Manual room correction for stereo - best practice ?

[Baku]

Hi Everyone !

I need an advice for the my stereo setup.
I am using a 7.2.4 system that is used for HC and music in a dedicated room. Rough dimentions : 520 cm x 320 cm. Ceiling 260 cm.

The stereo path of my system :
DAC: Denafrips ARES 12TH
Pre/Pro: Emotiva XMC-2
Power: Emotiva XPA-2
Speakers : KEF R3
Subwoofers : 2 x SVS SB-2000 Pro combined with MiniDSP 2x4HD
HTPC with local lossless collection streamed to the DAC exclusively through JRiver with USB connection.

In the past I was using the automatic room correction for both the music and the HC with some tweaks here and there. Audyssey (Denon), ARC (Anthem) and Dirac since Emotiva XMC-1.

Recently I have added the second sub with minidsp and some pressure acoustic treatment.
I understand that nowadays with the available tools (hardware and knowledge) one may achieve pretty good results manually measuring and building the RC filters.
Following the outstanding guides of Serkan ( OCA ) I would like to try to make a best possible digital filters correction manually instead of the Dirac for the music listening.
I have some knowledge and experience with REW over the years and the calibrated Umik-1 mic I have purchased in the past from the Cross Spectrum.
The MLP for the music is a static single position ( focused ).
I can of course make my measurements, time align and create the IIR filers with REW and upload it to the XMC-2. This part is what I am familiar with. And maybe I will achive even better customized resule in FEQ that Dirac, I am convinced that Dirac will do a better job treating the phase.
The OCA guides include several ways to create both the IIR and the FIR filters.
The FIR filters are a learning curve for me. I am trying to get my head over the Virtual Bass Array, Phase inversion, Convolution filters among other interesting phase related tools.

Since the guides are not structured between each other, please advice what would be the best practice for me to go with in order to combine the FIR with IIR filters ?

1. Should I measure and EQ with IIR filters first applying in XMC-2 ( full range - subs + L / subs + R ) and afterwards measure & create the convolution filter for the JRiver ? - probably not the optimal order.

2. Should I first create the convolution filter for full range - subs + L / subs + R, apply it in JRiver and then run the measurements with the pink noise/sweeps files generated by REW through the JRiver for the IIR filters EQ in order to apply them into the XMC-2 ? - probably better result.

3. Or should I split the FIR and IIR filters between the subs and the speakers ?

In other words - first create the FIR filters for each of the subwoofers and apply in the minidsp ? Then create the convolution filter for the speakers response only (no subs) for JRiver ? Then IIR EQ for the subs only to apply in minidsp and different IIR EQ for the L and R speakers only, applied in the XMC-2 DSP ? With both EQed responses crossed properly according to the Harman curve ?

As you may understand I am a bit lost. As a matter of fact I love what Dirac does to the room correction, but I do not like the black box implementation with no details of the changes made besides the impulse delay. Dirac treats the phase pretty nice though. Yet I want to try the manual path for thee stereo I never did before for the enthusiasm or maybe for even a better result than Dirac...

What do you think ?

 

[RayDunzl]

The miniDSP 2x4HD doesn't have enough taps to provide much control of bass frequencies using FIR filters, does it?

1024 per channel is the count, if I remember correctly.

(Or maybe it was just me having problems)

 

[Baku]

Yeah. miniDSP 2x4 is quite limited in this area. Might be enough though. I have not tried. 2048 per output for 2 channels...

 

[ZolaIII]

Make sure you make good decoupling of speakers, sub's from floor use silicone pads/bolsters under them.
Plug the R3 port's, do a white noise calibration to 83 dB mono at listening position. Enable loudness, move them over to determine best placement doing measurements on very loud levels (76~80 dB am @ 700~1000 Hz with loudness enabled and not full 86 dB to keep the room spirits there but in order). Take your time and do it more than once. Add sub's, put them to the left and right speaker and to do left and right chenel (one each) and same as speakers. Crossover them at 100 second order 12 dB slope like this:

Woofer crossover & offset

( - 100 Hz, -12 dB/oct, LR2 crossover without phase inversion) again at the white noise (mono 83 dB calibration point without loudness normalisation).
Show us the measurements at this point to determine manual by hand PEQ's.
Afterwards you will do RIF filter REW generation and export it to JRiver which has good convolver capabilities (I still call them convolution kernels).

Convolution - wiki.jriver.com

wiki.jriver.com

After which you are pretty much done expect some personal touches (as female uper overtones for example) but that is up to you and your taste.

 

[Baku]

Zolalll, thank for the good pointers. The speakers placement is done a while ago. R3 are on the good old filled Atacama stands with spikes with the silicone pads under the speakers. I am planning to plug the ports and to try the 90/100/110 XO once I progress with the alignment issue. ATM the speakers are configured as large with no XO therefore the plugs are off.
The subs are located in opposite side of the room. One is in front on the plane of the mains - about 1/4 room width, and the second is about middle back wall. Considering the room response this gives the best result on the MLP area.
The subs are aligned between each other in order to maximize the frequency response with 16ms delay for the back wall sub in the miniDSP_as simulated inside the REW with the alignment tool.

For a couple of days now I was trying to align the combined subwoofers response with the mains - but can't figure out what is it exactly I am aiming for and how to achieve it. This and this are very helpful yet I feel stuck. I got used that Dirac manages the phase issues and today after about 20 hours of reading the theoretical papers, posts here and there, watching youtube, then trying and getting lost again again and again I was thinking to forget about the manual path, to kick in Dirac live and get back to the music enjoyment I am missing... Yet I am enjoying the learning curve and maybe with some help I'll progress sooner.

I understand that the IR is not the tool for the bass alignment. Phase alignment is the way to go. But how exactly ? My left and right speakers phase is not equal in the XO region. I beleive that the reason for that is the difference in the side walls in thee room' Onee side is openings and the door, the other has windows with curtains.
I saw many suggestions from simply using the REW time alignment tool to more complex ones but nothing I tried brought the result to the the aligned step/phase/ir. Focusing on one taking the result far from another and vice versa. Remeasuring with the delays applied after the REW simulation brings wrong results in the time domain. Maybe this does not work for my case because I have aligned 2 subs with a quite big delay between each other and now trying to align 1 virtual sub makes all the calculation predictions to collapse. But on the other hand what is the difference in terms of the response and the time domain ? The phase is the phase is the phase. Maybe I am mistaken...

Here are my measurements - made some variety. Maybe is is as simple as lack of experience and the perfectionism I should put aside. Is there a right way to do it anyway ?

edit - measurements screenshots attached - 1/48 smoothing

 

[ZolaIII]

@Baku lol make screenshots of the REW measurements don't post native files of them.
Thing is how you improve time domain by putting corks in R3 port's you lower delay of them as well. One step at a time! Room placement pictures and diagram of the room would be appreciated. Let's do placement and basic stereo sub's placement proper (whole positioning as much as possible). We will worry about ISO 3382-1 and phase alignment later to the extent it's possible (without renovation, moving or doing a 20+ inch bass traps which are the only thing that will actually work)

 

[Baku]

@ZolaIII , I have edited the post with the screenshots.
Current state room picture is attached taken from near the back sub placement.
There is not much freedom in the room as regarding the speakers placement.
I tried of course to place the second sub MLP equidistant - near the left speaker, but the combined response is more SPL almost similar to the single right sub, so I koved it back with the 16ms delay to fill the deeps as you may see on the measurements.
That's the way I am trying to get them combined aligned with the mains...

 

[ZolaIII]

I assume that's a open doorway on the right side and responsible for the deep in 33 to 60 Hz region which is least pronounced with back sub.
I assume you can't do anything regarding open doorway.

 

[ZolaIII]

Otherwise it's a nice little room.
Can you lower bass traps or they are wall fixed? Would like to see sub's alone and second one placed similar to front one to opposite speaker, both closer to the wall behind them (that's why I ask about lowering the bass traps). Then we will try to have them over the level of main speakers in order to get headroom to address peaks and fill the hole. And speakers won't contribute when crossed and with plugs in port's and 100 to 200 Hz area will look better.
Edit: actually move right speaker little more (5~10 cm) from side wall.

 

[Baku]

I assume that's a open doorway on the right side and responsible for the deep in 33 to 60 Hz region which is least pronounced with back sub.
I assume you can't do anything regarding open doorway.

There is a doorway in the right and with the wall nearby it influences the response of the right speaker. However, the near 30Hz peak is a front-back wall room mod doubling later on - mid room and so on.

Otherwise it's a nice little room.
Can you lower bass traps or they are wall fixed? Would like to see sub's alone and second one placed similar to front one to opposite speaker, both closer to the wall behind them (that's why I ask about lowering the bass traps). Then we will try to have them over the level of main speakers in order to get headroom to address peaks and fill the hole. And speakers won't contribute when crossed and with plugs in port's and 100 to 200 Hz area will look better.

Thanks !
Those ( vicoustic wavewood ) traps are weak for the low bass. As you mentioned, for this type of a heavy lifting much deeper traps are needed, but this is not my intention. They are wall fixed.
Placing both subs on the front wall will force to push the subs gain to get the flat zone higher in order to cut the peaks after with fir/iir. Tried it, figured out the back wall sub makes much more sense with this room.
With the attached graphs I can cut the peaks in a low base area pretty nice to get the harman target curve response with the mains. But I can't get to this chapter, since I do not know how to get the subs aligned properly with the mains beforehand )

 

[ZolaIII]

Back wall sub shows that it makes more sense in readings but I want you to try stereo bass with crossovers at 100 Hz and R3 port's closed. You will need more gain from sub's in the first place to narrow deaps, cutting peaks is easier. That way you will cut peaks more but feel up the hole better. Once again you will worry about phase alignment and ISO 3382-1 later on to possible extent. You start by making less problems to start with with full closed enclosure setup (closing R3 port's and cutting them so that they don't contribute additional to problematic area).

https://www.audiosciencereview.com/forum/index.php?attachments/04-png.266014/

 

[Baku]

Maybe I am missing something.
Here are the screenshots of the integrated Front+Back subs with the Mains vs the Front only with the mains in frequency domain - 1/12 smoothing. Not the proper way, since the speakers ports are open and it the the only single sub in the front, but as I have mentioned adding second sub on the front plane adds some juice but the frequency response of both of them combined in the front is similar to the single front. Therefore you get the idea.
Moreover, I have tried the front subs for a while, and using them front + back ( both ways managed by Dirac Live ) . Comparing the representation in the room - I like the front + back more. Nevertheless on my previous pre/pro - XMC-1 there was a way to configure 2 subs as stereo - the left speaker crossed data to the left sub and the right to the right. It was nice to experiment with. The XMC-2 has only DUAL MONO (following the Dolby standarts) configuration for 2 subs, so the benefits of the actual front equidistant placement is somehow unclear to me in this case. Is it necessary ? May we suceed with the the next step with the current front/back placement or is it some kind of acoustics taboo that I am unfamiliar with ?
I am planning to use the R3 plugged after the XO is applied - they are open now in order to run full range sweeps since I believe it is the way to get a time domain integration simulation. But the proper time domain is even more unclear to my mind at this point... Assistance is greatly appreciated.

 

[ZolaIII]

Well then cross them @80 Hz.
Seams your old processor preamp whose miles ahead the new one.

Emotiva XMC-2 Review (AV Processor)

This is a review and detailed measurements of the Emotiva XMC-2 Audio/Video Processor (AVP). It is on kind loan from a member and costs US $3,000. The XMC-2 is surprisingly light for a high-end processor which my back appreciated as I carried it around: Ironically, the remote is made out of...

www.audiosciencereview.com

 

[ZolaIII]

Can you explain me the signal path? What's the use of either Emotiva XMC-2 or that R-2R DAC (God forbid)? Especially if you intend to use JRiver which has equal loudness normalisation and 64 bit volume controls?

 

[Baku]

Well then cross them @80 Hz.
Seams your old processor preamp whose miles ahead the new one.

Emotiva XMC-2 Review (AV Processor)

This is a review and detailed measurements of the Emotiva XMC-2 Audio/Video Processor (AVP). It is on kind loan from a member and costs US $3,000. The XMC-2 is surprisingly light for a high-end processor which my back appreciated as I carried it around: Ironically, the remote is made out of...

www.audiosciencereview.com

I may try 80 as well. Yet 40-60 deep is too big with the front location of 2 subwoofers. As for the review, it is still controversial topic. It would be nice if Amir could test another unit, but it is what it is. The XMC-1 was a great piece of hardware. XMC-2 is pretty good in many aspects. Buggy though.

Can you explain me the signal path? What's the use of either Emotiva XMC-2 or that R-2R DAC (God forbid)? Especially if you intend to use JRiver which has equal loudness normalisation and 64 bit volume controls?

The XMC-2 as you understand manages both the stereo as a pre/pro and the HC for 7.2.4 setup.
The stereo path is:
PC with Jriver -> usb to R2R ( suites my taste ) -> balanced to XMC-2 as a pre Amp with bass management -> bass goes to miniDSP with mono single unbalanced connection (both subwoofers connected to miniDSP) - stereo balanced to XPA-2 amp with R3 connected.
Since XMC-2 has several slots for speakers setup, this way I manage the easiest transition between music and HC when needed with a push of a button. Considering every setup is configured properly.

 

[ZolaIII]

I may try 80 as well. Yet 40-60 deep is too big with the front location of 2 subwoofers. As for the review, it is still controversial topic. It would be nice if Amir could test another unit, but it is what it is. The XMC-1 was a great piece of hardware. XMC-2 is pretty good in many aspects. Buggy though.

The XMC-2 as you understand manages both the stereo as a pre and the HC for 7.2.4 setup.
The stereo path is:
PC with Jriver -> usb to R2R ( suites my taste ) -> balanced to XMC-2 as a pre Amp with bass management -> bass goes to miniDSP with mono single unbalanced connection (both subwoofers connected to miniDSP) - stereo balanced to XPA-2 amp with R3 connected.
Since XMC-2 has several slots for speakers setup, this way I manage the easiest transition between music and HC when needed with a push of a button. Considering every setup is configured properly.

And you expect to get either good THD+N or good timing with multiple DAC's with different clock's and multiple ADC-DSP-DAC conversions? Every analog interconnect costs cuple dB and every conversion adds cuple ms.
Good luck with that, I am out.

 

[Baku]

Good THD+N/good timing comparing to what ? Yes I am trying to do my best.
ATM my question is what is the best effort practice considering the described path to get the SUBs aligned to the mains.

 

[ZolaIII]

Simplify the chain, use only one DAC the Mini DSP 2x4HD fead both power amplifier and sub's from it and even so for basic filters (high/low pass) and use JRiver for everything else (including as preamp and with loudness normalisation).

 

[Baku]

If my goal was to get a proper stereo only setup, my approach would be close to your suggestion indeed. However the purpose of my setup is to be flexible for both the HC and stereo with the best representation I can possibly achieve.
So back to the topic, what are my current best effort options considering the described path to get the SUBs aligned to the mains manually in time domain.

 

[Sokel]

A sane good use indicator is to have a look at the distortion tab of REW,sometimes it reveils issues that other tabs don't.