Problems with MSO & Dirac - need help

[Smithg97]

Really starting to pull my hair out with both MSO & Dirac Live when trying to EQ my system....maybe there's a simple explanation that i'm missing but looking for answers to see where i'm going wrong!

So my setup is;
* Source: WiiM Mini Network Streamer (PC when calibrating the setup)
* Pre: Minidsp Flex unbalanced
* Power: Audiolab 6000A
* Subs: 2x SVS PB1000 pro
* Mains: Klipsch RP-6000F
* Mic: Minidsp UMIK-1

I'm trying to make Dirac and MSO work but on both software programs it seems to be making the FR worse and by quite a lot vs what i achieved using REW and the minidsp 2x4 HD plugin.

I've tried a few different configs as follows (all FR pics use VAR smoothing);
1)
Measured subs individually at 3 positions (MLP, Left, Right) using REW
Imported measurements into MSO & followed the guide provided by Jeff Mery on YouTube
The optimised curve looked good so i exported the filters for each sub, imported into the minidsp PEQ filter options for each sub and it made barely any difference to FR (Pic below of before & after)

2)
Measured subs individually in REW then added small gains/delays in minidsp plugin to get favourable summing which worked
Set gains/delays to sum subs
Calculated PEQ filters in REW then imported to minidsp 2x4 HD plugin (Sub outputs)
Measured full setup to see if xover & 120Hz to 500Hz would benefit from additional PEQ (which it did)
Refined the PEQ filters again in REW and imported to minidsp (input PEQ)
Remeasured and the FR looks good & speakers sound good - so far this yields the best sound but completely circumvents MSO & Dirac which i thought would be beneficial rather than detrimental
Pic of FR (all measurements for this config were from 1 position; MLP);

3)
Measured again in REW, set sub gains/delays
Measured in Dirac using the 13-point config
Set my desired FR trying to keep the line as much in the raw FR as possible to limit cut/gain then exported to the minidsp
Remeasured in REW and proceeded to pull hair out (have kept the FR from config 2 above for context and to highlight diff to Dirac line in pink);

Is this problem I'm facing due to the fact that the MLP is being slightly sacrificed by MSO & Dirac to accommodate the other listening positions or am I missing some other important step in the process?

Before anyone states the obvious that if option 2 looks & sounds good then why not just settle....I have a sunk cost (Dirac) that I want to get value out of due to minidsp's non-refund policy

 

[kiwifi]

The combined response of the sub's that is predicted by MSO should closely match the measured response with the filters in place. If it doesn't, then something went wrong transferring the filter parameters to the DSP. Or you didn't use an acoustic timing reference when you took the measurements that were used as the inputs to MSO.

 

[My adventures in stereo]

I gave up on dirac with my minidsp SHD, it was killing the bass response
using REW only currently, much better results

 

[Ricwa]

Make sure MSO is set for 48k sample rate. Also verify your sample rate in Windows control panel, sounds, playback, advanced.

 

[Smithg97]

I didn't use a timing reference! How do i do that? So far dirac hasn't done much although dialogue does seem clearer

 

[Smithg97]

Make sure MSO is set for 48k sample rate. Also verify your sample rate in Windows control panel, sounds, playback, advanced.

So apply 24 bit 48khz in windows?

 

[My adventures in stereo]

I didn't use a timing reference! How do i do that? So far dirac hasn't done much although dialogue does seem clearer

In the REW measurement window, on the top left, there is a drop down box, in which you can choose to set timing reference

 

[Ricwa]

So apply 24 bit 48khz in windows?

Yes

 

[Smithg97]

Ok so I'll try 48khz setting and add a timing reference which will hopefully fix MSO, any guidance on Dirac? It seems whatever pre-settings i apply in the DDRC-24 plugin (delays, gains, PEQ, Xover) before running Dirac then get wiped after i export a filter from Dirac, am i doing something wrong?

 

[Ricwa]

Ok so I'll try 48khz setting and add a timing reference which will hopefully fix MSO, any guidance on Dirac? It seems whatever pre-settings i apply in the DDRC-24 plugin (delays, gains, PEQ, Xover) before running Dirac then get wiped after i export a filter from Dirac, am i doing something wrong?

I have a 2.2 setup. I use MSO to 300Hz on both mains and subs. As above, you have to be sure to use the “acoustic timing reference” in REW. I use HP and LP crossovers in MSO and import all 4 biquads into miniDSP. You will know if it worked if you get a fairly flat response from 30 to 300 hz. Once that is done (it’s a bit of a learning curve at first), I simply run Dirac as if it were a 2 channel setup.

 

[Smithg97]

I have a 2.2 setup. I use MSO to 300Hz on both mains and subs. As above, you have to be sure to use the “acoustic timing reference” in REW. I use HP and LP crossovers in MSO and import all 4 biquads into miniDSP. You will know if it worked if you get a fairly flat response from 30 to 300 hz. Once that is done (it’s a bit of a learning curve at first), I simply run Dirac as if it were a 2 channel setup.

Is there any difference in setting the xover in mso vs the ddrc24 plugin? I like to lpf the subs at 100hz and also hpf the subs at 10hz so that the start of the response line is really smooth

 

[Ricwa]

Yes, and it's a good question. There are a couple reasons to prefer to let MSO handle the LP/HP filters. (1) MSO will utilize/optimize the LP/HP filter parameters to assist in correcting the overall low-frequency response, and (2) if you were to instead setup your LP/HP filters in the ddrc24 plugin (independent of MSO), Dirac will have a harder time correcting the frequency response. By having MSO be in charge of the crossover filters, Dirac has an easier time downstream because the low frequency response has already been mostly optimized by MSO before it gets to Dirac. In this scenario, I've found that Dirac further flattens the frequency response (and corrects the phase) beyond what Dirac alone or MSO alone can accomplish.

 

[Smithg97]

Yes, and it's a good question. There are a couple reasons to prefer to let MSO handle the LP/HP filters. (1) MSO will utilize/optimize the LP/HP filter parameters to assist in correcting the overall low-frequency response, and (2) if you were to instead setup your LP/HP filters in the ddrc24 plugin (independent of MSO), Dirac will have a harder time correcting the frequency response. By having MSO be in charge of the crossover filters, Dirac has an easier time downstream because the low frequency response has already been mostly optimized by MSO before it gets to Dirac. In this scenario, I've found that Dirac further flattens the frequency response (and corrects the phase) beyond what Dirac alone or MSO alone can accomplish.

Thanks Ricwa is it possible in MSO to define the crossovers you want applied? I'm still new to all this but learning quickly, I'm reading a lot of different posts on various forums and a recurring comment seems to be not to boost eq by too much db, is this because it eliminates headroom and create distortion? Is there a general rule for subs and mains on max boost overall and for each freq? I run my subs as -25db on the svs app so have a fair bit of room to play with (assuming my logic holds up)

I'll have another stab at mso tomorrow, reducing seat to seat variation isn't so critical for me as the setup I'm discussing in this thread is purely for my music listening as i run a denon x3700h for movies/tv and audyssey xt32 has done a pretty good job there but isn't as capable for music as the flex + 6000a combo

 

[Ricwa]

Yes, MSO is highly configurable. You can setup Linkwitz, Butterworth, etc. filters having 12db/octave, 24db/octave, slope etc. MSO provides other filters such as allpass as well. More variety than the ddrc24. Unless you are concerned about efficiency (i.e., using flea watt amps and high efficiency speakers), there's seldom a need to boost gains. If' you're willing to only cut gains (that is, you have a powerful enough amp), you'll be better off. However, if you want to target a specific SPL (I use 75 db), MSO allows that as well. Best to start small and then add features as you master the basics. So I'd suggest first using MSO to optimize 30-300 hz and then running Dirac to optimize the full frequency range. Do this without worrying about crossover filters, headroom, etc. Once you've mastered this, you're a long way toward Nirvana. And don't forget to add the gains and delays from MSO analysis to the DDRC. This is critical.

 

[Smithg97]

Yes, MSO is highly configurable. You can setup Linkwitz, Butterworth, etc. filters having 12db/octave, 24db/octave, slope etc. MSO provides other filters such as allpass as well. More variety than the ddrc24. Unless you are concerned about efficiency (i.e., using flea watt amps and high efficiency speakers), there's seldom a need to boost gains. If' you're willing to only cut gains (that is, you have a powerful enough amp), you'll be better off. However, if you want to target a specific SPL (I use 75 db), MSO allows that as well. Best to start small and then add features as you master the basics. So I'd suggest first using MSO to optimize 30-300 hz and then running Dirac to optimize the full frequency range. Do this without worrying about crossover filters, headroom, etc. Once you've mastered this, you're a long way toward Nirvana. And don't forget to add the gains and delays from MSO analysis to the DDRC. This is critical.

Yes i check the closing comments on the mso filter report to see gains and delays recommended then plumb them into ddrc24 plugin. Another question on xover, presumably this is somehow embedded in the biquad filters that i export from mso and import to ddrc24? Would i still need to apply a HPF to the mains in ddrc24 as mso is only affecting the subs? One other question, I've been measuring output with umik1 with the mic in horizontal orientation, is this correct or should it be vertical for mso?

 

[Ricwa]

When you start MSO, choose to optimize both mains and subs. Add an HP filter to each main (for example Linzwitz 12db) and optionally an LP filter to each sub. When you export from MSO you'll end up with 4 files containing the biquads for four speakers, main left, main right, sub left, and sub right. You import each biquad into ddrc separately using PEQ, Advanced, Import, etc. So yes, in this scenario the mains xovers are embedded in the biquads and you won't use the crossover functionality of the miniDSp.

 

[Smithg97]

When you start MSO, choose to optimize both mains and subs. Add an HP filter to each main (for example Linzwitz 12db) and optionally an LP filter to each sub. When you export from MSO you'll end up with 4 files containing the biquads for four speakers, main left, main right, sub left, and sub right. You import each biquad into ddrc separately using PEQ, Advanced, Import, etc. So yes, in this scenario the mains xovers are embedded in the biquads and you won't use the crossover functionality of the miniDSp.

Perfect, thank you! Using your approach to run MSO upto 300hz for mains & subs with crossovers set at 100hz, i guess logically that MSO is not only setting a crossover for the mains in the relative biquads but also creating some peq for the mains (100hz to 300hz range)? Also when i import these filters to ddrc24 and subsequently calibrate using dirac live, will the MSO peq, gains, delays and crossovers be preserved? I ask as when i ran dirac the other day and exported the filter it created i noticed that the gains and delays I'd previously set in ddrc24 had disappeared!

 

[Ricwa]

Yes to all. Gains and delays need to me set in minidsp manually. Also, running Dirac or importing biquad files will not overwrite these gain and delay settings. Maybe you imported a minidsp config (.xml) file. That’s the only thing that could overwrite these settings.

 

[Smithg97]

Yes to all. Gains and delays need to me set in minidsp manually. Also, running Dirac or importing biquad files will not overwrite these gain and delay settings. Maybe you imported a minidsp config (.xml) file. That’s the only thing that could overwrite these settings.

Appreciate your guidance, hoping I'll get time tomorrow to re do all the measurements and run through the optimisation processes again as i'm keen to reach the audio nirvana you referred to earlier and take a break from the obsessive tinkering that's taken up most of this week's spare time

In your opinion would 3 different measurement positions (at the same height) suffice for the MSO & Dirac measurements considering my listening position is a 3 seater sofa about 1.5 meters from the speakers/subs and I'd read on a different ASR thread that to achieve better results with Dirac to only measure the critical positions

 

[Ricwa]

Can't offer guidance as I've only done single measurement at a single listening position. I'd suggest you plan do at least a couple "throw away" single position measurements to begin with; until you get the hang of it.