a Purifi Eigentact amp (exactly like the one Amir reviewed ) powering the front and a Behringer A500 powering the surrounds. I have an Octo 8 Pro on order (since Jan 6th) which should arrive any day now, and I have a very basic but quiet Dell I3 which functions as the server, running JRiver, Dirac Live 3 Multichannel Studio, Netflix, Kodi, and Qobuz.

I feed all the AV (netflix, Qobuz and Kodi) into JRiver via the WDM driver. I am planning to create two use cases: one will be a 2.1 system crossed over at 100-120 hz for music, and a 4.1 system on another zone in JRiver which decodes all the legacy surround formats.

The 2.1 use case seems kind of easy. I'll uses Jriver as a pass through with it set to 2.0 stereo and send everything to Ekio to do some parametric eq (to optimize the LS 50 Metas Listening Window FR), and then to a 24 db per octave LW High/Low pass filter. The two channel system will be measured and everything corrected with Dirac live with the final eq'd, crossed over, and Dirac'd output sent to the Octo.

The 4.1 use case seems to present me with more questions. Specifically, I'm wondering if I should just do this all in JRiver, or if Ekio can play a role.
I like Ekio b/c its interface makes a lot of sense to me in that it resembles the miniDSP 2 x 4 interface almost exactly. I also like that in the parametric EQ "Q" actually seems to mean "Q" and not "S". The LS 50 both Meta and original versions have some exceptionally well thought out and devleloped EQ by Maiky 76 which goes a long way toward optimizing their performance, and I want to take advantage of that work to take them to the next level. But I'm not at all sure how I could give myself an LFE channel on Ekio. The LFE seems to be created almost automatically on JRiver by just by going to Bass Management in Room Correction, selecting "Move Bass to subwoofer" and setting the channels to 5.1 in output format and setting the hi-lo pass in JRiver at 80 hz. Using JRiver for that use case will necessitate I translate all the Q values on the Parametic EQ filters but I do have the Wolfram software, and I can cut and paste the formula.

In any case I would like to know how to do an LFE channel in Ekio, and I also wonder how much latency all this will add. I've only read through the first 6 or so pages, so I don't know if anyone has discussed any of these issues, but I'm very interested in things like how to fight latency which seems an intermittent challlenge using these programs (maybe moving to an I7 10 gen INuc would remedy the latency?), and how exactly it would be best to use either JRiver and/or Ekio to set up a computer based surround sound front end.
 
Last edited:
OP
dualazmak

dualazmak

Major Contributor
Forum Donor
Joined
Feb 29, 2020
Messages
2,820
Likes
2,950
Location
Ichihara City, Chiba Prefecture, Japan
Hello phoenixdogfan,

Thank you for your continuing attention and interests on this thread.

Yes, so far, until recently, I have been sticking to 192 kHz operation of JRiver, EKIO and DAC8PRO. After my recent intensive studies and experiences on the "UHF noises (ultra-high frequency noises)" in some Hi-Res music files (my post #362 through #386), and my on-going very careful "Fq response measurements" at various levels of the audio system, however, now I believe 96 kHz operation, or even 48 kHz, would be just perfectly OK for really excellent Hi-Fi sound quality.

As you know well, 96 kHz stereo operation covers up to 48 kHz Fq sound, and 48 kHz stereo operation covers up to 24 kHz sound. As I decided and I am using - 48 dB/Oct low-pass (high-cut) filters at 25 kHz to cut-off the possible UHF noises, and also my listening capability over 10 kHz slightly declined depending on my age, I believe the current EKIO configuration with the -48 dB/Oct low-pass filters at 25 kHz should be quite nice and feasible.

Consequently, 96 kHz operation in JRiver, EKIO and DAC8PRO should be just OK for Fq coverage and also in terms of stable "handshakes" between the physical gears and software components involved.

From the beginning of this project, I have been also very much careful about the "latency, delay and synchronization issue" in the system. OKTO DAC8PRO is the perfect solution for synchronization of the all the 8 channels given by EKIO. I believe there should be no (audible?) latency/synchronization problems within EKIO's crossover software processing.

Guillaume of LUPISOFT kindly informed, as I shared in my post #140;
"EKIO uses IIR filters. The processing is done using a cascade of second order transposed direct form II biquad sections. Every calculation is done using 64 bit floating point numbers."

And I also confirmed that EKIO gives no audible post-ringing nor pre-ringing at all, as shared in my post #143, at least in my rather simple and straightforward utilization of EKIO for just crossover filter operations.

Furthermore, I do trust Yamaha NS-1000's rigid SP driver alignment, and I can physically adjust the positioning of my L and R sub-woofers (Yamaha YST SW1000) and super-tweeters (Fostex T925A), I do not set any delay value in my EKIO configurations.

Since I am still sticking to L and R stereo multi-way audio system, and I have little interest on 2.1, 4.1, 5.1 or higher AV system, I am not in the position to answer your inquiries on these. I agree with you, however, the "latency and synchronization issue" would be critical and important in establishing our audio system, especially if you would like to synchronize all the visual and audio signals in AV system.
 
Last edited:
OP
dualazmak

dualazmak

Major Contributor
Forum Donor
Joined
Feb 29, 2020
Messages
2,820
Likes
2,950
Location
Ichihara City, Chiba Prefecture, Japan
Hello phoenixdogfan,

Let me add one point.

I do not use crossover/EQ functions of JRiver, but use EKIO for all the crossover and relative gains configuration. This is simply because of that I do "love" the very much flexible and simple GUI operation of EKIO including the I/O setting as well as the "Mute" and "Solo" button on each of the output panel, and also the phase "invert" on-off check in each of the filters.
 

phoenixdogfan

Major Contributor
Forum Donor
Joined
Nov 6, 2018
Messages
3,294
Likes
5,072
Location
Nashville
Hello phoenixdogfan,

Let me add one point.

I do not use crossover/EQ functions of JRiver, but use EKIO for all the crossover and relative gains configuration. This is simply because of that I do "love" the very much flexible and simple GUI operation of EKIO including the I/O setting as well as the "Mute" and "Solo" button on each of the output panel, and also the phase "invert" on-off check in each of the filters.
At this point I'm leaning toward using JRiver/Ekio (Jriver for Dirac, Ekio for PEQ and Xover) for the 2.1 use case, and JRiver only for the 4.1 use case (Xover, multichannel codecs, PEQ, and Dirac Live) with each use case having its own JRiver Zone with all the appropriate settings.

At some point Dirac will put out its Bass Management module which includes Xovers and advanced bass correction, and at that point I'll use JRiver + Dirac Live 3 w BM for both use cases. Probably pick up at least one more sub as well.
 
Top Bottom