The point of using the LFE channel is because the LPF for LFE can often be set higher than a crossover. The objective is to get a flat measurement from the subs with no LPF in play for an octave above the maximum optimised frequency. So if you want to EQ up to 120Hz then you shouldn't have any filters in place until 240Hz, which, with a steep LPF at 250Hz is about right.

You can bake a target curve in to follow the shape of your sub's native response but MSO does a better job when given a flat target to aim for IMO.
 
Totally agree, makes sense to go for a flat target. Depending on your sub you might be able to bypass any inbuilt crossover and just have it play full range. The DIY gang certainly can. Then you don't need to worry about the LPF setting and needn't use HDMI 4.

For example can just use HDMI 2, swap cable from sub pre out to R main channel pre out disconnect the R speaker cables and run as large speaker. Free to do what you like then and eq as high as you want, without restrictions.

Actually typing it out now I realise that it is in fact way more complicated than just using HDMI 4, ie no need to swap any preouts or speaker cables. :facepalm:
 
Totally agree, makes sense to go for a flat target. Depending on your sub you might be able to bypass any inbuilt crossover and just have it play full range. The DIY gang certainly can. Then you don't need to worry about the LPF setting and needn't use HDMI 4.

For example can just use HDMI 2, swap cable from sub pre out to R main channel pre out disconnect the R speaker cables and run as large speaker. Free to do what you like then and eq as high as you want, without restrictions.

Actually typing it out now I realise that it is in fact way more complicated than just using HDMI 4, ie no need to swap any preouts or speaker cables. :facepalm:

Yeah that’s definitely a more complicated way of doing it.

If you can set the channel crossover up at 250 (or higher) then that’ll work on a mains channel as well, but you have to disconnect the main or turn its amp off

The only reason to not use the LFE channel would be with a stereo amp with pre outs, but that’s pretty rare.
 
Last edited:
I thought it’s been a while since anyone posted some recent information on MSO and figured it would be useful to people who are not aware of it.

Before you go an buy more subs I highly recommend calibrating your dual subs with this free software call Multi Sub Optimiser, or MSO. It can do incredible things to even out your in room response. Purchasing a MiniDSP 2x4HD and MSO is far cheaper.

Here's the link for it.

Here are pictures of the response of my 2 subs in my standard 4m x 5.5m living room. They face each other on the opposing short walls.


View attachment 1870609View attachment 1870610View attachment 1870611View attachment 1870612

Here is the MSO corrected response across the sofa's two seating positions.
MSO has a learning curve but is actually quite easy once you get used to it. Just follow the steps one by one.View attachment 1870613
Hi @JamesBaby,

Out of interest what number of filters did you end up with, and at what level did you initially measure your subs at? I know a number of factors play into this, but I'd still be interested.

MSO still isn't working out for me and I'm sure I must be doing something wrong. Whenever I apply the predicted config to my MiniDSP and rerun sweeps in REW, my output level is always at least 10db lower than the stated reference level in MSO (approx 60-65db).

Any help would be much appreciated.
 
Hi. Well I've just re-done mine again after upgrading the amps in by SVS subs.

I'm currently in 6 shared filters, and 3 on the outputs.
I set target to 75db, and maximum Q of 8.651 for any filters.
I set and constrained any filter gain to a max of 2db.

However, I want to understand all pass filters, so may tweak it again. But this is the measured response I currently have:
IMG_7786.jpeg


Edit: My initial REW measures with timing ref were around 63-65db. Post MSO you can see in the graph they’re around 72 I think.
 
Last edited:
Hi. Well I've just re-done mine again after upgrading the amps in by SVS subs.

I'm currently in 6 shared filters, and 3 on the outputs.
I set target to 75db, and maximum Q of 8.651 for any filters.
I set and constrained any filter gain to a max of 2db.

However, I want to understand all pass filters, so may tweak it again. But this is the measured response I currently have:
View attachment 1881821

Edit: My initial REW measures with timing ref were around 63-65db. Post MSO you can see in the graph they’re around 72 I think.

Cheers! That's a great looking response!

So as a test I tried the following
  • 4 shared filters, 5 on the outputs
  • db target set to 75, and like you a max Q of 8.651
  • 0 gain on all filters
I initially measured each sub (LFE channel) with timing ref at around 78-79db in REW (orange and red lines). This was with the level of the subs in the SVS app set to -22db, and my AVR set to -28.5db.

I got the following response (blue line) post MSO, which shows around 57db.

1686634775089.png


What maximum cuts if any do you apply?

@Conrad did mention in a previous post that I possibly measured too high, and you also mentioned your initial REW measurements at around 63-65db. So I'll try taking my initial REW measurements at a lower db.

I feel like I'm still missing some basic piece of this MSO puzzle, but I'll have another attempt at solving this at point this week! :confused::thumbsup:
 
Cheers! That's a great looking response!

So as a test I tried the following
  • 4 shared filters, 5 on the outputs
  • db target set to 75, and like you a max Q of 8.651
  • 0 gain on all filters
I initially measured each sub (LFE channel) with timing ref at around 78-79db in REW (orange and red lines). This was with the level of the subs in the SVS app set to -22db, and my AVR set to -28.5db.

I got the following response (blue line) post MSO, which shows around 57db.

View attachment 1882530

What maximum cuts if any do you apply?

@Conrad did mention in a previous post that I possibly measured too high, and you also mentioned your initial REW measurements at around 63-65db. So I'll try taking my initial REW measurements at a lower db.

I feel like I'm still missing some basic piece of this MSO puzzle, but I'll have another attempt at solving this at point this week! :confused::thumbsup:
What do the filters look like for that response?
I would pay attention to the target level and the shared gain in MSO. I tend to set the target level high, 100dB or so, so that it forces MSO to try and maximise output rather than cut it.

Post up your MSOP file if you can.
 
What do the filters look like for that response?
I would pay attention to the target level and the shared gain in MSO. I tend to set the target level high, 100dB or so, so that it forces MSO to try and maximise output rather than cut it.

Post up your MSOP file if you can.

Cheers Conrad,

Here's what the config looks like in MSO

1686643607264.png


And I was expecting something a bit closer to this when remeasuring in REW.

By MSO file, are you referring to the miniDSP Biquad Text file?
 
No, the MSO project file.

I find it's often informative to create a few graphs:
Raw responses (usually at one seat, the MLP)
Filtered responses (again, use the same seat)

That way you can see how much MSO is changing the raw responses. Ideally you want the least amount of change.

Your response looks good, it looks like the MSO prediction, it's all +-2.5dB, but it's just very low. My guess is that it's cutting the response hard with PEQ and then lifting it with 10/15dB in the share gain filter to get you back to the target of 75dB.
 
No, the MSO project file.

I find it's often informative to create a few graphs:
Raw responses (usually at one seat, the MLP)
Filtered responses (again, use the same seat)

That way you can see how much MSO is changing the raw responses. Ideally you want the least amount of change.

Your response looks good, it looks like the MSO prediction, it's all +-2.5dB, but it's just very low. My guess is that it's cutting the response hard with PEQ and then lifting it with 10/15dB in the share gain filter to get you back to the target of 75dB.

Ah ok, yes I did create this

1686644371832.png


The MSO project itself if pretty large, I ran over 30+ configs, playing with various filters, so it might be a bit too big to post here. I will have a go at setting higher target levels though.

Looking at the properties tab in MSO, you're correct in that there are at least 5 filters (including shared), that have cut PEQ by over 17db. :(

Is there any way to get around this in MSO, or is this just the nature of my room and due to number of subs?

When EQ'ing in REW only, I'm able to get a really good response, but of course this is only for the MLP.
 
There will be ways to make it better. A target of 90dB might be more appropriate here but it might just do the same EQ and then add 24dB shared gain which won't really help.

I'd need to see the project really.

I just worked with @JamesBaby's project and got a slightly worse response but only using 2 PEQ and 2 AP filters per sub. It really is a much better way of doing it.
 
There will be ways to make it better. A target of 90dB might be more appropriate here but it might just do the same EQ and then add 24dB shared gain which won't really help.

I'd need to see the project really.

I just worked with @JamesBaby's project and got a slightly worse response but only using 2 PEQ and 2 AP filters per sub. It really is a much better way of doing it.
Wow! That would be amazing!

Not at home at the moment, but happy to DM the project later this eve! 😄
 
I just came across this thread by accident, I would love to try MSO, shame there isn't a Mac version, Does anyone know of any plans to introduce a Mac Version?
 
I just came across this thread by accident, I would love to try MSO, shame there isn't a Mac version, Does anyone know of any plans to introduce a Mac Version?
I doubt there are plans. Andy (the author) flits between wanting to pack it all in and making updates. He does fix bugs and add new features/improve the workflow, but he doesn't have much interest in making it more widely adopted.

I run it on a Mac in parallels, but it'll run in a free VM as well.
 
Big thanks to @Conrad for helping me with my MSO. It's crazy how you got a similar response but with hardly any filters. I'm currently doing listening tests. It sounds smooth and clean, but think I need to increase the sub levels.

Will report back. :)
 
@Conrad would you beable to help me? ive tried to get assistance on AVS but no luck. So ive watched the tutorials and read the doc on MSO and for some reason no matter what i do i cant get my sweeps to match in REW post MSO.

I have done many new measurements fresh and always do MLP last so it doesnt move.

I have just even tried now with just 2 subs, 1 position so essentially just time align and eq to simplify and to aid troubleshooting but something is not right as no matter what i cant get (even close) to the MSO measurements.


1687534070088.png

1687534078883.png



Im using a MINIDSP , i am importing the quad files and adding the delay and fain from the final report , and i have used the shared filters on the input and and without and its made no difference.

AVR has all room correction off, LFE xover is maxed out. Im using L as the acoustic reference

I must be missing something. I just dont know what
 
Can you post up your MDAT and your MSOP file, as well as the minidsp config xml?
I would also measure each sub individually and confirm that they measure the same as the MSO prediction. If they don't then that's a pointer.

How close to a wall are you measuring?
 
Can you post up your MDAT and your MSOP file, as well as the minidsp config xml?
I would also measure each sub individually and confirm that they measure the same as the MSO prediction. If they don't then that's a pointer.

How close to a wall are you measuring?
Hi , i will do a fresh file as i have done a few . Annoyingly, earlier i manged to get 1 measurement the same by changeing my max Q length in INPUT and GENERAL to 8 like the guy in this thread. Then when i changed the SPL target , it just went wrong again.

then i went back and reused the same settings and it didnt measure the same, just not nonsistency.

What is MDAT? and if you let me know what optimization settings to use , ill create that sweep now and send you the files

My measurement is about 20 inch from back wall. I get consistent measurements in REW normally
 
Can you post up your MDAT and your MSOP file, as well as the minidsp config xml?
I would also measure each sub individually and confirm that they measure the same as the MSO prediction. If they don't then that's a pointer.

How close to a wall are you measuring?
oh yea MDAT is my REW , if you give me some "safe settings" ill start from scratch and send you fresh copies now as the mic is in position and everything is ready to go
 
Hi , i will do a fresh file as i have done a few . Annoyingly, earlier i manged to get 1 measurement the same by changeing my max Q length in INPUT and GENERAL to 8 like the guy in this thread. Then when i changed the SPL target , it just went wrong again.
What are the Qs of the filters that MSO generates? 8 is pretty narrow and you should probably review those with a view to removing them. But let's get things behaving consistently first.

then i went back and reused the same settings and it didnt measure the same, just not nonsistency.

What is MDAT? and if you let me know what optimization settings to use , ill create that sweep now and send you the files
MDAT is the file that REW uses, just do Save all measurements and save the file, then share it.

My measurement is about 20 inch from back wall. I get consistent measurements in REW normally
Ok, and when you apply filters from REW your measurements match the predictions?

You don't need to re-take the measurements or anything, just share the files that you've used to generate the MSO project, the MSO project, the "after" measurements, and the XML file from the minidsp. From that I or someone else will be able to look and see if everything has been applied correctly.
 
What are the Qs of the filters that MSO generates? 8 is pretty narrow and you should probably review those with a view to removing them. But let's get things behaving consistently first.


MDAT is the file that REW uses, just do Save all measurements and save the file, then share it.


Ok, and when you apply filters from REW your measurements match the predictions?

You don't need to re-take the measurements or anything, just share the files that you've used to generate the MSO project, the MSO project, the "after" measurements, and the XML file from the minidsp. From that I or someone else will be able to look and see if everything has been applied correctly.
When i use the alignment tool in REW , its not perfect but usually close enough.

I changed the Q to narrow to see if it would work and it did (for 1 measurement) was just troubleshooting ideas.

Anyhow, here is a fresh copy of everything I have just taken. 2 Subs just using MLP (will want to use other positions when i can get this working).

I have used the default optimization settings (previously i have tried changing these settings and nothing has worked to get close to the predicted response.

Please let me know if you need anything else but i think i have covered it.

This was one of the closer attempts, its still not right . Some of the others have been much much worse but will be good to know if i am doing anything wrong!

Thanks in advance
 

Attachments

  • MSO - AVFORUMS.zip
    372.6 KB · Views: 9
When i use the alignment tool in REW , its not perfect but usually close enough.
Hmmm. That, to me, suggests that there might be something else up.
If you don't move the mic, the alignment tool is incredibly accurate and should be within -=0.5dB or less across the range. I've had situations where I've ended up re-measuring as I didn't think it'd worked because the measurements completely overlapped.

It might be worth moving the mic to the middle of the room, measuring each sub, using the alignment tool to apply a delay and then applying that delay and confirming. It doesn't need to be a good response, you're just checking the measurements.

I changed the Q to narrow to see if it would work and it did (for 1 measurement) was just troubleshooting ideas.
Again, this is a red flag for me. I don't see why limiting the Q would lead the measurements to align when the Q is lower. I get that you're not saying that that's the solution, but it does point to inconsistency. Boundary zones can and do cause inconsistency in measurements though.

Anyhow, here is a fresh copy of everything I have just taken. 2 Subs just using MLP (will want to use other positions when i can get this working).
Thanks, I'll have a look.

I have used the default optimization settings (previously i have tried changing these settings and nothing has worked to get close to the predicted response.

Please let me know if you need anything else but i think i have covered it.

This was one of the closer attempts, its still not right . Some of the others have been much much worse but will be good to know if i am doing anything wrong!

Thanks in advance
At the moment I think the best plan is to put a workflow together that gives you consistent results. Then we can work on tweaks to the actual calculations. It's no good getting a perfect response in MSO if you can't apply it.
 
I'm going to have to go back on my word completely, those measurements look fine.
Sure, you have about a 1 or 2dB variation through some of the response, but I would consider that a match.

The alignment tool is spot on, MSO slightly less so. One reason is the precision of the delay. In MSO it's two decimal places - 19.72ms in your case - but in the minidsp that would be 19.7. That's enough to cause the difference.

That said, I think you can do better in MSO. You're giving up a huge amount of headroom and 19ms is a lot of delay. I'll take a look.
 
I'm going to have to go back on my word completely, those measurements look fine.
Sure, you have about a 1 or 2dB variation through some of the response, but I would consider that a match.

The alignment tool is spot on, MSO slightly less so. One reason is the precision of the delay. In MSO it's two decimal places - 19.72ms in your case - but in the minidsp that would be 19.7. That's enough to cause the difference.

That said, I think you can do better in MSO. You're giving up a huge amount of headroom and 19ms is a lot of delay. I'll take a look.
thanks, well its good to know i have not made any rookie errors. i was fairly confident i had understood the workflow and ensuring i had applied everything right.

What is interesting though , is that , that was my 2nd closest measurement. The others have been way off and i know i did everything the same lol

Thanks for taking the time thus far to assist!
 
I'm going to have to go back on my word completely, those measurements look fine.
Sure, you have about a 1 or 2dB variation through some of the response, but I would consider that a match.

The alignment tool is spot on, MSO slightly less so. One reason is the precision of the delay. In MSO it's two decimal places - 19.72ms in your case - but in the minidsp that would be 19.7. That's enough to cause the difference.

That said, I think you can do better in MSO. You're giving up a huge amount of headroom and 19ms is a lot of delay. I'll take a look.
each time im getting completeley different delays. Usually the sweet spot from my own eye balling in REW and from most measurements is 10-13 but MSO is changing that signficantly .

I have level matched the 2 SVS SB3000 subs to 85db at MLP using the LFE channel. I am around -15 and -18db in the SVS app
 

The latest video from AVForums

Is 4K Blu-ray Worth It?
Subscribe to our YouTube channel
Back
Top Bottom